Correspondence: Marc Najjar (mn2594@cumc.columbia.edu)

Background

Hepatocellular carcinoma (HCC) is the most common liver malignancy and the 5th cause of cancer-related mortality worldwide. Though previous studies have found that serum neutrophil-to-lymphocyte ratio (NLR) is predictive of survival post liver transplant (LT), peritumoral neutrophil (PMN) infiltration in the tumor microenvironment (TME) of HCC has not been thoroughly investigated yet. In this study we sought to evaluate tissue based PMN infiltration in HCC post LT using quantitative multiplex immunofluorescence (qmIF), previously used to study the TME of several other tumor types[1].

Methods

A database of 634 patients was created at Columbia University Irving Medical Center (CUIMC) including adult patients with available clinical follow up who underwent liver transplantation (LT) for HCC between 1998 and 2018. We evaluated a preliminary cohort of 10 patients using qmIF, excluding patients with viral hepatitis. FFPE tumor sections were pre-selected by a GI pathologist. Slides were stained using qmIF for MPO (PMNs), CD3 (T cells), CD8 (cytotoxic T cells), CD68 (macrophages), HLA-DR (immune activation), and Hep-Par1 (hepatocytes/tumor). Multiplex images were visualized using Vectra (Akoya) and processed using inForm (Akoya). Data was analyzed using R Studio for concatenation, density, nearest neighbor and statistical analysis. Serum NLR was calculated using complete blood counts collected prior to LT(Figure 1).

Results

Preliminary cohort of 10 patients includes 4 with recurrence at a median of 2.4 years and 6 with no recurrence at a median of 12 years post-LT. We found that patients with recurrence post-LT have significantly higher densities of MPO+ PMNs compared to those with no recurrence. This difference is primarily driven by PMNs located within the peritumoral stroma (Median [interquartile range [IQR] 2.46 [1.99 - 2.92] vs 1.23 [0.723 -1.78], p=0.019). Intratumoral PMN infiltration was not associated with recurrence (Median [IQR] 0.91 [0.59 - 1.20] vs 1.33 [0.56 – 1.90], p=0.308). Moreover, density of CD3, both intratumoral and peritumoral, did not correlate with recurrence, nor did the tissue-derived NLR. Further, we found that the tissue-derived NLR did not correlate with NLR in blood.

Conclusions

Higher densities of peritumoral PMNs are associated with post-LT HCC recurrence. Evaluation of TME using qmIF can be used to predict recurrence in post-LT HCC. Further, tissue based analysis of PMNs does not correlate with serum NLR allowing potential for composite biomarkers. As this is preliminary, further analysis is underway and will be validated on the larger cohort of patients.

Reference

1. Gartrell RD, Marks DK, Hart TD, et al. Quantitative Analysis of Immune Infiltrates in Primary Melanoma. Cancer Immunol Res 2018;6:481-93.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P1).

Quantitative multiplex immunofluorescence images of HCC

Correspondence: Tessa DesRochers (tessa.desrochers@kiyatec.com)

Background

Knowledge of immune responses that correlate with clinical outcome is essential for the development of strategies to harness a patient’s immune system to eradicate cancer. Pre-clinical platforms that recapitulate the immune response in the context of cancer are necessary for adequate understanding and detection of clinical efficacy, however, the technology to accurately test immuno-oncology (I/O) therapy response is lacking. Despite the value animal models provide in a pre-clinical setting, they lack matched patient tumor and immune cell interactions. To address this shortcoming, we developed in vitro 3D tissue models that maintain autologous patient tumor cells and immune cells for the testing and prediction of immune cell responses. We hypothesize that these 3D tissue models will recapitulate the patient tumor microenvironment and detect response to I/O agents.

Methods

Tumor cells and T-cells were obtained from seven melanoma patient biopsies and screened for PD-L1 and lymphocyte populations prior to incorporation into 3D culture. Effector cell to Tumor cell (E:T) optimization assays were conducted with expanded T-cells at different densities and co-cultured at different time points with tumor cells. Viability was measured using CellTiter-Glo® 3D. T-cell response was determined using flow cytometry following 24-hour co-culture with tumor cells. Microtumors were established using a biologically inert scaffold and extracellular matrix components. Microtumor viability was determined using PrestoBlue and T-cell infiltration was determined via flow cytometry. Analyte secretion was determined from supernatant using Milliplex MAP Human CD8+ T-cell Panel.

Results

We detected pembrolizumab binding to T-cells in a dose dependent manner and an increase in the activation marker CD69 on T-cells following tumor cell and pembrolizumab treatment in three of four patients tested. We devised an initial E:T optimization screen to identify a patient-specific ratio which renders our subsequent therapy response profiling highly personalized. CD3+CD8+ T-cell mediated tumor cell death and enhanced killing was detected in the presence of pembrolizumab. Immune cell infiltration as well as therapy related cell death was observed in our 3D microtumors. Altered patient specific cytokine secretion was measured when the cultures were treated with pembrolizumab and significantly correlated with pembrolizumab induced reduction of microtumor growth rates.

Conclusions

The data generated from these two complex 3D in vitro models allows us to better understand immune responses to autologous tumor cells and checkpoint blockade. Our models are therefore ideal and complimentary for preclinical testing of new I/O agents as well as patient response predictions to I/O based therapies.

Ethics Approval

Tissue was acquired with approval from Prisma Health's Institutional Review Board, PRO# 00069834.

Correspondence: Michael Curran (MCurran@mdanderson.org)

Background

The widely used glioblastoma multiforme (GBM) model GL261 is highly immunogenic and readily cured by checkpoint blockade limiting its use for pre-clinical modeling of immunotherapy for human GBM [1,2]. We developed four novel murine immunocompetent glioblastoma stem cell (QPP) lines derived from Nestin-CreER^T2 Quaking (QKI)^L/L; P53^L/L; PTEN^L/L mice, reflecting a common set of alterations in patients [3-5]. The four QPP cell lines are syngeneic to C57BL/6J mice and exhibit distinct responses to T-cell checkpoint blockade.

Methods

The differential responsiveness of each QPP line was assessed through analysis of tumor growth in the brain versus the flank in untreated, αPD-1, or αCTLA-4 treated mice. The impact of tumor genomic landscape on responsiveness at each site was measured through whole exome sequencing. To understand cellular factors modulating responsiveness of these GBM lines to checkpoint blockade, the immune microenvironments of sensitive (QPP7) versus resistant (QPP8) lines were compared in the brain using high parameter flow cytometry. Drivers of flank sensitivity versus brain resistance were also measured for QPP8.

Results

QPP GBM lines demonstrate a range of sensitivities to CTLA-4 and PD-1 blockade when implanted on the flank ranging from complete sensitivity (QPP7) to complete resistance (QPP4). In the brain, QPP7 remains sensitive to both antibodies, but QPP4 and QPP8 fail to respond to blockade of either checkpoint (Figure 1). Analysis of the QPP8 immune infiltrate in skin reveals enhanced ratios of CD8s to Treg and myeloid suppressors in response to checkpoint blockade; however, none of these benefits manifest in the brain QPP8 except a very specific increase in CD8s relative to granulocytic suppressors (Figure 2). Brain-implanted QPP8 reacts adaptively to checkpoint blockade by upregulating PD-L1 expression across its myeloid stroma. In contrast, immune-responsive QPP7 does not induce PD-L1 and shows markers of enhanced CD8 T cell fitness. Consistent with these observations, genomic analysis reveals a higher mutation density in QPP7 versus the other QPP lines. Using checkpoint-insensitive QPP4/8, we have now identified agonists of the Stimulator of Interferon Genes (STING) pathway as highly promising therapeutics for treating these tumors in the brain.

Conclusions

We have developed novel syngeneic models of GBM with relevant genetics and immune sensitivities relative to human disease. Through comparing T cell checkpoint blockade sensitive versus insensitive variants of these QPP lines, and through comparing variant sensitivity dictated by site of implantation, we have begun to identify the genetic and cellular components that govern immunotherapeutic sensitivity of GBM.

References

1. Reardon DA, Omuro A, Brandes AA, Rieger J, Wick A, Sepulveda J, Phuphanich S, de Souza P, Ahluwalia MS, Lim M, Vlahovic G, Sampson J (2017) OS10.3 Randomized Phase 3 Study Evaluating the Efficacy and Safety of Nivolumab vs Bevacizumab in Patients With Recurrent Glioblastoma: CheckMate 143. Neuro-Oncology 19: iii21-iii21.

2. Reardon DA, Gokhale PC, Klein SR, et al. Glioblastoma Eradication Following Immune Checkpoint Blockade in an Orthotopic, Immunocompetent Model. Cancer Immunol Res. 2016;4(2):124-135.

3. Hu J, Ho AL, Yuan L, et al. From the Cover: Neutralization of terminal differentiation in gliomagenesis. Proc Natl Acad Sci U S A. 2013;110(36):14520-14527.

4. Brennan CW, Verhaak RG, McKenna A, et al. The somatic genomic landscape of glioblastoma. Cell. 2013;155(2):462-477.

5. Shingu T, Ho AL, Yuan L, et al. Qki deficiency maintains stemness of glioma stem cells in suboptimal environment by downregulating endolysosomal degradation. Nat Genet. 2017;49(1):75-86.

Ethics Approval

All experiments were conducted according to protocols approved by the University of Texas MD Anderson Cancer Center Institutional Animal Care and Use Committee.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P4).

Orthotopic QPP survival and immune sensitivity

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P4).

Immune landscape of QPP8 TME in different niches

Correspondence: Namyong Kim (namyong@curiox.com)

Background

The naturally occurring tumor infiltrating lymphocytes (TILs) exist in a complex microenvironment containing the extracellular matrix, blood vessels, and stromal and endothelial components in addition to various immune cells. While a growing number of preclinical mouse models is under development, the heterogeneity of the cell composition in a solid tumor poses considerable technical challenges in isolating and characterizing the TILs for the downstream analysis. One common problem with TILs preparation occurs during solid tumor dissociation, whereby the TILs are left in a mixture with tissue debris and dead cells in suspension. Consequently, a preparation of autologous TILs often requires further costly and laborious processing such as density gradient centrifugation, immune cell sorting and enrichment, and dead cell and debris removal in combination with multiple centrifugation steps. We introduce a novel Laminar Wash™ technology, which can help overcome these technical challenges.

Methods

We performed pilot studies on syngeneic (MC38, CT26, CloudmanS91, 4T1) and humanized mouse tumor models as well as with human PBMCs and tumor biopsies using the Laminar Wash™ technology. Briefly, we evaluated various functional parameters of TILs such as polyfunctional CD8+ T cell responses and glucose update efficiency (2-NBDG) as well as conducting a side-by-side comparison of the TIL recovery rate and immunophenotypic characteristics of lymphoid and myeloid subsets on the Laminar Wash™ and the centrifugation-based systems. In addition, the cell retention rate, cell viability, debris removal, epitope preservation and the overall processing time were assessed and compared. Furthermore, we introduce a complete walk-away approach to sample preparation that eliminates operator-based variability while significantly enhancing reproducibility and consistency of downstream analysis.

Results

Our data demonstrate that the Laminar Wash™ method resulted in higher cell retention and viability, more clearly defined immune subsets, a lowered background signal, and an enhanced yield of the TILs from freshly dissociated tumor samples compared to the centrifugation-based counterparts. The Laminar Wash™ system can effectively remove the floating debris in suspension while keeping the live cells unperturbed, allowing the cell surface architecture and epitopes better retained for improved downstream analysis with flow cytometer. Additionally, the Laminar Wash™ AUTO system offers a completely automated sample processing solution for dissociated tumor samples, simplifying and expediting cell preparation with enhanced consistency and reproducibility.

Conclusions

Laminar Wash™ results in healthy, viable, and well defined population of TILs, while improving the overall quality of data. The AUTO station provides an automated, centrifuge-free, and walk-away workflow for dissociated tumor samples for cytometry-based assays.

Acknowledgements

Laminar Wash™ results in healthy, viable, and well defined population of TILs, while improving the overall quality of data. The AUTO station provides an automated, centrifuge-free, and walk-away workflow for dissociated tumor samples for cytometry-based assays.

Correspondence: Bruce Ruggeri (bruggeri@championsoncology.com); Amy Wesa (awesa@championsoncology.com)

Background

Humanized immune system (HIS) mouse models enable in vivo studies in the context of the human immune cells with a human tumor and are critical for the development of next generation immune-oncology (IO) agents. Humanization of immunodeficient mice through the adoptive transfer of normal adult PBMC leads to rapid engraftment of human T cells to study immune-modulatory agents in the context of human tumor xenografts, but is limited by the development of xenogeneic graft-versus host disease (xGVHD). In this study, we evaluated the engraftment of PBMC in β2microglobulin null super-immunodeficient mice NSG-B2M mice, that lack MHC Class I on host tissues. A cell line-derived xenograft model (CDX) co-engrafted with PBMC (PBMC-ImmunoGraft) was characterized for humanization, tumor infiltrating leukocytes (TIL) phenotype and tumor response to checkpoint inhibitors.

Methods

PBMC from healthy donors (N=7) were implanted and engraftment in peripheral blood was assessed by flow cytometry. T cell memory phenotypes were assessed over time in a small cohort, and costimulatory and inhibitory T cell subsets were evaluated at the terminal time point in blood and secondary lymphoid organs. Next, NSG-B2M mice were co-implanted with MDA-MB-231 breast cancer cell line s.c, humanized with PBMC and tumor growth kinetics were monitored. Efficacy studies evaluating check point inhibitors are currently ongoing.

Results

Successful PBMC engraftment without xGVHD was observed in NSG-B2M mice up to 8 weeks, in contrast with MHC Class I expressing immunodeficient mice that developed xGVHD within 4-5 weeks. Dose and donor- dependent chimerism was observed. T cells were detectable in the periphery starting at 2 weeks with stable levels (up to 40% of live cells to 8 weeks) with increasing T effector memory cells over the course of study. All tumors evaluated had high levels of TIL as measured by flow cytometry and immunohistochemistry. Costimulatory and inhibitory molecules evaluated on CD4 and CD8 included 4-1BB, TIM-3, LAG-3, OX-40, as well as PD-1, which was expressed on both peripheral blood cells and in TIL. Tumor growth kinetics was unaltered by PBMC humanization through a 5-week study window.

Conclusions

PBMC-humanized NSG-B2M mice may represent a model for evaluating of IO therapeutics with a long study window due to the lack of xGVHD. While PBMC engraftment kinetics are donor dependent, similar phenotypes are observed and T cell subsets expressing several relevant therapeutic targets, including PD-1 are present. This model may permit a rapid in vivo method to study checkpoint blockade and other T-cell-directed IO therapeutics.

Ethics Approval

The study was approved by Champions Oncology's Institutional Animal Care and Use Committee (IACUC).

Correspondence: Bruce Ruggeri (bruggeri@championsoncology.com); Amy Wesa (awesa@championsoncology.com)

Background

Harnessing NK cell anti-cancer cytotoxicity has gained interest as a therapeutic strategy, and consequently improved preclinical models supporting the translation of NK cell–mediated therapies to the clinic are desired. Reproducible models with human NK engraftment into immunodeficient mice co-engrafted with cell line-derived xenograft or patient-derived xenograft tumor models have been lacking due to an inability to support NK cell engraftment and persistence. Here we evaluated IL-15-NOG mice for the engraftment and sustained survival of both ex vivo expanded and primary human NK cell isolates for establishing models that engraft effectively with both human NK cells and a PDX or CDX tumor.

Methods

NK cells from normal adult peripheral blood mononuclear cells (PBMC) donors (N=3) were expanded using two different commercially available kits and evaluated for NK phenotype, expansion rates and yields. Titrated doses of ex vivo expanded NK cells were adoptively transferred into IL-15-NOG mice for human chimerism, and the persistence and survival of NK cells and their immunophenotype were assessed. In separate studies, naïve NK cells enriched from PBMC were also evaluated for NK cell persistence and expansion in vivo. To establish an NK ImmunoGraft, NK cells were engrafted in xenograft tumor bearing mice and tumor growth kinetics were characterized.

Results

Donor dependent NK expansion was observed ex vivo, with 28 to 50-fold expansion by two weeks. NK cells expanded ex vivo were CD3-CD16+CD56± and varied based on the expansion kit utilized. Nearly all CD45+ cells in circulation were NK cells, and these peaked by week 2, and were maintained for up to 10 weeks in IL-15-NOG mice. Primary NK cells engrafted with slower kinetics, with peak abundance at 3-4 weeks. NK cells expressed granzyme B, and further functional studies are in progress. For all NK cell populations, cell density-dependent engraftment was observed with a largely stable NK phenotype observed across the study. In the absence of any therapeutic treatment, NK cell persistence and expansion in vivo did not inhibit tumor xenograft growth kinetics in IL-15-NOG mice

Conclusions

IL-15-NOG mice support the survival and persistence of human NK cells from both ex vivo expanded and naïve NK cells, suggesting the universality of this platform for human NK engraftment. Our preliminary studies support IL-15 NOG mouse model as a suitable system for evaluation of NK cellular therapies or NK cell-modulating therapies in the context of patient-derived or cell-line derived xenograft (PDX or CDX) mouse models

Ethics Approval

The study was approved by Champions Oncology's Institutional Animal Care and Use Committee (IACUC).

Correspondence: Takashi Shimada (tashimada@shimadzu.com)

Background

With the development of immune checkpoint inhibitors, the focus of cancer therapy is shifting to immunotherapy. Our purpose is to develop the drug efficacy index by rapid analysis of antibodies accumulating in cancer tissue using liquid chromatography and mass spectrometry (LC-MS/MS) and by characterization of the antibody distribution in the tumor microenvironment. Using a novel proteolysis method in which antibody molecules are collected on a 100 nm resin pore and trypsin is immobilized on a 200 nm nanoparticle surface, we have developed a method for physicochemically limiting trypsin access to antibody and identifying the structural specificity of complementarity-determining regions while minimizing extra peptides and protease without depending on the type of antibody. Using this method to detect antibodies from formalin-fixed paraffin-embedded (FFPE) tumor tissues, we aim to develop novel diagnostics that can aid in therapeutic dosing and predicting responses to antibody-based therapies.

Methods

To demonstrate the feasibility of these approaches, the human breast and epidermoid carcinoma cell lines SKBR3 and A431 were incubated with either trastuzumab and cetuximab, which bind to erbB2 and EGFR, respectively. FFPE cell blocks were then prepared and proteins were extracted from 8 μm sections after deparaffinization and decrosslinking. The extracted proteins were subjected to the Fab-selective proteolysis nSMOL, and the signature peptides of each antibody, IYPTNGYTR for trastuzumab and SQVFFK for cetuximab, were detected via triple-quadrupole LC-MS/MS. SCID mice were subcutaneously implanted with BT474 cells and 5 days later were infused with 10 mg/kg or 20 mg/kg trastuzumab. 24 h after administration, tumor and other tissues were harvested and FFPE block were prepared for trastuzumab quantitation in FFPE tissues.

Results

As a result of the pretreatment protocol using the cell block, the conditions of deparaffinization, decrosslinking, and protein extraction were optimized. Mass spectra of the signature peptides from trastuzumab and cetuximab could be detected using 20,000 cells. This condition was also applied to xenograft tissue and the degree of trastuzumab accumulation was detected in FFPE tumor tissue in a dose-dependent manner.

Conclusions

We show that these approaches can be utilized to quantify antibody concentrations in typically-challenging FFPE specimens with good sensitivity and as such could be utilized to assess efficacy of the monoclonal antibody administered. There are also potential applications related to rapid drug screening using the patient-derived xenograft model. Our future plans are focused on adapting these solutions to the characterization of immune checkpoint inhibitor therapeutics in standard-of-care FFPE tissues obtained from patients undergoing immunotherapy.

Correspondence: Hisani Horne (hisani.madison@astrazeneca.com)

Background

Programmed cell death ligand-1 (PD-L1) expression and tumour mutation burden (TMB) have been shown to be predictive of response to anti-PD-1/PD-L1 immunotherapies in various cancers. The prevalence and distribution of PD-L1 expression and/or tumour mutations in HCC and correlation with clinical characteristics are poorly understood. A better understanding of these biomarkers may help inform appropriate patient selection strategies in HCC.

Methods

PD-L1 expression was evaluated on tumour cells (TC), immune cells (IC), or combined TC and IC using the VENTANA PD-L1 (SP263) Assay in three independent HCC sample sets: 2 from commercial tissue banks (n = 500 and n = 2417) and 1 from patients enrolled in NCT02519348, encompassing a wide range of stage and grade of disease. NCT02519348 is a phase 2 study evaluating safety, efficacy outcomes of durvalumab with or without tremelimumab in advanced HCC. TMB was assessed in tissue by whole exome sequencing in a subset of 70 patients from NCT02519348.

Results

At a cut-off date of Feb 28, 2019, 282/335 (84.2%) patients enrolled in NCT02519348 were successfully evaluated for PD-L1 (Table 1). Significant expression was seen in ICs relative to TCs. Patients in NCT02519348 showed higher PD-L1 expression in TCs than commercial cohorts. In a univariate analysis using 1% cut offs, higher TC (but not IC or combined TC and IC) PD-L1 expression was associated with patients with HCV infection (p=0.003). Of 70 study patients tested, 55 were evaluable for TMB (median 2.59, range 0.46 - 5.61 Mut/Mb). Among patients with available TMB data there was no observed correlation between TMB and PD-L1 expression.

Conclusions

PD-L1 expression was observed in both TC and ICs in HCC, with the latter being more prevalent. Viral status and disease stage may impact PD-L1 expression in this setting, but further work is needed to confirm this. TMB and PD-L1 appear to identify distinct patient subsets in HCC.

Trial Registration

NCT02519348

Ethics Approval

The study (NCT02519348) is performed in accordance with ethical principles that have their origin in the Declaration of Helsinki and are consistent with ICH/GCP, and applicable regulatory requirements.

Correspondence: Kyle Smith (kyle@micromedicine.com)

Background

Cell separation plays a vital role in research and clinical settings for the development and monitoring of cutting-edge therapies. Despite its labor-intensiveness and variability, density gradient centrifugation-based method (DGM) has remained the primary method of upstream cell isolation for decades due to a lack of viable alternatives. This is problematic as DGM is a non-scalable, manual process. To address this lack of innovation, we have developed an automated Microfluidic System based on inertial focusing that enables label-free white blood cell (WBC) separation and concentration from 3-75mL of whole blood in short timescale with high consistency, providing reliable sample preparation for downstream functional assays.

Methods

WBCs were isolated from 15% ACD-A anticoagulated peripheral human blood using the Microfluidic System or DGM. Cell number, viability, and immune phenotype were evaluated by hematology analyzer and flow cytometry. To assess B-cell function, cells were cryopreserved post separation, thawed, and stimulated with IL-2 and R848, followed by Human IgG and IgM ELISPOT. To assess T-cell function, thawed cells underwent bead-based granulocyte depletion and stimulation with CEF peptide pool, followed by Human IFNγ ELISPOT.

Results

The prototype Microfluidic System consistently processed 40mL of anticoagulated blood in approximately 20 minutes with minimal hands-on time as opposed to 60–90 minutes for DGM with significant hands-on time. While DGM collects only peripheral mononuclear cells (PBMCs), the System isolates the total WBC population and may be beneficial for immunophenotyping. As shown in Table 1, the Microfluidic System consistently provided improved WBC or PBMC recovery, viability, purity, RBC depletion, and platelet depletion as compared to DGM. Immune phenotyping shows that the Microfluidic System also consistently resulted in improved recovery of lymphocyte subsets, including CD19+, CD3+, CD4+, and CD8+ cells (Table 2). B-cell and T-cell functionality were found to be equivalent between the two cell isolation methods based on IgM/IgG and IFNγ secretion, respectively. With the improved cell recovery using the Microfluidic System, more target cells from the same blood sample may be collected for downstream assays.

Conclusions

The Microfluidic System offers a faster, more reliable method than DGM for upstream cell separation from whole blood. The System consistently recovers more cells, including functional lymphocytes of different subsets, compared to DGM, potentially allowing more assays to be executed from the same blood sample. Overall, this technology has the potential to transform cell separation by automating a variable and labor-intensive processes, and therefore has utility in applications that require consistent cell quality and functionality.

Correspondence: George Dominguez (george@anixa.com)

Background

Myeloid-derived suppressor cells (MDSCs) are contributors in supporting tumor progression and escape [1,2]. Studies have quantified MDSCs to detect tumor development, monitor progression, and/or predict therapeutic responses [3, 4]. Here, we compared several machine learning (ML) approaches to analyze flow cytometry data to detect breast cancer (stage I/II) through manual gating and hypervoxelation of cell events.

Methods

We used standard multiparametric flow cytometry techniques to measure myeloid-derived suppressor cell (MDSC), myeloid, and lymphocyte cell populations found in the peripheral blood of 99 biopsy-confirmed early stage BCa patients and 88 healthy donor female (HDF) controls. Manual gating was performed to generate gated values, and raw flow cytometry data were transformed using HyperVOX to generate hypervoxelated cytometry event counts. The ML algorithms used were: support vector machine (SVM), Bayes SVM, Ensemble SVM, k-nearest neighbor (kNN), and pattern recognition neural network (PRNN). All algorithms were trained using data from 64 BCa patients and 69 HDF controls. Predictions were evaluated using the performance of each trained ML algorithm on 35 early stage BCa patients and 19 HDF that were not used for training (holdout test set).

Results

Using manually gated counts, the resulting accuracies were: SVM = 75.4%, Bayes SVM = 71.3%, Ensemble SVM = 65.6%, and kNN = 69.7%. Using hypervoxelated event counts, the resulting accuracies were: SVM = 78.7%, Bayes SVM = 77.1%, Ensemble SVM = 57.4%, kNN = 67.2%, and PRNN = 92.6%. Hypervoxelated data analyzed using PRNN resulted in the highest accuracy with a sensitivity of 91.4% and a specificity of 94.7%; the resulting AUC = 0.9098 (95%CI = 0.8031 to 1.000). Additionally, we tested 26 samples collected from patients with confirmed ductal carcinoma in situ (DCIS) using hypervoxelated counts with a PRNN. Even though they are clinically deemed as pre-cancerous (stage 0), 18 out of 26 (AUC = 0.8421; 95%CI = 0.7163 to 0.9679) were classified as BCa suggesting utility for detecting the existence of even a non-invasive cancerous lesion.

Conclusions

Although further study is needed, we believe that using PRNN with MDSC immunophenotyping, in conjunction with other known clinical risk factors, would allow for clinicians to make a more informed diagnosis and treatment recommendation when screening and for recommending subsequent interventions for early stage breast cancer.

References

1. Kumar V, Patel S, Tcyganov E, Gabrilovich D. The nature of myeloid-derived suppressor cells in the tumor microenvironment. Trends Immunol. 2016; 37:208-220.

2. Marvel D, Gabrilovich D. Myeloid-derived suppressor cells in the tumor microenvironment: expect the unexpected. J Clin Invest. 2015; 125:3356-3364.

3. Elliott L, Doherty G, Sheahan K, Ryan E. Human tumor-infiltrating myeloid cells: phenotypic and functional diversity. Front Immunol. 2017; 8:86.

4. Okla K, Wertel I, Wawruszak A, Bobinski M, Kotarski J. Blood-based analyses of cancer: circulating myeloid-derived suppressor cells – is a new era coming? Crit Rev Clin Lab Sci. 2018.

Ethics Approval

The study was approved by the Virtua Oncology (#20161), University of Pennsylvania (#826544), and Cooper Health (#17-174) IRBs.

Correspondence: Nicholas Dupuis (nicholas.dupuis@biognosys.com)

Background

Measurement of circulating biomarkers in cancer has proven utility for early detection, differential diagnosis, and predicting pre-treatment response to therapy. More recently, circulating proteomic biomarkers for pre-treatment prediction of therapeutic response have received additional attention due to the heterogeneous responses to immunotherapies. To develop a greater understanding of the circulating plasma proteome in subjects with cancer we have optimized a depleted plasma proteomic workflow, based on label-free data independent acquisition (DIA) mass spectrometry, and applied it to plasma from subjects with late stage NSCLC. This approach provides a deep and unbiased description of the plasma proteome and the dysregulated biological pathways associated with lung cancer.

Methods

Plasma samples from subjects with Stage III-IV non-small cell lung cancer (NSCLC, n = 15) and age matched healthy donors (n = 15) were depleted of 14 high abundance proteins using MARS Hu-14 spin columns (Agilent). All samples were prepared for mass spectrometric acquisition using two-hour gradients on a C18 column coupled online to a Thermo Scientific Q Exactive HF-X operated in DIA mode. Targeted data extraction was performed using Spectronaut (Biognosys) with a hybrid library approach. Statistical analysis was conducted to identify disease associated biomarker candidates and pathway analysis highlights dysregulated biological functions.

Results

A comprehensive protein spectral library was created containing 1,827 unique proteins. In DIA acquisition, in total 1,304 proteins were quantified across all samples (1,105 average per sample). Univariate statistical testing identified 162 dysregulated proteins (125 up-regulated and 37 down-regulated; q-value > 0.05 and log2 fold change > 0.58). In addition to the acute phase proteins (e.g. CRP and SAA1) which were previously verified to be elevated in subjects with NSCLC, partial least squares discriminant analysis helped identify additional proteins that are differentially expressed between the sample groups. Most relevant to immune function was CLC (Galectin-10), which was elevated in NSCLC samples and has been identified as key component supporting the suppressive function of Tregs.[1] Furthermore, F13A1 was suppressed in the NSCLC samples which is known to be associated with macrophage activation.

Conclusions

162 proteins were identified as candidate biomarkers and reflect the host immune response via acute phase response signaling, innate immune response, and other proinflammatory stimuli. Several of these markers have been linked to patient outcomes and poor prognosis.

Reference

1. Kubach, J., et. al.; Blood 2007 110:1550-1558

Correspondence: Ahmad Tarhini (tarhiniaa@gmail.com)

Background

Interleukin 2 (IL-2) plays a key role in antitumor immunity by enhancing survival of antitumor cytotoxic T lymphocytes and natural killer (NK) cells and promoting proinflammatory cytokines, that can lead to durable responses in patients with melanoma. High levels of vascular endothelial growth factor (VEGF) are associated with non-response to IL-2 and combination biotherapy with Ziv-aflibercept (inhibitor of the VEGF pathway) and high-dose IL-2 may lead to improved antitumor efficacy. Mechanistic studies utilizing peripheral blood of melanoma patients treated with this biotherapy may illuminate the underlying mechanisms of immune susceptibility and resistance [1].

Methods

Patients with stage III or stage IV inoperable melanoma were treated with high-dose IL-2 alone or in combination with Ziv-aflibercept in a phase 2 clinical trial [1] (NCI8628; Tarhini et al. Cancer. 2018). Peripheral blood mononuclear cells (PBMC) from treated patients (N=89) on this trial were tested at baseline (before initiating systemic immunotherapy), and 6-weeks (following immunotherapy initiation). High complexity (14-color) flow cytometry designed to detect key immunological biomarkers such as myeloid-derived suppressor cells (MDSCs), regulatory T cells (Tregs), proliferating T-cells, PD-1 and TIM3 expression on T-cells, and differentiation of T-cells into Th1, Th2 or Th17 phenotype were used to evaluate the correlation between immunological biomarker expression and efficacy. Statistical significance was determined using ANOVA or paired student’s t-test.

Results

Treatment with high dose IL-2 resulted in significant immune activation as detected by significant increases in both proliferating CD4+ (p<0.0001) and CD8+ (p<0.0001) T-cells at 6-weeks post-treatment in both treatment arms in addition to increase in Tregs (CD4+ CD25+ Foxp3+ T-cells; p<0.0001). Addition of VEGF inhibition showed a general trend towards decrease in classical monocytes (CD14+ CD16-; p=0.0769) as well as Th17 cells (defined as CD45RA- CCR6+ CXCR3- CCR4+; p=0.0597). In patients receiving combination therapy, a higher proportion of subjects experienced CBR (Clinically Beneficial Response = CR+PR+SD) compared to monotherapy and this CBR correlated with a decrease in CD4+ ICOS+ (p=0.0219), classical monocytes (CD14+ CD16-; p=0.0141), Th17 cells (CD45RA- CCR6+ CXCR3- CCR4+; p=0.0445) as well activated CD4+ T-cells (CD4+ CD38+ HLA-DR+; p=0.0285).

Conclusions

VEGF inhibition with Ziv-aflibercept adds significant immunomodulatory effects when combined with IL-2. Further correlative analyses determining the effect of combination therapy on progression-free survival and identifying predictive biomarkers of therapeutic efficacy are ongoing and will be presented at the meeting.

Acknowledgements

This United States (U.S.) National Cancer Institute (NCI)-sponsored study was initiated by the California Cancer Consortium under N01 contract NO1-CM-2011-00038. Laboratory correlatives were supported by Navigate BioPharma.

Trial Registration

https://clinicaltrials.gov/ct2/show/NCT01258855

Reference

1. Tarhini AA, Frankel P, Ruel C, Ernstoff MS, Kuzel TM, Logan TF, et al. NCI 8628: A randomized phase 2 study of ziv‐aflibercept and high‐dose interleukin 2 or high‐dose interleukin 2 alone for inoperable stage III or IV melanoma. Cancer. 2018;124(22):4332-41.

Ethics Approval

The study was initiated after approval by the ethics committee at the participating sites and was conducted in accordance with the Declaration of Helsinki.

Correspondence: Ben Sidders (benjamin.sidders@astrazeneca.com); Kris Sachsenmeier (kris.sachsenmeier@astrazeneca.com)

Background

Quantification of immune cell abundance using gene signatures from mRNA profiling has the potential to inform clinical studies of cancer immunotherapy. However, few of the signatures reported in previous studies have been validated therefore the concordance of signature scores with corresponding immune cell abundance is unknown.

Methods

To tackle this challenge we designed a two-stage validation strategy. Firstly we validate signatures computationally using previously published datasets. Secondly we generate expression profiling data from an immune cell spike-in experiment with human PBMCs. As a proof of concept experiment, we implemented the method to validate two gene signatures for CD141+ dendritic cells (DC) and CD56+ natural killer (NK) cells.

Results

We demonstrate gene signatures for both CD56+ NK and CD141+ DC cell types show high and significant agreement to the corresponding immune cell abundance.

Conclusions

This work establishes a starting point for validating gene signatures through an approach that is tractable yet recapitulates real-world variability we might expect in clinical use.

Correspondence: Ariane Lozac’hmeur (ariane.lozachmeur@tempus.com)

Background

Microsatellite instability is a clinically actionable genomic indication for cancer immunotherapy. In microsatellite instability-high (MSI-H) tumors, defects in DNA mismatch repair (MMR) can cause a hypermutated phenotype where alterations accumulate in the repetitive microsatellite regions of DNA. MSI detection is typically performed by subjecting tumor tissue (“solid biopsy”) to clinical next-generation sequencing or specific assays, such as MMR IHC or MSI PCR. Circulating cell-free tumor DNA (cfDNA) testing (“liquid biopsy”) is rapidly emerging as a less invasive method for cancer detection and monitoring disease progression. Here, we explore the possibility of detecting MSI in cfDNA and develop a novel cfDNA MSI detection assay with high specificity.

Methods

The Tempus cfDNA targeted panel contains 39 highly informative microsatellite loci previously used by the clinically validated Tempus xT 595-gene panel. For each microsatellite locus, we identified all sequencing reads that mapped to the corresponding microsatellite region and quantified the number of repeat units contained within the sequencing read. Next, three distinct summary statistics were calculated to characterize the distribution of the number of repeat units for each locus. Finally, using 54 labeled patient samples (17 MSI-H, 37 microsatellite stable) sequenced with the Tempus cfDNA panel, a k-Nearest Neighbor (k-NN) classifier was trained to classify each locus for a new sample. Patient samples with more than 50% unstable loci were classified as MSI-H.

Results

We validated the ability of our model to detect MSI on a new independent validation dataset. MSI-H status was detected in 6 patient samples. In 3 of these patients (2 colorectal, 1 skin cancer), abnormal MMR IHC confirmed the detected MSI-H status. In the other 3 patients (1 colorectal, 1 non-small cell lung cancer, and 1 endometrial cancer), MSI-H status was confirmed by our clinically validated solid tumor MSI assay. Furthermore, the reliability of the model was validated in 10 technical replicates from 2 MSI-H patients in our training dataset. The results were 100% concordant with all 10 replicates classified as MSI-H.

Conclusions

These results demonstrate the ability of our assay to detect MSI in cfDNA with high specificity, providing a transformative opportunity to report a clinically actionable insight alongside other somatic changes detected from cfDNA.

Correspondence: Sheeba Irshad (sheeba.irshad@kcl.ac.uk)

Background

Triple negative breast cancer (TNBC) accounts for 10-20% of breast cancer and is associated with particularly poor prognosis. Patients are commonly treated with neoadjuvant chemotherapy (NAC) and response to treatment is a strong predictor of overall survival. Recently, the ability of chemotherapeutics to stimulate an anti-tumour immune response has been appreciated as an important mechanism of action; possibly contributing to the elimination of distant micro-metastatic disease by resetting of the attenuated functional immunity. In TNBCs, higher levels of tumour-infiltrating lymphocytes correlate with response to NAC and high intra-tumoral levels of immune-related genes, including those associated with type I interferon responses, and the presence of CD8+ cytotoxic T lymphocytes, correlate with improved disease outcome.

Methods

The underlying hypothesis of this study is that phenotypic profiling of peripheral blood cells have the potential to inform clinical decisions and help predict therapeutic response, with lower costs and higher compliance than serial tumour biopsies, due to their minimal invasiveness. Whilst significant research efforts have been made to assess circulating markers such as circulating tumour cells and circulating tumour DNA as potential biomarkers; understanding the evolving peripheral “immunological status” of TNBC patients on NAC is warranted.

We therefore set out to analyse serial blood samples from TNBC patients receiving NAC to monitor the changes in the peripheral immune response through deep analysis of functional and phenotypic immune markers. We investigated (1) whether chemotherapy affects the immune phenotype; and (2) whether a defined peripheral blood immune phenotypic profile relates to treatment response.

Results

Here we present preliminary results from 10 TNBC patients receiving NAC. Analysis of 39 PBMC populations using mass cytometry by time-of-flight (CyTOF), highlighted phenotypic changes in B cell populations in response to treatment, in particular a dramatic increase in circulating regulatory B cells (CD19+CD24+CD38+) post-chemotherapy (5.4% and 46.2% of B cells pre- and post-chemotherapy, respectively, p=0.0004). We also detected an increase in expression of exhaustion markers (CD38+CD39+) on CD8+ T cells which was associated with poor response to chemotherapy (0.8 and 2.7 fold increase from baseline in exhausted CD8+ T cells in patients with pathological complete response and residual disease, respectively, p=0.008).

Conclusions

We now plan to integrate these data with Luminex profiling of 36 serum cytokines, mass spectrometry analysis of circulating exosomes and clinicopathological and standard of care blood monitoring. Taken together, this study aims to provide a comprehensive analysis of the utility of immune monitoring to understand TNBC patient response to NAC.

Ethics Approval

The study was approved by NRES Committee London - Chelsea, approval number 13/LO/1248.

Correspondence: Arnav Mehta (nawi214@gmail.com)

Background

Immune checkpoint blockade (ICB) has revolutionized the treatment of many solid tumors, including metastatic melanoma. Despite recent successes, many patients fail to respond or are overcome by severe toxicities that limit further treatment. To date, there are no non-invasive predictors of response and toxicity that can guide treatment decisions. In this work, we perform whole plasma and plasma-derived exosome proteomic profiling to construct a predictive model of immunotherapy response and toxicity, and to glean further biologic insight into the mechanisms underlying resistance to ICB.

Methods

Whole plasma was analyzed in a cohort of 55metastatic melanoma patients receiving anti-PD1 antibodies (MGH IRB #11-181) at baseline, and on-treatment at 6 week and 6 month time-points. Exosomes were analyzed in 15 of these patients for all time-points. Proteomic analysis was performed using an innovative multiplex proximity extension assay that enabled detection of more than 1000 proteins simultaneously. A linear mixed model with maximum likelihood estimation for model parameters was used to analyze differences between patient groups, and significant differences were determined after Benjamini and Hochberg multiple hypothesis correction.

Results

Between plasma baseline and on-treatment time-points, 67 differentially expressed proteins were identified including markers of inflammation such as PD1, CXCL9, CXCL10, CXCL11, IL10, CCL3 and TNFR2. Exosome samples had a distinct protein signature over the treatment period compared to plasma, including differential expression of CXCL16, CCL18, CCL20, and IL6, among others. 41 proteins were differentially expressed in plasma between ICB responders and non-responders including several inflammatory proteins such as CD28, TNFb, MCSFRa and IL8, and others implicated in melanoma resistance, such as MIA and ERBB2. Similarly, exosome revealed a distinct protein signature between responders and non-responders compared to plasma consisting of CXCL9, CXCL13, CXCL16, CCL19, CD8a, GZMA and CD5 expression. Whereas plasma proteins reflected a myeloid signature, exosome proteins reflected a lymphoid signature, suggesting that the two compartments may capture elements of different immune processes. Integrating data from both plasma and exosome proteomics, we applied machine learning tools to build a predictor of response. Further analysis to look for predictors of toxicity is currently underway.

Conclusions

Overall, our work suggests that plasma and exosome protein signatures are distinct and may reflect unique immunological processes. Proteomic analysis of these compartments may be an effective way for non-invasive liquid biopsy to predict ICB response.

Correspondence: Marijana Rucevic (m.rucevic@olink.com)

Background

The response of metastatic melanoma to anti-PD1 is heterogeneous. We performed proteomic profiling of patient plasma samples to build a predictor of immunotherapy response and uncover biological insights underlying primary resistance.

Methods

An initial cohort comprised 55 metastatic melanoma patients receiving anti-PD1 (Pembrolizumab or Nivolumab) at Massachusetts General Hospital (MGH), and 116 additional patients comprised a validation cohort. Plasma samples were collected baseline and on-treatment, at 6 weeks and 6 months’ time-points, and profiled for 1000 proteins by a multiplex Proximity Extension Assay (PEA, by Olink Proteomics). A subset of patients had single-cell RNA-seq (Smart-Seq2 protocol) performed on tumor tissue. Group differences and treatment effects were evaluated using linear mixed models with maximum likelihood estimation for model parameters, and Benjamini and Hochberg multiple hypothesis correction.

Results

At the baseline, 6 differentially expressed proteins were identified between responders (R) and non-responders (NR) whereas immune suppression marker ST2 and IL-6 were found significantly higher among NR. Kaplan-Meier survival curves stratified by the baseline differentially expressed proteins were highly predictive of overall survival (OS) and progression-free survival (PFS). At 6-weeks on-treatment time point, 80 proteins were found differentially expressed between R and NR including several proteins implicated in primary or acquired resistance (IL8, MIA, TNFR1 among others). Several 6-weeks differentially expressed proteins were highly predictive of survival (ICOSL, IL8, MIA). Furthermore, 160 significantly differentially expressed (DE) proteins were identified across the treatment period majority of which are reflective of immune activation under the pressure of the immunotherapy. Analysis of single-cell RNA-seq data of tumor tissue from a subset of these patients revealed that gene expression of most proteins predictive of response were enriched among tumor myeloid cells, with the remainder of proteins being reflective of exhausted T cell states.

Conclusions

These results unveil a putative role of myeloid cells within the tumor microenvironment in anti-PD1 response or primary resistance. Whole plasma proteomic profiling of anti-PD1 treated patients revealed DE proteins between R and NR that may enable a liquid biopsy to predict anti-PD1 response. Importantly, we demonstrate the relationship of serum biomarkers to OS and PFS and are currently attempting to build machine learning classifiers as predictors of response to checkpoint therapy leveraging early and late on-treatment time points.

Correspondence: Rania Younis (ryounis@umaryland.edu)

Background

There is an urgent need for immune biomarkers that can monitor the status of inflammation of cancer patients. Soluble biomarkers represent a convenient prognostic and diagnostic method. Semaphorin 4D (Sema4D) is a glycoprotein that can function as a transmembrane protein or a cleaved soluble form (sSema4D), that we previously detected in peripheral blood [1]. The role of Sema4D as an inflammatory mediator in several pathological aspects and its role in tumor immune suppression [2,3], highlights its significance as a molecule to be further investigated for translational potential. The objective of this work was to investigate the level of sSema4D in plasma in relation to the histological pattern of tumor inflammation of head and neck squamous cell carcinoma (HNSCC) patients in real time.

Methods

Under University of Maryland institutional review board approval and upon patient consent, we obtained paired peripheral blood and tumor tissue of thirty-nine HNSCC patients, collected at the same time point to allow for real time correlative analysis. Thirty eight patients of classic autoimmune conditions, thirteen allergy patients, seven osteoarthritis patients and thirty-one healthy donors were included as controls. The level of Sema4D in plasma was detected using tailored direct ELISA assay. The histological pattern of tumor inflammation [4] was analyzed by three pathologists using the immunohistochemical staining of Sema4D of the tumor associated inflammatory cells (TAIs).

Results

sSema4D levels in plasma of HNSCC and the autoimmune individuals (p=0.18, independent-samples Mann-Whitney test), were not statistically significantly different, but sSema4D levels were significantly higher in the HNSCC and the autoimmune groups compared to healthy donors (p<0.001 for both comparisons). Three histological patterns of tumor inflammation were defined according to the extent of stromal inflammation and TAIs infiltrate into the tumor islands. First; the inflamed type (TAIs infiltrated the tumor cells), second the TAIs excluded type (inflamed stroma but TAIs did not infiltrate the tumor islands and/or were excluded by a thin peri-tumoral fibromyxoid zone) and third as deserted ( minimal to no TAIs in the peri-tumroal stroma or the tumor islands). The paired tumor tissue and blood samples collected at the same time point, showed that high levels of sSema4D in plasma, correlated directly with TAIs excluded histological pattern of tumor inflammation (p= 0.04).

Conclusions

Our data presents a novel role of Sema4D as a soluble immune biomarker that can read in real time the histological pattern of tumor inflammation. This opens new avenues for personalized immunotherapy and HNSCC patient stratification.

References

1. Derakhshandeh R, Sanadhya S, Lee Han K, Chen H, Goloubeva O, Webb TJ, Younis RH. Semaphorin 4D in human head and neck cancer tissue and peripheral blood: A dense fibrotic peri-tumoral stromal phenotype. Oncotarget. 2018; 9:11126-11144.

2. Younis RH, Han KL, Webb TJ. Human Head and Neck Squamous Cell Carcinoma-Associated Semaphorin 4D Induces Expansion of Myeloid-Derived Suppressor Cells. J Immunol. 2016; 196:1419-29.

3. Clavijo PE, Friedman J, Robbins Y, Moore EC, Smith E, Zauderer M, Evans EE, Allen CT. Semaphorin4D Inhibition Improves Response to Immune-Checkpoint Blockade via Attenuation of MDSC Recruitment and Function. Cancer Immunol Res. 2019; 7:282-291.

4. Ayers M, et al. IFN-gamma-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest. 2017; 127:2930–40

Ethics Approval

The study was approved by University of Maryland institutional review board, Institutution‘s Ethics Board, approval number (HCR-HP-00073603)

Correspondence: Adil Daud (daudai@gmail.com)

Background

Background: Activation markers such as PD-1 and PDL-1 as well as tumor mutation burden and IFN-gamma gene expression profiling have been explored as markers for response in melanoma and in other cancers. PD-1 inhibition activates checkpoint positive cytotoxic T lymphocytes (cpCTLs) inducing tumor regression. We have previously demonstrated that baseline peCTL frequency predicts response to anti–PD-1 monotherapy and combination CTLA4/PD-1 blockade in metastatic melanoma. We evaluated the frequency of this CD8+ T cell subset at baseline and after immunotherapy treatment and evaluated the utility of the intensity of expression activation marker expression as a surrogate for tumor response as assessed by flow cytometry.

Methods

We identified 490 patients with melanoma biopsied pre and post PD-1 therapy and available for analysis. Of these 148 patients had unresectable stage III or stage IV melanoma and were treatment naïve and started PD-1 therapy following biopsy. An additional 61 patients were identified with PD-1 resistant melanoma. Approximately 2 × 106 cells were stained with anti-hCD3, anti- hCD8, anti-hCD45, anti-CD4 , anti-Foxp3, anti–hCTLA-4 (14D3), anti–PD-1 , anti–HLA-DR, anti–PD-L1, and LIVE/DEAD Fix- able Aqua Dead Cell Stain (Life Technologies). Data were acquired by an LSRFortessa (BD Biosciences) and analyzed using FlowJo software (Tree Star, Inc.). Objective Responses were evaluated by RECIST 1.1, CR/PR were classified as “responders” and SD/PD as “non-responders.”

Results

: cpCTL percentage correlated with response. The mean cpCTL was 27.1% for treatment naïve responders (R), 16.52% for treatment naïve non-responders (NR) and 8.59% for PD-1 resistant patients post treatment (ANOVA p=0.0003 for R/NR, 801 (ANOVA p=0.0002).

Conclusions

PD-1 progressive patients are significantly depleted in cpCTL even compared to treatment naïve non-responders, suggesting that additional T cell influx may be needed for effective checkpoint blockade in these patients. In treatment naïve melanoma, CD8+ activation as shown by CTLA-4 MFI has an optimal range along the activation-dysfunction spectrum, and strongly correlates with response to PD-1 checkpoint therapy.

Acknowledgements

We gratefully acknowledge the patients who participated in this study

Ethics Approval

The study was approved by UCSF's Ethics Board approval number 138510

Correspondence: Sarah Warren (swarren@nanostring.com)

Background

The efficacy of immune response in solid tumor settings is driven by many factors including the biology of the tumor, the immune system, and the microenvironment. The Tumor Inflammation Signature (TIS) is an 18-gene Research Use Only (RUO) signature that measures the presence of a preexisting immune response on the nCounter platform and enriches for response to pembrolizumab [1]. We have incorporated TIS into the PanCancer IO 360 panel, a 770-gene RUO expression assay containing 48 additional signatures of tumor-immune biology. To accompany this panel, we have created analysis software that associates the gene expression and signature scores with annotations of the samples to characterize the immune system, tumor, and stroma within the tumor microenvironment to give insight into underlying biology of response to treatment, disease progression, survival, and other sample characteristics.

Methods

The PanCancer IO 360 assay relies on gene signatures to describe biological processes, measure the presence of 14 different immune cell populations, or report the expression of key therapeutic targets. Data from The Cancer Genome Atlas (TCGA) was used for signature training and development. Signatures are either single genes, weighted linear sums of multiple genes with coregulated expression, or algorithms to determine under-expression of genes in a coregulated pathway [2,3]. The analysis software leverages differential expression analysis and Cox proportional hazard modeling to associate gene expression and signature scores with the clinical annotations.

Results

In the PanCancer IO 360 analysis, genes and signatures are compared to clinical annotations through heat maps, volcano plots, forest plots, box plots, waterfall plots, swim lane plots, Kaplan Myer plots, scatter plots, and the IO 360 wheel plot. The report is delivered in an HTML format that provides interactive visualizations, quality control, and downloadable results. Data are analyzed individually and as part of larger treatment groups.

Conclusions

The PanCancer IO 360 assay is a tool for characterizing transcriptional patterns associated with tumor-immune interactions that can be applied across a wide range of cancer types. Gene signatures enable robust characterization of immune activity from small sample cohorts, and the report simplifies the interpretation of results. This combination enables researchers to have insight into clinically relevant biology that will ultimately lead to help drive the immune-oncology field.

References

1. Ayers M, Lunceford J, Nebozhyn M, et al. IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest. 2017;127(8):2930-2940.

2. Danaher P, Warren S, Dennis L, et al. Gene expression markers of Tumor Infiltrating Leukocytes. J Immunother Cancer. 2017;5:18.

3. Danaher P, Warren S, Lu R, et al. Pan-cancer adaptive immune resistance as defined by the Tumor Inflammation Signature (TIS): results from The Cancer Genome Atlas (TCGA). J Immunother Cancer. 2018;6(1):63.

Correspondence: Jose Villasboas (Villasboas@mayo.edu)

Background

CyTOF produces high dimensional single cell data allowing simultaneous monitoring of multiple immune cell subsets. This enables characterization of the immune system in normal and disease states. We developed a standardized pipeline to study human peripheral blood mononuclear cells (PBMCs) of cancer patients. Here we detail our process and present early findings on a cohort of 40 patients with early-stage triple-negative breast cancer (TNBC) treated with neoadjuvant chemotherapy.

Methods

Thirty commercially-available metal-tagged antibodies were optimized to identify major cell subsets using a 4-point titration scheme. Replicates of cryopreserved PBMCs from a pool of 4 healthy donors were created for panel titration and used as longitudinal references. We studied 40 cryopreserved PBMCs from patients with TNBC. We stained samples individually using standard protocol, barcoded overnight during DNA intercalation, and pooled for acquisition. Debarcoded output data was normalized on a per-batch basis to the median intensity of EQbeads. We uploaded files to an automated platform for unbiased processing. Patient-level meta-data was added to experiment matrix to determine differential abundance of immune subsets across clinical and pathological groups.

Results

We required 7 rounds of titration to optimize antibody concentrations. Data was collected on over 23 million live single-cell events (Table 1) assigned to 31 canonical populations (Figure 1A). The median frequencies of main populations were: B cells (11.9%), T-CD4+ cells (34.3%), T-CD8+ cells (11.7%), NK cells (8.6%), and monocytes (11.3%). At the profiling level, 76 subsets were agnostically identified, with B, T, NK, and monocytes broken down into 10, 32, 8, and 13 subsets respectively. Activated (CD38+CD161+) CD16+NK cells (Figure 1B) were more prevalent in TNBC samples (median 5.2%, range 0.5%-11.9%) compared to normal blood (median 0.76%, range 0.1%-2.4%). A population with phenotype suggestive of myeloid derived suppressor cells (LineagenegHLA-DRLowCD66b+CD24+CD16+; Figure 1C) was also more prevalent in TNBC samples (median 1.2%, range 0.1%-17.3%) compared to normal blood (median 0.6%, range 0.2%-1.0%). These populations demonstrated opposite association trends when patients were stratified by clinical outcomes. Activated NK cells were more frequent in patients achieving pathological complete response while MDSC-like cells were more frequent in those with residual disease (Figures 1D-1E).

Conclusions

We demonstrated the feasibility of a complete pipeline for deep phenotyping of cryopreserved PBMCs in cancer patients. Our approach identified rare cell subsets using an unbiased analysis tool, linking specific populations to opposite clinical outcomes. High dimensional immune monitoring is feasible and should be applied to study the immune system of cancer patients at large.

Acknowledgements

This work was part of the Mayo Clinic Cancer Immunome Project which is supported by the Wohlers Family Foundation. Samples were obtained in collaboration with investigators from the Mayo Clinic Breast Cancer Genome Guided Therapy (BEAUTY) study. The BEAUTY study is funded in part by the Mayo Clinic Center for Individualized Medicine; Nadia’s Gift Foundation; John P. Guider; the Eveleigh Family; George M. Eisenberg Foundation for Charities; generous support from Afaf Al-Bahar; and the Pharmacogenomics Research Network (PGRN). Other contributing groups include the Mayo Clinic Cancer Center and the Mayo Clinic Breast Specialized Program of Research Excellence (SPORE).

Ethics Approval

The study was reviewed approved by the Mayo Clinic Institutional review board (IRB).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P23).

High dimensional immune monitoring of breast cancer PBMCs

Correspondence: Xiao-Jun Ma (xiao-jun.ma@bio-techne.com)

Background

The tumor microenvironment (TME) is a network of complex interactions between the tumor and surrounding immune cells. Immunotherapies including immune checkpoint blockade have demonstrated therapeutic efficacy and durable responses for several tumor types, however most patients are nonresponsive or develop resistance to such immunotherapies. To identify new predictive biomarkers to better stratify patients, it is essential to comprehensively characterize the immune cells within the TME at the molecular level. Traditional methods to assess gene expression in tissues lack either spatial information or sensitivity/specificity. To address this, we have developed a novel workflow combining the single molecule and single cell visualization capabilities of the RNAscope in situ hybridization (ISH) assay with the highly multiplexed spatial profiling capabilities of the GeoMx™ Digital Spatial Profiler (DSP) RNA assays (Research Use Only).

Methods

The fully automated RNAscope Multiplex Fluorescent assay was used to visually identify CD3E (T-cell)-enriched regions and CD19 and CD20 (B-cell)-enriched regions within FFPE human lung cancer tissues. Using the GeoMx DSP, 10 CD3E-enriched regions of interest (ROI) and 10 CD19-enriched ROI were spatially profiled for 78 genes related to immune-oncology research. The RNAscope Multiplex Fluorescence assay was used again to visually confirm the differentially expressed genes between the T and B-cell-enriched regions with single cell resolution.

Results

To show a workflow combining RNAscope molecularly guided visualization and GeoMx DSP profiling is feasible, we confirmed that both assay protocols are compatible. We then examined concordance between GeoMx DSP and RNAscope ISH data, demonstrating that RNAscope and GeoMx DSP data can be obtained on the same section. To test the full automated workflow, we compared the differentially expressed genes within the T cell and B cell-enriched ROI. The RNAscope assay confirmed that, while the expression of the immunoregulatory molecules CTLA4, PD-L1, PD-1, and ICOSLG were detected in both ROI, the CD3E (T-cell)-enriched ROI demonstrated significantly higher expression of these checkpoint markers. Compared to the CD19-enriched ROI, the CD3-enriched ROI also showed increased inflammatory signature, demonstrated by elevated levels of cytokines and chemokines such as CCL5, CXCL9 and IFNG.

Conclusions

We present a robust workflow that overcomes the historical limitations of ISH and IHC by combining high resolution imaging with high plex profiling. With this workflow, the RNAscope ISH technology can molecularly guide the GeoMx DSP to precisely profile ROI while retaining the morphological context of heterogenous tumors. Furthermore, RNAscope assays can be used to confirm GeoMx DSP-identified gene expression signatures at single cell resolution.

Correspondence: Jeffrey Weber (jeffrey.weber@nyulangone.org); Jonathan Schneck (jschnec1@jhmi.edu)

Background

Since the introduction of checkpoint blockade inhibitors for cancer immunotherapy, numerous studies have sought to identify biomarkers predictive of patient response [1]. However, the relevance of antigen-driven responses to the tumor has yet to be investigated. To address this question, we examined T cell responses to MART-1, an antigen overexpressed in melanoma cells and a target for melanoma clinical trials that have had variable degrees of success. We hypothesized that features of patients’ MART-1 CD8+ T cell repertoires could predict their response to checkpoint blockade.

Methods

To understand the MART-1 T cell repertoire, MART-1 CD8+ T cells were expanded from HLA-A2+ melanoma patients and healthy donors using artificial antigen presenting cells (aAPC) or peptide-pulsed dendritic cells. Tetramer positive cells were sorted after 14-22 days and CDR3β sequenced. Motif analysis based on sequence homology was performed using the Immunomap algorithm by clustering 11,252 unique MART-1 CDR3β sequences from 33 samples and 20 donors, including five nivolumab responders and five non-responders [2].

Results

No significant difference in the frequency of MART-1 expanded T cells was seen between healthy donors and melanoma patients with or without checkpoint therapy. There was no immunodominant Vβ gene usage and limited clonotype overlap between donors. However, sequence homology showed extensive overlap between donors, driven by two clusters present in 60% and 80% of samples at average frequencies of 10% and 14%, respectively. These clusters were homologous to each other as well as the DMF4 T cell receptor (TCR), one of the first clinically used genetically engineered T cells, with a known crystal structure [3,4]. The core region of these clusters contained a conserved amino acid motif that was identical to contact residues between the DMF4 TCR and MART-1 peptide bound to HLA-A2. The motif identified from the core region of these clusters was highly conserved across samples, present almost exclusively in the junctional region between the D and J genes of the CDR3β, and encoded by a diverse range of nucleotides, all evidence of selective pressure. Despite its conservation, the frequency of this motif was nearly six times lower in pre-therapy samples expanded from non-responders compared to responders (40% vs. 7%, p=0.0045, Figure 1).

Conclusions

Since the frequency of the identified MART-1 TCR motif is significantly lower in non-responders compared to responders, it could potentially be used as a biomarker to predict response of HLA-A2+ melanoma patients to checkpoint blockade prior to the onset of therapy.

References

1. Zappasodi R, Wolchok JD, Merghoub T. Strategies for Predicting Response to Checkpoint Inhibitors. Curr Hematol Malig Rep. 2018;13(5):383-395.

2. Sidhom J-W, Bessell CA, Havel JJ, Kosmides A, Chan TA, Schneck JP. ImmunoMap: A Bioinformatics Tool for T-Cell Repertoire Analysis. Cancer Immunol Res. January 2017; 6(2):151-162.

3. Rosenberg SA, Packard BS, Aebersold PM, et al. Use of Tumor-Infiltrating Lymphocytes and Interleukin-2 in the Immunotherapy of Patients with Metastatic Melanoma. N Engl J Med. 1988;319(25):1676-1680.

4. Borbulevych OY, Santhanagopolan SM, Hossain M, Baker BM. TCRs used in cancer gene therapy cross-react with MART-1/Melan-A tumor antigens via distinct mechanisms. J Immunol. 2011;187(5):2453-2463.

Ethics Approval

The protocol was approved by the NYU Institutional Review Board, i16-01975

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P25).

See text for description

Correspondence: John O'Rourke (John.ORourke@Sartorius.com)

Background

Immunotherapy is an actively growing arena in oncotherapeutics research and development. In this context, whether testing CAR-T cells, checkpoint inhibitors, or novel bispecific antibodies, the ultimate goal is to modulate the immune system to harness its tumor killing power. T cells play a critical role in immune-regulated clearance of both liquid and solid tumors. Upon antigenic stimulation and activation, T cells rapidly expand, secrete cytokines, and differentiate to various functional subsets (e.g. effector T cells, memory T cells). On the other hand, suppression of T cells (i.e. exhaustion) leads to immune escape and the spread of tumor cells. Mouse models remain the most commonly used animal system for in vivo and in vitro cancer biology research and drug discovery. As researchers move forward to either better understand the role of T cells in cancer biology or to develop novel immunotherapies, there is a need for improved methods to quickly gather comprehensive data on T cell biology in this model. To address this, we demonstrate a multiplexed, high-throughput, robust assay workflow capable of measuring multiple murine T cell biology endpoints quickly and reproducibly in a single-well format.

Methods

In our workflow, stimulated mouse T cells were assayed in a 96-well plate using fluorescent antibodies against CD3, CD4, CD8, CD69, CD44, CD62L, and PD-1, QBeads for cytokine detection, and markers for cell viability and proliferation. Data were acquired on the iQue3 technology (VBR configuration) and analyzed on a plate-based level using the integrated ForeCyt software.

Results

Our assay workflow enabled simultaneous evaluation of viability, interrogation of helper and cytotoxic T cells for markers of activation and exhaustion, and identification of key memory subsets. Proliferation and secreted cytokines (IFN-gamma and IL-2) were also quantified. Data analysis and visualization of multiple endpoints was streamlined and performed in real time using the ForeCyt software.

Conclusions

The assay was completed in four hours, including data analysis. This workflow saves the end user’s time and resources by combining multiple experiments into a single, multiplexed workflow, and helps minimize subject-to-subject variability. Altogether, our workflow allows for easy phenotype and functional profiling of murine T cells in a single-well format while generating actionable results in a matter of hours.

Correspondence: Soner Altiok (soner@nilogen.com)

Background

The tumor stroma consists of various components of the tumor microenvironment including tumor cells, fibroblasts, immune cells and the extracellular matrix. Spatial organization and dynamic interplay of the complex cell-to-cell interactions play an important role in cellular phenotypes that can result in permanent alterations in cellular functions and response to oncology as well as immuno-oncology drug treatments. While informative, conventional 2D tumor dissociated models do not maintain the stromal-stoichiometry of the tumor microenvironment, lacking vital support mechanisms necessary to accurately assess ex-vivo tumor cell viability and immune-cell activation after drug treatment. Here, we describe a functional quantitative multiplex immunofluorescence platform, 3D-plEX, to quantify drug-mediated changes in tumor immune microenvironment and tumor cell viability in intact 3D tumor organoids of patient tumor samples.

Methods

All patient tumor samples were obtained with patient consent and relevant IRB approval. Unpropagated live 3D tumoroids measuring 100-150 micron in size were prepared from fresh patient tumors using a proprietary technology, pooled together to represent tumor heterogeneity and equally distributed to different treatment groups including nivolumab, ipilimumab, atezolizumab and urelumab singly or in different combinations. Cell media was collected for multiplex cytokine release assay. Tumoroids were fixed and embedded for multiplex immunofluorescence studies. In addition to tumor cell killing, treatment-mediated changes in TME was analyzed in each treatment group side-by-side using multiplex immunofluorescence markers including CD4, CD8, FoxP3, CD68, Pan-CK, PD-L1 and Ki67.

Results

Our results demonstrated that 3D-plEX platform using clinically relevant intact, uniformly sized tumoroids of fresh patient tumor tissue is highly versatile and reliable approach to quantify drug-mediated changes in cellular composition and spatial organization of the tumor immune microenvironment.

Conclusions

Combination of this approach with multiplex cytokine release assay allows a comprehensive understanding of dynamic changes within the tumor tissue upon drug treatment. The impact of different immuno-oncology drug treatments ex vivo on TME will be discussed. Application of this platform in the clinical studies may also allow determining the most effective combinatorial therapeutic strategies for individual patients.

Correspondence: Sizun Jiang (sizunj@stanford.edu); Xavier Rovira Clave (xrovira@stanford.edu); Garry Nolan (gnolan@stanford.edu)

Background

In tumor microenvironment, tumor-immune interactions are indicated by cell surface proteins such as T cell receptor (TCR) and PD-L1. The key workhorse for studying these cellular interactions is via imaging; conventional imaging methods are limited by the number of channels and the spatial resolving capabilities.

A new modality of imaging, Multiplexed Ion Beam Imaging (MIBI) [1,2], can resolve >40 parameters simultaneously in biological samples. MIBI can current attain single cell resolutions but has difficulties in resolving fine subcellular features. Here, we present Expansion MIBI (ExMIBI), which combines a physical expansion of a biological sample with the MIBI imaging method. ExMIBI will be critical for the scientific community to obtain previously inaccessible insights into the fine details of tumor microenvironment and cancer-immune cell interactions, and promises to unravel fundamental insights in patient immunotherapy responses.

Methods

Expansion microscopy (ExM) [3,4] is a technique that can physical expansion of biological specimens 4 to 10 folds through polymer chemistry, three-color fluorescent imaging of cellular features with an apparent lateral resolution of 70 nm in diffraction-limited confocal microscopes has been achieved. However, the expanded gel is fragile and contains up to 99.9% water, which limits its usage in imaging method that requires high vacuum condition. We explored a way to collapse the tissue-containing gel on a complementary charged substrate to achieve a vacuum-compatible gel that can be imaged by the MIBIscope, with lateral resolution <100 nm. Various methods for sample charging removal are systematically tested for imaging a non-conductive gel in MIBI.

Results

We have established a robust method, ExMIBI, that allows ExM hydrogels to be compatible with the high vacuum imaging conditions of the MIBI. This method can achieve 40 parameters. A validated panel of MIBI compatible antibodies, focusing on the immune system, is being tested and established for ExMIBI in FFPE tissues (Figure 1).

Conclusions

The combination of ExM and MIBI, termed ExMIBI, permits highly multiplexed super resolution imaging of tissue samples. We will now be able to map previously inaccessible, finer details of the tumor microenvironment. The application of ExMIBI to dissect cellular immune interactions, in their spatial biological context, will allow a better understanding into the basic principles of our immune system in healthy and disease states.

Acknowledgements

We thank Matt Newgren for tireless technical support on the MIBI instrument. This research has received advice and help from Prof. Michael Angelo, Prof. Sean Bendall and their research group. B.Z. is supported by the Stanford Graduate Student fellowship. S.J is supported by a Stanford Dean’s Fellowship and the Leukemia & Lymphoma Society Career Development Program. X.R.-C. is supported by a long-term EMBO fellowship. This work was supported by grants from the FDA, NIH, Parker Institute for Cancer Immunotherapy, the Bill and Melinda Gates Foundation, as well as the Rachford and Carlota A. Harris Endowed Professorship to G.P.N.

References

1. Angelo M, Bendall SC, Finck R, et al. Multiplexed ion beam imaging of human breast tumors. Nature Medicine. 2014;20(4):436–42.

2. Keren L, Bosse M, Marquez D, et al. A Structured Tumor-Immune Microenvironment in Triple Negative Breast Cancer Revealed by Multiplexed Ion Beam Imaging. Cell. 2018;174(6):1373-87.E19.

3. Chen F, Tillberg PW, Boyden ES. Expansion Microscopy, Science. 2015;347(6221):543-8.

4. Tillberg PW, Chen F, Piatkevich KD, et al. Protein-retention expansion microscopy of cells and tissues labeled using standard fluorescent proteins and antibodies. Nature Biotechnology. 2016;34(9):987–92.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P28).

The workflow and sample images of ExM-MIBI

Correspondence: Charles Caldwell (ccaldwell@flagshipbio.com)

Background

Manual pathology assessments of Immunohistochemistry (IHC) markers in immune oncology (IO) is often challenging and results can be highly variable[1,2]. Measuring biomarker presence in IO must take in to account both immune and tumor environments and provide contextual information on the interaction between tumor and immune biomarker landscapes [3]. Due to the complex nature surrounding tissue biomarker interpretation in IO, digital image analysis (IA) solutions have been developed that layer complex artificial intelligence (AI) and machine learning algorithms to obtain full tissue biomarker profiles necessary for drug development and patient stratification[4].

Here, a comprehensive tissue analysis solution is presented in monoplex PD-L1 and CD8 stained slides that includes precise digital biomarker scoring in tumor and stromal compartments, recapitulation of common scoring paradigms, analysis of biomarker expression at the tumor/stroma interface (margin), and quantification, scoring, and spatial localization of leukocytes in the tumor and stroma. Aggregation of all cellular and biomarker data generates tissue phenotypes that characterize the IO landscape of each tissue.

Methods

Serial sections of 20 NSCLC samples were IHC stained for PD-L1 and CD8 expression. Stained slides were scanned at 20x magnification and analyzed using Flagship Biosciences’ image analysis solutions. Image analysis algorithms which quantify biomarker expression, separate tumor and stromal compartments, detect tumor/stroma margins, and identify leukocytes in immunostained tissues were implemented in each tissue analyzed. Resulting image markups of cell detection and biomarker expression measured by image analysis were reviewed by an MD pathologist for acceptance. Tissues not meeting acceptance criteria were re-analyzed until acceptable to the reviewing pathologist.

Results

We demonstrate the synergistic value of layered image analysis algorithms which provide context to biomarker expression in NSCLC tissues. Samples were grouped in to immune desert, excluded, and inflamed phenotypes based on total leukocyte and CD8 expression patterns in the tumor, stroma, and margin. PD-L1 expression was scored based on percentages of tumor and stromal expression, as well as digital representations of common PD-L1 scoring paradigms. Additionally, samples were stratified by PD-L1 patterns of constitutive, induced, immune, or ignorant expression.

Conclusions

Digital image analysis of IHC stained tissues creates comprehensive tissue biomarker profiles that are useful in assessment of tumor and immune interactions in IO drug development and patient stratification. Complex algorithms that utilize AI and machine learning can be overseen by MD pathologists to create clinically acceptable digital analysis solutions.

References

1. Rimm DL, Han G, Taube JM, et al. A Prospective, Multi-institutional, Pathologist-Based Assessment of 4 Immunohistochemistry Assays for PD-L1 Expression in Non–Small Cell Lung Cancer. JAMA Oncol. 2017;3(8):1051–1058.

2. Hendry S, Salgado R, Gevaert T, et al. Assessing Tumor-infiltrating Lymphocytes in Solid Tumors: A Practical Review for Pathologists and Proposal for a Standardized Method From the International Immunooncology Biomarkers Working Group: Part 1: Assessing the Host Immune Response, TILs in Invasive Breast Carcinoma and Ductal Carcinoma In Situ, Metastatic Tumor Deposits and Areas for Further Research. Adv Anat Pathol. 2017;24(5):235–251.

3. Taube JM, Galon J, Sholl LM, et al. Implications of the tumor immune microenvironment for staging and therapeutics. Mod Pathol. 2018;31(2):214–234.

4. Silva MA, Ryall KA, Wilm C, Caldara J, Grote HJ, et al. PD-L1 immunostaining scoring for non-small cell lung cancer based on immunosurveillance parameters. PLOS ONE 2018; 13(6): e0196464

Correspondence: Sarah Church (schurch@nanostring.com)

Background

In colorectal cancer (CRC) there have been many recent advances in immune-related biomarkers that are both prognostic and predictive of response to immunotherapy. Microsatellite instability (MSI)/mismatch repair deficiency (dMMR) is present in ~15-20% of CRCs and corelates with increased immunogenic mutations that often augment lymphocyte infiltration into the tumor microenvironment (TME). Additionally, location of tumor infiltrating T-cells in two areas of the TME, the tumor center (CT) and invasive margin (IM) has also been shown to be prognostic and predictive of response to immunotherapy. Here we use multiplexed protein and RNA digital spatial profiling to elicit the immune landscape of MSI-MSS characterized CRC tumors.

Methods

Forty-eight CRC tumors were analyzed for gene expression (GX) using the NanoString® nCounter® PanCancer IO 360™ Research Use Only (RUO) Gene Expression Panel and assessed for 48 cell typing and biological signatures, including MMR loss/MSI predictor and the Tumor Inflammation Signature (TIS). A subset of 18 CRC tumors (6 MSI-TIS-hi, 6 MSS-TIS-hi, 6 MSS-TIS-lo) was selected for analysis with the RUO GeoMx™ Digital Spatial Profiler (DSP) using 40 antibodies, 84 or 1,600+ in situ probes. Selection of regions of interest (ROIs) in two locations, CT and IM were guided by staining with fluorescent markers (CD45, CD3, pan-CK, DNA). 300-600 μM diameter circle ROIs were selected, and in some cases segmented by pan-CK+/pan-CK-.

Results

Using whole tissue GX, we first confirmed MSI/dMMR characterization and TIS status of 48 CRC tumors using PanCancer IO 360 signatures. We selected 18 tumors within this cohort based on TIS status to further dissect the location-dependent immune contexture of the TME, with a particular emphasis on differentiating MSI-TIS-hi and MSS-TIS-hi CRCs. DSP confirmed loss of dMMR markers (MSH2/MLH1) and identified an increased amount of potentially suppressive macrophages (CD163+PD-L1+) in MSI-TIS-hi versus MSS-TIS-hi tumors. Segmentation of ROIs based on tumor versus stroma (pan-CK+/-) identified samples with high proportions of tumor-invading TILs. These samples were then further profiled using probes against 1600+ mRNA targets revealing distinct pathways related immune cell orientation within the TME.

Conclusions

Here we show the use of novel high-plex spatial profiling to profile location and pathways in the TME of MSI and MSS CRC tumors. These findings elicit unique biology related to the location and signaling of immune cells, which have the potential to unveil targets for therapeutic combinations.

Correspondence: Yi Zheng (YZheng@akoyabio.com)

Background

The tumor microenvironment hosts a myriad of cellular interactions that influence tumor biology and patient outcomes. Multiplex immunofluorescence (mIF) provides the ability to investigate a large number of these interactions in a single tissue section, and has been shown to outperform other testing modalities for predicting response to immunotherapies [1].

Multispectral imaging (MSI) improves the capabilities of mIF by providing the ability to spectrally unmix fluorescence signals. This increases the number of markers that can be probed in the same scan and allows for separation of true immunofluorescence signals from tissue autofluorescence background.

Here, we apply MSI to explore spatial interactions observed in lung cancer samples using an end-to-end translational workflow based on the PhenopticsTM platform. The workflow includes a pre-optimized 7-color staining panel kit along with a pre-configured analysis algorithm for cell phenotyping.

Using tissue microarrays (TMA), we demonstrate the heterogeneity of spatial interactions observed among different lung cancer samples and the improved sensitivity of detection afforded by unmixing multispectral scans.

Methods

A lung cancer TMA was created using the 3DHistech TMA Master II from five formalin-fixed paraffin-embedded lung cancer tissue blocks. The TMAs were stained using the Opal Polaris 7-Color PD1/PD-L1 Lung Cancer Panel Kit on the Leica BOND RXTM automated stainer using the associated preloaded Opal 7-Color Panel Kit protocol. Whole slide MOTiFTM multispectral scans were acquired on Vectra Polaris® with pre-defined acquisition parameters. Scans were unmixed and analyzed with inForm® software using a pre-configured algorithm tailored to the PD1/PD-L1 Lung Cancer Panel Kit. Spatial analyses and visualizations were performed using the phenoptr and phenoptrReports R-based packages and custom scripts.

Results

The pre-optimized Opal Polaris 7-Color PD1/PD-L1 Lung Cancer Panel Kit was able to visualize the panel targets (PD-L1, PD-1, CD8, CD68, FoxP3, and Cytokeratin) across the variety of lung cancer samples in the TMA. Cell phenotyping and spatial analyses revealed core-to-core variations in cell densities and proximities among different markers. Measurement of the dynamic range of PD-L1 expression across different cores also revealed the improved sensitivity in PD-L1 detection provided by unmixing.

Conclusions

The end-to-end Phenoptics staining, imaging, unmixing, and spatial analysis workflow described here provides a robust and sensitive platform for exploring the immune landscape within the tumor microenvironment.

Reference

1. Lu S, Stein JE, Rimm DL, et al. Comparison of Biomarker Modalities for Predicting Response to PD-1/PD-L1 Checkpoint Blockade: A Systematic Review and Meta-analysis. JAMA Oncol. Published online July 18, 2019. doi:10.1001/jamaoncol.2019.1549

Correspondence: Kurt Schalper (kurt.schalper@yale.edu)

Background

Tumor cells accumulate deleterious genomic alterations through sustained mutagenic exposure and defective DNA repair. Approximately 15% of human colorectal carcinomas (CRCs) display mismatch repair deficiency (MSI-H) associated with increased somatic mutations and sensitivity to immune checkpoint blockers. Advanced tumors can harbor additional DNA-repair alterations with functional/therapeutic implications. Increased double strand DNA breaks have been reported across solid tumors and can be detected by changes in Serine139-phosphorylated histone H2AX (γH2AX). We studied the immune composition of human CRCs with MSI-H and elevated DDR.

Methods

Using multiplexed quantitative immunofluorescence (QIF), we studied the level of major adaptive and innate immune markers in a retrospective collection of 265 stage I-IV CRCs from Yale represented in tissuemicroarrays. We used previously validated QIF panels including the markers DAPI, cytokeratin, γH2AX, CD3, CD4, CD8, CD20, PD-L1, CD15, myeloperoxidase (MPO), IL-8, Ki-67, granzyme-B (GZB), Beta-2 microglobulin (B2M), HLA-class I and HLA-class II. The MSI status was determined using clinical-grade immunohistochemistry detection of MLH1, MSH2, MSH6 and PMS2. We analyzed the association between localized measurement of markers and with major clinicopathologic variables/survival.

Results

From 252 evaluable cases, 12.1% were classified as MSI-H. Relative to MSS tumors, MSI-H CRCs showed significantly higher levels of PD-L1, lower CD20 and non-significant increases in CD3, CD4, CD8, T-cell Ki-67 and T-cell GZB. MSI-H cases displayed lower tumor-cell B2M and increased stromal HLA-class II expression. MSI-H tumors also showed significantly higher levels of IL-8 and MPO+ cells than MSS counterpart. The level of γH2AX was comparable between MSI-H and MSS malignancies. Cases with increased tumor-cell γH2AX (> cohort median) showed significantly higher levels of PD-L1 and all studied lymphocyte markers than cases with lower γH2AX. In addition, these tumors displayed significantly higher T-cell proliferation, mild increases in T-cell GZB and higher levels of HLA-class I/class II proteins. The levels of IL-8 and MPO+ cells were comparable across the γH2AX groups. The DNA repair and immune markers were variably associated with 5-year overall survival in the cohort.

Conclusions

DNA repair deficiency defines human CRCs with distinct innate and adaptive immune contexture. While mismatch repair deficiency is associated with mild/moderate intratumor T-cell responses and prominent myeloid cell features; elevated DDR display prominent adaptive immunity and unaltered myeloid-cell changes. Our data indicate that MSI-H and DDR phenotypes are independent features in human CRC and this could be used to design optimal therapeutic strategies.

Ethics Approval

All tissues were used after approval from the Yale Human Investigation committee protocol #9505008219 which approved the patient consent forms or waiver of consent.

Correspondence: Kurt Schalper (kurt.schalper@yale.edu)

Background

Tumor cells accumulate genomic alterations as a consequence of sustained mutagenic events and defective DNA repair mechanisms, collectively called DNA damage response (DDR). Targeting DDR pathways can induce synthetic lethality and prominent anti-tumor responses in neoplasms with DNA repair deficiency. In addition, increased DNA damage could favor anti-tumor immune responses by increasing the neo-antigenic load and T-cell recognition. Despite its therapeutic implications, the frequency and significance of DDR alterations in human non-small cell lung cancer (NSCLC) remains poorly understood.

Methods

Using irradiated cell line preparations and expression controls, we standardized a multiplexed quantitative immunofluorescence (mQIF) panel for simultaneous and localized measurement of DAPI (all cells), cytokeratin for tumor epithelial cells (AE1/AE3, DAKO), γH2AX to map active DNA damage/repair responses (JBW301, Millipore), CD3 for T-lymphocytes (Rabbit polyclonal, DAKO) and PD-L1 (E1L3N, CST) in formalin-fixed paraffin-embedded (FFPE) tissue samples. We used this panel to interrogate 4 retrospective NSCLC cohorts from Yale represented in tissue microarray format including immunotherapy-naïve cases (Cohort#1: n=297 and #2:n=175); lung adenocarcinomas tested for major oncogenic mutations (Cohort #3, n=139); and baseline NSCLC samples from patients treated with immune checkpoint blockers (Cohort #4, n=84). We analyzed the levels of the markers in different tumor tissue compartments and their association with major clinicopathological variables.

Results

Detectable nuclear tumor-cell γH2Ax was recognized in 37-58% of NSCLCs. Elevated tumor-cell γH2Ax expression was consistently associated with smoking history, increased intratumor CD3+ T-cells and PD-L1 protein expression across the cohorts. The level of γH2Ax was significantly lower in KRAS mutant lung adenocarcinomas than in EGFR mutant or EGFR/KRAS wild type tumors. No additional clinicopathologic associations were found. γH2Ax was not prognostic as single marker. However, elevated simultaneous expression of γH2Ax and CD3 was associated with longer 5-year overall survival in the immunotherapy-naïve cohorts. In patients treated with PD-1 axis blockers, elevated baseline γH2Ax/CD3 was associated with a clear trend toward longer survival but did not reach statistical significance.

Conclusions

Active DDR as measured by tumor-cell γH2Ax expression occurs in a high proportion of human NSCLCs and is associated with T-cell inflamed tumors. Despite their association with smoking, lung adenocarcinomas harboring activating mutations in KRAS display lower DDR markers than EGFR mutant or EGFR/KRAS wild type malignancies. Collectively, our results support the use of combination therapy targeting DDR and immunostimulatory therapies in a fraction of NSCLC.

Ethics Approval

All tissues were used after approval from the Yale Human Investigation committee protocol #9505008219 which approved the patient consent forms or waiver of consent.

Correspondence: Peter Miller (pmiller@akoyabio.com)

Background

Understanding cellular heterogeneity and spatial relationships between biomarkers within the tumor microenvironment (TME) is a key component to translational research in immuno-oncology. Multiplex immunofluorescence (mIF) on formalin-fixed, paraffin-embedded (FFPE) tissue is the multiparameter assay most frequently chosen across all current I/O clinical trials, as it allows for quantitative assessment of these relationships in situ. Running medium to large scale translational studies on FFPE tissue demands an assay that is reproducible, quantitative, easy-to-use, and standardized, yet still allows for flexibility when detecting differentially expressing biomarkers across samples. In this study, we demonstrate a fully developed, flexible, end-to-end workflow solution for tissue biomarker discovery by applying miF in lung cancer and melanoma. This newly developed Phenoptics™ solution provides an integrated MOTiF™ workflow including primary antibodies and image analysis algorithms enabling a more comprehensive and specific TME analysis with minimal user optimization.

Methods

FFPE samples from human lung cancer and melanoma were stained using Opal Polaris 7-Color PD1/PD-L1 Lung Cancer and Melanoma Panel Kits. Staining was performed on the Leica BOND RX™ automated stainer with the pre-loaded MOTiF protocol. Multispectral scans were acquired on Vectra Polaris® with pre-optimized acquisition parameters and analyzed with a pre-configured phenotyping algorithm in inForm®. Spatial analyses and visualizations were performed in R using phenoptr and phenoptrReports.

Results

This simplified end-to-end solution results in better quantification of cancer-immune interactions by providing:

• Well-optimized Opal Polaris 7-Color PD1/PD-L1 Lung Cancer and Melanoma Panel Kits. Along with the pre-loaded Leica BOND RX automation protocol, we provide a staining workflow with pre-defined primary antibody concentration, fixed staining order, and Opal™ dye-antibody pairs, leaving Opal concentrations as a flexible dial. Recommended image acquisition parameters on the Vectra Polaris® that significantly simplify visualization of multiple markers via multispectral isolation. Pre-configured image analysis algorithms that make quantitative analysis at a per-cell and per-slide level streamlined and standardized.

Conclusions

The 7-Color PD1/PD-L1 panel kits utilizing MOTiF whole slide scanning enable visualization of multiple biomarkers at the whole slide level, revealing distribution patterns and their spatial context across the entire tissue section. With these new assays, we have demonstrated an easy-to-use yet comprehensive end-to-end Phenoptics research workflow. We have radically simplified the Opal method and facilitated the development and optimization of translational multiplex fluorescent assays by providing pre-defined staining conditions while still giving researchers the flexibility to balance signals based on their tissue samples. Complementary pre-configured phenotyping provides researchers faster access to quantitative data across study samples.

Correspondence: Tad George (tgeorge@rarecyte.com)

Background

Pick-Seq is a novel workflow uniquely enabled by the RareCyte CyteFinder® Instrument that combines visualization of multiple protein markers with investigation of gene expression from selected micro-regions on tissue slides, providing spatial and contextual investigation of tumors and their microenvironment.

Methods

Frozen breast carcinoma and formalin-fixed, paraffin-embedded tonsil sections were stained by multi-parameter immunofluorescence (IF) for markers of T cells, B cells, and cytokeratin. Slides were imaged with the CyteFinder® Instrument and 40 μm micro-regions were retrieved with the integrated CytePicker® Retrieval Module. RNA was isolated and whole transcriptome amplified (SMART-seq v4), followed by Nextera XT library preparation, sequencing on Illumina MiSeq, and gene expression analysis. Differentially expressed genes were selected to create a Pick-Seq-informed IF staining panel to confirm RNA expression results. Cell compositions of each micro-region were deconvolved with CIBERSORT.

Results

Tonsil micro-regions from one T cell zone and two adjacent follicles were retrieved for RNA sequencing. Transcriptomic analysis confirmed increased expression of B cell markers in follicles and T cell markers in the T cell zone. CIBERSORT analysis revealed distinct cellular compositions between T cell zones and the B cell follicles. Principle component analysis of gene expression found that micro-regions retrieved from the two follicles clustered independently from each other, and from the T cell zone micro-regions. Differential expression analysis between the adjacent follicles revealed distinct patterns of CD21 expression, a marker which was not present in the original IF staining panel. Subsequent staining confirmed differential protein expression of CD21, indicating that only one follicle contained a germinal center. In breast carcinoma, ROI were identified for micro-region retrieval that included tumor cells, tumor cells with interspersed tumor infiltrating lymphocytes (TIL), or adjacent lymphoid aggregates. Micro-regions were picked and sequenced. Hierarchical clustering and differential expression analysis differentiated the three micro-region types and revealed tumor- and T cell-specific expression signatures. CIBERSORT demonstrated the presence of T cell-associated transcriptomic profiles in lymphoid aggregates and in TIL-containing micro-regions that were proportional to the number of T cells retrieved. Aligned RNA-seq reads were further analyzed via TraCeR to identify TCR α and β chain sequences from retrieved TILs.

Conclusions

These data establish the potential of combining multi-parameter IF microscopy with highly focused RNA sequencing as a powerful tool for investigation and biomarker discovery for immuno-oncology.

Correspondence: Renaud Burrer (rburrer@histalim.com)

Background

Lung cancer is the most common cause of cancer-related deaths worldwide with non-small cell lung cancer (NSCLC) representing the gross majority of the cases. The immune microenvironment of NSCLC is diverse with many players that can impact tumor development and clinical outcomes. In particular, myeloid-derived suppressor cells (MDSC) are important components of the immunosuppressive network that can hinder the activity of T cells, natural killer cells, and dendritic cells. MDSC in the blood may represent prognostic markers for NSCLC patients and for monitoring a patient’s response to immunotherapies. There is a gap in the relevance of MDSC within the tissue context due to limitations with conventional immunohistochemistry. Multiplex immunofluorescence offers a technical advantage by allowing the detection of co-expression and spatial organization of multiple targets within a preserved tissue architecture on a single slide.

Methods

We have developed the multiplex immunofluorescence Histoprofile-MDSC panel to identify monocytic MDSC (M-MDSC) and polymorphonuclear MDSC (PMN-MDSC) in situ. Five human NSCLC tissue samples were investigated by multiplex immunofluorescence and H&E staining. After multispectral acquisition, the MDSC populations were evaluated with the imaging software HALO. Paired peripheral blood was analyzed for circulating M-MDSC by flow cytometry.

Results

The development and verification of the multiplex panel are presented. The NSCLC subtype of the samples was determined by a pathologist from the H&E sections. Monocytes, neutrophils, M-MDSC, and PMN-MDSC were evaluated in the five tissue samples. The neutrophils, monocytes, and M-MDSC in the peripheral blood could be assessed by flow cytometry. A varying distribution of the cell populations in the lung tissue and the peripheral blood of the different NSCLC subtypes can be appreciated. The two approaches are compared.

Conclusions

We present an in-depth combined approach for MDSC investigation in lung tissue and the peripheral blood of NSCLC patients. The approaches presented here demonstrate the power of multiplex immunohistochemistry and flow cytometry in the identification and quantification of multiple immune cell populations with a limited quantity of patient sample and the potential application of this method in both preclinical and clinical studies.

Correspondence: Gary Kohanbash (gary.kohanbash2@chp.edu)

Background

Gliomas are the most common primary central nervous system tumor, with malignant gliomas causing significant morbidity and mortality. Thirty percent of a glioma’s cellular mass may be attributed to immunosuppressive and pro-tumoral tumor-associated myeloid cells (TAMCs), primarily myeloid-derived suppressor cells (MDSCs) and tumor-associated macrophages (TAMs) [1-4]. Multiple preclinical studies and clinical trials have attempted to target these cells; however, monitoring responses to these therapies remains a challenge. Quantifying TAMCs within gliomas using an antibody-based tracer for non-invasive positron emission tomography (immunoPET) may allow for better patient stratification, monitoring of treatment efficacy, and ultimately improve survival rates [5-9]. Integrin CD11b is a cellular marker expressed on the surface of TAMCs frequently used to identify macrophages and microglia. We therefore hypothesized that radiolabeled anti-CD11b antibody (Ab) could be used for immunoPET imaging of TAMCs in a preclinical orthotopic syngeneic glioma model.

Methods

The human/mouse cross-reactive anti-CD11b Ab (clone M1/70) was conjugated with p-NCS-Bz-DFO chelator and radiolabeled with 89Zr for PET imaging with specific activity of 2 μCi/μg. PET/CT imaging, with or without a blocking dose of anti-CD11b Ab, was performed in mice bearing established orthotopic syngeneic GL261 gliomas. Flow cytometry and histology in tissues collected from post-imaging biodistribution validated targeting of CD11b+ TAMCs.

Results

Standard uptake values (SUV) indicated significant 89Zr-anti-CD11b Ab uptake in the tumor ipsilateral right brain (SUVmean = 2.6 ± 0.24) compared to contralateral left brain (SUVmean = 0.6 ± 0.11). Blocking with 10-fold lower specific activity 89Zr-anti-CD11b Ab reduced the SUV in right brain with (SUVmean = 0.11 ± 0.06). Spleen and lymph nodes also showed high uptake, while bone and muscle showed low uptake. Biodistribution analysis confirmed these results. Additionally, no uptake was observed in the brain of non-tumor bearing mice that received 89Zr-ant- CD11b. Flow cytometry with QuantiBRITE Fluorescence Quantitation Kit demonstrated that the majority of tumor-infiltrating immune cells expressed CD11b at an average of 54,076 CD11b molecules per cell in GL261.

Conclusions

Imaging TAMCs with 89Zr-labeled anti-CD11b Ab may be feasibility for preclinical studies, patient stratification, and monitoring of immunotherapy.

References

1. Gabrusiewicz K, Rodriguez B, Wei J, et al. Glioblastoma-infiltrated innate immune cells resemble M0 macrophage phenotype. JCI insight. 2016; 1(2).

2. Kennedy BC, Showers CR, Anderson DE, et al. Tumor-associated macrophages in glioma: friend or foe? J oncol; 2013. 2013.

3. Kohanbash G, Okada H. Myeloid-derived suppressor cells (MDSCs) in gliomas and glioma-development. Immunolo invest. 2012; 41:658–679.

4. Lapa C, Linsenmann T, Lückerath K, et al. Tumor-associated macrophages in glioblastoma multiforme-a suitable target for somatostatin receptor-based imaging and therapy? PloS one. 2015; 10.

5. Kohanbash G, McKaveney K, Sakaki M, et al. GM-CSF Promotes the Immunosuppressive Activity of Glioma-Infiltrating Myeloid Cells through Interleukin-4 Receptor-α. Cancer Res. 2013; 73:6413–6423.

6. Raychaudhuri B, Rayman P, Huang P, et al. Myeloid derived suppressor cell infiltration of murine and human gliomas is associated with reduction of tumor infiltrating lymphocytes. J Neurooncol. 2015; 122:293–301.

7. Okada H, Kohanbash G, Zhu X, et al. Immunotherapeutic approaches for glioma. Crit rev immunol. 2009; 29:1–42.

8. Otvos B, Silver DJ, Mulkearns-Hubert EE, et al. Cancer Stem Cell-Secreted Macrophage Migration Inhibitory Factor Stimulates Myeloid Derived Suppressor Cell Function and Facilitates Glioblastoma Immune Evasion. Stem Cells. 2016; 34:2026–2039.

9. Meyer C, Cagnon L, Costa-Nunes CM, et al. Frequencies of circulating MDSC correlate with clinical outcome of melanoma patients treated with ipilimumab. Cancer Immunol Immunother. 2014; 63:247–257.

Ethics Approval

The study was approved by University of Pittsburgh's Institutional Animal Care and Use Committee (IACUC).

Correspondence: Brooke Helfer (brooke@celsense.com)

Background

Cancer immunotherapies have made a great progress and hold much promise in the treatment of cancer. Specifically, in the case of B-cell malignancies (such as Acute Lymphoblastic Leukemia, or ALL), CAR (chimeric antigen receptor) and TCR (T-cell receptor) therapies have demonstrated encouraging clinical results. As we begin to target solid tumors with TCR and CAR T-cells, the hurdle of being able to select a suitable target and achieve successful cellular delivery/homing to the site of disease remains. With this in mind, being able to visualize a rapidly dividing cellular population is another obstacle to consider.

Methods

Here we demonstrate the application of two clinically applicable perfluorocarbon (PFC) tracers, one commercially available and a next-generation magnetic resonance imaging (MRI) probe called FETRIS. Both of these agents enable the migration and persistence of cellular therapies to be noninvasively imaged by 19F MRI, while the FETRIS reagent adds additional detection sensitivity.

Results

Using a general T-cell expansion protocol, we show that adding a cellular label does not alter the viability or release characteristics of T cells. By pairing the PFC signal with conventional proton MRI from the same imaging session, the images are able to be overlaid, allowing cells to be traced to their anatomical location. With nominal exogenous fluorine naturally present in tissue, labeled cells appear with little background.

Conclusions

Images of both reagents show the detection and sensitivity of the method and how they can be applied to monitor the distribution of cells over time.

Correspondence: Ana Hidalgo Sastre (ahidalgo@definiens.com)

Background

Given the heterogeneity of tumors and the variety of potential biomarkers in immune oncology, there is a need for quantitative standardized assays to reliably assess the immune status of a patient’s tumor to be able to extract the true biological information across cohorts. Here, two different tissue-based approaches have been compared: multiplex immunofluorescence (mIF) and multiplex chromogenic immunohistochemistry (mIHC). Independently of the technique used, assay reproducibility and standardized quantification of staining intensity are a prerequisite for obtaining consistent results. Using a cohort of non-small cell lung carcinoma (NSCLC) patients, we identified patterns of immune cell infiltration that were comparable, independent of the assay applied.

Methods

Formalin-fixed paraffin-embedded (FFPE) true consecutive slides from 7 NSCLC resections were stained with a multiplex chromogenic panel (including CD3, PD-L1, CD68, CD8, PD-1) at Mosaic Laboratories [1] and with the UltiMapper kits (I/O PD-L1 and I/O PD-1) from Ultivue. mIHC scans were acquired with an Aperio AT Turbo scanner (Leica), while mIF scans were acquired with a Zeiss Axio Scan.Z1 scanner (Zeiss) both as whole slide images. mIHC and mIF images were co-registered, and Definiens custom algorithms for digital image analysis were applied [2,3].

Results

Densities of immune cell populations and their locations in different compartments (invasive margin vs tumor center and tumor epithelium vs tumor stroma) were measured (Figure 1). For instance, CD3 cell density had a Pearson correlation of 0,91 and a Spearman correlation of 0,89 between both assays (mIHC vs mIF). Differentiation between tumor epithelium and tumor stroma was based on a histology-driven deep learning approach for mIHC and on pan Cytokeratin for mIF (Figure 1).

Conclusions

By applying mIHC and mIF in true consecutive tissue slides we retrieved the information of tumor immune cell infiltrates that was consistent across the different assays and distinguished it from information that is specific to either of the assays. We believe that being able to relate across staining techniques could help pathologists and research centers draw conclusions across cohorts that were stained with the same markers but with different assays.

References

1. Lisa M. Dauffenbach, Christopher A. Kerfoot, et al. Characterization of inflammatory cell patterns and densities using multiplex immunohistochemistry immuno-oncology assays [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl): Abstract nr B069.

2. Lorenz Rognoni, PhD; Vinay Pawar, PhD; Tze Heng Tanet, et al. Automated quantification of whole-slide multispectral immunofluorescence images to identify spatial expression patterns in the lung cancer microenvironment. SITC Annual Meeting; 2018 Nov 7-11; Washington, DC. Poster nr P442.

3. Brieu, Nicolas & Meier, Armin & Kapil, Ansh & Schönmeyer, Ralf & Gavriel, Christos & Caie, Peter & Schmidt, Günter. (2019). Domain Adaptation-based Augmentation for Weakly Supervised Nuclei Detection.

Ethics Approval

Ethical approval was granted by the Liverpool Research Ethics Committee, reference number 97/141.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P39).

Consecutive slides from NSCLC resection

Correspondence: Daniel Buchanan (daniel.buchanan@unimelb.edu.au)

Background

Deficiency in the mismatch repair (dMMR) can result from inherited mechanisms (Lynch Syndrome (LS)) or from somatic inactivation caused by hypermethylation of the MLH1 gene promoter (MLH1 methylated). A third subtype of dMMR colorectal cancer (CRC) and endometrial cancer (EC) have neither LS nor MLH1 promoter methylation and are referred to as suspected Lynch syndrome (SLS). There remains a knowledge gap as to whether the tumour microenvironment (TiME) is different between LS, MLH1-methylated and SLS dMMR CRC and EC. The aim of this study was to characterise and identify immune patterns within the TiME that may enhance the current clinical triaging of LS, SLS and MLH1-methylated subtypes of dMMR CRC and EC.

Methods

Ten FFPE samples from seven individuals were studied: CRC (N=5; 1xLS, 2xMLH1 methylated, 1xSLS and 1x proficient MMR (pMMR)) and EC (N=2; 1xLS and 1xpMMR) and where available adjacent normal tissue (N=5: 3xcolon and 2xendometrium) were characterized using the Ultivue UltiMapper I/O portfolio (InSituPlex) on a Leica Bond autostainer and digitally acquired using the Zeiss Axio Scan Z1. We evaluated CD3, CD8, CD11c, CD20, CD45RO, CD68, CD163, Granzyme B, Ki-67, MHC II, PD-1, PD-L1, and pan-cytokeratin. Tissue phenomic approaches were developed to spatially characterize, quantify immune cell patterns and visualize heterogeneity within the TiME (Figure 1).

Results

InSituPlex technology enabled the visualization of the heterogenous infiltration and co-localization patterns across LS dMMR CRC (Figures 2&3). Tissue phenomic approaches demonstrated the following; within the SLS category the colon cancer had higher mean areas of intraepithelial (IE) PD-L1 (6% vs. 2%), CD8 (18% vs. 8%) and CD68 (28% vs. 12%) compared to the pMMR EC. The MLH1 methylated tumour with a high tumour mutation burden (33.84 mutations/MB) had a higher mean area of IE PD-L1 (8% vs. 2%) and CD8 (30% vs. 5%) while the tumour with low TMB had a higher mean area of IE CD68 (15% vs. 8%). Within LS, the EC had a higher mean area of IE PD-L1 compared to the CRC (14% vs. 0%), in contrast the CRC had a higher IE CD8 area (27% vs. 10%) (Figures 4&5).

Conclusions

This study evaluated the immune contexture within inherited and sporadic subtypes of dMMR CRCs and ECs, highlighting differing immune infiltration patterns and phenomic densities. Integration of multiplex technologies and Tissue Phenomics can enhance the understanding of the dMMR TiME and with potential utility in clinical triaging and to inform immune-oncology clinical trials.

Acknowledgements

We thank the investigators and participants of the ACCFR and ANGELS studies.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P40).

Phenotypes

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P40).

TiME regions of immune infiltration

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P40).

TiME regions of immune infiltration

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4 (abstract P40).

TiME regions of immune infiltration

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5 (abstract P40).

TiME regions of immune infiltration

Correspondence: Jacob Estes (estesja@ohsu.edu); Garry Nolan (gnolan@stanford.edu)

Background

Multiplexed Ion Beam Imaging (MIBI) is a novel imaging modality capable of resolving >40 parameters simultaneously in biological samples. Here, we developed viralMIBI, a highly sensitive method capable of detecting down to single copies of nucleic acids, in addition to protein epitopes. ViralMIBI enables the functional dissection of the immune landscape in viral driven diseases, such as that of tumor viruses (HBV, EBV, LCV) and others (HIV, SIV, Zika, Ebola). The combination of viralMIBI and cutting-edge cell neighborhood analytical methods will be paramount to better understand the immunological host-pathogen interactions for viral diseases, revealing insights into virus-induced immunodeficiency as well as virus-driven cancers.

Methods

To allow for the sensitive detection of nucleic acids, we took advantage of a customized branched DNA amplification method that can be easily adapted to a variety of multiplexed imaging platforms. Formalin-Fixed and Paraffin-Embedded (FFPE) tissue samples from Rhesus macaque animal models for a number of viral diseases were processed for viralMIBI nucleic acid and protein marker detection. Imaging was performed with the MIBIscope, a secondary ion mass spectrometry based device.

Results

We have established a robust method for highly multiplexed nucleic acid and protein epitope detection in FFPE tissue samples. As a proof of concept, we were able to detect down to single integrated copies of SIV. The establishment and validation of a Rhesus macaque specific antibody panel allowed for the in-depth characterization of cellular identities at the single-cell level, while maintaining their tissue geopositions.

Conclusions

ViralMIBI enables the MIBI to achieve highly sensitive nucleic acid detection, in addition to its multiplexed protein capabilities. Here, we leveage this method for the detection of various viral pathogens. ViralMIBI is also applicable to other targets, such as genomic amplifications frequently seen in cancers, or gene expression studies. The ability to image >40 parameters in tissue samples will vital for a better understanding of immune regulation of diseases, such as the establishment of viral related cancers as well as latent tissue reservoirs of pathogens. These discoveries can then be translated to better immunotherapy treatments against viral driven diseases.

Acknowledgements

We thank Matt Newgren for tireless technical support on the MIBI instrument, Rachel Finck, Xiao-Jun Ma and Bingqing Zhang for helpful discussions. S.J was supported by a Stanford Dean’s Fellowship and the Leukemia & Lymphoma Society Career Development Program. X.R.-C. was supported by a long-term EMBO fellowship. This work was supported by grants from the FDA, NIH, Parker Institute for Cancer Immunotherapy, the Bill and Melinda Gates Foundation, as well as the Rachford and Carlota A. Harris Endowed Professorship to G.P.N.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P41).

Validation of viralMIBI: Detection of SIV in infected tissue

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P41).

Detection of single integration events of SIV with viralMIBI

Correspondence: Anna Juncker-Jensen (anna.juncker-jensen@neogenomics.com)

Background

Ovarian granulosa cell tumors (GCTs) are rare tumor accounting for 2-5% of all ovarian cancers. The main current treatment for GCT is surgery, however a subset require chemotherapy for residual and recurrent disease. GCT malignancies are often low-grade, however a clinical characteristic of these tumors is a tendency for late recurrence which is the most critical factor for GCT death. As the onset of recurrence is unpredictable, future research should focus on identifying both biomarkers for prognosis prediction, as well as targets that could help guide clinical trials in the development of targeted therapies for this rare indication. As GCTs are rare tumors making tissue availability very limited, we used a dual multiplexing approach in order to maximize the data output from a total of 14 FFPE tumor samples (6 primary tumors, and 8 recurrent tumors).

Methods

For protein multiplexing we have used MultiOmyx™, an immunofluorescence (IF) multiplexing assay utilizing a pair of directly conjugated Cyanine dye-labeled (Cy3, Cy5) antibodies per round of staining (Figure 1). Each round of staining is imaged and followed by dye inactivation enabling repeated rounds of staining and deactivation, while deep learning based cell classification algorithms identify positive cells for each biomarker. We generated a 15-marker panel consisting of CD3, CD4, CD8, FoxP3, CD68, CD163, HLA-DR, CD34, CTLA-4, PD-1, PD-L1, Ki67, vimentin, S100, and Pan Cytokeratin. For the gene expression analysis RNA was extracted from the adjacent 10 μm section and then analyzed using the Nanostring nCounter assay, specifically the 770 gene PanCancer Immune Panel. Hybridization, purification and immobilization and counts were based on manufacturer’s protocol.

Results

On protein level we confirmed previous findings that ovarian GCTs are so-called “cold” tumors, with a very low density of T cell infiltration. When we analyzed the presence of macrophages in the tumor microenvironment however, we found a 113% increase in TAM density in recurrent tumors compared to primary tumors. When searching for markers differentially expressed between primary and recurrent tumors we detected 4 genes in our PanCancer immune panel that were either significantly down-regulated (CCND3 or TOLLIP), or up-regulated (MAP3K and TNFSF4) in recurrent tumors. TNFSF4 encodes the protein OX40L, and interestingly a high expression of its receptor OX40 has previously been shown to be indicative for response to chemotherapy in recurrent ovarian cancer [1].

Conclusions

We have used a dual multiplexing approach on both gene and protein level in order to immunoprofile the tumor microenvironment of ovarian rare granulosa tumors.

Reference

1. Ramser M, Eichelberger S, Däster S, Weixler B, Kraljević M, Mechera R, Tampakis A, Delko T, Güth U, Stadlmann S, Terracciano L, Droeser RA, Singer G. High OX40 expression in recurrent ovarian carcinoma is indicative for response to repeated chemotherapy. BMC Cancer. 2018;18:425-433.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P42).

Immunofluorescent overlay image of GCT recurrent tumor

Correspondence: Won-Mean Lee (wmlee@akoyabio.com); Gabriel Mercado (gmercado@akoyabio.com)

Background

Analyzing populations at the single cell level has become increasingly important in the study of cancer and autoimmune disorders due to high levels of population heterogeneity and rare cell phenotypes that can drive disease pathogenesis and progression. Until recently, characterizing protein markers on single cells was limited to a handful of markers due to the technical and logistical challenges of flow cytometry platforms. New advancements in single cell analysis technologies have enabled researchers to study more than 30 parameters per cell. But these platforms are expensive and require significant panel design, thereby limiting access and usability. Here we demonstrate the use of the recently launched CODEX System to generate an in-depth immune profile of human PBMC samples.

Methods

The CODEX® System is an affordable, benchtop instrument that integrates with existing fluorescence microscopes and enables highly multiplexed imaging of over 40 markers in fresh frozen and FFPE tissue samples. The CODEX technology uses a DNA-based barcode library to label antibodies and iterative cycles of adding and removing cognate dye-labeled oligonucleotides to reveal the staining of three markers per cycle. Data acquisition is fully automated by the CODEX instrument. We tested a custom panel of more than 25 markers on the PBMC samples and acquired the images using a Keyence benchtop microscope.

Results

The CODEX system was used to generate highly multiplexed immune profiles of human PBMC samples using an optimized custom panel of CODEX antibodies. The images were processed using the CODEX Software Suite and cell phenotypes were clustered and annotated using the Multiplexed Analysis Viewer (MAV). Antibody specificity and panel performance were evaluated by assessing co-expression and mutually exclusive expression of relevant immune markers with the CODEX analysis pipeline.

Conclusions

Simultaneous analysis of tens of markers in blood or plasma samples can have several applications in the discovery of cellular biomarkers, immune monitoring and drug discovery and development. This preliminary study shows the compatibility of the CODEX system with cell suspensions for highly multiplexed, single-cell analysis and offers a more cost-effective method for immune profiling of blood samples.

Correspondence: Jin Miao Chen (Chen_Jinmiao@immunol.a-star.edu.sg)

Background

Hepatocellular carcinoma (HCC) is a lethal cancer, being the fourth leading cause of cancer-associated mortality worldwide due to its low five-year survival and high reoccurrence rates [1], and identifying indicators of prognosis is key in developing novel treatments and improving survival of HCC patients. With the advent of digital pathology, the immune-architecture of solid tumours has become a central interest of cancer research and has been studied for the development of predictive and diagnostic applications. Here, we have assessed if intercellular Euclidean distances in the tumour immune microenvironment can possibly be used to predict patient prognosis.

Methods

In this study, biopsies were taken from 110 HCC patients who underwent surgical resection. The solar-IHC pipeline involves arranging the liver biopsies into tissue arrays and subsequently studying them using automated multiplex immunohistochemistry/immunofluorescence (mIHC/IF) protocol developed in Singapore General Hospital, with biomarkers Ecadherin, CD3, CD8, CD103 and PD1 [2], followed by image analysis software inForm version 2.4.2.

Results

Ecadherin was adopted as the tumour cell marker, and 26 immune cell phenotypes are defined by variable levels of immune markers CD8, CD103, and PD-1 (as shown in Table 1) Dimensionality reduction and unsupervised clustering of the distances between tumour and immune cell phenotypes showed distinct clusters of patients with significant differences in clinical outcomes. Long cell-cell distances between immune cell phenotypes and tumor cells was associated with an improved overall survival (p-value = 0.02) and disease-free survival (p-value = 0.01), while the opposite was true for short cell-cell distances between immune cell phenotypes and tumor cells. This was observed in the analysis of all CD8+, CD8-, CD103+, CD103-, PD1+ and PD1- cells, possibly due to the suppressive immunomodulatory effect Ecadherin+ tumour cells has on neighbouring immune cells [3]. Furthermore, machine learning enabled the prediction of clusters with considerable accuracy, with a K-fold cross-validation of 91%. This indicates a strong association of cell-cell distances with patient survival, and a robust reproducibility of distance pattern-based predictors.

Conclusions

In this study, our data suggests that the analysis of intercellular distances has the potential to be used as a prognostic indicator in HCC. Coupled with next generation machine learning techniques, this novel approach to cell-to-cell distance analysis has the potential to be an easily implementable algorithm to predict patient prognosis in HCC. This bioinformatics approach can also be utilized in the analysis of other biomarkers and cancer types, and this brings exciting prospects for the future of cancer research.

References

1. McGlynn KA, Petrick JL, London WT. Global epidemiology of hepatocellular carcinoma: an emphasis on demographic and regional variability. Clinics in liver disease. 2015;19(2):223-38.

2. Lim JCT, Yeong JPS, Lim CJ, Ong CCH, Wong SC, Chew VSP, Ahmed SS, Tan PH, Iqbal J: An automated staining protocol for seven-colour immunofluorescence of human tissue sections for diagnostic and prognostic use. Pathology. 2018 Apr;50(3):333-341

3. Nagl S, Haas M, Lahmer G, Büttner-Herold M, Grabenbauer GG, Fietkau R, et al. Cell-to-cell distances between tumor-infiltrating inflammatory cells have the potential to distinguish functionally active from suppressed inflammatory cells. Oncoimmunology. 2016;5(5):e1127494.

Ethics Approval

This study was approved by the Institutional Review Board (IRB), approval number 2014/590/B.

Correspondence: Kate Lillard (kate@indicalab.com)

Background

The immune cell milieu that comprises the tumor microenvironment (TME) is highly heterogeneous and complex. Depending on biological interactions and functional state, immune cell populations can either promote or suppress tumor progression. CD8+ T cells, for example, are the primary mediators of anti-tumor immunity; however, they are often ineffective either because they are unable to infiltrate the tumor or because they become functionally exhausted [1,2], Pathologically activated myeloid-derived suppressor cells (MDSCs) which infiltrate the tumor are also associated with tumor progression [3,4]. Multiple biomarkers are required to accurately identify these individual immune cell types and their functional states. In this work, we employ advanced multiplexing techniques to observe biologically and functionally distinct T cell and MDSC populations and to quantify their density and distribution within the TME of several tumor types.

Methods

UltiMapper assays were used to perform multiplex immunofluorescence on multiple tumor FFPE samples (lung, colorectal, breast). Three multilpex panels were run in this study: UltiMapper PD-1 [CD3, CD45RO, PD-1, CK/Sox1], UltiMapper T-reg[CD4, CD8, FoxP3, CK, Sox10], and UltiMapper MDSC[CD11b, CD14, CD15, HLA-DR].FFPE slides were stained using the BOND RX autostainer from Leica Biosystems and scanned on the CyteFinder® II HT Instrument from RareCyte, Inc. This instrument performs high-speed, whole-slide scanning in the 5 channels used in the UltiMapper Kits and outputs an open source, stitched, pyramidal TIFF. Image analysis was conducted using HALO 3.0 software to perform cell phenotyping, proximity analysis, image registration, and density mapping.

Results

Cell counts for relevant phenotypes were obtained for each panel to identify exhausted T cells, T-regs, cytotoxic T cells, M-MDSCs, and PMN-MDSCs. Spatial analysis was employed to map the degree of T-cell infiltration and exhaustion correlating to T-reg and MDSC expression in the tumor microenvironment.

Conclusions

Here we present a workflow for tackling the complexity of the tumor immune microenvironment by leveraging high-quality multiplex panels, high-speed whole-slide imaging, and quantitative spatial analysis. UltiMapper assays used in this study (PD-1, T-reg, and MDSC) were able to identify single-cell phenotypes through co-localization and negative selection of markers. Using HALO image analysis, cell populations were enumerated and quantified to measure the level of T-cell exhaustion caused by T-cell regulation and myeloid-derived immune cell suppression.

References

1. Fridman WH, Zitvogel L, Sautes-Fridman C, Kroemer G. The immune contexture in cancer prognosis and treatment. Nat. Rev. Clin. Oncol. 2017; 14:717–734.

2. Thommen DS, Schumacher TN. T Cell Dysfunction in Cancer. Cancer Cell. 2018; 33:547-562.

3. Gabrilovich DI, Ostrand-Rosenberg S, Bronte V. Coordinated regulation of myeloid cells by tumours. Nat. Rev. Immunol. 2012; 12:253–268.

4. Kumar V, Patel S, Tcyganov E, Gabrilovich DI. The Nature of Myeloid-Derived Suppressor Cells in the Tumor Microenvironment. Trends Immunol. 2016; 37:208–220.

Correspondence: Mael Manesse (mael.manesse@ultivue.com)

Background

Innovative and efficient translational research tools enabling a better understanding of the tumor and its microenvironment are a keystone of the development of digital pathology. Current immunohistochemistry (IHC) methods limit the depth of information from a single tissue sample to a single target in the case of chromogenic staining, or to sample morphology and general cell identification in the case of hematoxylin and eosin staining (H&E). True phenotyping requires the use of a single section, as serial sections may not contain the same cells, especially small immune cells such as T-cells. Multiplex immunofluorescence (mIF) methods have been established to provide insights into a wide number of markers of interest and their spatial context in a single sample. Here, we demonstrate a new research approach combining multiplexed detection of protein markers with standard H&E pathology review in tumor samples, in a streamlined, single-day sample-to-answer workflow.

Methods

InSituPlex technology was used to perform multiplex immunofluorescence staining of formalin-fixed, paraffin-embedded (FFPE) samples from human tonsil and primary tumor biopsies on the Leica Biosystems BOND RX autostainer. The tissues were then imaged in five distinct fluorescent channels (DAPI, FITC, TRITC, Cy5, Cy7) before being stained using standard H&E protocols and imaged again. Fluorescent and brightfield whole-slide images were acquired on a ZEISS AxioScan.Z1 slide scanner. Images of the same tissue section were co-registered and fused into a single image for analysis using Indica Labs HALO software.

Results

The InSituPlex technology enables deep phenotyping of immune cells through colocalization and co-expression of multiple protein markers in tumor samples. Phenotypic information was then overlaid with the H&E image of the same section to facilitate identification and immuno-profiling of specific cells in the tumor and its environment. The fused images were also analyzed to provide cell counts, distance mapping, and expression levels of each of the markers.

Conclusions

In this work, we present a new modality for pathology research with a convenient workflow that enables fast tissue review and deep immuno-profiling and phenotyping of the tumor via fusion of H&E and mIHC staining of the same tissue section.

Correspondence: James Mansfield (jim@jmansfield.com)

Background

The rapid growth of research into immuno-oncology research has fueled a need to track be able to determine the location of a variety of immune cells systemically and in solid tumors. However, existing methods for cell tracking that have generally been insufficient. Magnetic Particle Imaging (MPI) is a novel tomographic molecular imaging technique that can be used to non-invasively track iron-oxide tagged immune cells in 3D in vivo, with contrast similar to nuclear medicine but without the complex workflow, safety, and half-life limitations. In this study, we compared the behavior of monocytes loaded with nanoparticles in vitro and nanoparticles injected intravenously and subsequently taken up by phagocytic cells (in situ loading) and imaged using MPI the differences in biodistribution and migration of monocytes in in naïve and tumor-bearing and naïve (control) mice.

Methods

Twenty mice were implanted with 300,000 4T1 tumour cells in the 4th mammary fat pad. CD11b+ mouse monocytes were harvested using EasySep® Mouse CD11b positive selection kit II (StemCell Technologies). The isolated monocytes were prelabeled with Vivotrax® (100 μg/mL). On day 7 post-implantation, either 5 million prelabeled cells or free Vivotrax (6 mg/kg) were intravenously injected into normal or tumor-bearing mice for in vitro or in situ targeting experiments (N=5 mice for all four groups). 3D MPI images using a MOMENTUM MPI system (Magnetic Insight) were acquired 1, 4, 7 and 10 days after injection. MicroCT images (CT120, Trifoil Imaging) were acquired and co-registered using VivoQuant (Invicro). Tumors, liver, spleen and draining lymph nodes were then harvested, imaged, fixed, and stained with Prussian blue and analyzed for iron contents.

Results

Tumor-bearing mice showed a significant accumulation of nanoparticles for both the in situ and in vitro targeting methods, although the time and amount of accumulation was different. For both experiments, nanoparticles were predominately detected in the expanding margins of the tumor. For the in vitro labeled monocytes, accumulation was rapid, with the maximum accumulation being at 24 hours post-injection, while for the in situ labeled cells, accumulation was slower.

Conclusions

By combining the sensitivity, specificity as well as accurate quantitation potentials of MPI, information can be obtained on labeled monocytes and their biodistribution in tumour models. Other cells can also be labeled (dendritic cells, MDSCs, NKs, and T cells) and this information can be utilized to better understand the factors influencing immune cell migration in and around tumors.

Correspondence: James Mansfield (jim@jmansfield.com)

Background

Cancer immunotherapy is now the “fifth pillar” of cancer therapeutics [1]. Although hugely successful, there are limitations. In many studies, less than half the patients are responsive to therapy. One hypothesis is that refractory tumours are immunologically ‘cold’ – i.e., there are insufficient immune cells in the tumour for the therapy to be efficacious [2]. Thus, methods to stimulate an immunogenic response in solid tumours to improve immunotherapy efficacy are desirable.

Hyperthermia is known to induce a local immunogenic response, making it a potential adjunct to radiation and immune therapies. One hyperthermia method is Magnetic Fluid Hyperthermia (MFH), which is based on electromagnetic heating of magnetic nanoparticles (MNPs) [3,4]. , However, poor control of heating localization and magnitude have prevented MFH’s widespread clinical adoption.

Magnetic Particle Imaging (MPI) is an emerging tracer imaging technique that directly detects and quantitates superparamagnetic iron-oxide nanoparticles with exceptional contrast and high sensitivity at millimeter-scale resolutions [5]. MPI’s contrast is similar to nuclear medicine, but without the complex workflow, safety, and half-life limitations of a radioactive tracer.

Methods

Here we describe how MPI and MFH can be combined to produce spatially localized heating and accurate control of heating magnitude. Spatial localization is achieved using a unique mechanism, magnetic localization. Localization is effected by using a strong magnetic field gradient to produce a “field-free region” (FFR) where nanoparticles are heated, while nanoparticles outside the FFR are quenched and do not heat. The use of an FFR thus enables millimeter-scale control over which MNPs are heated [6-8].

Results

MPI is first used to quantitate the MNPs prior to heating, to enable treatment planning and prediction of the heating dose. MFH can then be induced in target regions of interest located anywhere in the body while avoiding regions containing MNPs that should not be heated, such as the liver or lymph nodes.

Conclusions

Combined MPI-MFH enables new treatment workflows that exploit spatially localized MFH and accurate control of heating magnitude. These workflows may resemble image-guided radiation therapy or image-guided high-intensity focused ultrasound. Combined MPI-MFH also prevents damage to nearby healthy tissue while enabling new applications such as targeted immunogenic stimulation. MPI-MFH also enables new heat-actuation applications involving systemic injection of MNPs followed by local targeting such as local release of a drug [9] (break thermally labile bonds/nanocarriers) without requiring active chemical targeting. While currently only available for small animal use, its underlying physics does not prevent its translation to human sizes

References

1. Zaidi, N; Jaffee, E. J Clin Invest (2018).

2. Sharma, P. et al. Curr Opinion Immunol (2016).

3. Jordan, A. et al. Journal of Magnetism and Magnetic materials, (1999).

4. Latorre, M. Puerto Rico health sciences journal, 28(3) (2009).

5. Gleich, B. & Weizenecker, J. Nature 435,1214–1217 (2005).

6. Murase, K. Physica Medica, 29(6), 624-630 (2013).

7. Hensley, D. Physics in Medicine & Biology, 62(9), 3483 (2017).

8. Tay, Z. ACS nano, 12(4), 3699-3713 (2018).

9. Liu, J. F., Small, 14(44), 1802563 (2018).

Correspondence: Shawn O’Neil (llospo@gmail.com)

Background

The inflammatory bowel diseases ulcerative colitis (UC) and Crohn’s disease (CD) are chronic, relapsing inflammatory disorders of the gastrointestinal tract (GIT) that affect millions of individuals worldwide [1]. The pathogenesis of these disorders is thought to involve dysregulation of mucosal immune homeostasis in the GIT in response to environmental factors in genetically susceptible individuals [2]. Regulatory T cells (Treg) are CD4+ T lymphocytes that play a central role in peripheral immune tolerance, actively inhibiting inflammation upon antigenic stimulation. There are two major populations of Treg: conventional Treg and TR1 cells [3]. Conventional Treg arise from the thymus (tTreg) or can be induced in the periphery (pTreg); both tTreg and pTreg constitutively express FoxP3 and CD25 (IL-2Rα). An imbalance in conventional Treg and effector T cells in the GIT microenvironment is thought to play a part in the pathogenesis of inflammatory bowel disease (IBD) [4]. Thus, we sought to quantify conventional Treg and CTL populations in GIT tissue sections from IBD patients versus normal individuals by multiplex immunofluorescence.

Methods

Conventional Treg are typically defined as lymphocytes with a CD3+/CD4+/CD25+/FoxP3+ immuno-phenotype. This complex antigenic signature has made it difficult to definitively label Treg populations in tissue sections by immunohistochemistry. We combined a 5-plex (CD3, CD4, CD8α, CD25, FoxP3) immunofluorescence assay using Ultivue InSituPlex® multiplex technology with image analysis using Indica Labs HaloTM software to identify, localize and enumerate: 1) total CD3+ T cells, 2) CD8α+ cytotoxic T lymphocytes (CTL) and 3) CD3+/CD4+/CD25+/FoxP3+ conventional Treg in formalin-fixed paraffin-embedded (FFPE) sections of GIT from patients with UC and CD versus controls. Using this approach, we were able to definitively identify and enumerate these immune cell populations on single FFPE tissue sections from each specimen.

Results

We found greater Treg and CTL cell densities (cells/mm2) in colon from CD and UC patients versus controls and higher densities of Treg and lower densities of CTL in small intestine from patients with CD versus controls.

Conclusions

The Ultivue InSituPlex© assay was capable of discretely localizing conventional Tregs and CTL in human tissues. This multiplex platform could be used to simultaneously localize Tregs and CTL in FFPE surgical resections and biopsies of neoplastic tissue as well.

References

1. Ng SC, Shi HY, Hamidi N, Underwood FE, Tang W, Benchimol EI, Panaccione R, Ghosh S, Wu JCY, Chan FKL, Sung JJY, Kaplan GG. Worldwide incidence and prevalence of inflammatory bowel disease in the 21st century: a systematic review of population-based studies. Lancet. 2018; 390:2769-78.

2. Corridoni D, Arseneau KO, Cominelli F. Inflammatory bowel disease. Immunol Lett. 2014; 161:231-5.

3. van Herk EH, Te Velde AA. Treg subsets in inflammatory bowel disease and colorectal carcinoma: Characteristics, role, and therapeutic targets. J Gastroenterol Hepatol. 2016; 31:1393-404.

4. Yamada A, Arakaki R, Saito M, Tsunematsu T, Kudo Y, Ishimaru N. Role of regulatory T cell in the pathogenesis of inflammatory bowel disease. World J Gastroenterol. 2016; 22:2195-205.

Ethics Approval

Human tissues were obtained from the National Disease Research Interchange (NDRI) with support from NIH grant U42OD11158. Tissues were collected for research purposes under IRB-approved informed consent and collection procedures and provided to Pfizer in accordance with applicable government regulations and guidelines.

Correspondence: Benjamin Pelz (benjamin.pelz@lunaphore.com)

Background

The tumor microenvironment plays a vital role in cancer development. Multiplex immunostainings allow studying the interaction of different cell types in the tumor microenvironment using a single tissue slide. Though several techniques are available to perform high-plex stainings, they require intensive manual handling, are highly time consuming or not compatible with tissue sections on standard microscope slides. Here we present a fully automated microscope integrated method for rapid high-plex sequential fluorescent immunostaining and imaging of tissue sections.

Methods

Formalin-fixed, paraffin-embedded tissue sections underwent manual dewaxing and antigen retrieval step. All subsequent steps of staining, antibody elution and imaging were automated on the microscope integrated microfluidic device. A single tissue section was stained sequentially for 24 different immunophenotyping and tissue structural markers. Each staining cycle consisted of incubation of the tissue section with a pair of mouse and rabbit primary antibodies, followed by the corresponding fluorescently labelled secondary antibodies and DAPI. The section was imaged after each staining cycle and subsequently eluted before staining the next pair of markers.

Results

Our microscope integrated microfluidic system allowed automated 24-plex staining with conventional primary and fluorescently labelled secondary antibodies in less than five hours, including image acquisition steps. The microfluidic tissue processor enabled fast fluidic exchange and thereby resulted in reduced staining time down to 10-12min per marker. Integration of a window into the microfluidic chip allowed direct tissue imaging under the microscope avoiding the removal and mounting of the slide. Protocol optimization resulted in a high signal to background noise ratio for each marker and complete elution of antibodies from the previous staining step. A comparison of a 10-plex staining with standard chromogenic stainings on sequential sections showed high concordance for the stained area on tonsil as well as lung cancer tissue sections (Figure 1).

Conclusions

With the microscope integrated microfluidic system, it is possible to perform fast multiplex stainings including image acquisition without the need to handle the tissue slide. Moreover, due to the sequential nature of the system it would be easily possible to further increase the number of markers in the multiplex staining. We foresee this technique to greatly facilitates the execution of high-plex stainings and thereby the discovery of novel tumor-microenvironment interactions.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P50).

Automated microfluidics-assisted multiplexing

Correspondence: Jessica Finn (jessica.finn@ionpath.com)

Background

Elucidating both the cell types present in the tumor microenvironment and the spatial relationship between immune and cancerous cells is at the forefront of immunotherapy research. To address this, MIBI has been developed to image up to 40 markers at single cell resolution.

Methods

Staining of 10 NSCLC formalin-fixed paraffin embedded (FFPE) tissue sections was performed similarly to traditional IHC except that a panel of 20 metal labeled antibodies were stained simultaneously. The tissue was imaged at subcellular resolution using an ion beam and time-of-flight secondary ion mass spectrometry (ToF-SIMS). The masses of detected species were then assigned to target biomolecules given the unique label of each antibody and multi-step processing and segmentation were performed to create images of the TME and enable quantitative metrics of different cell subsets.

Results

Control samples imaged at study start and end showed consistent marker quantification (inter-run R2>0.99), indicating MIBI staining and acquisition is reproducible and robust. Each tumor sample was imaged across 10 regions of interest (ROIs) to assess heterogeneity of the TME. Highly expressed nuclear, membrane, and cytoplasmic markers were utilized in conjunction to accurately determine cell boundaries in tissue images. The resulting single cell segmentation enabled quantitative analyses of both marker expression and the spatial relationships between cells of different types. At the highest level, cells were classified as positive for markers that are indicative of immune and tumor cells based on measured intensities of marker expression, such as CD45+ and keratin+ cells in epithelial cancers, respectively. Co-expression of markers were used to classify immune cells into subsets, including T cells and macrophages (Figure 1). Cell types and their frequency were compared within the 10 ROIs collected per sample as well as between samples. Finally, distances between tumor and the closest immune cell were measured as a means for describing the spatial organization of the TME, which has been linked to patient survival.

Conclusions

MIBI offers high-parameter capability, at sensitivity and resolution uniquely suited to understanding the complex tumor immune landscape, including the spatial relationship of immune and tumor cells and the expression of immunoregulatory proteins.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P51).

Cell segmentation and classification

Correspondence: Jie Zhang-Hoover (jie.zhang-hoover@merck.com)

Background

ILT3 on human monocytic myeloid cells is linked to immune tolerance in transplantation and immune suppression in cancer. Anti-ILT3 mAb is being developed as a cancer immunotherapy to reverse the suppression and increase T cell activation. To evaluate the effect of an anti-ILT3 mAb in vivo, we sought to select an appropriate humanized tumor model and identify immune activation signatures in the model that are associated with the treatment efficacy.

Methods

Humanized tumor models were generated by subcutaneously implanting Panc 08.13 or SK-MEL-5 tumor cells in NSG mice engrafted with human cord-blood CD34+ hematopoietic stem cells (hu-NSG). Tumor-bearing mice were treated with a human-mouse chimeric anti-ILT3 mAb. Single cell mass cytometry (CyTOF) that simultaneously quantifies over 40 cell surface and intracellular markers was used to phenotype tumor infiltrating cells (TILs) in these tumor models.

Results

The CyTOF phenotyping of untreated mice showed an overall immune suppressive environment in the tumor in both Panc 08.13 and SK-MEL-5 hu-NSG models. ILT3 expression was detected in both models. However, the levels of ILT3 expression on CD14+ myeloid cells and percentage of CD14+ myeloid cells among TILs were higher in SK-MEL-5 compared to Panc 08.13 tumors, which led to the selection of the SK-MEL-5 model for further exploration. Anti-ILT3 mAb treatment in the SK-MEL-5 hu-NSG model increased activation of CD14+ myeloid sub-populations in TILs by viSNE CyTOF clustering analysis. Furthermore, the treatment increased levels of CD69 and HLA-DR expression on CD4+ T cells, while reducing the percentage of naïve CD4+ T suppressor cells in CD45+ TILs.

Conclusions

Anti-ILT3 mAb treatment induced a conversion from intra-tumoral immune suppression to activation in a SK-MEL-5 hu-NSG tumor model. CyTOF in anti-ILT3 drug discovery holds promise to effect a paradigm-shift in our ability to understand MOA and evaluate the impact of therapeutic interventions that can accelerate biomarker discovery and drug development.

The anti-ILT3 mAb activity in human immune cells in vitro and tumor efficacy in vivo is presented in a companion poster.

Ethics Approval

The study was approved by Merck Institutional Animal Care and Use Committee, approval number 2022-200518-FEB.

Correspondence: Sripad Ram (sripad.ram@gmail.com)

Background

Immunohistochemistry (IHC) assays play a central role in evaluating biomarker expression in tissue sections for diagnostic and research applications. Manual scoring of IHC images, which is the current standard of practice, is based on qualitative criteria and are known to have several shortcomings in terms of reproducibility and scalability to large scale studies. While digital image analysis (DIA) based approaches hold significant promise to overcome these limitations, current DIA methods pose several challenges that have limited their widespread use in analyzing clinical samples.

Methods

We introduce a novel DIA metric, the pixelwise H-score (pix H-score), that quantifies biomarker expression from whole-slide scanned IHC images. Pix H-score is unique in that it does not rely on the detection of individual cells or the delineation of subcellular compartments (e.g. nucleus and cell membrane) which are necessary for traditional scoring algorithms such as the H-score. All DIA metrics are calculated using either commercially available (HALO, Visiopharm) or open-source (QuPath) digital pathology software packages.

Results

We compute the pix H-score, the ATM score [1] and the traditional H-score [2] from IHC images for several biomarkers including PD-L1. Our results show that the pix H-score exhibit tight concordance to multiple orthogonal measurements such as mRNA levels and pathologist score, and provide consistently better performance over other DIA metrics.

Conclusions

We anticipate that the new metric introduced here will be broadly applicable to quantify biomarker expression from a wide variety of IHC images. Although not shown here, the new metric can also be applied to immunofluorescence images. Moreover, these results underscore the benefit of digital image analysis-based approaches which offer an objective, reproducible and highly scalable strategy to quantitatively analyze IHC images.

References

1. Choudhury KR, Yagle KJ, Swanson PE, Krohn KA, Rajendran JG, A Robust Automated Measure of Average Antibody Staining in Immunohistochemistry Images. J Histochem Cytochem. 2010; 58: 96-107.

2. Hatanaka Y, Hashizume K, Nitta K, Kato T, Itoh I, Tani Y, Cytometrical image analysis for immunohistochemical hormone receptor status in breast carcinomas. Pathol Int. 2003; 53: 693-699.

Correspondence: Clifford Hoyt (choyt@akoyabio.com)

Background

In cancer research, advancing our understanding of the underlying mechanisms driving disease progression is key to developing new therapeutic regimens and improving patient outcomes. Over the past several years, multiplex immunofluorescence (mIF) has played a vital role in elucidating novel immune-tumor interactions and identifying targets of interest for drug discovery and development.

Emerging studies utilizing mIF have revealed complex cell-to-cell interactions within the tumor microenvironment (TME), however, greater interrogation of the biology comprising these interactions, including cellular composition and functional status, require higher levels of multiplexing. With the rapidly increasing number of available multiplexing approaches, there is an inherent tradeoff between capability and throughput.

In this study, we demonstrate a streamlined workflow to develop and optimize a 9-color assay on the Leica BOND RX™ autostainer. This methodology offers an optimal balance between multiplexing and sample throughput to facilitate research and support translational studies on whole formalin-fixed paraffin-embedded (FFPE) tissue.

Methods

For the 9-color assay, Opal™ fluorophores were used on serial sections of lung cancer FFPE tissue. The panel was designed on Akoya’s Mantra 2 semi-automated multispectral microscope, which allows for rapid analysis of staining performance. Once optimized, multispectral images were acquired on both the Mantra 2 and Vectra Polaris of the same tissue regions and analyzed to show equivalence between the platforms. Cell counts, densities, and spatial parameters were generated using Akoya’s inForm image analysis software and the R script package phenoptrReports, which produces quick, summarized outputs of the image analysis data. These same analyses were also used to evaluate reproducibility of all markers when run in a high-throughput process.

Results

Dynamic range of measured per cell signals for all markers had a median of 200:1. Agreement between the Mantra 2 and Vectra Polaris-based measurements was generally >95% when comparing cellular expression signals and cell counts based on cell phenotyping classifiers. Cross talk was undetectable after spectral unmixing despite significant spectral overlap inherent in a 9-color assay. Reproducibility across three batches of five serial sections of lung cancer tissue was generally <10% coefficient of variation for all markers in the assay, supporting a high-through process of approximately 20 resection samples per day.

Conclusions

We have successfully established a standardized process for 9-color multiplexing that offers a balance between elucidating the intricate cellular biology driving disease progression and therapeutic responsiveness within the TME while simultaneously providing a practical and reliable assay that can be implemented to support translational, high-throughput studies in clinical research.

Correspondence: Ana Hidalgo Sastre (ahidalgo@definiens.com)

Background

One of the biggest challenges in multiplex chromogenic IHC (mIHC) is to accurately identify and quantify double positive cells. Multiplex immunofluorescence (mIF) instead, allows for visualization of plenty of biomarkers at once with true co-localization. However, visualizing tissue morphology in mIF images can be challenging and the vast color combinations overwhelming. Pathologists are key to retrieve biological information from multiplex assays and provide annotations for assay validation. To support pathologist analysis, resections of non-small cell lung carcinoma (NSCLC) patients were stained with mIF and displayed as pseudo mIHC images. Additionally, consecutive slides were stained with a mIHC panel. PD-L1 positive macrophages from the pseudo mIHC images were quantified and compared to the readouts identified in the real chromogenic IHC.

Methods

7 formalin-fixed paraffin-embedded (FFPE) resections from NSCLC patients were stained using Ultivue’s UltiMapper I/O PD-L1 kit and I/O PD-1 kit and whole image scans were acquired with a Zeiss Axio Scan.Z1 scanner (Zeiss). Consecutive slides were stained with a multiplex chromogenic panel (including CD68, CD8, PD-1) at Mosaic Laboratories (1) and scanned with an Aperio AT Turbo scanner (Leica). Images were analyzed using an automated workflow for quantitative multiplex image analysis developed at Definiens (2). Afterwards, mIF images were converted into pseudo mIHC images. Pathologists annotated double positive macrophages for CD68 and PD-L1 on both images. Results were compared with automatically detected double positive cells and across assays. In addition, pathologists qualitatively assessed visual similarity of real and artificial chromogenic images.

Results

Pathologists annotated double positive macrophages for CD68 and PD-L1 markers on both images (mIF and pseudo mIHC). Results were compared with those obtained using artificial intelligence to automatically detect double positive cells and across assays. In addition, pathologists qualitatively assessed visual similarity of real and artificial chromogenic images (Figure 1).

Conclusions

Transferring mIF into pseudo mIHC images helps to combine the advantages from both approaches: true colocalization of biomarkers whilst maintaining tissue morphology, facilitating visual evaluation of digital images by pathologists. This technology could be used to complement research, clinical routine diagnostic, drug development and biomarker discovery.

References

(1) Lisa M. Dauffenbach, Christopher A. Kerfoot, et al. Characterization of inflammatory cell patterns and densities using multiplex immunohistochemistry immuno-oncology assays [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl): Abstract nr B069.

(2) Lorenz Rognoni, PhD; Vinay Pawar, PhD; Tze Heng Tanet, et al. Automated quantification of whole-slide multispectral immunofluorescence images to identify spatial expression patterns in the lung cancer microenvironment. SITC Annual Meeting; 2018 Nov 7-11; Washington, DC. Poster nr P442.

Ethics Approval

Ethical approval was granted by the Liverpool Research Ethics Committee, reference number 97/141.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P55).

mIF and pseudo mIHC

Correspondence: Aaron Goldman (goldman1@mit.edu)

Background

The role of myeloid cell populations within a tumor and their contribution to effective cancer immunotherapy is emerging with considerable interest. However, assessing the role of intratumoral myeloid cells under therapy pressure has it’s challenges. Here, we implemented a multiplex immunohistochemistry (mIHC) panel and a human ex vivo system to interrogate key myeloid subsets as they affiliate with infiltration and activation of an emerging immunotherapy for glioblastoma multiforme (GBM) – oncolytic Herpes simplex virus-1 (oHSV-1). mIHC combined with advancements in digital pathology and machine learning algorithms have enabled identification, quantification and spatial orientation of multiple cell types in a single field of view (FOV).

Methods

Cell Signaling Technology antibodies (CD3e,D7A6E™, ID: 85061), (CD68, D4B9C, ID: 76437), (CD11c, D3V1E, 49420), (MHC Class II (HLA-DRB) LGII-612.14), (Pan-Keratin, C11, 4545) were optimized for mIHC staining, using a tyramide signal amplification approach to pin-point, in a single FOV: intratumoral T-cells, defined by CD3e+; macrophages, defined by CD68 and conventional dendritic cells defined by CD11c+MHCII +, in relation to the surrounding tissue architecture defined by pan cytokeratin. These biomarkers were integrated with incidences of oHSV-1 infiltration and replication (via expression of green fluorescent protein).

Results

First, we confirmed an optimized protocol for treating GBM ex vivo with oHSV-1 such that tissue viability, infiltration and replication of the virus are optimal. The staining of the mIHC panel was optimized using matched 3,3′-Diaminobenzidine chromogenic and single biomarker fluorescent controls tonsil tissue, which were validated using tumor samples from patients with high grade glioma. We characterized the spatial arrangement of myeloid subpopulations, ex vivo and correlated the changes in spatial orientation, quantity and localization of cells to the tumor. We determined that dynamic re-arrangement of myeloid cells can be observed under pressure of immunotherapy within the tumor, and confirm both a time-dependent and dose-dependent effect of oHSV-1 on this immune cell modulation.

Conclusions

These data suggest a unique, multiplexed approach to study spatial arrangement of myeloid and T-cell populations and their spatial distribution within tumors under basal growth conditions or in the presence of anticancer immunotherapies, which may implicate the activity of myeloid cells with treatment responses. These findings could impact personalized cancer immunotherapy for patients receiving care.

Ethics Approval

The samples were collected under IRB approval.

Consent

The samples were collected under written patient consent for publication of this abstract.

Correspondence: Aaron Goldman (goldman1@mit.edu)

Background

Response and resistance to cancer therapy relies on the presence of active immune cells in the tumor microenvironment, which recalibrate the body’s own defense largely by modulating exhaustion of cytotoxic lymphocytes including T cells and natural killer (NK) cells. However, there is a critical gap in our understanding for the role of immune cells to drive response or resistance to drugs and immunotherapies at the individual patient level. This is primarily due to limitations in complex tumor-immune interfaces that exist in many current tumor models.

Methods

Here, we deployed an ex-vivo human system that uses an explant of native, patient-derived solid tumors including autologous immune cells. Utilizing biopsied tumor tissue from patients diagnosed with triple-negative (ER- PR- HER2-) breast cancers (TNBC, N=7), we studied drug-induced cell death (cleaved caspase-3) and spatial heterogeneity of NK cells (CD3-CD56+PanCK-) using multiplex immunohistochemistry (mIHC). Spatial orientation of cells in the microenvironment, including proximity of NK to tumor and NK cell density within regions of the tumor vs. stroma were performed using HALO-based quantitative analyses. Finally, we deployed in-vitro co-culture studies using 3-D TNBC organoids and human-derived NK cells (NK-92MI).

Results

First, we report the ability of the ex-vivo human system to retain the spatial orientation and total population of natural killer cells and T-cells over the course of a 72h explant culture. Next, using Spearman correlation analyses and principal component analysis (PCA), we determined that drug response to both immunotherapy and conventional cancer drugs, indicated by high incidence of cleaved caspase-3 after drug pressure, is directly associated with changes to the tumor-NK cell proximity and density of NK cells within the tumor bed vs. the stroma. Finally, using the 3-D tumor organoid cultures with NK cells, we determined that activity and tumor cytolysis by NK cells is hampered through cancer cell-activated cytokines, which diminish expression of activating biomarkers including NKG2D/C.

Conclusions

Taken together, these results provide a method to study the spatial arrangement of immune cells in an entirely human system, which can be perturbed with anticancer drugs to reliably influence the expression and growth patterns of immune cells. We further demonstrate that this strategy can help to guide in-vitro studies to further elucidate mechanisms of action of drugs, which influence response vs. resistance via immune cell activity.

Ethics Approval

Anonymous breast cancer tissue samples were collected under IRB approval with due written consent from each patient.

Correspondence: Ammar Sukari (sukaria@karmanos.org)

Background

PD-1 inhibitors aim to re-instate the natural anti-cancer immune-mediated cytotoxicity. Although PD-1 inhibitors are now considered part of standard of care treatment in advanced metastatic NSCLC [1], little is known about the effects of PD-1 inhibitors on asymptomatic central nervous system (CNS) metastases. We hypothesized that early MRI brain imaging due to the development of neurological signs and symptoms following the initiation of PD-1 inhibitor may help delineate a subset of NSCLC patients with asymptomatic and undiagnosed CNS metastases prior to initiation of therapy and may predict for worse outcomes.

Methods

Data from NSCLC patients who received at least one dose of PD-1 inhibitors between September 2013 through the data cut-off of May 2017 were captured from our institution’s pharmacy database. The primary objective was to describe the characteristics of patients with MRI brain being performed within 4 weeks of the first dose of PD-1 inhibitors and the secondary objectives were estimation of progression free survival (PFS) and overall survival (OS) for the same population.

Results

140 NSCLC patients received at least one dose of PD-1 inhibitors prior to data cut-off. Median age was 64 (range: 24-86). 83 (59%) were male. 64 (46%) were treated on a clinical trial. There were 92 (66%) adenocarcinoma, 41 (29%) squamous cell carcinoma (SCC) and 7 (5%) poorly differentiated NSCLC. 84 (40%) had a pre-PD1 inhibitor MRI brain performed and 25 (18%) had been diagnosed with baseline CNS metastases. 128 (91%; Group 1) did not have an MRI brain performed within 4 weeks of starting PD-1 inhibitors, while 12 (9%; Group 2) patients did. 9 out of 12 patients had new or worsening CNS metastases. Of the 9, 1 had WBRT, 1 had gamma knife, 1 went onto hospice while a decision was made to monitor imaging and symptoms in 6 patients. The median PFS was 5.28 months (95% CI, 3.90 to 8.03) and 1.75 months (95% CI, 1.08 to NE) for Group 1 and Group 2, respectively. The median OS was not reached (95% CI, 15.38 to NE) and 5.77 months (95% CI, 2.85 to NE) for Group 1 and Group 2, respectively.

Conclusions

In this retrospective analysis, patients who had MRI brain within 4 weeks of starting PD-1 inhibitors had worse outcomes.

Reference

1. NCCN Clinical Practice Guidelines in Oncology. Non-Small Cell Lung Cancer. Version 5. 2019- June 7, 2019. https://www.nccn.org/professionals/physician_gls/default.aspx#site, last accessed 7/30/2019.

Ethics Approval

The study was approved by the Wayne State University Institution's Ethics Board, approval number 062616M1E.

Correspondence: Jaime Rodriguez-Canales (rodriguezcanalesj@medimmune.com)

Background

Novel multiplex immunofluorescent (mIF) platforms have been developed for immunoprofiling of solid tumors to understand the tumor microenvironment and to identify biomarkers for immunotherapy. One of these methods employ IF and tyramide signal amplification (TSA) to generate between 4 to 9-marker multiplex panels analyzed with multispectral imaging. Although this method can provide reliable data, they can show variability in consistency depending on the markers [1], which affects the reliability of mIF for use in clinical trials. The goal of this study was to develop and validate a highly consistent mIF method for its use in clinical trials in the pharma and academic environments.

Methods

A mIF panel for the analysis of carcinomas was optimized using an automated stainer (Leica) and automated multispectral scanner (Polaris, Akoya Biosciences). The markers included keratins (AE1/AE3), CD68, PD-L1, PD1, CD8 and Ki67. Each primary antibody was first performed on standard chromogenic IHC according to previously validated protocols. The mIF panel was developed with a secondary antibody detection and TSA, using specific fluorophores for each marker. Once optimized, the miF panel was tested on serial sections of formalin-fixed human tonsil controls and six non-small cell lung carcinomas, including replicates, together with standard IHC staining of each individual marker for comparison. A set of three repeats on different days for each mIF with all cases was performed to test consistency and reproducibility of the mIF method. All slides were digitally scanned and analyzed using Automated Definiens Insights Platform with custom algorithms (Definiens AG, Munich, Germany), comparing the cell populations in the serial section slides between standard IHC and mIF, and between mIF repeated rounds. The data was statistically analyzed using Pearson’s correlations.

Results

Using an automated workflow, the mIF data compared with standard IHC showed correlations between 0.83 to 0.99. The data from the three rounds of mIF performed on different days showed correlations between 0.89 and 0.99. The marker that showed the lowest correlation was CD68 (r=0.83), the possible cause was the difficulties on cell segmentation due to morphologic irregularity shown by macrophages. Overall our present data showed a much higher consistency than our previously published results using a non-automated mIF protocol, which correlations ranged from 0.17 to 0.87[1].

Conclusions

Our data demonstrates that using an automated workflow including automated staining, scanning and analysis, and with properly validated IHC markers, mIF becomes a highly consistent methodology and it is compatible for its use with clinical trial tissue specimens.

Trial Registration

Not applicable

Reference

1. Parra ER, Uraoka N, Jiang M, Cook P, Gibbons D, Forget MA, Bernatchez C, Haymaker C, Wistuba II, Rodriguez-Canales J. Validation of multiplex immunofluorescence panels using multispectral microscopy for immune-profiling of formalin-fixed and paraffin-embedded human tumor tissues. Sci Rep. 2017; 7:13380-13391.

Correspondence: Steve Benz (Steve.Benz@nantomics.com)

Background

Multiple methods to characterize immune-cell populations in tumor microenvironment (TME) are being assessed as potential biomarkers of immunotherapy response. These include manual pathological assessment of lymphocyte infiltration, immunohistochemical (IHC) staining for specific adaptive response markers such as CD8, and more recently transcriptomic-based deconvolutions of immune populations such as xCell and TIMER. Here we combined digital masking using deep-neural nets with transcriptomic deconvolution to infer where immune-subpopulations may reside in the TME.

Methods

An unselected set of 187 clinical samples from the ImmunityBio database were analyzed. Each had H&E stained diagnostic slides with pathologist-annotated tumor regions, as well as deep whole-transcriptomic sequencing (>200M reads). Deep neural networks previously trained on TCGA slide images were used to generate digital spatial masks for 3 characteristics: tumor-content, lymphocytes, and stroma. Patients were scored based on the presence of intratumoral lymphocytes (iTIL) and stromal lymphocytes (sTILs). Immune subpopulations were inferred from RNAseq expression of published immune-cell-specific genesets [1,2], as was Wnt-signaling level [3]. Significant associations between immune subpopulations and level of infiltration were analyzed.

Results

Manually annotated positive tumor regions were accurately digitally masked as >83% tumor or lymphocyte. Wnt signaling was strongly associated with overall stromal content (Rho=0.47, p<0.0001). Strong anti-correlation was observed between levels of sTILs and iTILs (Rho=-0.42, p<0.0001), and remained significant when including overall stroma area as a covariate. Digital lymphocyte masks somewhat correlated with RNAseq-based deconvolution of lymphocyte classes (Rho=0.30, p=0.0001) in line with reports from others [4], however this decreased when comparing lymphocyte count within annotated tumor regions only (Rho=0.17, p=0.03), despite high concordance of lymphocyte counts within and outside of annotated regions overall (Rho=0.82, p<0.0001). RNAseq-based lymphocyte levels were more associated with sTILs than iTILs (Rho=0.19 vs. -0.28, p<0.01 respectively).

Conclusions

Adaptive response effectors such as NK and T-cells are found more resident in surrounding stromal tissue than infiltrating tumor tissue. Increased Wnt/B-catenin signaling in stromal regions, reported by others as immunosuppressive, may sequester immune effectors and aid in immune escape.

References

1. Bindea G, et al. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer. Immunity 2013; 39:782-795.

2. Danaher P, et al. Gene expression markers of tumor infiltrating leukocytes. J Immunother Cancer. 2017; 5:18.

3. Slattery ML,et al. Expression of Wnt-signaling pathway genes and their associations with miRNAs in colorectal cancer. Oncotarget. 2018; 9:6075.

4. Pai SG, et al. Wnt/beta-catenin pathway: modulating anticancer immune response. J Hematol Oncol. 2017; 10:101.

Correspondence: Jay Tarolli (jay@ionpath.com)

Background

When studying the tumor microenvironment, knowing not only the types of immune cells present but also the spatial distribution and relationship of these immune cells to other immune and tumor cells provides crucial information. In the past, techniques used to analyze these spatial relationships have been limited by the number of biomarkers that could be simultaneously measured. Recently, with the development of multiplexed ion beam imaging (MIBI), 40+ biomarkers can be simultaneously measured in a single scan [1]. By probing with an ion beam, tissue sections can be imaged at a spatial resolution on the same order of magnitude as light based techniques, providing subcellular resolution. This combination of multiplexed biomarker measurements and subcellular spatial resolution enables segmentation of the image into individual cells, making possible subsequent cell type classification and quantification.

Methods

Samples of placenta, lung, tonsil, lymph node, thymus, and liver were imaged with MIBI. Segmentation of these images was performed in two steps. First, a MaskRCNN [2] model was trained to utilize multiplexed MIBI data to predict the location of cell instances in a MIBI image for a single class of objects by learning features from a set of nuclear, cytoplasmic, and membrane markers. The centroids of each predicted cell instance were used as seed points and, after manual refinement of these seed points, watershed segmentation was performed to determine boundaries between instances. Both the summed intensity of a marker as well as a weighted cell score which accounts for the spatial distribution of a marker’s expression throughout a cell instance were calculated and were used for cell type classification.

Results

Cell population and densities were calculated for a number of different cell types, including T cells, B cells, and macrophages based on a combination of one or more coexpressed biomarkers present within segmented cells. Figure 1 shows an example FOV with several cell types classified in a single image. Expression of immunoregulatory proteins including PD-1 and PD-L1 were quantified and assigned to specific cell types. Finally, nearest neighbor distances between various cell types were determined to characterize the spatial organization of cell populations within each tissue image.

Conclusions

The ability to characterize the many different cell types within the tumor microenvironment is made possible by the highly multiplexed nature of MIBI data, the subcellular spatial resolution of the image data, and downstream analysis tools, including computer vision approaches, which enable cell segmentation, classification, and spatial analysis.

References

1. Keren L, Bosse M, Marquez D, Angoshtari R, Jain S, Varma S, Yang S, Kurian A, Van Valen D, West R, Bendall S, Angelo M. A Structured Tumor-Immune Microenvironment in Triple Negative Breast Cancer Revealed by Multiplexed Ion Beam Imaging. Cell. 2018; 174:1373-1387.

2. He K, Gkioxari G, Dollár P, Girshick R. Mask R-CNN. IEEE International Conference on Computer Vision (ICCV). 2017; 2961-2969.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P61).

See text for description

Correspondence: Mike Spencer (mspencer@bethyl.com)

Background

For the advancement of immunotherapeutics, the need to understand the tumor microenvironment has never been more pressing. Recent advances in mIF and multispectral imaging facilitate accurate simultaneous analysis of multiple tissue markers. This is critical in instances where sample is limited, such as a tumor biopsy or other clinical specimen. The applications of mIF are numerous, and span clinical, translational, and basic research applications. A seven-color mIF can take eight weeks, or more, to develop. Herein, we describe a simplified, faster approach.

Methods

FFPE human tissue was stained with PathPlex™ Panel 4 IHC validated primary antibodies (Bethyl Laboratories [A810-004]), mouse or rabbit HRP-conjugated secondary antibodies (Bethyl Laboratories [A90-116P, A120-501P]) and detected using Opal™ Polaris 7-color IHC kit fluorophores (Perkin Elmer [NEL861001KT]). Primary antibody order was optimized utilizing tissue microarray serial sections, and three slides per target by staining after the first, third, or sixth heat-induced epitope retrieval (HIER). All three slides were imaged using the same exposure time and analyzed for target/nucleus counts, signal intensity, and background. Finally, the order was tested in the seven-color mIF and compared to single stain for confirmation. Whole slide scans were generated using the Vectra Polaris® and analyzed using InForm® image analysis package.

Results

Development time of a seven-color mIF was reduced using IHC validated antibodies and the optimized dilution. Antibody order was guided by results of three slides stained after first, third or sixth HEIR. The ratio of target staining/DAPI nuclear counts, average intensity and overall background predicts the optimal order of staining. Some targets reveal larger average area staining, higher intensity and lower background when stained last, for example FOXP3 (Table 1, Figure 1), while the inverse may be true, or no effect for other targets. There are 720 possible combinations for a seven-color panel. Using this method, the number of slides was reduced to three per target (18) plus confirmation seven-color slides resulting in a panel containing CD3, CD8, CD68, Cytokeratin, FOXP3 and PD-L1 (Figure 2).

Conclusions

Multiplex IF is a powerful technique that allows for examination of spatial arrangement of proteins of interest as well as protein interaction/co-localization of multiple targets within a single tissue specimen. MIF panels can take eight or more weeks to optimize, however, researchers can save time and resources using validated antibodies and this antibody order guide.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P62).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P62).

See text for description

Correspondence: Omar Laterza (omar_laterza@merck.com); Charo Garrido (charo.garrido@merck.com)

Background

Myeloid-derived suppressor cells (MDSCs) are a group of leucocytes with myeloid origin and immune-suppressive function. Ample recent evidence supports key contributions of MDSC to tumor progression through immune-mediated mechanisms. MDSCs include two major subsets based on their phenotypic and morphological features: polymorphonuclear (PMN)-MDSC and monocytic (M)-MDSC. However, these cells remain less studied than T lymphocytes as their phenotypical, morphological and functional heterogeneity generate confusion in investigation and analysis of their role.

Methods

With the progresses on multiplex IHC assay technology, we are now able to develop multiplex immunofluorescent marker panels to evaluate the expression and localization of the main subpopulations of myeloid cells in the tumor microenvironment.

Results

We have developed and validated CD14, CD66b, CD163 and MHCII (HLA-DR) IHC multiplex marker panel to evaluate the main populations of myeloid cells, along with their activation status. After completing the validation for individual markers in a single chromogenic IHC platform, we optimized the incorporation of each marker into the multiplex platform. In parallel, a multiplex image analysis algorithm (APP) was generated and validated to quantify each subpopulation of myeloid cells in the tumor area. Last, fit for purpose analytical validation, including sensitivity, specificity and precision was successfully carried out.

Conclusions

This validated assay is currently being used to support multiple ongoing and future clinical trials.

Correspondence: Julia Kennedy-Darling (j.kennedy@akoyabio.com)

Background

Characterizing the complexities of the tumor microenvironment is elemental to understanding disease mechanisms. The spatial relationships between infiltrating immune cells and the remodeling of the cellular matrix is widely recognized as a key component to defining tumor heterogeneity. Current methodologies for analyzing the spatial dimension in tissues, like traditional immunofluorescence (IF) and immunohistochemistry (IHC), are limited to a few parameters at a time, restricting the scope of identifiable cells. Conversely, single-cell technologies like mass cytometry and NGS-based tools provide multiplexing capabilities, but at the expense of the associated spatial information. Here, we present the analysis of human lung cancer FFPE tissues with CODEX using a panel of more than 20 markers targeting the tumor microenvironment.

Methods

The CODEX technology, developed by Akoya, is comprised of a fluidics instrument that interfaces with existing microscope hardware, as well as a suite of reagents and associated control and analysis software. The CODEX technology involves labeling antibodies with oligonucleotide-based Barcodes followed by a single staining step. Around 40 parameters can be measured within a single tissue through fully-automated, iterative cycles of adding and removing corresponding dye-conjugated Reporters. Here, we apply this technology using a panel of antibodies targeting immune, cancer and other architectural features to measure cell subsets in cancer FFPE tissues. Image data is processed using the CODEX analysis pipeline, including clustering, annotation and mapping of cell types to the original image data with the Multiplexed Analysis Viewer (MAV).

Results

The CODEX technology was used to ascertain complex cellular niches and spatial associations between multiple cell types based on the staining pattern of more than 20 parameters. The high-parameter antibody panel used was optimized with human FFPE tonsil tissues. Human lung cancer FFPE tissues were then analyzed with this same panel. Cell clustering using the MAV software identified tens of cell types within the tumor tissues. Immune cell sub-types were mapped onto the original image data to assess infiltration and spatial associations.

Conclusions

Unlike other cyclic IF approaches involving multiple antibody staining and stripping steps, the CODEX platform involves a single initial staining step and subsequent gentle and relatively fast manipulation of the tissue thereafter. This provides a superior workflow and prevents tissue degradation. CODEX data from various normal and cancer human FFPE tissue types is shown here with corresponding single-cell analysis of key tissue features. Overall, the CODEX platform is an accessible and versatile technology for high parameter, spatial profiling of tissue specimens.

Correspondence: Songming Peng (speng@pactpharma.com)

Background

T cells targeting tumor-exclusive neoepitopes (neoE) have been postulated to represent the primary mediators of clinical benefit for patients with solid tumors treated with immunotherapies. Identifying and tracking these T cells in patients can help to understand the mechanism for immune checkpoint inhibitor therapies, as well as provide new therapeutic candidates for personalized adoptive cell therapies. However, this has been hampered by the low frequency of neoE-specific T cells in peripheral blood. To this end, we demonstrate the use of the imPACT Isolation Technology®, an ultra-sensitive high-throughput technology, to capture neoE-specific CD8 T (neoE-T) cells from peripheral blood. In addition, this technology can be utilized to quantify and monitor neoE-T cells longitudinally during therapy. We show here preliminary data applying the imPACT technology to clinical trial samples for the characterization of mutation-targeted T cell responses from patients associated with clinical benefit.

Methods

Peripheral blood mononuclear cells (PBMC) from patients with non-small cell lung cancer (NSCLC) and treated with combinations containing an anti-PD-1 antibody were analyzed. Briefly, tumor-exclusive neoE-HLA target candidates were predicted and barcoded snare libraries comprising personalized neoE-HLA reagents were produced for capture of neoE-specific CD8+ T cells from PBMCs. Longitudinal analysis of neoE-T cells responses throughout the duration of treatment was performed to obtain valuable information on neoTCR sequences and neoE-T cell quantification & phenotype.

Results

A baseline neoE-specific CD8 T cell profile was identified in all subjects prior to treatment. Among NSCLC subjects exhibiting objective responses to therapy, some neoE-T cell clones identified at baseline persist in the blood and/or diversify in clonality over the course of treatment. In some circumstances, new neoE-T cell clones have emerged on treatment with anti-PD-1. Furthermore, phenotype analysis suggested the neoE-T cells captured from blood have been activated, indicating previous encounter with their respective neoE-HLA targets.

Conclusions

The imPACT technology was used to assess the phenotype & quantity of neoE-specific T cells in blood of trial participants over time. This approach revealed the evolution of mutation-targeted T cell responses in participants with clinical benefit and may prove to be a powerful tool to provide mechanistic understanding of immune responses associated with clinical benefit. These data support further testing of the neoE-T cell capture technology, with the potential to uncover the identity of neoE-specific T cells pre-existing in the blood of patients and the evolution of immune attack to cancer.

Ethics Approval

The study was approved by the institutional review boards or ethics committees of the participating sites in Arcus Biosciences’ clinical studies.

Correspondence: Sabine Mueller (sabine.mueller@ucsf.edu); Hideho Okada (hideho.okada@ucsf.edu)

Background

Diffuse midline glioma, including diffuse intrinsic pontine glioma (DIPG) constitutes up to 20% of pediatric brain cancer and has a median survival of 9-10 months. Given the proximity of DIPG to parenchymal regions that play vital homeostatic functions, surgical resections are often restricted in size and scope, leaving irradiation and chemotherapy as the primary management options. The ongoing development of immunotherapy has shown significant promise in many fields, including that of gliomas. Genetic studies revealed that greater than 70% of DIPG cases harbor an amino acid substitution from lysine (K) to methionine (M) at position 27 of histone 3 variant 3 (H3.3). We previously identified a novel HLA-A*02:01-restricted neoantigen epitope encompassing the H3.3K27M mutation. Accordingly, we have implemented a pilot vaccine through the Pacific Pediatric Neuro-Oncology Consortium (PNOC).

Methods

Twenty-nine newly diagnosed DIPG patients who are HLA-A2+ and H3.3K27M+ underwent radiation therapy, and then received the H3.3K27M peptide vaccine and tetanus toxoid (TT) peptide emulsified in Montanide in combination with poly-ICLC every 3 weeks for a total of 24 weeks. Our objective is to characterize vaccine-induced H3.3K27M-specific CD8+ T-cell and myeloid-derived immunosuppressive subpopulations in peripheral blood mononuclear cells utilizing a novel H3.3K27M-specific dextramer-based mass cytometry (CyTOF) method [1,2].

Results

Through this approach, the temporal expansion of vaccine-reactive CD8+ T-cells was observed in all patients who completed a minimum of 24 weeks on the study (n = 4). Simultaneously, this expansion was not observed in 4 of 5 patients who withdrew from the regimen due to progression. These T-cells were clustered on a tSNE plot using canonical CD8+ T-cell activation markers and further classified by their expression profiles, revealing distinct effector memory, central memory and transitional effector subpopulations. Chronological monitoring of these groups indicates the time course-dependent development and persistence of vaccine-reactive exhausted and effector memory CD8+ T-cells in 3 of the 4 initial patients analyzed. Furthermore, an analogous clustering and phenotyping approach was used for myeloid cells, allowing for the identification of myeloid-derived suppressor cell (MDSC) subpopulations. A comparative analysis revealed a positive correlation between two monocytic myeloid-derived suppressor cell (M-MDSC) subpopulations and progression-free survival.

Conclusions

Future plans include analyzing the remainder of patients enrolled in the trial and the utilization of CD8+ and MDSC-specific CyTOF panels to further classify the aforementioned subpopulations to further elucidate this relationship. This methodology offers insight into the progression of vaccine-induced patient immune responses and exhibits promise as a platform that may be extrapolated to other immunotherapies.

References

1. Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, Ohgaki H, Wiestler OD, Kleihues P, Ellison DW. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: A summary. Acta Neuropathol. 2016; 131:803–820.

2. Chheda Z, Kohanbash G, Okada K, Jahan N, Sidney J, Pecoraro M, Yang X, Carrera D, Downey K, Shrivastav S, Liu S, Lin Y, Lagisetti C, Chuntova P, Watchmaker P, Mueller S, Pollack I, Rajalingam R, Carcaboso A, Mann M, Sette A, Garcia C, Hou Y, Okada H. Novel and shared neoantigen derived from histone 3 variant H3.3K27M mutation for glioma T cell therapy. J Exp Med. 2017; 215:141-157.

Ethics Approval

The study was approved by UCSF IRB #: 16-20574

Correspondence: Theresa Walunas (t-walunas@northwestern.edu)

Background

Immune related adverse events (irAE) occur in >80% of patients receiving immune checkpoint inhibitors (ICI). Currently, most data about the incidence of irAE comes from clinical trials with restrictive eligibility requirements. With the wide use of ICI therapy as standard of care for many cancers, it is important to assess incidence of irAE in a general patient population. The Chicago Area Patient Reported Outcomes Research Network (CAPriCORN) is a clinical data research network containing medical records for >9.5M patients who receive care in 11 institutions spanning diverse patient populations and healthcare settings [1]. Using CAPriCORN, we wanted to determine whether we could identify a large, diverse cohort of patients who received ICIs as a foundation for exploring the incidence of irAE in a real-world data source.

Methods

We identified all patients within CAPriCORN who were 19-88 years old, had a diagnosis for an ICI-approved cancer, and received an ICI from 1/1/2011 through 12/31/2018. Clinical experts identified the International Classification of Disease 9 and 10 codes used to document cancer diagnosis and the RxNorm [2] codes for each ICI documented as a medication ordered in the medical record (Table 1). The query was developed against the PCORnet Common Data Model version 4.1 [3], validated on the Northwestern University site node in CAPriCORN and distributed to all CAPriCORN sites. Six of 9 sites returned counts. Data was centrally aggregated and stratified by age, race, sex and therapy. All data are aggregated counts.

Results

As shown in Table 2, we identified 6,541 patients within CAPriCORN who received ICI therapy for cancer. 45% are female, 75% identify as white, 13% African American, 2% Asian and 1% Native American, and 86% are 51-83 years of age. The most well represented cancers were Non-small Cell Lung Carcinoma (50%) and Metastatic Melanoma (18%). Overall, 67% received anti-PD1 therapies, followed by combination ICI therapies, anti-PDL1 and anti-CTLA4, though usage varied within cancer types.

Conclusions

Our results demonstrate that a large cohort of cancer patients who have received ICI therapy can be identified in an integrated medical record data environment that spans 11 institutions in a major urban center. This population is racially diverse, represents both sexes, a wide range of ages and includes all cancer types approved for ICI therapy as of 2018. This real-world cohort will be an effective foundation on which to explore the incidence of irAE, particularly rare irAE that require large sample sizes to investigate.

Acknowledgements

The investigators would like to acknowledge the CAPriCORN network for query support. This work was supported by Patient Centered Outcomes Research Institute (PCORI) CDRN-1306-04737.

References

1. Kho AN, Hynes DM, Goel S, et al. CAPriCORN: Chicago Area Patient-Centered Outcomes Research Network. J. Am Med Inform Assoc. 2014; 21:607-611.

2. https://pcornet.org/download/pcornet-common-data-model-v4-1-specification-5-may-2018/?wpdmdl=1919&refresh=5d42596fab3e11564629359

3. https://www.nlm.nih.gov/research/umls/rxnorm/

Ethics Approval

The study was approved under the CAPriCORN IRB, CHAIRb: Research Protocol #14120201 “CAPriCORN Clinical Data Research Network Master Protocol”.

Correspondence: Suhendan Ekmekcioglu (sekmekcioglu@mdanderson.org); Chantale Bernatchez (cbarnatchez@mdanderson.org)

Background

Immune infiltration of T cells (TIL) into the melanoma microenvironment has been associated with improved survival for some patients, and also has been exploited to grow TIL in vitro for adoptive therapy. However, prognostic significance of immune infiltrating cells in melanoma and other tumors remains a relatively new concept, and markers related to suppressive versus active functional TIL remain unclear. We previously reported that in Stage III melanoma patients’ tumors, positive expression of CD74 together with low or absent Macrophage Migration Inhibitory Factor (MIF) associates with favorable prognosis [1].

Methods

From an ongoing clinical trial using TIL intended for adoptive immunotherapy, we have studied the melanoma patient tumors specimens (FFPE) from 20 patients whose autologous TIL lines grew to sufficient number for possible use clinically. We also examined another 20 sets of melanoma tumor from which the TIL did not grow or not grow well. We analyzed the differences in the two groups of tumors (40 total FFPE) for CD74 regulated pathway features and inflammatory marker expression.

Results

CD74 regulated markers included CD44, MIF, and downstream inflammatory targets including inducible Nitric Oxide Synthase (iNOS) and Nitrotyrosine (NT). Our findings confirm our previous report in that tumor CD74 expression significantly associates with favorable OS and PFS (both, p=0.0038) and provides new data that in this set of patients the CD74 also correlates with best irRC of TIL treated patients. New findings include that the NT expression in tumor cells associated with poor TIL growth (p=0.014), as well as lack of clinical response to TIL treatment (p=0.02). We have also found that tumor cell-derived MIF and iNOS expression correlate with unfavorable prognosis for both OS and PFS (p=0.016 and 0.018, respectively).

Conclusions

We have identified the protein expression of CD74, MIF and of iNOS as providing survival information, and proposed that CD74+/MIF-/iNOS- together be considered to form a "signature" of good prognosis in general melanoma outcomes as well as TIL growth and favorable responses for these patients. Use of this signature for selecting patients for entry into TIL and possibly other immunotherapy trials, as well as research on the differential pathways of IFN-γ signaling in melanoma appear as important areas for future mechanistic research to improve patient outcome.

Reference

1. Ekmekcioglu S, Davies MA, Tanese K, Roszik J, Shin-Sim M, Bassett RL Jr, Milton DR, Woodman SE, Prieto VG, Gershenwald JE, Morton DL, Hoon DS, Grimm EA.Inflammatory Marker Testing Identifies CD74 Expression in Melanoma Tumor Cells, and Its Expression Associates with Favorable Survival for Stage III Melanoma. Clin Cancer Res. 2016 Jun 15;22(12):3016-24.

Correspondence: Akil Merchant (akil.merchant@cshs.org)

Background

Diffuse large B cell lymphoma (DLBCL) being the most subtype of non-Hodgkin lymphoma. Despite evidence of expression of PDL-1 on lymphoma cells, less than 10% of DLBCL patients respond to PD1 therapy [1]. We hypothesize that a better characterization of spatial architecture of the tumour microenvironment (TME) in lymphoma will help explain differences in responses to PD1/PDL-1 inhibitors and guide future targeted immunotherapies for these patients.

Methods

Here we characterized the TME in DLBCL using imaging mass cytometry (IMC), which allows high-dimensional, single-cell and spatial analysis of FFPE tissues at sub-cellular resolution [2]. Using a panel of 32 antibodies, IMC was performed 41 tissue microarray cores from 33 DLBCL cases. IMC images were analyzed for relevant immunophenotypes, the spatial architecture of those phenotypes and compared to clinical outcomes to identify immune contexture based biomarkers.

Results

Phenograph was used to cluster tumor and immune cells based on phenotype (Figure 1A). Immune cell represented 33% of the cells represented by CD4 (36%), CD8 (30%), macrophages (26%) and TREG (8%) (Figure 1B). Immune cell infiltration in individual tumor samples ranged from 7% to 75% with marked heterogeneity. (Figure 1C-D. Analysis of immune marker expression on tumor cells identified co-expression of PD-L1/CCR4/TIM3 to be highly prognostic for overall survival (p=0.003, Figure 1E)

To characterize the patterns of spatial interaction in the TME, we developed an unsupervised multivariate model to construct spatial meta-clusters based on average distances from CD8 to the centroids of 5 nearest endothelial cells, TREG, CD4 T cells, macrophages, and tumor cells (Figure 2A). Spatial analysis revealed 11 meta-clusters for CD8 T cell interactions (Figure 2B). Meta-clusters 2, 6, 8 and 11 were the 4 most dominant patterns of CD8 spatial interaction in the TME. Each CD8 spatial interaction pattern is distinctive with case to case heterogeneity (Figures 2C-D). Risk assessment analyses of spatial clusters 1, 2 and 4 (“hazardous”) had almost 3 times higher odds of being identified in refractory cases compared to clusters 3, 5 and 6 (“protective”) (Figure 2E). In the “protective” spatial neighborhoods, we observed the presence of activated CD8, Th1-like CD4, and less suppressive TREG phenotypes, with opposite in “hazardous” areas (Figures 3A-B). TIM-3 expression was high both on T cells and tumor cells in the “hazardous” neighborhoods.

Conclusions

Our novel approach to spatial analysis of the immune architecture reveals clinically relevant insights into the TME.

References

1. Ansell, S. M. et al. Nivolumab for Relapsed/Refractory Diffuse Large B-Cell Lymphoma in Patients Ineligible for or Having Failed Autologous Transplantation: A Single-Arm, Phase II Study. J. Clin. Oncol. 37, 481–489 (2019).

2. Giesen, C. et al. highly multiplexed imaging of tumor tissues with subcellular resolution by mass cytometry. Nat. methods | 11, 417 (2014).

Ethics Approval

The study was approved by USC IRB, approval number HS10-260

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P69).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P69).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P69).

See text for description

Correspondence: Alexander Klimowicz (alexander.klimowicz@boehringer-ingelheim.com)

Background

T cell exhaustion and the PD-L1/PD-1 checkpoint axis has been extensively characterized in peripheral blood mononuclear cells and in human tumor tissues. This has provided a better understanding of the role this pathway plays in tumor immunology and of its clinical utility in predicting responsiveness to checkpoint inhibitor therapies. T cell exhaustion is not only associated with tumor progression, but has recently been associated with better prognosis and milder course of disease for a number of autoimmune and autoinflammatory disorders. We set out to characterize and contrast the T-cell exhaustion environment between colonic Crohn’s disease (CD) and colorectal cancer (CRC). We applied the Ultivue UltiMapper multiplex fluorescence IHC platform to capture complex immune cell phenotypes and provide a more in depth characterization than traditional IHC.

Methods

Commercially sourced FFPE surgical resections from n=5 colonic CD patients (matched lesional and non-lesional tissue) were compared to n=5 CRC tumor resections (3 hot and 2 cold tumors) using the Ultivue UltiMapper multiplex fluorescence immunohistochemistry platform. Two UltiMapper kits were used to evaluate the T cell environment in these tissues: UltiMapper I/O PD-L1 panel included the markers CD8, CD68, PD-L1, and pan-Cytokeratin/Sox10; UltiMapper I/O PD-1 panel included the markers CD3, CD45RO, PD-1, and pan-Cytokeratin/Sox10. All assays were stained on Leica BOND RX autostainers. Whole-slide images were acquired on a ZEISS Axio Scan.Z1 slide scanner. Image analysis was performed using Indica Labs HALO software.

Results

Contrasted to non-lesional CD tissues, several similarities were observed between CD lesional tissue and CRC, including the presence of PD-L1+ immunoreactivity in epithelial/tumor cells, increased immunoreactivity for PD-L1 in CD68+ cells, and a closer relationship between intra-epithelial and stromal CD8+ cells with PD-L1+ cells. In addition, areas in CRC and CD heavily infiltrated by immune cells or with tertiary lymphoid structures contained clusters of PD-L1+ cells that were negative for both CD68 and pan-Cytokeratin. Most of the CD3+ cells in non-lesional CD were PD-1 negative, except around tertiary lymphoid structures. In contrast, a greater percentage of CD3+ cells were also PD-1+ in CRC, and more so in lesional CD tissue.

Conclusions

The Ultivue UltiMapper multiplex fluorescence immunohistochemistry platform was effective in characterizing the PD-L1/PD-1 axis and T cell exhaustion environment in FFPE tissue, in part due to the ability to clearly identify more complex immune cell phenotypes than traditional multiplex IHC. The application of the UltiMapper assays demonstrated many similarities between marker and cell type distribution between lesional colonic CD and CRC.

Correspondence: Jeffrey Gulcher (jgulcher@wuxinetcode.com)

Background

Tumor mutation burden (TMB) is used as a surrogate marker for the neoantigen load of a tumor, and many studies have shown that TMB predicts the success of immune-oncology (IO) treatments for cancers, such as anti-PD-1 or anti-CTLA4 therapy. While IO treatment of patients with high TMB has led to success and excitement in the field, not all high TMB patients respond to IO treatment. TMB can be determined using next-generation sequencing, and both panel and whole-exome DNA sequencing have been used to measure the mutation load of a tumor.

Because all genes are not expressed in every cell, inclusion of the mutations found in unexpressed genes may confound the utility of TMB to predict neoantigen load. In this study, we explore improving TMB by taking into account both DNA mutation and RNA expression with the hypothesis that using both criteria will lead to a TMB biomarker that correlates better with neoantigen load.

Methods

We examined DNA and RNA sequencing data for different cancer types in The Cancer Genome Atlas (TCGA) to determine refined TMB values. The data were assessed to identify mutations with and without expression.

Results

We found that a significant faction mutations included in a standard TMB calculations reside in genes that are not expressed, and this fraction varies significantly among samples. A corrected TMB that incorporates gene expression is likely to be a better predictor of neoantigen load. In addition, our group has previously identified TCGA samples with allele specific expression in which up to 25% of the tumor mutations are not expressed despite significant expression of the gene. The fraction of unexpressed mutations is much higher in some cancer types. Adding this correction to the TMB calculation, including only variants that are expressed in the tumor in the TMB calculation, changes the average TMB values found among different cancer types, and corrects the TMB in individual samples. This refined TMB value provides a biomarker that reclassifies samples scored as high TMB to low TMB and is likely to better predict response to IO treatment.

Conclusions

As our results suggest, adding RNA sequencing can be used to improve the TMB biomarker to better separate treatment groups. The initial analysis was performed with TCGA exome data and is now being extended to refine TMB values generated from gene panels including TSO 500. We are also evaluating differences in tumor-infiltrating lymphocytes in based on the refined TMB value.

Correspondence: Sean Boyle (sean.boyle@personalis.com)

Background

Comprehensive detection of potential neoantigens and accurate prediction of their MHC presentation are critical prerequisites for selecting neoepitopes that can be used for creating personalized cancer vaccines. However, prediction models developed using in-vitro MHC-peptide binding assays cannot model upstream presentation machinery, such as proteasome cleavage and peptide loading. Advances in immuno-affinity purification followed by mass spectrometry (IP-MS) have enabled direct detection of MHC-bound peptides and can therefore be used for modelling native MHC-peptide presentation. Further, genetically engineered cell lines that express a single HLA allele enable unambiguous HLA-peptide assignment. Here, we present an overview of our MHC presentation prediction framework based on a large collection of such mono-allelic cell lines and discuss its utility in conjunction with ImmunoID NeXT, our commercially available exome scale DNA and RNA sequencing and analytics platform specifically designed to enable the development of immuno-therapies.

Methods

Mono-allelic cell lines were generated from K-562 null-HLA parental cells by transfecting each of the selected alleles. Cells were grown, screened for surface expression, lysed and immuno-affinity purified using a column coated with HLA class I (W6/32) antibody. Peptides were gently eluted and analyzed using LC-MS/MS. Peptide-to-spectrum assignment was performed and filtered at 1% false discovery rate.

Results

The training data for our MHC presentation prediction framework were generated using a large collection of genetically engineered mono-allelic cell lines, encompassing approximately 60 HLA Class I alleles that are frequently present across various populations. The resulting immuno-peptidomics data were comprehensive and of high quality - the peptide yields were high (median of approx. 1600 unique peptides per allele) and the dominant motifs were in agreement with published motifs. Our prediction framework is based on multiple modelling algorithms, including a multi-layer neural network, and uses proprietary and standard features such as peptide sequence, peptide length, binding pocket sequence and abundance (measured by transcripts per million). We created allele-specific and pan-allele models and evaluated them on an independent hold-out dataset. Both our allele-specific and pan-allele models had superior performance compared to other public tools, with a higher precision across a range of recall (sensitivity) values.

Conclusions

Our integrated pipeline for neoepitope discovery, which includes the comprehensive profiling of putative neoantigens using ImmunoID NeXT and accurate and sensitive prediction of MHC presentation of such neoantigens across all HLA Class I alleles (using our pan-allelic models) enables the effective generation of neoepitopes that are critical for developing personalized cancer vaccines.

Correspondence: Alex Powlesland (alex.powlesland@gmail.com)

Background

The repertoire of HLA-peptides presented to the immune system which derive from cancer-associated, viral, and mutated proteins are attractive targets for immunotherapy. Identifying the full complement of peptides derived from a protein presented on a major class-I HLA restriction provides a vital step toward increasing the speed and viability of many immunotherapeutic strategies. Advances in next-generation sequencing (NGS) and single-cell technologies have enabled the accurate capture of somatic mutations accumulated by a tumour, yet a significant hurdle remains how this information can be utilized for immunotherapeutic benefit. Identifying which somatic mutations produce neoantigens is crucial in providing the link between genetic change and immunological impact.

Methods

Directly identifying potential neoantigens using mass spectrometry offers a significant improvement over traditional approaches based on prediction. However, the relatively high sample requirement of this approach inherently limits the depth of analysis that can be performed, with a significant risk that low abundance neoantigens are not detected.

By integrating multiomics data from over 1000 experiments in 200 immortalised cell lines, we have generated a database of over two million unique HLA-peptide sequences that offers near total coverage of the protein-coding genome. Our comprehensive HLA class-I peptide atlas has been used as a reference tool to aid direct identification of neoantigens by targeted mass spectrometry, to probe indirectly for the presence of neoantigens, and to explore how many common driver mutations associated with cancer interact with the immunopeptidome.

Results

We have identified hundreds of neoantigens directly by mass spectrometry and found that mutated proteins follow the same pattern of antigen processing and presentation as their unmutated equivalents. As a result, our HLA peptide atlas offers significant value in predicting the likelihood of a somatic mutation creating a neoantigen. Comparing predicted neoantigens with those directly identified by mass spectrometry, we show effective prioritization of mutations by accurately predicting the presence and relative abundance of neoantigens. Applying this process toward the five most commonly mutated genes in cancer reveals a marked bias toward mutations that either act negatively or are in ‘quiet’ areas of the immune landscape. As all mutated peptides contain novel amino acid sequence, and are hence able to elicit an immune response, this ability to convert ‘potential’ into ‘actual’ is crucial in establishing a mechanism for identifying false positive results observed in cell-based assays.

Conclusions

An integrative multiomics approach to neoantigen identification has delivered a powerful reference for developing novel immunotherapies

Correspondence: Gad Berdugo (gberdugo@epivaxonco.com)

Background

We hypothesized that neo-epitope-based prediction using an advanced in silico T cell epitope screening system (Ancer™) may better identify patients with improved prognosis than tumor mutation burden. Analysis of genomic data derived from the muscle-invasive bladder cancer (BLCA) cohort of The Cancer Genome Atlas (TCGA) database for CD4, CD8, and Treg neo-epitopes was performed to determine whether Ancer™ would improve prognostic stratification compared to tumor mutational burden (TMB).

Methods

BLCA patient mutanomes (n=412) were retrieved from the TCGA and evaluated with Ancer™, an innovative and automated neo-epitope screening platform that combines proprietary machine learning-based HLA I and HLA II neo-epitope identification tools with removal of inhibitory regulatory T cell epitopes for neo-epitope ranking and personalized cancer vaccine design. BLCA patients were separated based on median TMB or neo-epitope burdens. We investigated the effect of integrating both CD8 and CD4 neo-epitope burdens as most mutanome pipelines exclusively focus on the identification of Class I neo-epitopes. Overall survival was analyzed using the Kaplan-Meier method and differences analyzed by log-rank testing.

Results

Compared to low TMB, high TMB was significantly associated with improved survival (p = 0.0001, difference of 38.5 months in median survival, Figure 1). Improved differentiation of median survival times was obtained when separating patients based on their Class I neo-epitope content, as estimated by Ancer™ (p < 0.0001, difference of 59.8 months in median survival). Adding Class II neo-epitope burden further increased separation of OS times, showcased by a 69.6-month increase in median survival for BLCA patients with both high CD8 and high CD4 neo-epitope contents compared to other patients (p = 0.0001). Since we discovered that Class II neo-epitopes can induce inhibitory responses, we further evaluated whether the screening of these detrimental sequences could improve our analysis. Upon identifying Class II neo-epitopes likely to induce T effector (Teff) responses, we found that the median survival of patients with high CD8 and high CD4 Teff contents was extended by nearly 4 months to 73.4 months compared, to the remainder of the cohort (p < 0.0001, Figure 2).

Conclusions

Our analysis suggests that optimal host-immune recognition of CD8+, CD4+, and Treg epitopes plays a key role in cancer survival. While defining CD8 neo-epitope burden enhanced associations with OS, the inclusion of CD4 Teff neo-epitope burden substantially helped identify long-term survivors. These results suggest that defining the number of true neo-epitopes using Ancer™ may represent a novel prognostic or predictive biomarker.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P74).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P74).

See text for description

Correspondence: Alessandra Nardin (alessandra.nardin@immunoscape.com)

Background

During clinical trial immune monitoring, especially in the field of immunotherapy, it is critical to collect in-depth phenotypic information from multiple immune cell populations in order to assess the biological activity of the immunotherapy, to identify biomarkers of response or progression, and/or to identify new drug targets. However, patient samples, for example peripheral blood mononuclear cells (PBMC) or tissues, are often only available in small amounts and current methods face limitations in either depth of analysis and/or cell throughput.

Methods

In order to identify therapy-relevant antigens and to facilitate a concurrent in-depth characterization of cells directed towards these targets, immunoSCAPE leverages the high-dimensional immune profiling capabilities of cytometry by time of flight (CyTOF) and a unique methodology allowing the identification and characterization of rare antigen-specific T-cell subsets (targetSCAPE). By implementing a new technology (ultraSCAPE) that combines flow and mass cytometry together with a combinatorial live cell barcoding strategy, we further increased the high-dimensional phenotyping capacities to over 100 different marker molecules through simultaneous in-depth profiling of up to three additional immune cell subsets from the same sample.

Results

We isolated 4 different immune cell populations from a single sample and combined 3 different phenotypic panels consisting of 35 makers each together with a combinatorial tetramer multiplex and phenotyping panel for deep profiling of myeloid cells, NK cells, B cells and T cells. We demonstrate the potential of this novel immuno-phenotyping method, by tracking virus-specific T cells while simultaneously characterizing 4 immune cell subsets with over 100 distinct phenotypic markers from a single sample, which is currently impossible employing modern flow cytometers or classical mass cytometry methods.

Conclusions

With its ability to provide an unprecedented picture of the immune status within a single sample, including T cell specificity information and in depth profiling of relevant immune cell subsets, ultraSCAPE in combination with targetSCAPE can provide detailed insights on the effects of immunotherapy on the immune cell population. Information learned from in-depth immune phenotyping of several immune cell subsets such as T, NK and myeloid cell subsets can be leveraged for the development of novel diagnostics, for biomarker discovery and for monitoring therapeutic strategies in immunotherapy.

Correspondence: Takashi Shimada (tashimada@shimadzu.com)

Background

The highly complex population of peptides associated with human leucocyte antigens (HLA) is the human immunopeptidome. Comprehensive characterization of the immunopeptidome is key in predicting immunotherapeutic responses by evaluating targets of T cell interaction and in developing the next generation of cancer immunotherapies. Mass spectrometry (MS) is a technology that holds significant promise for untargeted and complete identification of the immunopeptidome. MS acquisition is mainly used an electrospray ionization (ESI) combined with nanoflow liquid chromatography (LC). However, the analysis time and retention reproducibility could be issues. Therefore, we tried to develop an MS platform that can ensure the coverage while increasing throughput using a microflow LC.

Methods

HLA class I complexes were purified from A431 cell lysate by W6/32 immunoaffinity. Purified HLA peptides were eluted with 5% formic acid. The peptides were fractionated by 10 kDa ultrafiltration. HLA peptides were separated with L-column2 ODS (0.3x150 mm) using a trap-elute protocol of microflow LC (Nexera-Mikros) and quadrupole time-of-flight MS (LCMS-9030). Flow rate was set at 5 μl/min with a gradient of acetonitrile in 0.1% formic acid for 18 min. MS/MS spectra were acquired using a data-dependent manner of top-10 precursor intensities. The precursor scan was first set from 400 to 600 Da. The charge states of precursors were set between 1 to 4, and MS/MS scan was from 200 to 1200 Da. The data were analyzed by Mascot proteome server and PEAKS sequencing software on SwissProt database. The mass tolerances of precursors and fragments were set at 0.05 Da and 0.3 Da. Minimal peptide length was set to 8 amino acids.

Results

An initial round of optimizations was performed to establish optimal parameters for immunopeptidome identifications. Using tryptic peptides from A431 lysate, we optimized the 50-100 msec repeat of MS/MS scanning, top-10 of data-dependent acquisitions per scan, 50 Da scan range, 35V±10V spread of collision electrode voltage, and 3.0 kV of electrospray voltage. These parameters were then applied for identification of HLA-associated peptides from A431 cells. From this, we identified 4,217 MS/MS and 801 sequences from 34,042 spectra.

Conclusions

From these data, we demonstrate that similar sensitivity can be sufficiently achieved with microflow platform as has been demonstrated preciously for nanoflow LC-MS. This has significant advantages in terms of throughput, instrument maintenance, and widespread applicability. Our future directions are to determine whether cancer neoepitopes identified by these approaches may be recognized and therapeutically targeted by patient T cells.

Correspondence: Sean Boyle (sean.boyle@personalis.com)

Background

Checkpoint inhibitor therapy has demonstrated meaningful antitumor activity for many patients, though the majority fail to achieve complete response. Thus, it is of particular interest to identify biomarkers and mechanisms that promote positive response to immunotherapy. In the present study, we apply our comprehensive tumor immunogenomics platform (ImmunoID NeXT), integrating data from the tumor, tumor microenvironment and immune system to create a comprehensive biological signature of patient response to therapy.

Methods

We characterized the immunogenomics of 52 unresectable, stage III/IV melanoma patients who underwent anti-PD-1 therapy to assess factors influencing response. RECIST criteria were used to evaluate tumor response to therapy, with a median follow-up of 12 months. For each patient, a single paired FFPE tumor and normal blood sample was collected and profiled using Personalis’ ImmunoID NeXT platform; an augmented exome/transcriptome platform and analysis pipeline, which produces comprehensive tumor mutation information, gene expression quantification, neoantigen characterization, HLA typing and LOH, TCR repertoire profiling and tumor microenvironment profiling. Tumor molecular information was then analyzed together with clinical outcome.

Results

Comprehensive profiling demonstrated that elevated pretreatment neoantigen burden was predictive of response to PD-1 blockade, and significantly associated with progression-free survival. Additionally, we observed increased response to anti-PD-1 therapy in patients with elevated pretreatment TCR clonality. Patients with high neoantigen burden and TCR clonality that failed to achieve complete response revealed potential resistance mechanisms to anti-PD-1 therapy. Specifically, we identified two patients with high expression of IDO1 or CTLA4, which may facilitate PD-1-independent immune escape. Additionally, we found two patients with antigen presentation machinery (APM) mutations. The first patient had independent HLA-A and HLA-B mutations, likely leading to loss of surface expression of the proteins. In the second APM mutation patient we observed a high frequency (80% AF) frameshift variant in B2M, which potentially prevents proper HLA class I folding and antigen presentation. These APM mutations suggest reduced neoantigen presentation in these patients, which are probable mechanisms for tumor escape. By integrating neoantigen burden, HLA-LOH and APM mutational data into a corrected neoantigen burden, we were able to increase the predictive strength of this biomarker.

Conclusions

In summary, our comprehensive cancer immunogenomic analyses demonstrate that genomic and immune profiling of pretreatment patient samples can identify biomarkers and resistance mechanisms to immune checkpoint blockade, suggesting the potential efficacy of these as an integrated biomarker to optimize anti-PD-1 therapy patient selection.

Correspondence: Seth Sadis (Seth.Sadis@thermofisher.com)

Background

Tumor mutation burden (TMB) measures the number of somatic mutations and is a positive predictive factor for response to immune-checkpoint inhibitors in multiple cancer types. While whole exome sequencing (WES) is the gold standard for TMB measurement, it is not practical for routine use. TMB values measured using targeted sequencing have been shown to have good correlation with WES. However, during FFPE preservation, DNA may undergo cytosine deamination, resulting in false C>T substitutions and elevated TMB values. We have assessed the effect of DNA damage and repair on TMB values using the Oncomine Tumor Mutation Load Assay (OTMLA), a targeted next generation sequencing assay.

Methods

We measured TMB from 37 FFPE colon, lung, endometrial and gastric tumors using the OTMLA panel on Ion GeneStudio with 20ng of input DNA from tumor only samples. The informatics workflow utilizes a custom variant calling and germline variant filtering algorithm to accurately estimate somatic variants in tumor tissue. In parallel, TMB was measured by Whole Exome Sequencing (WES) targeting 50Mb using 100ng of tumor and matched normal DNA on a HiSeq X instrument. We examined factors that affect OTMLA measurements: Deamination signature, degree of deamination and allele ratio identify DNA samples with high levels of damage due to FFPE preservation. A Uracil-DNA glycosylase (UDG) repair step was introduced to eliminate damaged targets and improve usable TMB values of DNA from FFPE tumor tissue. At the variant level, samples with high deamination scores were analyzed dynamically as a function of allele frequency to study TMB values for correlation with WES.

Results

OTMLA TMB values showed good correlation with WES-derived TMB; however ~10% of tumor DNA samples had high TMB and deamination values outside the expected range. These samples were included as a subset of samples tested with and without the UDG repair step. UDG treatment decreased TMB and deamination scores, resulting in higher correlation with WES TMB values. Some samples with very high deamination scores were unable to be rescued; however, TMB values in samples with low deamination and minimal damage were not affected.

Conclusions

We show that deaminated cytosine bases can be enzymatically removed by treatment with UDG. In a subset of FFPE samples tested, UDG treatment was demonstrated to reduce the OTMLA estimated SNP proportion consistent with deamination. This results in consistent and effective reduction of C>T artifacts without affecting true variants and can provide TMB values in a biologically relevant range.

Correspondence: Gennaro Ciliberto (gennaro.ciliberto@ifo.gov.it)

Background

The importance of tumor-host interactions during cancerogenesis and metastatic progression has been now widely appreciated. More recently, the impact of the tumor immune microenvironment (TIME) to mold tumor evolution was convincingly demonstrated [1]. Solid tumors systemically reprogram the lung unique immune environment, dominated by intravascular neutrophil functions [2], to colonize this site. The concept of ‘oligoprogression’ has recently received mounting attention, due to its relevance and because it represents an interesting in vivo model to study TIME, although the specific mechanisms of oligometastatic process are relatively underinvestigated [3, 4].

Methods

RNA sequencing was performed on a retrospective collection of tissue samples from primary renal cell carcinoma, melanoma, and NSCLC and paired lung oligometastases of untreated patients (Figure 1). Enrichment of tumor-related pathways and transcripts that reflect the enrichment of immune cell subsets was assessed by single sample gene set enrichment analysis. Differentially expressed genes between primary tumors versus the corresponding lung metastases were used for pathway analysis. Neutrophils extracellular traps (NETs) were revealed by immunofluorescence, assessing extracellular DNA and citrullinated H4 histone co-localization and/or myeloperoxidase [5]. Autophagy was assessed the CYTO-ID® kit [5].

Results

While tumor-related pathway enrichment differed mostly according to the primary tumor histology, perturbations of immune-regulatory pathways was observed during oligoprogression in the lung. Deconvolution of immune cell subpopulations identified increased immature dendritic cells and reduced T cell abundance in oligometastatic lesions. Strikingly, a large proportion of differentially modulated pathways were “immune” rather than “cancer-cell”-related. Core analysis confirmed that the main transcriptomic network that is affected during disease progression is immune-based, centered on a cross-link between innate and adaptive immunity. Specifically, it was associated with decreased HLA, iCOS, IL-9, and IL-17 pathway activity and downregulation of interferon signaling. During progression, we observed coherent modulation of transcripts associated with NET generation, related to upregulation of key autophagic genes, to competition of the HMGB1 molecule with CXCL12 and CXCR4 and RAGE receptor (AGER) activation. Accordingly, NET expression was strikingly more abundant in lung metastases than in primary tumors.

Conclusions

Our results identify evident molecular mechanisms associated with suppression of the immune milieu during disease oligoprogression in the lung across different tumors. They include innate-adaptive immune dysfunction HLA-mediated, and interferon dysregulation associated with neutrophil-mediated immune suppression. Since these tumors are targeted by immune checkpoint blockade (ICB) our data highlights the relevance of characterizing the TIME composition in paired primary and oligometastatic lesions during ICB treatment to optimize treatment approaches.

Acknowledgements

Paola Nistico' and Gennaro Ciliberto are co-last authors.

References

1. Angelova M, Mlecnik B, Vasaturo A, Bindea G, Fredriksen T, Lafontaine L, et al. Evolution of Metastases in Space and Time under Immune Selection. Cell. 2018 Oct;175(3):751-765.e16.

2. Granton E, Kim JH, Podstawka J, Yipp BG. The Lung Microvasculature Is a Functional Immune Niche. Trends Immunol. 2018 Nov;39(11):890-899.

3. Weichselbaum RR. The 46th David A. Karnofsky Memorial Award Lecture: Oligometastasis-From Conception to Treatment. J Clin Oncol. 2018 Sep27:JCO1800847.

4. Stephens SJ, Moravan MJ, Salama JK. Managing Patients With Oligometastatic Non-Small-Cell Lung Cancer. J Oncol Pract. 2018 Jan;14(1):23-31.

5. Maugeri N, Campana L, Gavina M, Covino C, De Metrio M, Panciroli C, Maiuri L, Maseri A, D'Angelo A, Bianchi ME, Rovere-Querini P, Manfredi AA.Activated platelets present high mobility group box 1 to neutrophils, inducing autophagy and promoting the extrusion of neutrophil extracellular traps. J Thromb Haemost. 2014 Dec;12(12):2074-88. doi: 10.1111/jth.P689.

Ethics Approval

The study was approved by IFO Institution’s Ethics Board, approval number 561/03.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P79).

See text for description

Correspondence: Carrie Brachmann (carrie.brachmann@gilead.com)

Background

The benefit of checkpoint blockade in advanced gastric cancer is limited and patient selection biomarkers are needed. In a randomized phase 2 study in >=2nd line advanced gastroesophageal adenocarcinoma (GEA) cancer in Europe, US and Australia, there was no clinical benefit for the addition of andecaliximab to nivolumab in the total population or evaluated subgroups (including PD-L1) [1,2]. Pharmacodynamic analyses demonstrated little to no impact of andecaliximab [3]. This exploratory biomarker analysis included all patients as a nivolumab-treated population.

Methods

Evaluation of archival tumor tissue was described [2,3]. Tumor mutation burden (TMB) was evaluated by whole exome sequencing with matched normal. Survival analyses (cox proportional hazards) were adjusted for age and sex.

Results

Overall survival (N=141, median 6.05 months) was not associated with PD-L1 positive (>=1% tumor or tumor+immune cells)/negative, diffuse/non-diffuse, gastric/gastroesophageal, prior therapy (median=2), tumor IFNgamma signature or CD8+ tumor infiltrate. Differential gene expression analysis identified GRB7, a downstream mediator of HER2 signaling and part of the HER2/ERBB2 amplicon in breast cancer [4], as one of two genes associated with survival >1 year (FDR=0.027). HER2-positivity (medical record) was associated with a 3.5-fold higher median expression of GRB7. Prolonged survival was associated with both HER2-positivity (n=43/132; HR=0.58, p=0.01) and the top quartile of GRB7 expression (n=25/94; HR=0.48, p=0.007). The median survival for HER2-positive patients was 10.1 months versus 5.95 months for HER2-negative. HER2 status was not associated with PD-L1 status or CD8+ infiltrate. Nearly all HER2-positive (n=40/43) and 2 HER2-negative patients received trastuzumab (median 62 days post-trastuzumab). Prior or best response was not related to 1 year survival and 2 of 3 HER2-positive patients that did not receive trastuzumab had >1 year survival. TMB was also evaluated and significantly associated with HER2-positivity (N=61, p=0.041). In the subset of 61 patients with TMB data, patients with high TMB plus HER2-positivity had the longest median survival of 15.4 months compared to all other patients at 6.7 months (N=14 vs 47; HR=0.47; p=0.04).

Conclusions

HER2 was associated with improved survival with checkpoint blockade in advanced GEA patients, regardless of response to prior trastuzumab. This study was limited by the lack of pre-treatment biopsies, but consistent with a recent report on Asian GEA patients [5]. The combination of HER2-positivity and relatively higher TMB in a limited dataset led to the greatest observed median survival time, suggesting an interaction between HER2 and TMB that warrants exploration in future GEA studies involving checkpoint inhibition.

Acknowledgements

The authors gratefully acknowledge the patients and their families who participated in this study.

Trial Registration

Clinicaltrials.gov NCT02862535

References

1. Shah M, Metges J-M, et al. A phase 2, open-label, randomized study to evaluate the efficacy and safety of andecaliximab combined with nivolumab versus nivolumab alone in subjects with unresectable or recurrent gastric or gastroesophageal junction adenocarcinoma. ASCO Gastrointestinal Cancers Symposium. 2019.

2. Metges J-M, Elboudwarej E, et al. Exploratory evaluation of baseline tumor biomarkers and their association with response and survival in patients with previously treated advanced gastric cancer treated with andecaliximab combined with nivolumab versus nivolumab. ASCO Gastrointestinal Cancers Symposium. 2019.

3. Brachmann C, Zhang Y, et al. Evaluation of intratumoral T cells in biopsies from advanced gastric patients treated with andecaliximab and nivolumab. ASCO Gastrointestinal Cancers Symposium. 2019.

4. Ferrari A, Vincent-Saloman A, et al. A whole-genome sequence and transcriptome perspective on HER2-positive breast cancers. Nature Communications. 2016.

5. Satoh T, Kang Y-K, et al. Exploratory subgroup analysis of patients with prior trastuzumab use in the ATTRACTION-2 trial: a randomized phase III clinical trial investigating the efficacy and safety of nivolumab in patients with advanced gastric/gastroesophageal junction cancer. Gastric Cancer. 2019.

Ethics Approval

This study was approved by the institutional review board or independent ethics committee appropriate for each site.

Correspondence: David Stenehjem (stene032@d.umn.edu)

Background

Tumor mutational burden (TMB) is emerging as a potential predictor of response to immunotherapy in various tumor types. However, the association of TMB data with clinical, demographic, genomic, and treatment characteristics warrants further investigation.

Methods

Nine U.S. Comprehensive Cancer Centers participated in this observational, cohort study; five centers are members of the Oncology Research Information Exchange Network (ORIEN). Adult patients with stage IV non-small cell lung cancer (NSCLC) with tissue-based TMB data from any testing platform were included and their treatment information was abstracted using a standardized case report form. TMB reporting ranged from September 2014 through March 2019. TMB-High and TMB-Low were defined as >10 mutations/megabase (mut/Mb) and <10 mut/Mb, respectively. Clinical, demographic, genomic, and treatment characteristics were compared by TMB level.

Results

There were 426 patients enrolled in the study across seven of the nine sites. TMB results from comprehensive genomic profiling (CGP) were available for 354 patients. CGP vendors included Foundation Medicine (79.9%), Caris Life Sciences (17.0%), Tempus (2.8%), and NantHealth (0.3%). The median time from diagnosis to CGP testing was 45 days. A comparison of clinical and demographic characteristics by TMB is presented in Table 1. TMB-High status was associated with male gender (p<0.01), and positive smoking history (p<0.01). No correlation was found between TMB and PD-L1 (Table 2). TMB-High was positively associated with multiple oncogenes including STK11, LRP1B, TP53, and KDM5C (Table 2). In addition, there were significant negative associations between TMB-High and individual occurrences of altered ALK (p=0.03), EGFR (p<0.01), and ROS1 (p=0.03). The proportion of patients receiving first-line immunotherapy increased yearly from 8.5% in 2015, 19% in 2016, 40% in 2017, and 46% in 2018.

Conclusions

These interim results demonstrate the feasibility of conducting multi-site observational electronic health record-based studies with CGP and TMB across a national cohort of comprehensive cancer centers. Immunotherapy utilization has been increasing in the first-line setting. Associations between TMB status and driver mutations are indicative of cancer etiology and informative for treatment decision-making. Updated results will be presented with an expanded cohort of patients and future publications with the final cohort (n~1000) will explore treatment, survival and response data.

Acknowledgements

This study was sponsored by Bristol-Myers Squibb. Recruitment efforts were supported by Mikaela Larson (Huntsman Cancer Institute) and M2Gen®

Correspondence: Jian Han (jhan@irepertoire.com)

Background

The T-cell receptor (TCR) is responsible for recognizing antigens as peptides bound to a major histocompatibility complex. TCRs typically contain both an alpha (α) and beta (β) chain that contribute to antigen specificity; however, we have seen multiple cases of a single cell containing dual α or dual β chains as well. When analyzing bulk repertoires, information about endogenous pairing of α and/or β chains is lost after bulk lysis of T-cell populations. Pairing α and β chains from a single cell while also analyzing the phenotypic expression allows us to track TCR specificity and T cell function. This information can provide direct calculations of clonal frequency in various cell subsets, allow tracking of specific lymphocytes with treatment, and reveal paired information for both chains of the receptor for downstream Car-T development.

Methods

Here, we developed a method for high-throughput pairing of TCR α and β chains along with expression profiling. We examined, on average, around 15,000 CD4+ cells loaded onto the BD Rhapsody Express system. The receptor information is amplified from the same cDNA using iRepertoire’s proprietary method that incorporates a multiplex mix of primers associated with both the TCR α and β loci; phenotyping of the cell is obtained using the BD Rhapsody RNA-seq kit. Alongside the high throughput data, we also performed FACS-based single cell sequencing on the same individual’s samples through our iPair method (presented previously) and examined the overlapping receptor sequences between both methods.

Results

With this mid throughput method, we are able to accurately assess the frequency of single cells containing dual alpha or dual beta TCRs, which can help to evaluate the high throughput data.

Conclusions

The described high throughput application should be applicable to any oligo-dT based single cell strategy.

Ethics Approval

This study was approved by New England IRB, IRB number 120160202

Correspondence: Eric Kong (ekong@pgdx.com)

Background

Therapeutic response to immune checkpoint inhibitors (ICIs) requires a prior, suppressed immune response that is released via the interaction of the checkpoint receptors with their cognate ligands. Microsatellite instability (MSI) and tumor mutation burden (TMB) have emerged as composite genomic metrics that may better predict patient response to ICI treatment, compared to conventional PD-L1 expression. However, testing for MSI and TMB separately is labor intensive, increases turnaround time, and consumes valuable tumor tissue samples. Furthermore, recent studies have demonstrated that antigen presentation mechanisms may also play a role in predicting outcomes, wherein loss of heterozygosity in MHC class I genes (LOH-MHC) suggests low likelihood of ICI benefit.

Methods

Here, we propose that these varied immune-oncology metrics can be measured together in a single assay, utilizing next-generation sequencing (NGS) technologies. To test this hypothesis, we analyzed >200 pan-cancer FFPE tumor tissue samples using the PGDx elio™ tissue complete assay (currently in development; >500 genes panel) to measure MSI, determine TMB (across 1.3 Mb), and assess MHC status in a single assay. Detection of MSI was assessed for accuracy against a validated PCR approach and TMB determination from our targeted panel was compared to whole-exome sequencing (WES) derived TMB. LOH-MHC status reported from PGDx elio tissue complete was compared to MHC status determined by WES for accuracy. Additionally, TMB and MHC status were analyzed together in FFPE samples from ICI treated patients and correlated to clinical outcomes.

Results

115 pan-cancer samples were analyzed for microsatellite status and demonstrated an overall agreement of 100.0% with a PCR-based method. TMB was determined in 118 pan-cancer samples and displayed a high level of concordance with WES-derived TMB (Pearson correlation, p=0.903) across a range of TMB scores (0.2-89.7 muts/Mbp). FFPE tissue samples from 98 cancer patients previously treated with ICIs were then tested for LOH-MHC and demonstrated 88% accuracy of detection when compared to WES. Furthermore, analysis of TMB and MHC status in tandem found that patients with both high TMB and normal MHC status were found to have a significantly higher PFS, suggesting greater efficacy of ICI therapy.

Conclusions

These data demonstrate the feasibility of measuring MSI, TMB, and evaluating LOH-MHC with high accuracy in a single >500 gene NGS assay. Additionally, the results presented herein suggest that measuring TMB and MHC status concurrently can provide added utility in predicting patient response to ICI therapy.

Correspondence: Yee chao (ychao@vghtpe.gov.tw)

Background

Tumor mutational burden (TMB) has been emerging as a relatively new biomarker that is independent of PDL1 for the prediction of response to the immune checkpoint inhibitor (ICI) treatment. A recent study has shown that whole exome sequencing (WES)-derived TMB correlates well panel-derived TMB that is estimated using targeted sequencing. Here, we evaluate the correlation between panel-derived TMB with response to ICI treatment in gastrointestinal cancers.

Methods

FFPE tumor and normal tissues from 18 patients with gastrointestinal cancers who had previously received ICI therapy at Taipei Veterans General Hospital were retrospectively underwent targeted next-generation sequencing (ACTOncoTM) for the identification of somatic variants across 440 genes and the calculation of TMB. NetMHC and IEDB were used to predict neopeptide bound to patient-specific HLA class one genotype. RECIST criteria were used to categorize tumor response.

Results

Patients were grouped into responder (PR or CR, n=10) and non-responder (PD or SD, n=7). Among all patients, responders had significantly higher TMB than non-responders (mean 7.37 muts/Mb vs. 1.24 muts/Mb, p=0.0007). The number of predicted neopeptides were significantly higher in responders than in non-responders (mean 10.8 vs. 3.6, p=0.0226). Notably, a non-responder harboring EGFR gain-of-function mutation did not respond to the ICI treatment despite high TMB. Furthermore, patients harboring MUC16 mutation demonstrated higher TMB than patients without MUC16 mutation (mean 17.38 muts/Mb vs. 3.9 muts/Mb, p=0.0005) (Figure 1).

Conclusions

Although the cohort size is small, our study showed that panel-based TMB is predictive of response to ICI. As in lung cancers, EGFR mutation is associated with decreased efficacy of ICI in the GI cancers.

Ethics Approval

The study was approved by TPEVGH intuition’s Ethics Board, approval number 2015-07-002BC.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P84).

See text for description

Correspondence: Seth Sadis (Seth.Sadis@thermofisher.com)

Background

Next-generation sequencing (NGS) is being applied to support routine clinical research in oncology with a primary focus on evaluating known oncogenic variants. However, the advent of cancer immunotherapies requires that clinical research solutions must also address biomarkers such as Tumor Mutational Burden (TMB) and Microsatellite Instability (MSI) for immune checkpoint inhibitors. Therefore, we developed a research use NGS solution for FFPE tissues that expanded upon our current Oncomine Tumor Mutation Load Assay by measuring biomarkers for both targeted and immune checkpoint therapies.

Methods

Gene content was prioritized based on the relevance and variant prevalence of biomarkers in solid tumors. Additional genomic regions were added to supplement the coding sequence footprint to support TMB. The assay used Ion AmpliSeq™ technology with automated templating on the Ion Chef™ system and sequencing on the Ion GeneStudio™ S5 sequencing platform. An automated tumor-only workflow for variant calling, TMB and MSI estimation and sample quality reporting was provided within Ion Reporter Software. Decision support tools were used for variant interpretation and evaluation of potential variant relevance.

Results

Over 500 genes with known DNA and RNA alterations were included. The panel has broad capability for variant calling, fusion detection, MSI status, in addition to TMB. Specifically, for TMB, DNA repair pathways were comprehensively represented as alterations in these genes may lead to high mutation burden. A coding sequence footprint to support TMB was generated. In development studies, the assay displayed high uniformity and consistent read depth to support robust variant calling. The automated workflow required minimal input of FFPE tumor only DNA and RNA material. Sample to report turnaround time was less than five days. In-silico assessment of TMB using publicly available whole-exome cancer sequencing data resulted in high correlation (R2 = 0.902, 0-40 mutations/Mb) and was parallel to the performance of our existing Oncomine Tumor Mutation Load Assay (R2 = 0.901, 0-40 mutations/Mb). Empirical analysis and performance of the assay on a common set of cell lines based on a universal reference standard also resulted in a positive correlation.

Conclusions

A larger tumor only NGS assay was developed to support comprehensive genomic profiling and routine clinical research in oncology. The assay design and informatics workflow support characterization of mutational signatures and provide normalized TMB estimates. Minimal input material requirement and rapid sample to report time will have a high impact on clinical research. More detailed information on the assay and an update on performance will be presented.

Correspondence: Natalie LaFranzo (natalie_lafranzo@cofactorgenomics.com)

Background

While cancer checkpoint inhibitors have garnered much attention due to their ability to generate durable responses and improved survival, the actual number of patients who are eligible for, receive treatment with, and respond to these therapies remains modest [1]. This is driven by a dependency on legacy diagnostics, built on single-analyte biomarkers such as PD-L1, which have failed to capture the complexity of disease [2]. Even in the case of non-small cell lung cancer (NSCLC), an indication where the benefit of IO therapies is considered significant, there is much to be learned about the biology of the patients who respond, or do not respond to these therapies.

Methods

Multidimensional RNA models have emerged to move beyond these legacy methods to reveal the full scope of disease complexity, resulting in increased predictive accuracy. Leveraging a database of gene expression models built using Predictive Immune Modeling, immune context of the tumor microenvironment is quantified. In this study, a cohort of NSCLC patients who received second-line immunotherapies (checkpoint inhibitors) were evaluated retrospectively. Pre-treatment solid tumor FFPE tissue samples were processed using the ImmunoPrism immune profiling assay to generate comprehensive, individual immune profiles. Pathological, demographic, and survival data (including overall survival and progression-free survival, indicative of therapy response), was used to group patients for predictive biomarker discovery.

Results

Individual immune profiles of the patients are compared, both within and between relevant cohorts, and statistically-significant biological signals are reported. Machine-learning derived multidimensional biomarkers were also generated, which are defined by the optimal combination of all analytes measured in the assay, enabling improvements in predictive accuracy. This study represents the first data generated using the ImmunoPrism assay with patients receiving checkpoint inhibitor therapies.

Conclusions

Predictive Immune Modeling enables us to build multidimensional models of disease. When combined with well-curated patient cohorts, such as the NSCLC patients described here, predictive biomarkers may developed which capture more facets of the complex immune contexture than previously possible.

References

1. Haslam A, Prasad V. Estimation of the Percentage of US Patients With Cancer Who Are Eligible for and Respond to Checkpoint Inhibitor Immunotherapy Drugs. JAMA Netw Open. 2019 May 3;2(5):e192535.

2. Nishino M, Ramaiya NH, Hatabu H, Hodi FS. Monitoring immune-checkpoint blockade: response evaluation and biomarker development. Nat Rev Clin Oncol. 2017 Nov;14(11):655-668.

Ethics Approval

The human tissue samples utilized for this study were provided by TriStar Technology Group and have written, informed donor consent permitting academic and commercial research for publication, as well as approval from a competent ethical committee.

Correspondence: Jian Han (jhan@irepertoire.com)

Background

Next generation sequencing of the immune repertoire is a comprehensive immune profiling methodology that allows detailed, sequence-specific insight into the adaptive immune response. While immune repertoire analysis of bulk RNA typically focuses on a single receptor chain, understanding of the variable rearrangements of the immune repertoire as a whole provides a broader view of the immune landscape with potential prognostic value. This is accomplished through the study of all seven TCR and BCR chains together (i.e., TCR-alpha, TCR-beta, TCR-delta, TCR-gamma, and BCR-IgK and -IgL). One of the key challenges during immune receptor amplification is the formation of dimers, which can compete with the immune amplicons of interest during library preparation.

Methods

We therefore developed a novel PCR technique, dimer avoided multiplex PCR (dam-PCR), that effectively avoids dimer formation during PCR and incorporates unique molecular identifiers for direct RNA quantification and error removal. With one sample, dam-PCR allows for the amplification of all seven TCR and BCR loci in a single, quantitative multiplex reaction. Here, we apply this method to the amplification of both PBMC and FFPE RNA from renal cancer patients undergoing treatment.

Results

We found that both TCR-alpha and -beta diversity prior to treatment along with the expression ratio between B cells and T cells are good predictors of treatment efficacy.

Conclusions

Our study suggests that examining multi-chain immune repertoire composition can be valuable for predicting treatment response and evaluating treatment protocols. Additionally, this method shows promise for future applications in both clinical settings and basic research, as it allows for a cost effective, all-inclusive, and quantitative immune-profiling analysis of immune repertoires from a range of sample types, including FFPE, where sample RNA may be both limited in quantity and degraded in quality.

Ethics Approval

This study was approved by the University of University of Pittsburgh's Ethics Board.

Correspondence: Sameh El-Difrawy (Sameh.El-Difrawy@thermofisher.com)

Background

Comprehensive genomic profiling using next-generation sequencing (NGS) has become an essential tool to support routine clinical research in oncology. Advent of cancer immunotherapies also requires assessment of immune checkpoint inhibitor biomarkers such as microsatellite instability (MSI) and tumor mutational burden (TMB).

MSI arises from defects in the mismatch repair (MMR) system and is associated with hypermutability of short DNA sequence repeats, microsatellite locations, throughout the genome. Such defects are commonly observed in colorectal, gastric and endometrial cancers and have been shown to be predictive of response to immunotherapy treatment. Traditionally MSI testing has been done using single biomarker tests such as PCR/fragment analysis or immunohistochemistry (IHC) that require high sample input and are time consuming. Therefore, we developed an RUO NGS solution appropriate for FFPE tissues that addresses biomarkers for targeted and immune checkpoint therapies.

Methods

The performance of our RUO NGS based MSI approach was tested in the context of a large Ion AmpliSeq™ panel composed of more than 13,000 amplicons covering 500+ genes. The content includes a diverse set of microsatellite markers targeting MSI locations comprised of mono- and di-nucleotide repeats that range from 7 to 34 bp. Sequencing was carried out on the Ion 550™ chip and the Ion GeneStudio™ S5 system. In-sample standards were designed and incorporated as internal references utilized by the analysis pipeline and a novel algorithm was developed that leverages the unique signal processing properties inherent in semi-conductor sequencing. The test provides results for individual microsatellites and generates an MSI score and status for the sample of interest.

Results

The performance of the MSI solution was tested using a set of over 400 FFPE and cell-line samples from different tissue types and showed excellent concordance with orthogonal tests. We report on the sensitivity and specificity of our tumor only approach and propose ideas to utilize generated MSI score in combination with other bio markers.

Conclusions

An NGS assay was developed to support comprehensive genomic profiling and routine clinical research in oncology. The assay design and unique informatics workflow support precise characterization of mutational signatures and provides normalized MSI and TMB estimates. The performance of the assay was verified over a large cohort of colorectal, gastric and endometrial cancer samples with MSI status independently assigned by orthogonal tests. [For Research use Only. Not for use in diagnostic procedures]

Correspondence: Alicia Gingrich (agingrich@ucdavis.edu)

Background

The TYRO3, AXL and MERTK (TAM) receptor tyrosine kinase (RTK) family have been associated with a number of human cancers, including melanoma.[1-3] Effects attributed to oncogenesis and metastasis (epithelial-to-mesenchymal transition) of the TAM receptors have been described.[2] Recent evidence correlating obesity with a paradoxical improved response to immunotherapy in melanoma suggests both tumor microenvironment and clinical phenotype play a role in response.[4] Therefore, we sought to build a predictive model of response to therapy from biomarkers, using TAM receptors and conventional markers of checkpoint inhibition such as PD-1. This model was tested in the normal weight, overweight and obese populations.

Methods

TCGA-SKCM melanoma tumor mRNA expression and clinical data for metastatic melanoma patients were downloaded from the GDC legacy archive (https://portal.gdc.cancer.gov/legacy-archive) (n = 471).[5] Biomarkers were defined as “high” or “low” expression in each patient. Differences in Kaplan-Meier survival curves based on level of expression were tested using G-rho family tests. Strength of relationships between biomarkers were measured using Pearson’s correlation. All statistical analysis were performed using R package “survival”.

Results

Normal weight, overweight and obese patients had markedly different biomarker profiles associated with survival (Figure 1). In the normal weight population, high CD8 (p=0.0093), PD1 (p=0.0093) and CD84 (p=0.022) were associated with improved survival. In the overweight population, high CD8 (p=0.0098), PD1 (p=0.0004) and CD84 (p=0.0081) were associated with improved survival, while high Gas6 (p=0.029) and MERTK (p=0.043) were associated with decreased survival. And in the obese population, high AXL expression was associated with improved survival (p=0.004), while CD8 (p=0.91) and PD1 (p=0.89) demonstrated no association. In correlation analysis, AXL expression was most closely associated with macrophage markers CD163 (r=0.52), CD84(r=0.56) and MS4A4A(r=0.53) in the obese but not the normal weight population.

Conclusions

Taken together, these data suggest that immunologic response in metastatic melanoma patients is driven by separate immune profiles for obese and non-obese populations. AXL appears to mediate response in the obese population by a macrophage-driven mechanism as opposed to T cell mediation. Collectively, the significant differences in the transcriptomic profiles between obese and non-obese patients suggest potential clinical implications regarding targets for treatment and application to patients based on clinical phenotype.

References

1. Dransfield I, Farnworth S. Axl and Mer receptor tyrosine kinases: distinct and nonoverlapping roles in inflammation and cancer? Adv Exp Med Biol. 2016;930:113-32.

2. Verma A, Warner SL, Vankayalapati H, et al. Targeting Axl and Mer kinases in cancer. Mol Cancer Ther. 2011;10:1763-73.

3. Wu X, Liu X, Koul S, et al. AXL kinase as a novel target for cancer therapy. Oncotarget. 2014;5:9546-9563.

4. McQuade JL, Daniel CR, Hess KR, et al. Association of body-mass index and outcomes in patients with metastatic melanoma treated wtih targeted therapy, immunotherapy, or chemotherapy: a retrospective, multicohort analysis. Lancet Oncol. 2018;19:310-322.

5. Guan J, Gupta R, Filipp FV. Cancer systems biology of TCGA SKCM: efficient detection of genomic drivers in melanoma. Sci Rep. 2015;5:7857.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P89).

KM curves by clinical phenotype and biomarker

Correspondence: Sean Glenn (sean.glenn@omniseq.com)

Background

Tumors often do not respond to PD-1/PD-L1 axis inhibitors due to immune escape mechanisms present in the tumor microenvironment. Bi-functional antibody-based immunotherapies that simultaneously target immune checkpoints and immunosuppressive cells are being developed to slow tumor growth. Anti-PD-L1/ TGF-β trap fusion proteins are one approach being tested to counter the traditional immune checkpoint inhibition via PD-1/PD-L1 axes and simultaneously inhibit the pro-tumor/anti-inflammatory effects of TGF-β. In this study, we not only describe the tumor immune microenvironment of tumors expressing PD-L1 and TGF-β, but also describe potential patient selection strategies based on gene expression measurements of these tumor immune microenvironments from clinical samples.

Methods

RNA-seq was performed for 395 immune transcripts on 1323 FFPE tumors of diverse histologies. To find true TGF-β high expressing tumors, TGF-β gene expression was normalized by a tumor inflammatory score (average expression rank of 161 inflammation genes derived from co-expression signature of 1323 tumors spanning 35 tumor histologies). Proportion of PD-L1 IHC positive, tumor mutational burden (TMB) high and cell proliferation categories was estimated for TGF-β high expressing tumors. Inclusion and exclusion criteria were developed based on PD-L1 and normalized TGF-β expression.

Results

Gene expression revealed varying degrees of TGF-β high tumors in all tumor types studied. Sarcoma, pancreatic cancer and breast cancer had the highest proportion of TGF-β high tumors. Within these TGF-β high tumors, 41% were PD-L1 IHC+ (TPS≥1%), and 28% were TMB-high. 11% (n=147/1323) tumors were both TGF-β high and PD-L1 high making these tumor microenvironments ideal for a potential PD-L1/TGF-β trap treatment. Interestingly, 47% (n=69/147) of these tumors presented with strong/moderate inflammation, with 53% (n=78/147) being non-inflamed tumors. Conversely, there were 11.7% (n=155/1323) tumors that were TGF-β low and PD-L1 low presenting suboptimal tumor micro-environment for a potential treatment. Notably, only 26% (n=40/155) of these tumors presented with strong/moderate inflammation with clear majority (74%; n=115/155) being strongly or moderately inflamed tumors.

Conclusions

This large clinically tested tumor cohort suggests an immune phenotype of potentially PD-L1/TGF-β trap responsive tumors exists across multiple histologies. PD-L1/TGF-β high tumors have distinct immune profiles compared to PD-L1/TGF-β low tumors. A clinical immune gene expression assay described in this study could not only improve patient selection for anti-PD-L1/TGF-β trap treatment, but for other bi-specific fusion protein based immunotherapies.

Ethics Approval

De-identified specimens and data were analyzed by OmniSeq under IRB approved protocol BDR 080316 (Roswell Park Comprehensive Cancer Center, Buffalo, NY).

Correspondence: Christopher Kubu (chris.kubu@fluidigm.com)

Background

RNA sequencing (RNA-seq) provides hypothesis-free profiling of transcript levels and isoforms. This profiling captures a comprehensive view of the peripheral immune system or the tumor microenvironment. The resulting profiles can be used to characterize differential gene expression patterns that can further the understanding of the immune system. To further enable these types of studies we have developed a highly cost-effective, nanoliter-volume microfluidics-based workflow and chemistry compatible with Illumina® sequencing instruments to simultaneously generate RNA-seq libraries from up to 48 samples. This method fully automates solid-phase capture of polyadenylated RNA, reverse transcription, and index PCR within a compact nanoscale integrated fluidic circuit (IFC) on our Juno™ system. The workflow includes reagents necessary to generate full-length, random-primed RNA-seq libraries from as little as 10 ng of total RNA, while preserving strandedness information.

Methods

Multiple replicates of 10 ng and 100 ng of total RNA from control samples spiked with ERCC RNA Spike-In Mixes were used to prepare RNA-seq library using the Advanta™ RNA-Seq NGS Library Prep Kit. The performance was compared to a conventional library preparation kit. We also used our platform to profile total RNA purified from FACS-sorted CD3+, CD8+, CD28–, and CD25+/hi T suppressor cells.

Results

RNA-seq libraries from control RNAs at both 10 and 100 ng input have less than 10% rRNA reads, replicate correlations greater than 99%, and gene-level and transcript-level detection rates that are highly concordant with a conventional library preparation kit. Additionally, the data confirms comparable dynamic range and linearity of response of the ERCC spike-in controls. Libraries prepared from FACS-sorted T cells show differential expression profiles consistent with the expected patterns.

Conclusions

The Advanta RNA-Seq NGS Library Prep workflow simplifies the high-throughput generation of RNA-seq libraries, significantly minimizing hands-on time and costly reagent consumption, which will facilitate the incorporation of RNA sequencing into the immune-oncology research toolkit.

For Research Use Only. Not for use in diagnostic procedures.

Correspondence: Komal Gupte-Singh (Komal.Singh@bms.com)

Background

Metastatic non-small cell lung cancer (mNSCLC) patients who receive comprehensive genomic profiling (CGP) at diagnosis may be more likely to receive optimal first line (1L) therapies than patients who receive panel testing (PT) with the enhanced ability to identify biomarkers with associated therapies. The incremental survival benefits of receiving optimal treatments following CGP testing at diagnosis have yet to be estimated.

Methods

A Markov simulation model of biomarker testing and treatment assignment for mNSCLC was built to estimate the survival outcomes associated with CGP versus PT. Biomarkers identified with PT were EGFR, ALK, ROS1, BRAF, and PD-L1 (≥50%). All biomarker tests were tested simultaneously using single gene testing or assay. CGP, which employed Next-Generation Sequencing, identified all the above biomarker changes and estimated tumor mutational burden (TMB). The model assumed that PD-L1 testing was conducted together with CGP. Biomarker identification, except for TMB and PD-L1, was assumed to be mutually exclusive and to occur at published prevalence rates. Incremental false-negative rates of each genetic test in PT relative to CGP were applied. Treatment pathways followed NCCN guidelines and current published clinical trial results. Key inputs and assumptions were tested in sensitivity analyses.

Results

Patient overall survival for each biomarker test within each testing strategy are shown in Table 1. Patients receiving CGP had 8.5% (1.4 months) longer survival on average than those who received PT. Patients receiving CGP testing at presentation spent more time on 1L therapies (40% vs. 33%), thereby less time on 2L therapies (23% vs. 26%) compared to patients receiving PT at presentation.

Conclusions

CGP testing among mNSCLC patients at the time of diagnosis resulted in survival gains in comparison to PT due to higher proportion of patients receiving optimal 1L treatment.

Correspondence: Davide Bedognetti (dbedognetti@sidra.org)

Background

The immune system has a substantial effect on the progression of colon cancer. Typically, an immune response defined by a polarized Th1 phenotype, characterized by expression of chemokine-receptor ligands, activation of interferon-stimulated genes, production of cytotoxic molecules by effector immune cells, and upregulation of immune regulatory genes, has been associated with immune-mediated tumor rejection. We have previously introduced a gene signature, called Immunology Constant of Rejection (ICR), that reflects these immune components.[1–4] This signature was able to differentiate quite well the patients with an active immune environment and improved survival vs those who did not[5].

Virtually, all correlative analyses integrating exome and transcriptomic data in colon cancer based on publicly available date use the TCGA cohort. Although it is broadly accepted that T-cell infiltration influences prognosis in colon cancer[6], the association between transcriptomic immune signature and patient survival could not be observed in the TCGA colon cancer cohort. This is likely due to the per protocol exclusion of samples with low tumor purity (i.e., higher stromal/immune infiltration), as at that time TCGA consortium focused on defining cancer genetic makeup. To gain more insight into the underlying mechanism of cancer tissue rejection by the immune system in colon cancer, we build an extensive data repository from high quality snap frozen colon cancer samples unbiased for tumor purity.

Methods

RNA and DNA were isolated from a cohort of 366 colon cancer patients collected over the last decade at the University of Leiden Medical Center (LUMC), Netherlands. Tissue sections flanking the corresponding samples were hematoxylin- and eosin-stained. RNA-seq (HiSeq4000) data was obtained using HISAT2 alignment[7] and quantile normalized after GC-correction of the raw counts.[8] Whole Exome Sequencing (WES) (>100X) was performed for normal and cancer tissue (366 RNA-seq and 608 WES). T-cell repertoire was analyzed using Adaptive immunoSEQ in 125 samples. Tumor immune phenotype classification was done using unsupervised consensus clustering based on the expression of ICR genes.

Results

We have built one of the most extensive high-quality datasets for immunogenomic alterations available so far in colon cancer. Our preliminary data supports a positive impact of ICR gene expression in colon cancer cohort: patients with a Th-1 polarized microenvironment display better survival. Integrative analysis encompassing somatic mutation, copy number variations, and transcriptome is ongoing and will be presented at the conference (Figure 1).

Conclusions

This newly generated immune centric NGS dataset, generated in Qatar, will contribute dramatically to elucidating the genetic determinants of immune responsiveness in cancer.

Acknowledgements

This work was supported by Qatar National Research Fund (QNRF) with grant JSREP07-012-3-005

References

1. Wang, E., Worschech, A. & Marincola, F. M. The immunologic constant of rejection. Trends Immunol. 29, 256–262 (2008).

2. Spivey, T. L. et al. Gene expression profiling in acute allograft rejection: challenging the immunologic constant of rejection hypothesis. J. Transl. Med. 9, 174 (2011).

3. Bertucci, F. et al. The immunologic constant of rejection classification refines the prognostic value of conventional prognostic signatures in breast cancer. Br. J. Cancer (2018). doi:10.1038/s41416-018-0309-1

4. Galon, J., Angell, H. K., Bedognetti, D. & Marincola, F. M. The Continuum of Cancer Immunosurveillance: Prognostic, Predictive, and Mechanistic Signatures. Immunity 39, 11–26 (2013).

5. Roelands, J. et al. Genomic landscape of tumor-host interactions with differential prognostic and predictive connotations. bioRxiv 546069 (2019). doi:10.1101/546069

6. Pagès, F. et al. International validation of the consensus Immunoscore for the classification of colon cancer: a prognostic and accuracy study. The Lancet 391, 2128–2139 (2018).

7. Kim, D., Langmead, B. & Salzberg, S. L. HISAT: a fast spliced aligner with low memory requirements. Nat. Methods 12, 357–360 (2015).

8. Risso, D., Schwartz, K., Sherlock, G. & Dudoit, S. GC-Content Normalization for RNA-Seq Data. BMC Bioinformatics 12, 480 (2011).

Ethics Approval

Sidra Medicine IRB approval : #1602002725

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P93).

The advanced immune-centric NGS cohort for colon cancer

Correspondence: Taigo Kato (kato@uro.med.osaka-u.ac.jp)

Background

With the spread of usage of Immune checkpoint inhibitors (ICIs), a certain number of patients face discontinuation of ICIs due to severe immune-related adverse events (irAEs). Recently, some reports have shown encouraging efficacy among patients who discontinued ICIs, leading to the hypothesis that irAEs-experienced patients have strong and long-lasting anti-cancer immune responses. So far, the molecular mechanisms of the immune response, particularly for T cells that play pivotal roles in attacking cancer cells, still remain unclear. Thus, characterization of T cell repertoire and immune signatures in peripheral blood mononuclear cells (PBMCs) and tumors before and after ICIs treatment should contribute to better understanding of irAEs-related anti-cancer immune responses.

Methods

In this study, we collected PBMCs from 4 urological cancer patients, before ICIs treatment and at the onset of severe irAEs. For 1 kidney cancer patient who had long durable response after discontinuation of ICIs, we also collected metastatic tissue sample and applied a next generation sequencing approach to characterize T cell receptor (TCR) repertoires using RNAs isolated from tumors and PBMCs. We also measured mRNA expression levels of immune-related genes in the PBMCs of pre- and post-ICIs treatment.

Results

We found that elevated transcriptional levels of CD3, CD4, CD8, GZMA, PRF1, and FOXP3 along with high GZMA/CD3 and PRF1/CD3 ratio in the peripheral blood at the onset of irAEs. TCR repertoire analysis revealed drastic expansion of certain T cell clones in metastatic tissue after irAEs (Figure 1). Interestingly, some of these abundant TCR clonotypes were also increased in peripheral blood at the onset of irAEs (Figure 2).

Conclusions

Our findings revealed that a certain number of expanded- and irAEs-related T cell clones in cancer tissue may also circulate systemically and then attack tumor cells in distant regions, leading to durable response in the patients with irAEs.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P94).

Clonal T cell expansion in pancreatic metastasis

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P94).

Expanded T cells in metastatic site are detected in systemic

Correspondence: Namit Kumar (Namit.Kumar@bms.com)

Background

Elucidating biomarkers associated with immunotherapy response and resistance will allow for better informed patient selection and treatment decisions as well as enhanced drug development strategy. Current biomarker strategies are based on cellular markers (eg, immunohistochemistry) or bulk molecular averages (eg, whole-exome sequencing). However, there is limited ability to integrate cellular and molecular data. Single-cell RNAseq (scRNAseq) is a promising technology allowing for an unbiased analysis of the tumor microenviroment (TME) at cellular resolution. Despite the immense potential, implementation of this technology in clinical trials has been limited due to lack of methodologies applicable to clinically relevant specimens such as core-needle biopsies (CNB). Here, we describe the development of clinically applicable scRNAseq technology and analysis.

Methods

Treatment-naïve commercially sourced tumor resections were used to generate ex-vivo CNB for scRNAseq analysis with 10X genomics. Post-clustering, unsupervised cell-type identification was performed (SingleR), and downstream analyses were carried out (Seurat v2, custom R). Cells from multiple patients/tumor types (endometrial, TNBC, NSCLC, ccRCC, gastrointestinal; n=8), and healthy donors (peripheral blood mononuclear cells; n=3) were combined, batch-corrected and aligned using canonical correlation analysis (CCA); and differential gene expression was performed (MAST algorithm).

Results

CNB scRNAseq was optimized across 5 tumor types, and the resulting data from ~43,000 cells allowed for the unbiased identification of TME cellular components (stromal, epithelial, immune-cell subtypes). The cellular resolution of this dataset allowed us to identify cell populations with distinct gene signatures. For example, we identified 2 macrophage subclusters—a lung tumor-specific cluster and a tumor-independent cluster. Lung-specific macrophages showed upregulation of genes including SPP1, G0S2, RGCC, PHLDA1, and TREM. Differential gene expression analysis evaluated similarities and differences between TME vs healthy PB cells and allowed for surrogate pharmacodynamics marker assessment. In our analysis, 1197 genes were differentially expressed; the most enriched genes in tumor-derived monocytes included HSPA1A, IL8, APOE, and SPP1 whereas PB monocytes were enriched for genes including LGALS2, S100A12, S100A9, AHNAK, and CSTA.

Conclusions

We have demonstrated the feasibility of scRNAseq from single CNB through the development of protocols to enable identification of biomarkers related to pharmacodynamics, therapeutic response, or disease progression. Further, we have optimized the bioinformatics workflow to derive meaningful biological insights from these scRNAseq datasets, such as mechanisms involved in immune response or resistance that are tumor extrinsic or intrinsic. Our pilot study sets the groundwork to explore including scRNAseq in future prospective clinical studies.

Acknowledgements

Bristol-Myers Squibb.

Correspondence: Richard Chen (richard.chen@personalis.com)

Background

The promise of immunotherapy has revealed the need for comprehensive profiling of the tumor and its immune microenvironment, including analysis of the T-cell receptor (TCR) repertoire. To address this challenge, we developed ImmunoID NeXT to provide a more comprehensive view of the tumor and tumor microenvironment (TME) from limited FFPE tumor biopsies. This includes profiling both the TCR alpha and beta chains. We show that ImmunoID NeXT accurately and reproducibly profiles abundant clones and provides information on the diversity of T-cells in tumor samples.

Methods

We first analyze the reproducibility of ImmunoID NeXT using replicates of PBMCs. Then, we compare the concordance of clones from ImmunoID NeXT to the top clones from a standalone TCR sequencing approach. We also analyze the reproducibility of clones in patient-derived FFPE samples, and compare to IHC quantification of CD3+ cells to highlight the intra-sample heterogeneity of T-cell abundance and diversity. We then analyze the clonal diversity of pre-treatment tumor samples in a cohort of melanoma patients who underwent PD-1 blockade. Finally, we use ImmunoID NeXT to profile the clonal diversity across over 100 solid tumor samples.

Results

Abundances of clones shared between replicates of PBMC samples have a very high concordance (R2>0.99 with both TRA and TRB). Compared to the standalone TCR approach, we identify over 96% of the top 1000 TRA clones, and over 99% of the top 1000 TRB clones, both with highly concordant abundances (R2>0.95 and R2>0.94 in TRA and TRB, respectively).

Subsequent curls of a tumor FFPE sample also have a high concordance of clonal abundances (R2>0.89 and R2>0.91 in TRA and TRB, respectively). TCR sequencing also provides a view of the clonal diversity of T-cells in a sample, which is not available with quantification via IHC. Finally, in a melanoma cohort, clonality based on either TRA or TRB is significantly different in responders to checkpoint inhibition.

Conclusions

The ImmunoID NeXT platform can provide insight into the diversity of the immune repertoire, highlighting the platform’s ability to provide comprehensive analysis of both the tumor and tumor microenvironment. We demonstrate that ImmunoID NeXT is reproducible, sensitive, and accurate at profiling high-abundance TRA and TRB clones, as well as feasible with FFPE samples. We also highlight how immune repertoire results from ImmunoID NeXT can be used to gain understanding about the immunological composition of the TME. Finally, we show how ImmunoID NeXT can profile the diversity of the TCR repertoire in tumor samples.

Correspondence: Philip Jermann (philipmartin.jermann@usb.ch)

Background

There is an outstanding need to identify predictive biomarkers for response to anti-PD-1 monotherapy for NSCLC. Here we investigated TCRB clonal expansion and TCR convergence within the pretreatment tumor microenvironment as predictors of response in a cohort of 37 FFPE-preserved biopsies. For context, we compared the predictive value of these features with TMB values from the same tumors.

Methods

Total RNA from FFPE-preserved pretreatment NSCLC biopsies (11 responders, 14 non-responders) was extracted for TCRB repertoire sequencing via the Oncomine TCRB-SR assay (15-265ng RNA input; average 164ng) and the Ion Torrent Gene Studio S5. TMB values were obtained from FFPE-preserved gDNA from the same biopsies using the Oncomine Tumor Mutation Burden Assay. TCR convergence and clonal expansion were evaluated independently or in a combined model as predictors of response.

Results

TCRB sequencing revealed increased TCR convergence (p = .02, Wilcoxon) and clonal expansion (p = .06, Wilcoxon) in those who benefited from anti-PD-1 therapy. A logistic regression classifier combining both features was able to discriminate responders from non-responders with a sensitivity of .91 and specificity of .71 at the optimal cutoff, per the Youden’s J method. The TCR-based classifier was able to identify responders who otherwise had low to intermediate (<10muts per Mb) TMB.

Conclusions

TCRB clonal expansion and convergence warrant further evaluation as potential predictive biomarkers of response. Importantly, TCRB sequencing may allow for identification of responders who are otherwise missed by TMB-based stratification.

Correspondence: Timothy Looney (timothy.looney@thermofisher.com)

Background

Identifying the optimal input amount and sequencing depth for B and T cell receptor repertoire profiling is challenging owing to variation in material quality and lymphocyte diversity in blood and FFPE preserved specimens. Rarefaction analysis has emerged as a potential approach for assessing whether immune repertoire libraries have been sequenced to saturation. Here we present a novel automated method for saturation analysis of IGH and TCRB chain libraries derived from sequencing of peripheral blood leukocytes (PBL) and FFPE-preserved RNA and DNA.

Methods

Human TCRB and IGH repertoire libraries were generated using the Oncomine TCRB-SR and BCR IGH-SR assays from: (1) 25ng PBL total RNA (2) 500ng PBL gDNA (3) 150ng RNA from FFPE preserved NSCLC and (4) 200ng gDNA from FFPE preserved brain tissue. Mouse TCRB and IGH libraries were generated using the Ion Ampliseq TCRB-SR and BCR IGH-SR assays and 25ng RNA or 500 ngDNA derived from spleen or lymph node. Libraries were sequenced on the Ion Torrent Gene Studio S5 then analyzed with Ion Reporter to identify clonotypes, quantify clonal expansion and diversity, and for IGH chain libraries, identify B cell clonal lineages and assess isotype usage. We then repeated clonotyping and analysis of secondary repertoire features using data that had been downsampled to fixed read depths.

Results

We observed an asymptotic relationship between the sequencing depth and the number of B and T cell clones detected, clone Shannon diversity, and B cell clonal lineage richness and diversity, indicating that libraries had been sequenced to saturation. By contrast, T and B cell normalized Shannon entropy appeared robust to sequencing depth.

Conclusions

Automated downsampling analysis may serve as a convenient tool for optimizing sequencing depth and input amount for B and T cell repertoire sequencing studies. We expect this approach to become a routine component of immune repertoire analysis.

Correspondence: Luca Mazzarella (luca.mazzarella@ieo.it)

Background

Tumor mutational burden (TMB) is increasingly proposed as a predictive biomarker for immunotherapy response in cancer patients.

TMB assessed by Whole Exome Sequencing (WES) is considered the gold standard but remains confined to research settings. Targeted enrichment panels of various genomic sizes are emerging as a more sustainable methodology for assessing TMB in the clinical setting. However, panel-based TMB quantification has not been adequately standardized to date, leading to major heterogeneities in TMB measurement and a lack of uniformly accepted cutoff values, thus limiting the possibility to transfer results across settings. In particular, the choice of variants to include in TMB calculation (synonymous, cancer driver genes or low-allelic frequency mutations, or other features) may strongly affect results and in particular TMB predictive value [1]

Methods

We developed "TMBler", an R package to calculate TMB from targeted sequencing panels. TMBler allows to select multiple filters on mutation counts for TMB quantification. It also includes a set of functions to simulate custom panels on WES and calculate predictive value based on available data on immunotherapy response matched with sequencing data [2,3]. Finally, it allows to measure panel-based TMB concordance with WES-based TMB and its predictive value using Receiver Operating Characteristic (ROC) curves.

Results

By simulating custom and commercially available panels, we show that the application of specific filter combinations can significantly influence TMB calculation and its predictive value, and we identify instances where risk of erroneous assignment of patients to responder/nonresponder groups is highest

Conclusions

TMBler is a useful tool for quantifying TMB from targeted panels. It can analyze performance of existing panels, optimize analytical pipeline and design novel custom panels through simulations.

References

1. Fancello L, Gandini S, Pelicci PG, Mazzarella L. Tumor mutational burden quantification from targeted gene panels: major advancements and challenges. J Immunother Cancer. 2019 Jul 15;7(1):183

2. Hellmann MD, Nathanson T, Rizvi , et al. Genomic Features of Response to Combination Immunotherapy in Patients with Advanced Non-Small-Cell Lung Cancer. Cancer Cell. 2018

May 14;33(5):843-852

3. Samstein RM, Lee CH, Shoushtari AN. Tumor mutational load predicts survival after immunotherapy across multiple cancer types. Nat Genet. 2019 Feb;51(2):202-206

Correspondence: Sarbajit Mukherjee (sarbajit.mukherjee@roswellpark.org)

Background

Obesity is associated with an elevated risk of GEAC [1], but the molecular mechanism remains unknown. Paradoxically, however, obesity is associated with a superior response to anti-PD-1 treatment [2,3]. This may be explained by our recent observations that obesity enhances PD-1 mediated T-cell dysfunction in a mechanism involving leptin signaling. Prompted by this data, we aimed to identify obesity/leptin-regulated molecular biomarkers in GEAC.

Methods

Based on the body-mass index (BMI), we categorized patients into normal (BMI 18-24.9), overweight (BMI 25-29.9) and obese (BMI ≥30). We then retrospectively analyzed the clinical report of PD-L1 staining by IHC_22C3/Keytruda from metastatic GEAC patients treated at our institution between 2014-2019. Chi-squared test was used to determine the association between categorical variables. Next, we performed RNA-seq analysis of 13 gastric cancer FFPE specimens (8 obese and 5 normal weight) to identify differential gene expression between these two groups. Gene expression was quantified by log-fold changes. Differentially expressed genes were identified by using DESeq2. Then we looked at the association between the expression of leptin and immune-related genes from those specimens, using generalized linear model implemented in DESeq2. TCGA gastric cancer database (TCGA -STAD) was used to validate these associations (using Pearson test) independently. A p-value of <0.05.

Results

Our analysis of the clinical report of 77 patients with metastatic GEAC revealed that patients with a BMI =>25 were more likely to express PD-L1 than normal-weight individuals (p = 0.03)(Table 1). Our RNA-seq analysis identified the following genes to be up-regulated in the obese group: NOS2, FOXP3, IDO1, EOMES, CD160, and CXCR5 (p<0.05). Expression of these genes was positively correlated with leptin in our database; however, these associations did not reach statistical significance; possibly due to our small sample size. The same analysis within the TCGA-STAD database identified a strong positive correlation between the expression of all six genes and leptin (p <0.05)(Figure 1). GSEA identified several up-regulated immune-related pathways (Adaptive Immune System, Antigen Processing Cross Presentation etc.) in the obese group.

Conclusions

Our preliminary data suggest that obesity, and specifically leptin, is associated with several immune markers in GEAC. Our mechanistic studies will explore how obesity/ leptin regulates the immune system and promotes cancer. These studies may allow us to identify new leptin regulated pathways as therapeutic targets.

References

1. Garai J, Uddo RB, Mohler MC, Pelligrino N, Scribner R, Sothern MS et al. At the crossroad between obesity and gastric cancer. Methods Mol Biol.2015; 1238:689-707.

2. Ibrahimi S, Mukherjee S, Roman D, King C, Machiorlatti M, Aljumaily R. Effect of Body Mass Index and Albumin level on Outcomes of Patients Receiving Anti PD-1/PD-L1 Therapy. J Clin Oncol 36, 2018 (suppl 5S; abstr 213).

3. Wang Z, Aguilar EG, Luna JI, Dunai C, Khuat LT, Le CT et al. Paradoxical effects of obesity on T cell function during tumor progression and PD-1 checkpoint blockade. Nat Med. 2019 Jan; 25(1):141-151.

Ethics Approval

The study was approved by the Institutional Review Board at Roswell Park Comprehensive Cancer Center, approval number STUDY00000894 / BDR 109419.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P100).

See text for description

Correspondence: Xavier Solé (x.sole@iconcologia.net)

Background

Malignant pleural mesothelioma (MPM) is a rare and aggressive neoplasia. Immune checkpoint inhibitors in MPM demonstrated modest efficacy, partly due to lack of predictive biomarkers of clinical benefit from immunotherapy. The aims of this work were: to identify immune fractions associated with clinical outcome; to stratify MPM patients based on their immune contexture and to characterize the immune-based groups at the genomic and transcriptomic levels.

Methods

Seven gene-expression datasets of MPM were used to assess the immune microenvironment of 516 samples. The abundance of 20 immune fractions in each sample was inferred using Gene Set Variation Analysis. Identification of clinically-relevant fractions was performed with Cox Proportional-Hazards Models adjusted for age, stage, sex, and tumor histology.

Results

T-Helper 2 (Th2, HR=2.14, p=1.5x10-4) and cytotoxic T cells (CTC; HR=0.57, p=9.1x10-3) were found to be consistently associated with overall survival in multiple datasets. Three immune clusters (IG) were subsequently defined based on Th2 and CTC immune infiltration levels: IG1 (54.5% of samples) had high Th2/low CTC levels, IG2 (37%) had either low or high levels of both fractions, and IG3 (8.5%) had low Th2/high CTC levels. Immune clusters were associated with overall survival independently of tumor histology, with an improving survival from IG1 to IG3 (HR IG2=0.52, 95% CI 0.39–0.69; HR IG3=0.32, 95% CI 0.19–0.53; p=8.4x10-8; Figure 1). IG3 was significantly enriched in epithelioid tumors (90% IG3 vs. 62% IG1, p=0.001) and patients were younger compared to the other groups (60 years IG3 vs. 66 years IG1, p=0.021). These groups showed differential molecular profiles, being IG1 enriched for CDKN2A and IFN-related genes deletions. No statistically significant differences in the tumor mutational burden was observed, howerver IG3 tumours had fewer mutations than IG1 and IG2 groups. At the transcriptional level, IG1 samples showed upregulation of cell proliferation and DNA repair-related gene-sets, while IG3 samples presented upregulation of immune checkpoint inhibitors (Figure 2) and inflammation-related pathways. Finally, integration of gene expression with functional signatures of in vitro drug response showed that IG3 patients are more likely to respond to immune checkpoint inhibitors, while IG1 patients might be more sensitive to PARP inhibitors.

Conclusions

Analysis of publicly available gene-expression data of MPM reveals three major immune-based groups, based on Th2 and CTC composition. These clusters are associated with distinct genomic profiles and clinical outcome. Further validation of this classification is warranted in an independent cohort of MPM.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P101).

Overall survival analysis according to the immune groups

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P101).

Expression of immune checkpoint markers

Correspondence: Sarabjot Pabla (sarabjot.pabla@omniseq.com)

Background

Cancer testis (CT) antigens are tumor antigens that have a highly tissue restricted expression in germ cells but are often expressed in diverse malignancies. With their highly immunogenic expression limited to tumor cells, CT antigens have become a prime target for cancer vaccinations and T-cell based therapy with chimeric T-cell receptors. In this study, we investigated the association of two CT antigens (NY-ESO-1 and LAGE-1a) with the immune microenvironment of real-world clinical tumors spanning multiple histologies. Furthermore, we describe the association of CT antigens with traditional biomarkers of immunotherapy such as PD-L1 immunohistochemistry (IHC) and tumor mutational burden (TMB), with inflammatory status and cell proliferation status with confirmatory studies performed on a large TCGA pan-cancer cohort of 11,001 tumors.

Methods

Unsupervised clustering was performed on gene-expression data of 395 immune transcripts of 1323 FFPE tumors to reveal three inflammatory patient clusters and three distinct gene groups; CT-antigen, inflammatory and neoplastic clusters. Test for proportions was performed using Pearson’s chi-squared test to describe association of NY-ESO-1 and LAGE-1a with PD-L1 IHC, TMB, inflammatory cluster and cell-proliferation. A retrospective cohort (n=242) of checkpoint inhibition (CPI) treated tumors was utilized to perform overall survival (Kaplan-Meier curves) and response to CPI therapy for CT antigen+ tumors. Survival analysis was confirmed against the Pan-Cancer TCGA cohort (n=11,001).

Results

Unsupervised clustering showed clear co-expression sub-clustering of CTA genes differentiated from “immune” and from “neoplastic expression”. PD-L1 IHC status was not associated with NY-ESO-1 (p=0.71) or LAGE-1a (p=0.52) status. Interestingly, LAGE-1a positive cases were over-represented in TMB high cases (p=0.016), whereas, NY-ESO-1 status was not associated with TMB. NY-ESO-1 positive cases were highly over-represented in non-inflamed cluster (p=0.006), whereas, LAGE-1a status was not associated with inflammation status. Both NY-ESO-1 (p=0.031) and LAGE-1a (p=0.008) were significantly associated with cell-proliferation status. NY-ESO-1 positive tumors have significantly (p=0.014) higher response rate in retrospective cohort but this was not observed for LAGE-1a status. NY-ESO-1 and LAGE-1a status showed trend toward better (p=0.09 and p=0.06 respectively) survival in the retrospective and TCGA pan-cancer cohort.

Conclusions

This study presents an in-depth analysis of the immune landscape of CT antigen positive tumors across multiple histologies. CT antigen bearing tumors not only have unique immune profiles but also have significant associations with biologically relevant emerging biomarkers such as inflammatory signature, TMB and cell proliferation. CT antigens are a viable target for non-inflamed tumors for checkpoint inhibition therapy.

Ethics Approval

De-identified specimens and data were analyzed by OmniSeq under IRB approved protocol BDR 080316 (Roswell Park Comprehensive Cancer Center, Buffalo, NY).

Correspondence: Jason Perera (jason.perera@tempus.com)

Background

Human leukocyte antigen (HLA) class I proteins are expressed on the surface of all nucleated cells and are vital for immune surveillance. When tumor-specific mutations (neoantigens) are presented on HLA molecules to CD8+ T cells, this recognition can drive immune responses against the tumor and lead to tumor destruction. One mechanism of immune escape for tumors is loss of heterozygosity in HLA genes (HLA-LOH), which reduces the total number of neoantigens available for presentation to T cells. Due to the highly polymorphic nature of HLA, the copy number status of HLA genes is extremely challenging to assess by standard bioinformatics approaches. To investigate the prevalence of HLA-LOH, we developed a specialized pipeline to detect HLA-LOH by DNA next-generation sequencing (NGS).

Methods

A cohort of colorectal and non-small cell lung cancer samples underwent DNA sequencing on the Tempus xT panel using paired, formalin-fixed, paraffin-embedded tumor and normal (blood or saliva) samples. To detect HLA-LOH from NGS data, we used NGS-based HLA typing to resolve the patient’s most probable HLA haplotype. Based on this haplotype, we adaptively realigned reads, extracted a number of features describing the relative allele coverage in the tumor and normal samples, and used these features to make a confident determination of allelic loss in the patient’s tumor sample.

Results

Evidence of HLA-LOH was detected in 16% of non-small cell lung tumor samples and 17% of colorectal tumor samples. We did not observe a significant association between LOH status and tumor mutational burden or neoantigen load. In the colorectal cancer cohort, HLA-LOH was observed in tumor samples classified as microsatellite instability-high (MSI-H); however, the association between HLA-LOH status and MSI status was not statistically significant.

Conclusions

We developed a novel method of determining HLA-LOH by DNA NGS and demonstrated that HLA-LOH is a readily detectable feature in human tumors. These results highlight the complexity of antigen presentation, the potential importance of HLA-LOH as a biomarker of immunotherapy response and resistance, and lays the groundwork for future investigations.

Correspondence: David Page (david.page2@providence.org)

Background

Immune checkpoint blockade is only modestly effective in metastatic breast cancer. One potential contributing factor is chronic lymphodepletion associated with preceding curative-intent chemo. Here, we evaluate the short and long-term effects of chemo on peripheral T-cell counts and clonal diversity in a cohort of breast cancer patients.

Methods

Stage I-III subjects (n=24) receiving curative-intent chemo (doxorubicin, cyclophosphamide, paclitaxel) were monitored longitudinally (mixed effects linear model) with serial peripheral blood mononuclear cell flow cytometry and quantitative immunosequencing of the T-cell receptor β locus (TCRseq) using the immunoSEQ® assay (Adaptive Biotechnologies, Seattle, WA). To evaluate for long-term chemo effects, these analyses were repeated in a cohort of recurrent breast cancer patients who received chemo >12 months preceding analysis (n=9). Wilcoxon rank sum and tests of slope were employed to screen for associations with chemo response, defined as complete pathologic response (pCR) at surgical resection.

Results

By TCRseq, chemo resulted in an acute decline in T-cell fraction (0-8 weeks, p12 months following chemo.

Conclusions

Curative-intent chemo is associated with T-cell death followed by reconstitution, with the resulting T-cell repertoire being more clonal and less abundant in naïve T cells. These findings persist at the time of metastatic recurrence, and therefore may contribute to immunotherapy non-response in metastatic disease. Conversely, we identified T-cell reconstitution as a potential biologic modifier of chemo response. T-cell reconstitution can be therapeutically targeted with inhibitors of androgen receptor signaling, which in experimental models enables thymic maturation of naïve T-cell clones and an increase in peripheral T-cell count. This hypothesis is being evaluated in an ongoing phase II clinical trial of bicalutamide (androgen receptor antagonist) plus ipilimumab and nivolumab in metastatic breast cancer (NCT03650894).

Ethics Approval

The study was reviewed and approved by Providence Heath and Services Internal review board, approval number 15-162.

Correspondence: Nitika Sharma (sharman@ecu.edu)

Background

Tumor mutational burden has emerged as a potential biomarker predictive of response to Immune checkpoint blockade (ICB) in lung cancer. The utility of this biomarker in oncogenic driver mutations, that account for nearly 20-50% of NSCLC, is still unknown. KRAS mutation in lung cancer is a prognostic biomarker whereas EGFR and BRAF pathogenic mutations are predictive of response to tyrosine kinase inhibitors (TKI). ICB with bevacizumab has demonstrated clinical benefit in EGFR mutated lung cancers per IMpower150 clinical trial [1]. TMB analysis between actionable/pathogenic EGFR mutations (i.e. exon 19 del, exon 21 L858R, T790M) and EGFR uncommon/variants mutations may provide therapeutic implications [2]. To explore the immunological basis for these findings, we evaluated the immune biomarker profile of NSCLC patients using Caris next-generation sequencing (NGS) platform.

Methods

We studied tissue samples on 446 patients with NSCLC from 2016-18. TMB was measured by counting all non-synonymous somatic mutations per megabase of the genome coding area using targeted NGS (592 genes). High TMB was defined as ≥ 10 mut/Mb. The analysis was conducted using SAS 9.4. Variables were tested using a Wilcoxon signed-rank test.

Results

KRAS mutations were found in 85 pts (19%), BRAF in 9 pts (2%), EGFR mutation in 36 pts (8%), EGFR pathogenic mutation in 22 pts (5%), EGFR variants in 14pts (3%). The median TMB of KRAS mutant vs KRAS wt (wild type) was 10 vs 7 mut/Mb (range 0-31, p<0.01).

Conclusions

This study highlights the unique immune profile of certain oncogenic driver mutations in NSCLC. Our results show that KRAS and BRAF mutant subsets have a significantly higher TMB than KRAS and BRAF wild type. In addition, EGFR variants have a higher TMB as compared to actionable pathogenic EGFR mutations. These findings could have therapeutic implications in guiding patient selection for ICB and merit a prospective investigation.

References

1. Socinski M, Jotte R,Cappuzzo F. Atezolizumab for First-Line Treatment of Metastatic Nonsquamous NSCLC. N Engl J Med. 2018; 378:2288-2301.

2. Offin M, Rizvi H,Tenet M.Tumor Mutation Burden and Efficacy of EGFR-Tyrosine Kinase Inhibitors in Patients with EGFR-Mutant Lung Cancers. Clin Cancer Res. 2018;1102.

Ethics Approval

The study was approved by ECU Institutional Review Board, approval number UMCIRB 15-001311.

Correspondence: Mark Awad (Mark_Awad@DFCI.harvard.edu)

Background

High tumor mutational burden (TMB) has been associated with response to checkpoint blockade in non-small cell lung cancer (NSCLC) and other malignancies. However, the degree to which TMB changes over time, across anatomical sites, and with intervening treatment remains unknown. To evaluate TMB changes across time points, we compared TMB in tissue specimens from patients with serially-biopsied NSCLC.

Methods

Clinicopathologic characteristics and changes in TMB were analyzed from patients with NSCLC and more than one tissue specimen that had undergone targeted next generation sequencing (NGS, OncoPanel) at the Dana-Farber Cancer Institute. Those representing distinct primary tumors by histologic or genomic analysis were excluded.

Results

193 NSCLC patients with more than one interpretable NGS result were identified; 30 were excluded due to separate primary tumors. Of the 163 remaining patients included in the analysis, the median time between samples was 14 months (range: 0 to 114 months). TMB was higher in current and former smokers (median TMB 10.7 v 6.4 mutations/megabase(Mb); p < 0.0001), patients without an identifiable oncogenic driver mutation (median TMB 14.5 v 8.5 mutations/Mb; p = 0.004), and patients with locoregional disease at the time of diagnosis (median TMB 10.8 v 8.0 mutations/Mb, p = 0.02). TMB correlated closely across all matched tumor pairs (Pearson’s r = 0.85, Figure 1). Significant increases or decreases in TMB were uncommon in paired samples, and we observed no significant change in median TMB with increasing time between specimen collection or with intervening chemotherapy, immunotherapy, radiation therapy, or targeted therapy.

Conclusions

In NSCLC, TMB correlated closely across tumor pairs, and increasing time between sample collections and intervening treatments were not correlated with significant changes in TMB.

Ethics Approval

This study was conducted under Dana-Farber/Harvard Cancer Center Protocol 02-180.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P107).

See text for description

Correspondence: Kristen Strand-Tibbitts (kristen@oncologie.international)

Background

Immune therapies for cancer have generated an enhanced focus on controlling cancer through modulation of biologies associated with the tumor microenvironment, rather than the traditional approach of targeting cancer cell biology. As more and more targeted therapies are designed to modulate the tumor microenvironment, we need a better understanding of microenvironmental heterogeneity in human cancer. Similar to what has been done to describe patient subsets based on their cancer biology using DNA and RNA signatures, we are working to describe patient subsets based on their microenvironment biology. The ultimate goal is to find effective means of identifying patients for novel therapeutic treatments that target biological pathways that regulate the non-neoplastic cells and drive cancer progression.

Methods

RNA from publicly available sources including microarray and RNASeq were analyzed with respect to gene signatures that describe four different microenvironmental phenotypes.

Results

These four microenvironmental subtypes are prognostic, but also show evidence of being predictive to existing modalities of cancer drugs when analyzed in retrospective analysis. We examined the impact of cancer stage on the distribution of these subtypes and find little variation.

Conclusions

Future clinical trials to prospectively test these four unique signatures as predictive biomarkers for therapy need to be designed.

Correspondence: Peter Szabo (Peter.Szabo@bms.com)

Background

Gene expression profiling (GEP) has been used to identify biomarkers of response to immunotherapy. Using a GEP-based inflammation assay, we derived and analytically validated a CD8 signature to assess T-cell infiltration in the tumor microenvironment (TME) [1]. Here, we retrospectively explore the association of the CD8 signature, alone and in relation to established biomarkers PD-L1 and tumor mutational burden (TMB), with clinical response to nivolumab treatment.

Methods

In the phase 2 CheckMate 275 trial, 270 patients with platinum-resistant metastatic urothelial carcinoma (UC) and evaluable tumor PD-L1 expression received nivolumab treatment. Responses were determined by blinded, independent review committee assessments [2]. Minimal follow-up time for the current analysis was ~3 years. T-cell infiltration in the TME was assessed using the CD8 signature and by immunohistochemistry (IHC) using an automated commercial proprietary assay (Dako mouse clone C8/144B; Agilent Technologies Co) [1]. PD-L1 expression on tumor cells was independently assessed by IHC using the PD-L1 IHC 28-8 pharmDx assay (Dako). TMB was measured by whole exome sequencing [3]. Cox proportional-hazards regression assessed the dependence of progression-free survival (PFS) or overall survival (OS), and logistic regression assessed the dependence of objective response (OR) on biomarker values. The linear effects of biomarkers and their multiplicative interaction were included when multiple biomarkers were evaluated. Likelihood-ratio tests (2-sided) were used to assess biomarker interaction effects. Associations with PFS and OS were investigated using Kaplan–Meier analyses with biomarker scores categorized by tertile.

Results

GEP was evaluable in 205 (76%) and GEP+TMB in 113 (42%) of 270 treated patients. Baseline characteristics, OR, PFS, and OS were similar between all treated patients and the GEP-evaluable cohort. CD8 signature scores showed a positive association with OR (P=0.005), PFS (P=0.005), and OS (P

Conclusions

These results suggest that the GEP-based CD8 signature may have utility as a potential biomarker for predicting clinical response to nivolumab treatment. The CD8 signature may be used alone and/or in combination with other relevant and potentially independent biomarkers such as PD-L1 expression or TMB to identify patients likely to benefit from anti–PD-1 therapies.

Acknowledgements

Bristol-Myers Squibb. Professional medical writing and editorial assistance were provided by Katerina Pipili, PhD, and Jay Rathi, MA, of Spark Medica Inc, funded by Bristol-Myers Squibb.

Trial Registration

NCT02387996

References

1. Szabo PM, Qi Z, Zerba K, et al. Association of an inflammatory gene signature with CD8 expression by immunohistochemistry (IHC) in multiple tumor types. J Clin Oncol. 2019; 37(Suppl): Abstract 2593.

2. Sharma P, Retz M, Siefker-Radtke A, et al. Nivolumab in metastatic urothelial carcinoma after platinum therapy (CheckMate 275): a multicentre, single-arm, phase 2 trial. Lancet Oncol. 2017; 18:312-322.

3. Galsky M, Saci A, Szabo P, et al. Impact of tumor mutation burden on nivolumab efficacy in second-line urothelial carcinoma patients: exploratory analysis of the phase II CheckMate 275 study. Ann Oncol. 2017; 28(Suppl 5): Abstract 848PD.

Ethics Approval

The protocol was approved by site institutional review boards or independent ethics committees and conducted according to Good Clinical Practice guidelines, per the International Conference on Harmonisation. Patients provided written informed consent based on Declaration of Helsinki principles.

Correspondence: Peter Szabo (Peter.Szabo@bms.com)

Background

Close proximity of CD8+ T cells to cancer cells has been associated with improved outcome with immunotherapy. Using a gene expression profiling (GEP)-based inflammation assay, we previously derived gene signatures that defined CD8+ T-cell infiltration (CD8 signature) and localization to tumor parenchymal and stromal compartments (CD8-topology signatures) in multiple tumor types [1,2]. In patients with urothelial carcinoma (UC), high stromal/epithelial-mesenchymal transition (EMT) gene expression has been associated with T-cell exclusion and poor response to immunotherapy [3]. Here, we assess three CD8-derived signatures and compare them with a CD8 immunohistochemistry (IHC)-derived score combined with EMT gene expression (CD8.IHC_EMT) to evaluate associations between these biomarkers and with response to nivolumab in patients with UC in CheckMate 275 [4].

Methods

270 patients with platinum-resistant metastatic UC received nivolumab, with response assessed by blinded central review [4]. CD8+ T-cell infiltration in the TME (assessed using the CD8 signature [1], CD8-topology signatures (parenchymal, stromal) [2], and by IHC using a proprietary commercial assay [Dako mouse clone C8/144B antibody; Agilent Technologies Co]) and PD-L1 expression on tumor cells (Dako PD-L1 IHC 28-8 pharmDx) were assessed on baseline tumor samples. Predictive performance of the CD8 signature and CD8-topology signatures individually, the combined CD8-derived signatures (triple CD8), and CD8.IHC_EMT, was evaluated using Cox proportional-hazards regression for overall and progression-free survival (OS, PFS) and with logistic regression for objective response (OR). Odds ratios were scaled to reflect the difference between the 75th and 25th biomarker percentiles. Two-sided likelihood-ratio tests were used to assess biomarker and interaction effects. Associations with PFS and OS were also investigated using Kaplan–Meier analyses with biomarker scores categorized by tertile.

Results

GEP was evaluable in 205/270 (76%) patients. Baseline characteristics and clinical outcomes were similar in the overall population and the GEP-evaluable cohort. Response and survival predictions from the triple CD8 and CD8.IHC_EMT overlapped, and both biomarkers predicted benefit from nivolumab independent of PD-L1 expression. Odds ratios for OR were 2.59 (95% CI, 1.59–4.21) for triple CD8, 2.12 (1.47–3.07) for CD8.IHC_EMT, 2.51 (1.42–4.43) for the CD8 signature, and 1.74 (1.22–2.49) for the parenchymal CD8-topology signature.

Conclusions

Combined CD8 and CD8-topology gene signatures (triple CD8) showed similar performance to CD8.IHC_EMT for predicting response and survival in nivolumab-treated patients with UC. These data suggest potential utility of testing biomarker combinations and support further evaluation of gene signatures associated with parenchymal vs stromal CD8+ T-cell localization for predicting response to immunotherapy in patients with cancer.

Acknowledgements

Bristol-Myers Squibb. Professional medical writing and editorial assistance were provided by Bernard Kerr, PGDipSci, and Jay Rathi, MA, of Spark Medica Inc, funded by Bristol-Myers Squibb.

Trial Registration

NCT02387996

References

1. Szabo PM, Qi Z, Zerba K, et al. Association of an inflammatory gene signature with CD8 expression by immunohistochemistry (IHC) in multiple tumor types. J Clin Oncol. 2019; 37(Suppl): Abstract 2593.

2. Szabo PM, Lee G, Ely S, et al. CD8+ T cells in tumor parenchyma and stroma by image analysis and gene expression profiling: potential biomarkers for immuno-oncology therapy. J Clin Oncol. 2019; 37(Suppl): Abstract 2594.

3. Wang L, Saci A, Szabo PM, et al. EMT- and stroma-related gene expression and resistance to PD-1 blockade in urothelial cancer. Nat Commun. 2018; 9:3503.

4. Sharma P, Retz M, Siefker-Radtke A, et al. Nivolumab in metastatic urothelial carcinoma after platinum therapy (CheckMate 275): a multicentre, single-arm, phase 2 trial. Lancet Oncol. 2017; 18:312-322.

Ethics Approval

The protocol was approved by site institutional review boards or independent ethics committees and conducted according to Good Clinical Practice guidelines, per the International Conference on Harmonisation. Patients provided written informed consent based on Declaration of Helsinki principles.

Correspondence: Misuk Lee (misuklee55@gmail.com)

Background

Microsatellite instability status is currently used to predict susceptibility to immunotherapy. MANTIS score was originally developed to identify microsatellite instability through next-generation sequencing (NGS). Although the 0.4 cutoff identifies MSI-high status, there is insufficient data for this score's repercussion for MSI-stable patients [1]. MSI-high status rarely occurs in lung cancer patients representing less than 1% of the cases. Therefore, we aim to identify how MANTIS score correlates with immune profile and clinical outcomes in MSI-stable lung cancer.

Methods

MANTIS score was calculated for two TCGA (The Cancer Genome Atlas) cohorts: squamous cell carcinoma (SqCC, n= 501) and adenocarcinoma (ADC, n=517). After excluding MSI-high patients (n=3 and 1, respectively) we stratified each cohort into quartiles. The highest quartile was named MANTIS-high (M-H) and the lowest quartile MANTIS-low (M-L). Immune profile (immune cell infiltration and PD-L1 expression), tumor mutational burden (TMB), neoantigen burden and survival outcomes were compared between M-H and M-L. Tumor immune landscape was identified using signatures from immune metagenes predicting infiltration for 31 immune cells.

Results

M-H was associated with higher activated CD4, gamma delta and Th17 T cell infiltration when compared with M-L in lung SqCC (all p <0.05). No statistically significant difference in tumor T cell infiltration was found in ADC (Figure 1,2). M-H patients had a higher TMB when compared with M-L patients in ADC (p<0.05) and the same tendency was observed for SqCC (p=0.10) (Figure 3). Additionally, M-H correlated with lower PD-L1 (CD274) expression in both SqCC and ADC (each p<0.05) when compared with M-L. No significant differences in neoantigen burden were demonstrated. M-H patients showed a trend towards lower median overall survival in SqCC and ADC (75 vs 63 months p=0.21; 53 vs 50 months, p=0.14, Figure 3C,3D).

Conclusions

This is the first report that illustrates the implications of a microsatellite instability score on immune landscape, PD-L1 expression, TMB and clinical outcome from a pool of more than a thousand MSS non-small cell lung cancer patients.

Reference

1. Angelova M, Charoentong P, Hackl H, Fischer ML, Snajder R, Krogsdam AM, Waldner MJ, Bindea G, Mlecnik B, Galon J, Trajanoski Z. Characterization of the immunophenotypes and antigenomes of colorectal cancers reveals distinct tumor escape mechanisms and novel targets for immunotherapy. Genome Biol. 2015; 16:64.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P111).

Immune landscape in squamous cell carcinoma

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P111).

Immune landscape in adenocarcinoma

Correspondence: Tarjei Mikkelsen (tarjei@10xgenomics.com)

Background

Recent progress in cancer immunotherapy emphasizes the importance of understanding immune-regulatory pathways in cancer. It has been shown that immune cells play a crucial role in the tumor microenvironment and can be used for targeted therapeutics. Therefore, it is important to understand and characterize T cells and their antigen binding specificity and diversity in order to develop effective targeted immunotherapies. Recent technological advancements have enabled the integration of simultaneous cell-surface protein, transcriptome, immune repertoire and antigen specificity measurements at single cell resolution, providing comprehensive, scalable, high-throughput characterization of immune cells.

Methods

Using the 10x Genomics Single Cell Immune Profiling Solution with Feature Barcoding technology in conjunction with Biolegend oligo-conjugated antibodies and Immudex DNA barcoded peptide-MHC Dextramer® (pMHC), we performed multi-omic characterization of CD8+ T cell recognition of various virus and common cancer antigens in normal patients. Next generation sequencing libraries were made following the 10x Genomics workflow, where gene expression and immune repertoire libraries are generated alongside libraries from DNA barcodes conjugated to antibodies or pMHC, allowing quantification of cell surface proteins and identification of T cell receptor (TCR) specificities. Analysis was performed using the latest version of Cell Ranger (v3.0). The TCR-dist algorithm was used to identify clusters of related TCR sequences and enriched CDR3 motifs.

Results

We performed multi-omic characterization of ~100,000 CD8+ T cells from four MHC-matched donors. The multi-omic combination of gene expression, paired alpha/beta T cell receptor (TCR) repertoire, cell surface proteins and pMHC binding specificity allowed the identification of CD8+ T cell subpopulations with specificity for pMHCs within our panel. Within our data, we observed TCRs with cognate antigens that had been reported previously, while also identifying entirely new TCR–pMHC interactions. In addition, we observed specific expanded non-naïve T cell clones along with more diverse binding in the naïve compartment.

Conclusions

This rich and large dataset illustrates the power and scalability of the 10x Genomics Chromium Single Cell Immune Profiling Solution with Feature Barcoding technology and presents an exciting opportunity for researchers to explore and draw further conclusions about the mechanisms of TCR–pMHC interaction. Furthermore, this experiment serves as the next step on the path toward the even larger-scale experiments that will be necessary to fully comprehend the rules of antigen recognition in the adaptive immune system in response to cancer and infectious diseases and will be key in the development of successful immunotherapies.

Acknowledgements

This study was performed in collaboration with our 10x Genomics partners Immudex and Biolegend.

Correspondence: Jonathan Dry (Jonathan.Dry@astrazeneca.com); Ben Sidders (benjamin.sidders@astrazeneca.com)

Background

Understanding of the complex interplay between tumors and their immunologic microenvironment is critical for immune-oncology (IO) studies, which can facilitate the discovery of novel prognostic biomarkers, identification of new drug targets, and determination of drug resistance mechanisms. However, due to a lack of proper analysis tools and datasets, systematically exploring the tumor–immune interaction is still a big challenge.

Methods

Here, we deconvoluted the immune cell compositions and performed IO-related pathway/signature enrichment analysis for 9,721 primary tumor samples from 33 TCGA cancer types using transcriptomic data, and developed a web-based application, IO Browser.

Results

The browser allows the user to visualize the immune infiltrations of a sample or cohort, and to define disease segments or “immuno-types” based on the presence of single/multiple immune cell types or IO-related pathway/signatures. Users can then perform survival comparisons, explore gene expression of key cancer and IO genes as well as generate oncoprints in the different segments. The browser also provides statistical analysis to identify the gene or mutations enriched in the immuno-typed disease segment, and correlate gene expression or mutations with specific immune cell types in tumor microenvironment (TME).

Conclusions

In summary IO Browser enables comprehensive analysis and visualization of the dynamic interactions between tumor and immune landscape, and can aid our understanding of the interplay between tumor genomics and immune biology to facilitate line of sight and disease segmentation.

Correspondence: Bruna Cerbelli (bruna.cerbelli@uniroma1.it)

Background

In the neoadjuvant setting, pathological complete response (pCR) is more frequently achieved by triple negative breast cancer subtype and patients who attain this status show improved survival; However, standard neoadjuvant therapy results in pCR rates slightly over 30% of cases. The mechanism underlying the resistance to chemotherapy is still unclear and could be related both to the molecular heterogeneity of cancer cells and to the activation of the treatment-related immune response. For this reason, the search for immune biomarkers able to predict the response to chemotherapy represents a new promising frontier. Recent reports underscore the role of TILs, PDL-1 and CD73. The aim of our work is to define a novel tissutal immune profile (TIP) able to predict pCR [1,2].

Methods

We enrolled 61 pts who received NAC (EC for 4 cycles followed by Paclitaxel q7 for 12 cycles or q21 for 4 cycles) between Jan 2011 and June 2017 at Policlinico Umberto I and San Giovanni Addolorata Hospital of Rome. We performed, in basal paraffin-embedded biopsies, stromal TILS evaluation and immunohistochemistry for PD-L1 (Ventana SP142 clone) evaluated both on tumor cells (TC) and tumor-infiltrating immune cells (IC) and CD-73 assessed on TC. We defined “positive tissutal immune profile” (TIP+) the pts with “high TILS” (>50%), “PD-L1 positive” ( >1% both on TC and IC ) and “low CD73” (<40%), and the others as “negative tissutal immune profile” (TIP-).Statistical analysis was performed with T di Student test and χ2 test.

Results

We enrolled 61 females (median age: 50 y; range 28-75) affected by TNBC. The clinical stage before NAC was as follow: 3 pts cT3 (5%), 3 pts cT4 (5%) and 28 pts were cN+ (38%). Twenty-three patients (38%) showed pCR. No significant associations were found between pR and cT, cN, age, and KI-67. Seven patients (11%) were TIP+ and achieved pCR in 100% of cases; 54 patients were TIP- and pCR were showed in 16/54 of cases (30%) (p< 0,001).

Conclusions

TIP+ seems to be associated with higher pCR rate in TNBC patients .These preliminary results suggest the possibility of using novel profiles combining multiple immune- biomarkers.

References

1. Matsumoto H, Koo SL, Dent R, Tan PH, Iqbal J. Role of inflammatory infiltrates in triple negative breast cancer. J Clin Pathol. 2015;68:506–510. doi: 10.1136/jclinpath-2015-202944 139.

2. Jiang T, Xu X, Qiao M et al. Comprehensive evaluation of NT5E/CD73 expression and its prognostic significance in distinct types of cancers. BMC Cancer. 2018 Mar 7;18(1):267. doi: 10.1186/s12885-018-4073-7. PubMed PMID: 29514610; PubMed Central PMCID: PMC5842577.

Ethics Approval

CE 4181 Sapienza University of Rome

Correspondence: Paolo Antonio Ascierto (paolo.ascierto@gmail.com)

Background

Immune checkpoint inhibitors (ICI) have improved the treatment options for patients with advanced stage melanoma, with improved clinical responses and overall survival compared to standard systemic therapies. However, a large percentage of melanoma patients do not respond to ICIs, highlighting the need for a greater understanding of the tumor environment and host immune response. Here, we apply unbiased discovery proteomics, based on label-free data independent acquisition (DIA) mass spectrometry to deeply characterize global tumor proteomes to identify proteins and pathways that are associated with pre-treatment response to anti-PD-1 immunotherapy.

Methods

Unbiased, data-independent acquisition (DIA) mass spectrometry was used to analyze formalin fixed paraffin imbedded (FFPE) tumor tissue samples from subjects with Stage III-IV melanoma which were resected prior to initiation of first-line anti-PD-1 ICI therapy. The selected samples represent two distinct clinical subgroups; those who received clinical benefit, with a partial response or better (PR, SD and CR, n = 13), and those with no clinical benefit (PD, n = 9) and no observable response to therapy. Samples were prepared for mass spectrometry using standard procedures. All samples were analyzed using 2-hour gradients on a LC-MS/MS setup operated in DIA mode. Data was extracted using Spectronaut (Biognosys) with a sample specific spectral library which was combined with a large human tissue resource library. Statistical analysis was conducted to identify proteins that are either up- or down-regulated with respect to benefit group. Pathway analysis was also conducted to highlight dysregulated biological functions and pathways.

Results

7,590 proteins were quantified across all samples, with 6,627 quantified on average per sample. Univariate statistical testing between groups identified 254 proteins that are dysregulated (120 up-regulated and 134 down-regulated) in subjects who received clinical benefit. Through partial least squares discriminant analysis (PLS-DA) a set of 25 proteins was identified that describe the variance between the two sample groups. When annotated to their sub-cellular location, all up-regulated species are identified as mitochondrial proteins, indicating an enhanced metabolic environment, and the down-regulated species are cytosolic, lysosomal or membrane associated. This observation was also reflected in pathway analysis which identified up-regulation of arginine and citrulline metabolism and down-regulation of adhesion related processes driven by MHC-II and integrins.

Conclusions

Global profiling of the tumor proteome provides a unique characterization of melanoma tumor biology. A pathway level analysis shows increased metabolic processes combined with decreases in adhesion related proteins may underly the differences in benefit related to ICI therapy.

Ethics Approval

The study was approved by the Istituto Nazionale Tumori IRCCS Fondazione “G. Pascale” of Napoli Institutution‘s Ethics Board, approval number 33/17.

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

Correspondnce: Namyong Kim (namyong@curiox.com)

Background

Blood cells are prime indicators of immuno-surveillance, and the ease of blood sampling makes blood analysis a key interest for clinical and research applications. While current flow cytometry methods are high-throughput and provide fine resolution in the segregation of white blood cell (WBC) populations, WBC enrichment involving red blood cell (RBC) lysis are laborious and typically performed manually, contributing to experimental variability especially as blood cells are sensitive to physical and chemical stress.

Methods

We describe RBC lysis and leukocyte immunostaining on a centrifuge-less platform Laminar Wash™, using a novel wall-less plate and laminar flow washer. The Laminar Wash™ 24-well plate consists of an array of hydrophilic spots surrounded by hydrophobic surface, which functions as a virtual wall that separates each spot. Each well is capable of staining and lysing 100uL of whole blood. During lysis, WBCs settle to the surface of the spot, allowing the spent lysis buffer to be removed by a gentle and continuous laminar-flow washing process on the Laminar Wash™ system, eliminating centrifugation and resuspension that may stress cells and disrupt antibody binding.

Results

We observed improved retention of CD45+ lymphocytes while lysing on Laminar Wash™ plates compared to conventional centrifuge tubes. In studies comparing mouse whole blood lysis and antibody staining by conventional tube centrifuge and Laminar Wash, Laminar Wash achieved dramatically higher staining index and improved resolution of cell cluster by flow cytometry.

Conclusions

In summary, Laminar Wash system provides gentle, fast and convenient blood lysis, while improving data quality with superior antibody staining.

Correspondence: Jacques Fieschi (jacques.fieschi@haliodx.com)

Background

In the management of advanced Non-Small Cell Lung Carcinoma (NSCLC), PD1/L1 immune checkpoint inhibitors (ICIs) have increased overall survival (OS) over standard second-line chemotherapy. While this long-term increase in OS is driven by about 20% of patients, others display disease progression during the first weeks. PIONeeR workpackage 2 aims to understand and eventually predict response and/or resistance to those ICIs in stage IV or recurrent NSCLC patients. For that purpose, an Immunogram was designed that integrates a comprehensive set of biomarkers measured in the tumor microenvironment.

Methods

The immune contexture from the PIONeeR trial’s patients is being characterized in a prospective manner and will be confronted to clinical data at the end of the study. This multi-modal approach, encompassing a range of immune scoring assays, is applied to blood and tumor biopsy from each patient, both sampled before and throughout anti-PD1/L1 ICI treatment. This work aims at describing pre-treatment samples profiling.

Results

We assessed the feasibility of such a profiling and provide descriptive multi-modal immune profiles for the 10 first PIONeeR-included patients. These profiles combine raw results from more than ten tests, corresponding to the following technologies and biomarkers: Genomic Next Generation Sequencing for Tumor Mutational Burden (TMB), DNA mismatch-repair deficiency (MSI/MSS status) and T Cell Clonality assessments ; dualplex and multiplex immunohistochemistry coupled to digital pathology analyses to assess Immune Cells Infiltration and PD-L1 mediated inhibition (Immunoscore® IC), Immune Suppression through Regulatory T cells and Myeloid-derives suppressor cells quantification, T-Cell Exhaustion status ; standardized methods for assessment of endothelial activation markers ; flow cytometry for circulating immune cell subtypes quantification; ICI plasma exposure levels. A multimodal integrative Immunogram presentation is proposed for each patients.

Conclusions

This preliminary study shows that multimodal immune profiling is feasible and could be a new tool to understand the biology and pharmacology of lung cancer resistance to anti-PD1/L1 ICIs and potentially guide patient management décisions.

Acknowledgements

This work is supported by the French National Cancer Agency, Agence Nationale du Cancer, through the PIONeeR project financing.

Trial Registration

ClinicalTrials.gov Identifier: NCT03493581

Ethics Approval

The study was approved by the French Ethic Comitee CPP Ouest II Angers, approval number 2018/08.

Correspondence: Li Peng (lpeng@palleonpharma.com)

Background

The glyco-immune checkpoint (Siglec/sialoglycan axis) has emerged as a new mechanism of cancer immune escape and offers new therapeutic interventions to overcome resistance to current immunotherapies. Siglecs (sialic acid-recognizing Ig-superfamily lectins) are type I transmembrane sialoglycan binding proteins expressed on various immune cells (innate and adaptive). Humans express at least fourteen unique Siglecs which have distinct preferred sialoglycan ligands. Tumors upregulate certain sialoglycan patterns to facilitate immune cell evasion by engaging these inhibitory Siglec receptors. This tumor inhibitory “glyco-code” consists of a heterogenous mixture of numerous sialoglycans, binding to Siglecs through low affinity and high avidity interactions.

Methods

Deciphering the hypersialylation glyco-code of tumors is key to identifying cancer patients for glyco-immune checkpoint blockade therapies. However, the heterogeneity and complexity of sialoglycans make characterization of the tumor surface sialoglycome difficult with current technologies. To overcome this challenge, we developed a proprietary sialoglycan-probing reagent, HYDRA, to functionally detect inhibitory tumor sialoglycans engaging Siglecs. HYDRA mimics this natural avidity driven Siglec-sialoglycan interaction, consisting of multimeric fusions of a Siglec N-terminal extracellular domain containing the carbohydrate recognition domain (CRD), a trimerization motif, and a Fc dimerization domain.

Results

We have generated several HYDRA constructs with robust expression using a mammalian HEK293 system. Size-exclusion chromatography profiles of HYDRA demonstrate high purity and confirmed multimeric assembly. HYDRAs have greater than fifteen-fold increase in binding affinity compared to Siglec-Fc dimers as measured using bio-layer interferometry Octet. HYDRA also demonstrates sialoglycan-specific binding, as its binding was eliminated when cells were treated with sialidase (which removes terminal sialic-acids of sialoglycan) or using cells lacking sialoglycans from knocking out UDP-GlcNAc 2-Epimerase. Glycan array binding of HYDRA confirmed similar sialoglycan preferences of its Siglec counterpart as described in the literature, suggesting engineering did not alter glyco-recognition properties. These high-affinity and sialoglycan-specific HYDRAs enabled us to develop a robust immunohistochemistry (IHC) assay to analyze cancer patient samples. A cohort of tissues (>2,500 patients) from various indications were analyzed to enable indication prioritization for glyco-immune checkpoint therapies. HYDRA IHC on healthy and cancerous human tissues demonstrate unique binding patterns with concordance between duplicate primary tumor cores and primary tumor versus metastatic cores from the same patient in non-small cell lung, kidney and colon cancer samples.

Conclusions

In summary, the HYDRA technology distills the structural heterogeneity of tumor surface sialoglycans to a straightforward functional readout of immunosuppressive glyco-codes engaging inhibitory Siglecs, which may allow patient stratification based on deciphering a tumor-specific surface glycan pattern.

Correspondence: Diwakar Davar (davard@upmc.edu)

Background

PD1 blockade produces responses in 30-40% of metastatic melanoma (MEL) with durable relapse-free benefit [1,2]. Pre-existing tumor-infiltrating CD8+T cell infiltrates (TIL), neoantigen burden and IFN-γ gene expression signature (GES) correlate with clinical anti-tumor response [3-5] to PD1 blockade. However, neoantigen burden and IFN-γ GES are cost-prohibitive and time-consuming assays that are not available for clinical use; while CD8 T cell analysis by immunohistochemistry (IHC) is cost-effective and operator-independent. The aim of this study is to develop and validate an image analysis algorithm to automatically quantify CD8+ T cells (CD8 score) in patients with metastatic MEL treated with PD1 blockade.

Methods

Included patients had advanced metastatic MEL treated with PD1 blockade. Radiographic response assessed using RECIST v1.1. For the purposes of this analysis, patients were defined as responders (R; complete, partial response, stable disease) or non-responders (NR; progressive disease). Pre-treatment tumor biopsies from 58 patients were utilized. Brightfield image analysis results were cross-validated with fluorescence-based quantification (AQUA™). A nuclear image algorithm designed to run on whole slide images was optimized to manual count. The algorithm was locked down and used on a cohort of whole tissue sections from MEL patients. All images were reviewed by independent pathologist blinded to clinical outcomes. Response and outcomes were statistically correlated with image analysis results.

Results

There were 40 R patients and 18 NR patients. Median CD8 score was 101 cells/mm3 in R and 48.7 cells/mm3 in NR (p=0.098). Median PFS were greater in R compared to NR (18 months vs. 2 months, p100 cells/mm3 (64%).

Conclusions

We report the successful technical development and clinical validation of an image algorithm to automate CD8 score for metastatic MEL treated with PD1 blockade. Preliminary results demonstrate CD8 score was directly associated with response and improved PFS. CD8 score is an assay that could be carried out using existing technology in pathology departments. Further analysis will focus on validating these results in a larger cohort to permit clinical use.

References

1. Ribas A, Hamid O, Daud A, Hodi FS, Wolchok JD, Kefford R, Joshua AM, Patnaik A, Hwu WJ, Weber JS, Gangadhar TC, Hersey P, Dronca R, Joseph RW, Zarour H, Chmielowski B, Lawrence DP, Algazi A, Rizvi NA, Hoffner B, Mateus C, Gergich K, Lindia JA, Giannotti M, Li XN, Ebbinghaus S, Kang SP, Robert C. Association of Pembrolizumab With Tumor Response and Survival Among Patients With Advanced Melanoma. JAMA. 2016; 315:1600-9.

2. Larkin J, Lao CD, Urba WJ, McDermott DF, Horak C, Jiang J, Wolchok JD. Efficacy and Safety of Nivolumab in Patients With BRAF V600 Mutant and BRAF Wild-Type Advanced Melanoma: A Pooled Analysis of 4 Clinical Trials. JAMA Oncol. 2015;4:433-40.

3. Tumeh P, Harview C, Yearley J, Shintaku I, Taylor E, Robert L, Chmielowski B, Spasic M, Henry G, Ciobanu V, West A, Carmona M, Kivork C, Seja E, Cherry G, Gutierrez A, Grogan T, Mateus C, Tomasic G, Glaspy J, Emerson R, Robins H, Pierce R, Elashoff D, Robert C, Ribas A. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 2014;515:568-71

4. Cristescu R, Mogg R, Ayers M, Albright A, Murphy E, Yearley J, Sher X, Liu XQ, Lu H, Nebozhyn M, Zhang C, Lunceford JK, Joe A, Cheng J, Webber AL, Ibrahim N, Plimack ER, Ott PA, Seiwert TY, Ribas A, McClanahan TK, Tomassini JE, Loboda A, Kaufman D. Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy. Science. 2018;362:197

5. Rizvi NA, Hellmann MD, Snyder A, Kvistborg P, Makarov V, Havel JJ, Lee W, Yuan J, Wong P, Ho TS, Miller ML, Rekhtman N, Moreira AL, Ibrahim F, Bruggeman C, Gasmi B, Zappasodi R, Maeda Y, Sander C, Garon EB, Merghoub T, Wolchok JD, Schumacher TN, Chan TA. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science. 2015;348:124-8.

Ethics Approval

The study was approved by University of Pittsburgh‘s Institutional Review Board, approval number PRO18080253.

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

Correspondence: Jacques Fieschi (jacques.fieschi@haliodx.com)

Background

To tailor clinical care and personalized treatment of cancer patients, the scientific community together with the practitioners have focused into refining our understanding of cancer biology and resistance to treatments. In that perspective, the concept of Immunoscore proposed by Galon et al [1, 2, 3] has highlighted the crucial role of immune response to the tumor. In parallel, immunotherapies by immune checkpoint inhibitors (ICI) anti-PD-1/PD-L1 were approved in several cancer indications, such as Non-Small Cell Lung Cancer or melanoma, even if only a minority of these patients respond positively to the treatment. In addition, ICI are far less effective for other high-incidence indications like colorectal cancer (CRC), thus suggesting that multiple factors may be critical for capturing the exact nature of the tumour microenvironment (TME). In this context, the comprehensive identification and assessment of these factors could be key to stratify patients and allow the selection of the optimal treatment.

Methods

In order to support clinical researchers and biopharmaceutical companies in the evaluation of the efficacy of candidate drugs, HalioDx has developed the Cancer Immunogram, a solution based on Blank CU et al. [4]. Our multi-parameter approach encompassing a unique range of immune scoring assays is based on the analysis and the understanding of the immune contexture of tumors and offers a personalized and dynamic “fingerprint” of tumor-immune system interaction. To address this, the Cancer Immunogram combines different technologies and biomarkers to assess 1) the tumor characteristics (Tumor foreignness, MSI, PD-L1 expression, common mutation drivers), 2) the immune infiltration (Immunoscore®, CD8/PD-L1 proximity, TCR clonality, immune expression signature), 3) the immune checkpoint status (T Cell Exhaustion BrightPlex panel) and 4) the immune suppression status (Treg, MDSC and M1/M2 macrophage BrightPlex panels).

Results

Here, we consolidate our Proof of Concept for the Cancer Immunogram in the context of CRC [5] by leveraging this meta-analysis on a 20-patients cohort. Using machine learning algorithms to extract the most relevant features, we show that the Cancer Immunogram allows to identify patient-specific patterns which might improve the prediction of the response to therapy.

Conclusions

We believe that the Cancer Immunogram has the potential to facilitate drug development by providing a 360° vision of the tumour immune contexture and may also help clinicians to personalize advanced cancer patient care.

References

1. Galon J, Bruni D. Approaches to treat immune hot, altered and cold tumours with combination immunotherapies. Nat Rev Drug Discov. 2019;18:197-218.

2. Pagès F et al. International validation of the consensus Immunoscore for the classification of colon cancer: a prognostic and accuracy study. Lancet. 2018; 391:2128-2139.

3. Mlecnik B et al. Integrative Analyses of Colorectal Cancer Show Immunoscore Is a Stronger Predictor of Patient Survival Than Microsatellite Instability. Immunity. 2016; 44:698-711.

4. Blank CU et al. The "cancer immunogram". Science. 2016;; 352:658-60.

5. Sbarrato T et al, Combining multimodal biomarkers as an immunogram to guide immunotherapy use: A proof of concept. Proceedings: AACR Annual Meeting 2019; March 29-April 3, 2019; Atlanta, GA.

Correspondence: John Powderly (jpowderly@carolinabiooncology.org)

Background

Isolation of rare lymphocyte populations from peripheral blood products of cancer patients can be challenging due to technician variability and substantial cell loss through standard cell separation methods such as Mononuclear Cell Preparation Tubes™ (CPTs). An automated microfluidic approach was evaluated to determine lymphocyte recovery, processing time, and ease of use. Furthermore, increased yields of rare cells from cancer patients’ peripheral blood could potentially substitute for leukapheresis when leukapheresis is not a viable option.

Methods

White blood cells (WBCs) or peripheral blood mononuclear cells (PBMCs) were isolated from human peripheral blood using either MicroMedicine’s Microfluidic System (MS) or CPTs. Cell viability and lymphocyte recovery were compared using a hematology analyzer and flow cytometry. Further, a rare lymphocyte population was positively immunomagnetically selected from healthy volunteers and cancer patients. Immunophenotyping was performed pre- and post-cell selection, followed by in vitro expansion of the rare lymphocytes.

Results

Using cells collected from healthy volunteers, the automated MS prototype consistently recovered 83.5 ± 10.1% lymphocytes in a total of 31 ± 5.8 minutes, including hands-on time, compared to the standard CPT process, which recovered 43.2 ± 7.6% lymphocytes in 72.2 ± 4.1 minutes, from 32 – 34 mL blood samples (n = 5). The viability was comparable at 97.9 ± 1.6% (MS) and 96.7 ± 3.0% (CPT). In further studies with both healthy donors and cancer patients, a rare lymphocyte population was successfully selected from WBCs isolated with the MS, enabling immunophenotyping of the rare cell population and subsequent in vitro expansion. Expansion of this lymphocyte population from a colon adenocarcinoma patient was found to be suppressed post-immunotherapy compared to pre-treatment. Cells isolated from patients with other malignancies were successfully expanded. Finally, peripheral blood collected from cancer patients yielded a greater number of rare lymphocytes using the MS for the cell expansion study in comparison to the CPT isolation method, which correlates to having a higher lymphocyte recovery.

Conclusions

The MS consistently recovered approximately twice the number of lymphocytes in half the time compared to the traditional CPT method. The automated cell separation process improves the consistency of cell isolation while freeing up technician time. Rare lymphocyte populations could be reliably recovered from peripheral blood with significantly higher yield compared to the CPT method. While leukapheresis enriches for MNCs and MS isolates all WBCs, these results suggest that peripheral blood collection-based MS has the potential to complement leukapheresis, especially for small- to medium-scale studies.

Correspondence: David Rimm (david.rimm@yale.edu)

Background

Currently, there is no diagnostic test that can accurately predict response in melanoma patients treated with immunotherapy. NanoString® nCounter® PanCancer IO 360™ panel (Research Use Only) measures mRNA from 770 genes related to the tumor and host immune response. Here, we used this panel to assess the predictive value of individual genes and weighted gene signatures in a cohort of immunotherapy (ITx) treated melanoma patients.

Methods

We used pretreatment, formalin-fixed paraffin-embedded (FFPE) whole tissue sections from 59 melanoma patients that received single agent or combination immunotherapy (pembrolizumab, nivolumab, or nivolumab plus ipilimumab). Two slides from each patient were macrodissected and RNA was extracted. The mRNA transcripts were hybridized and tagged by unique probes for the 770-plex PanCancer IO 360 panel and then measured on the nCounter platform. RNA counts were correlated with best overall response (BOR), clinical benefit (CB), progression free survival (PFS) and overall survival (OS).

Results

Indoleamine 2,3-dioxygenase 1 (IDO1) was the best single gene predictor of BOR (Area under the curve (AUC) = 0.73) and CB (AUC = 0.70). Among other genes, IDO1 mRNA was also found to be significantly associated with longer PFS (P < 0.01, False discovery rate (FDR) = 0.18) and OS (P < 0.01, FDR = 0.052). The previously described 18-gene tumor inflammation score (Ayers TIS) validated for the prediction of BOR (AUC = 0.68), PFS (P < 0.05, FDR = 0.18) and OS (P < 0.001, FDR = 0.025). TIS also predicted CB (AUC = 0.67). Its predictive value remained the same irrespective to immunotherapy agent administered. Nevertheless, it decreased for patients harboring the BRAF and NRAS mutations (AUC = 0.76 versus 0.51 and 0.44 for patients with BRAF and NRAS mutations respectively). The best signatures for this cohort were for Cytotoxicity, Immunoproteasome and CD56dim Cells which were predictive for BOR (AUC = 0.72, 0.71 and 0.70 respectively), CB (AUC = 0.69, 0.68 and 0.70 respectively), and OS (all FDRs < 0.05). Further work is underway to compare these Yale melanoma results with other cohorts.

Conclusions

Pretreatment mRNA counts of single genes or weighted signature scores are related to immunotherapy outcomes in melanoma patients. This work validated the Ayers TIS signature and highlighted the role of the immune microenvironment, especially NK cells, in mediating antitumor response after immune checkpoint inhibition.

NanoString nCounter is Intended for Research Use Only. Not for Use in Diagnostic Procedures.

Ethics Approval

The study was approved by Yale University Human Investigation Committee, approval number 9505008219.

Correspondence: Kathlynn Brown (kathlynn.brown@sri.com)

Background

Cancer Immunotherapies designed to generate a cell-mediated immune response against tumors are emerging as frontline treatment options for cancer; however, concerns regarding efficacy, safety and cost efficacy have limited the use of these treatments.

Methods

To address these weaknesses, we developed a novel immunotherapy capable of delivering previously encountered antigenic peptides specifically to cancer cells and facilitating their presentation through the MHC class I pathway. It utilizes a synthetic nanoparticle delivery system comprised of three components: a neutral stealth liposome, encapsulated synthetic immunogenic HLA class I restricted peptides derived from measles virus (MV), and a tumor-targeting peptide on the external surface of the liposome. The targeting peptide results in accumulation of liposomes specifically inside cancer cells, and facilitates presentation of MV-derived immunogenic peptides in HLA class I molecules (Figure 1). We refer to this system as TALL (Targeted Antigen Loaded Liposomes). Therefore, TALL can generate a secondary immune response specifically against the targeted tumor cells in a patient who has been previously vaccinated against or infected by MV. In short, we are attempting to trick the immune system into responding as though the cancer cell is infected with MV without the use of viral particles.

Results

We synthesized liposomes encapsulating H250, an immunogenic HLA class I restricted peptide identified from measles hemagglutinin protein. These liposomes were targeted to breast and lung cancer cells via our targeting peptide, which was identified using phage-display methodology. Treatment of lung cancer cells with TALL results in functional presentation of H250 in both MHC and HLA class I molecules. Our in-vitro and in-vivo studies indicate that presentation of H250 is dependent on the cancer targeting peptide; liposomes that lack the targeting peptide did not accumulate in the cancer cells and presentation of H250 was abrogated. Treatment with TALL substantially reduced growth of LLC1 and 4T1 tumors in vaccinated C57BL/6 and Balb/c mice respectively.

Conclusions

The outcome of our therapy is a robust cytotoxic T lymphocyte response directed specifically against the tumor. It's advantages include: 1) Bypassing the need to identify tumor-associated antigens or educate the immune system through a primary immune response; 2) It is anticipated to be effective against tumors with a low mutational load, making it efficacious on early-stage and metastatic cancer; 3) It does not use a live virus or biologically-derived material, allowing for complete synthetic manufacturing. It also does not require isolation or ex-vivo manipulation of patient’s cells, reducing production time and costs.

Acknowledgements

Research reported in this work was supported by DOD under grant number W81XWH-16-1-0262 and SRI internal research funds

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P124).

TALL mechanism of action

Correspondence: Stephen Willingham (swillingham@corvuspharma.com)

Background

Extracellular adenosine in the tumor microenvironment generates an immunosuppressive niche that promotes tumor growth and metastasis by signaling through the A2A receptor (A2AR) on immune cells. Various agents targeting the adenosine pathway are now in clinical trials as cancer therapies. Ciforadenant is a selective A2AR antagonist and CPI-006 is an anti-CD73 antibody (Fc-mutant IgG1) that blocks the enzymatic conversion of AMP to adenosine and directly stimulates immunity. Both agents are now being studied in clinical trials (NCT02655822 and NCT03454451). In this report, we evaluate the role of adenosine and AMP-related gene expression profiles (GEPs) that may predict the response of patients receiving adenosine pathway therapies. Ex vivo studies reveal a requirement for dual blockade of CD73 and A2AR for optimal neutralization of AMP mediated immunosuppression.

Methods

Normal human PBMCs were stimulated ex vivo with NECA (stable adenosine analog) or AMP. RNA from tumor biopsies and PBMC was analyzed using NanoString. Renal cell cancer (RCC) tumor biopsies collected from patients treated with ciforadenant (100 mg BID) either as a single agent (n=18) or in combination with atezolizumab (n=14).

Results

Ex vivo A2AR agonism resulted in dose-dependent increases in CXCR2 ligands (CXCL1,2,3,5,8) and key mediators of neutrophil/MDSC biology (CSF3, IL-23). Increases in monocyte/macrophage inflammatory mediators such as IL-1beta and CCL2,3,7,8, 20 were also observed, as were increases in SERPINB2, S100A8, PTGS2, THBS1. Preliminary biomarker analysis suggests ciforadenant anti-tumor activity in RCC was associated with increased expression of select analytes (AdenoSig) in pretreatment biopsies (Figure 1).

Ex vivo AMP or AMPalphaS (a non-hydrolyzable AMP analog) stimulation induced a similar GEP (AMPSig), but included specific decreases in OAS3, BIRC5, CDK1, MX1, IFI27, and IFIT1. CD73 antibody and small molecule antagonists amplified the AMPSig by preserving AMP, which itself directly stimulates adenosine receptors. In contrast, ciforadenant inhibited induction of the AdenoSig and AMPSig in all experimental settings at the transcript and protein level.

Conclusions

A2AR agonists and AMP induce specific GEPs dominated by immunosuppressive mediators of MDSC and monocyte/macrophage biology. These GEPs may be used as biomarkers for patient selection. CD73 antagonists alone may be limited by the induction of compensatory immunosuppressive pathways mediated by AMP accumulation. Combination ciforadenant and CPI-006 treatment may synergize to activate anti-tumor immunity by 1) blocking adenosine production and signaling, 2) directly activating immune cells, and 3) blocking a compensatory induction of AMPSig. This combination strategy is being evaluated in an ongoing Ph1/1b clinical trial in patients with advanced solid tumors.

Trial Registration

NCT02655822 and NCT03454451

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P125).

See text for description

Correspondence: David Rimm (david.rimm@yale.edu)

Background

Only a minority of patients with advanced NSCLC truly benefit from single-agent PD-1 checkpoint blockade, and more robust predictive biomarkers are needed to optimally deliver these therapies. The GeoMx Digital Spatial Profiler (DSP) (NanoString, Inc.) allows high-plex protein expression analysis in a quantitative and spatially-resolved manner from single formalin-fixed paraffin embedded tissue sections. Here we use this technology as a discovery tool to find protein markers associated with benefit from single-agent PD-1 checkpoint blockade in NSCLC.

Methods

We used the GeoMx DSP in a cohort of 63 immunotherapy-treated NSCLC cases represented in a tissue microarray, 52 of whom had pre-treatment samples and received single-agent PD-1 checkpoint blockade. A panel of 40 photocleavable oligonucleotide-labeled primary antibodies (NanoString Human IO panel) was used for protein detection. Proteins were measured in 4 independent molecularly-defined tissue compartments by fluorescence co-localization (tumor [panCK+], leucocytes [CD45+], macrophages [CD68+], and non-immune stromal cells [CK-/CD45-CD68-/DNA+]). The photocleaved oligos were hybridized and digitally counted with the nCounter platform. Two cut-points (median and top tertile) were explored for each maker. All statistical testing was performed using a two-sided significance level of α=0.05 without correction for multiple hypothesis testing.

Results

160 protein variables were generated per case (normalized counts within molecularly defined compartments). In univariate analyses using pre-specified cut-points, 10 markers were associated with clinical benefit (CB) or non-CB, 6 markers with PFS, and 13 markers with OS. Of these, CD56 (top tertile) and CD4 (median) measured in the CD45 compartment were the only markers that significantly predicted either CB (OR 6.7, p = 0.014 and OR 8.5, p = 0.014, respectively) longer PFS (HR 0.38, p = 0.011 and HR 0.33, p = 0.002, respectively) and longer OS (HR 0.44, p = 0.044 and HR 0.31, p = 0.002, respectively). After adjusting for 3 baseline clinical prognostic factors (performance status, liver metastasis, dNLR) in a multivariate Cox proportional hazard model, both CD56 and CD4 remained predictive for PFS (HR 0.39, p = 0.020 and HR 0.37, p = 0.017, respectively), while only CD4 was predictive for OS (HR 0.28, p = 0.006).

Conclusions

This pilot scale, discovery study shows the potential of the DSP technology in the identification of spatially-informed biomarkers of response to PD-1 checkpoint blockade in NSCLC. This works highlights a previously undescribed role for CD56+ immune cells and CD4+ T-cells as potential predictors of immunotherapy outcomes in NSCLC.

Ethics Approval

All tissue samples were collected and used with specific consent or waiver of consent under the approval from the Yale Human Investigation Committee protocol #9505008219.

Correspondence: Sean Downing (sean.downing@ultivue.com)

Background

Current IHC methods limit the depth of information from a single tissue sample to a single target in the case of chromogenic staining, or to sample morphology and general cell identification in the case of H&E. Multiplex immunofluorescence (mIF) methods provide insights into a wide number of markers of interest and their spatial context in a single sample but limit the level of marker co-localization detection possible because of multiple antigen retrieval or photobleaching steps. Here, we demonstrate the utility of a new 12-plex mIF panel using InSituPlex technology that can identify thousands of phenotypes and spatial behavior through the co-localization of markers that was once limited to the domain of flow cytometry.

Methods

The 12-Plex marker panel was developed including: CD3, CD4, CD8, CD20, Granzyme B, CD56, CD68, CD163, FoxP3, PD-1, PD-L1, and pan-Cytokeratin/Sox10 used the InSituPlex and DNA-Exchange technology to perform mIF staining of FFPE samples from tonsil and tumor biopsies on the Leica Biosystems BOND RX autostainer. The tissues were then imaged in five distinct fluorescent channels (DAPI, FITC, TRITC, Cy5, Cy7) in 3 rounds of image acquisition on the ZEISS Axio Scan.Z1. HALO analysis software was used to identify cell phenotypes and spatial interactions across the whole slide images. UMAP and PSDM were also used to characterize cellular phenotypes and similarities amongst samples in the cohorts. Downstream H&E staining was performed on the same slides with a fourth imaging round to produce a fused 12-plex fluorescent and brightfield image.

Results

The multiplex panel was able identify all 12 markers in FFPE samples and immunophenotype single cells through co-expression of several biomarkers. Of the 4,096 possible phenotypes, the relevant phenotypes mapped included, but were not limited to: T cells, T-regs, Cytotoxic T-cells, Exhausted T-cells, B cells, NK cells, M1 and M2 macrophages, tumor cells, and expression along the PD-L1 and PD-1 immune checkpoint axis. Distance mapping and infiltration indexes were measured in tumor regions, stroma compartments, and along invasive margins.

Conclusions

In this work, we introduce a tumor and immune cell phenotyping multiplex immunofluorescence panel for the comprehensive characterization of the tumor microenvironment and its applicability across a range of carcinoma and melanoma FFPE tissue samples for support of deep pathology assessment in drug discovery research. The ability to colocalize markers in the same compartment for the identification of thousands of phenotypes when combined with brightfield pathological assessment within a single sample has the potential to accelerate immunotherapy research.

Correspondence: Jonathan Schoenfeld (jdschoenfeld@partners.org)

Background

PD-L1 expression is one of the most studied biomarkers to predict the efficacy of immune checkpoint inhibitors (ICIs), but its clinical significance is controversial. Several factors have limited the study of PD-L1 expression. Most trials use of hazard ratios (HRs) to measure treatment effects on survival outcomes, a questionable practice for immunotherapy studies. Additionally, trials use different cut-off values to dichotomize PD-L1 scores, complicating meta-analyses. Therefore, we performed a pooled analysis to: i) estimate the distribution of PD-L1 expression scores in clinical populations, and ii) assess the relationship between PD-L1 levels and ICIs’ effects on overall survival (OS). Instead of HRs, we used a more robust metric, i.e. differences in restricted mean survival times (ΔRMSTs).

Methods

Following PRISMA guidelines, we analyzed individual-level data reconstructed from the publications of 14 randomized clinical trials of ICIs. We used an imputation-based approach to estimate i) the distribution of PD-L1 scores, ii) the survival distribution in different PD-L1 classes, and iii) pooled ΔRMST estimates. We show the advantage provided by meta-analytic estimates such as ours for the design of future studies in a simulation study. We simulated 10,000 NSCLC trials (1:1 randomization; sample size: 500 patients) that compared ICIs with standard chemotherapy. Simulated trials followed either i) a design that does not use prior information on the distribution of PD-L1 levels and their association with ICIs’ effects, or ii) a design tailored to our meta-analytic estimates.

Results

We reconstructed data on 7,918 individual patients, 3,496 with NSCLC, 4,529 with other tumors. The estimated distribution of PD-L1 expression is U-shaped, with most patients presenting a low or high expression: only about 7% had an expression in the 5%-50% range. ΔRMST estimates suggest that i) ICIs provide an OS benefit to all patients, and ii) the magnitude of OS benefits increases along with PD-L1 score, although changes in ΔRMSTs were greater in NSCLC (Figure 1). In the simulations, the power to detect a positive treatment effect increased from 80% to 93% using a design tailored to meta-analytic information.

Conclusions

By highlighting that higher PD-L1 scores predict increasing OS benefits, our findings extend those of recent meta-analyses that evaluated PD-L1 expression scores as predictors of ICIs’ efficacy. They also illustrate how meta-analytic estimates like ours can improve the power of future trials to detect ICIs’ benefits. Our findings also suggest that the practice of dichotomizing the range of PD-L1 expression scores is inadequate for patient stratification.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P128).

See text for description

Correspondence: Jennifer Carlisle (jennifer.w.carlisle@emory.edu)

Background

Although immunotherapy with PD-1 or dual PD-1/CTLA-4 blockade can be a successful treatment for patients with advanced renal cell carcinoma (RCC), the majority do not respond [1, 2]. Prior peripheral blood immune profiling studies have shown that a transient rise in activated CD8 T cells correlates with clinical response to PD1 blockade in lung cancer [3, 4] and melanoma [5]. The effect of checkpoint blockade on circulating CD8 T cells in RCC is unknown, as is the T cell biology underlying clinical response.

Methods

Serial peripheral blood mononuclear cells were obtained from patients with RCC undergoing immunotherapy. Samples were obtained at baseline (cycle 1) and initiation of each subsequent cycle (up to cycle 6). Flow cytometry identified longitudinal changes in T cell subsets. Additionally, recently activated CD8 T cells, identified by surface expression of CD38 and HLA-DR, were sorted at baseline, post-cycle 1, and post-cycle 2, and analyzed by RNA seq. Clinical responses were determined at the first restaging scans using RECIST v1.1 criteria, to define those with clinical benefit (complete response, partial response, or stable disease) or no clinical benefit (progressive disease).

Results

Of 27 patients analyzed, 10 received nivolumab, 7 nivolumab + NKTR-214, and 10 nivolumab + ipilimumab. Median age was 58 years (range 33-78) with a male (70%), Caucasian (89%), and solely clear cell histology (83%) predominance. A burst in circulating activated CD8 T cells as defined by a ≥1.8 fold increase in CD38+HLA-DR+ CD8 T cells from baseline to post-cycle 1 (Figure 1A) was observed in 8/12 patients who had clinical benefit and 6/15 patients with no clinical benefit (Figure 1B). Transcriptional analysis revealed that in patients with the aforementioned immunological response, T-cells had upregulated TCR signaling, CD28 signaling, enhanced glycolysis and iron uptake, and reduced TGF-beta signaling compared to patients without an immunologic response.

Conclusions

Peripheral blood immune monitoring of RCC patients while on immunotherapy may provide an early predictor of response. One important limitation identified is the treatment specific cycle length defined sample collection timing and therefore may miss transient early immunologic changes. This study advances knowledge regarding the newly generated effector CD8 T cells that contain important information about the immunobiology underlying response to immunotherapy.

Acknowledgements

This work was supported by funding from the NCI grant 1-R00-CA197891 and Nektar Therapeutics. We would like to acknowledge The Yerkes NHP Genomics Core which is supported in part by NIH P51 OD011132, the Emory Flow Cytometry Core (EFCC) supported by the National Center for Georgia Clinical & Translational Science Alliance of the National Institutes of Health under Award Number UL1TR002378, and NIH/NCI under award number, 2P30CA138292-04.

References

1. Motzer RJ, Escudier B, McDermott DF, George S, Hammers HJ, Srinivas S, Tykodi SS, Sosman JA, Procopio G, Plimack ER, Castellano D, Choueiri TK, Gurney H, Donskov F, Bono P, Wagstaff J, Gauler TC, Ueda T, Tomita Y, Schutz FA, Kollmannsberger C, Larkin J, Ravaud A, Simon JS, Xu LA, Waxman IM, Sharma P, CheckMate I. Nivolumab versus Everolimus in Advanced Renal-Cell Carcinoma. N Engl J Med. 2015;373(19):1803-1813.

2. Motzer RJ, Tannir NM, McDermott DF, Aren Frontera O, Melichar B, Choueiri TK, Plimack ER, Barthelemy P, Porta C, George S, Powles T, Donskov F, Neiman V, Kollmannsberger CK, Salman P, Gurney H, Hawkins R, Ravaud A, Grimm MO, Bracarda S, Barrios CH, Tomita Y, Castellano D, Rini BI, Chen AC, Mekan S, McHenry MB, Wind-Rotolo M, Doan J, Sharma P, Hammers HJ, Escudier B, CheckMate I. Nivolumab plus Ipilimumab versus Sunitinib in Advanced Renal-Cell Carcinoma. N Engl J Med. 2018;378(14):1277-1290.

3. Kamphorst AO, Pillai RN, Yang S, Nasti TH, Akondy RS, Wieland A, Sica GL, Yu K, Koenig L, Patel NT, Behera M, Wu H, McCausland M, Chen Z, Zhang C, Khuri FR, Owonikoko TK, Ahmed R, Ramalingam SS. Proliferation of PD-1+ CD8 T cells in peripheral blood after PD-1-targeted therapy in lung cancer patients. Proc Natl Acad Sci U S A. 2017;114(19):4993-4998.

4. Kamphorst AO, Wieland A, Nasti T, Yang S, Zhang R, Barber DL, Konieczny BT, Daugherty CZ, Koenig L, Yu K, Sica GL, Sharpe AH, Freeman GJ, Blazar BR, Turka LA, Owonikoko TK, Pillai RN, Ramalingam SS, Araki K, Ahmed R. Rescue of exhausted CD8 T cells by PD-1-targeted therapies is CD28-dependent. Science. 2017;355(6332):1423-1427.

5. Huang AC, Postow MA, Orlowski RJ, Mick R, Bengsch B, Manne S, Xu W, Harmon S, Giles JR, Wenz B, Adamow M, Kuk D, Panageas KS, Carrera C, Wong P, Quagliarello F, Wubbenhorst B, D'Andrea K, Pauken KE, Herati RS, Staupe RP, Schenkel JM, McGettigan S, Kothari S, George SM, Vonderheide RH, Amaravadi RK, Karakousis GC, Schuchter LM, Xu X, Nathanson KL, Wolchok JD, Gangadhar TC, Wherry EJ. T-cell invigoration to tumour burden ratio associated with anti-PD-1 response. Nature. 2017;545(7652):60-65.

Ethics Approval

Samples are collected under an approved IRB protocol (The Urological Satellite Specimen Bank at Emory University, IRB00055316), and all patients provided informed consent.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P129).

See text for description

Correspondence: Tyler Hulett (tyler.hulett@cdi-lab.com)

Background

Autoantibody landscapes are very specific to the individual, can remain stable for many years, and contain unique features reported in association with cancer, autoimmunity, infection, neurologic conditions, CD8+ T cell behavior, and checkpoint blockade adverse events [1–11].

The goal of this work was to determine whether pre-existing antigen-specific features in melanoma patient autoantibody landscapes would associate with clinical outcomes following checkpoint blockade.

Methods

Pre-treatment serum samples were collected from 117 melanoma patients prior to checkpoint blockade with anti-CTLA4 (N=60), anti-PD1 (N=38), or both in combination (N=16). All data was collected with approval of the NYU Institutional Review Board at the NYU Perlmutter Cancer Center with informed consent [11].

Serum samples were run on HuProt Human Proteome Microarrays containing >19,000 human proteins by CDI Laboratories. Raw serum IgG signal intensities were processed across staining cohorts via interquartile range normalization.

Pre-existing antibody responses were defined as patient-specific IgG signals >3.5 median absolute deviations above cohort median IgG background (modified Z-score). Group statistics were computed (GraphPad Prism), and gene ontology enrichment analysis was performed (Enrichr) [12].

Results

Several pre-existing antigen-specific IgG autoantibody targets were observed to have associations with good outcomes (SD/PR) or objective clinical responses (PR/CR) versus patients with progressive disease (POD). While final determination of the most predictive subsets is ongoing, many targets represent genes in an axis surrounding immune signaling pathways, hereditary neurodegenerative disease, and the ubiquitin proteasome pathway (ie, UBQLN1, UBQLN2).

An exemplary example was observed in the autoantibody responses shared by >10% of all patients regardless of clinical outcome. Gene ontology enrichment analysis of these shared melanoma-patient autoantibodies versus KEGG 2019 [12] demonstrates this set of proteins is strongly enriched for neurotrophin signaling-associated proteins after multi-sample correction (P=0.004) (Table 1). Several other associations were observed cohort-wide for ontologies with tissue-specific enrichment in the brain, neurons, and neuronal processes.

Conclusions

In this pilot study, we found strong associations across the cohort for autoantibodies against nerve-growth-inducing neurotrophins and genes like UBQLN1 and UBQLN2 which have strong associations with amyotrophic lateral sclerosis, frontotemporal dementia, Parkinson’s, and Alzheimer’s – neurodegenerative diseases that are known to have incidences which correlate with melanoma [14–16]; this hints at a potential immunologic connection between the conditions, perhaps related to an antitumor / autoimmune axis involving the targets reported here.

Acknowledgements

We thank the patients and their families who consented to participate in this study. Funding support for the study was provided by the NYU Cancer Center and NIH/NCI Cancer Center Support Grant P30CA016087, the Marc Jacobs campaign to support melanoma research, Goldberg Charitable Trust, Wings for Things Foundation and Clayman Family Foundation to I. Osman; the American Medical Association foundation, the Melanoma Research Foundation and the American Skin Association grants to M. Gowen.

Trial Registration

Patient samples included in this study were not part of a randomized controlled clinical trial.

References

1. Nagele EP, Han M, Acharya NK, DeMarshall C, Kosciuk MC, Nagele RG. Natural IgG Autoantibodies Are Abundant and Ubiquitous in Human Sera, and Their Number Is Influenced By Age, Gender, and Disease. Tsokos GC, editor. PLoS ONE. 2013;8:e60726.

2. Larman HB, Zhao Z, Laserson U, Li MZ, Ciccia A, Gakidis MAM, et al. Autoantigen discovery with a synthetic human peptidome. Nature Biotechnology. 2011;29:535–41.

3. Meyer S, Woodward M, Hertel C, Vlaicu P, Haque Y, Kärner J, et al. AIRE-Deficient Patients Harbor Unique High-Affinity Disease-Ameliorating Autoantibodies. Cell. 2016;166:582–95.

4. Graff JN, Puri S, Bifulco CB, Fox BA, Beer TM. Sustained Complete Response to CTLA-4 Blockade in a Patient with Metastatic, Castration-Resistant Prostate Cancer. Cancer Immunology Research. 2014;2:399–403.

5. Gnjatic S, Ritter E, Büchler MW, Giese NA, Brors B, Frei C, et al. Seromic profiling of ovarian and pancreatic cancer. Proceedings of the National Academy of Sciences. 2010;107:5088–5093.

6. Wongkulab P, Wipasa J, Chaiwarith R, Supparatpinyo K. Autoantibody to Interferon-gamma Associated with Adult-Onset Immunodeficiency in Non-HIV Individuals in Northern Thailand. Rottenberg ME, editor. PLoS ONE. 2013;8:e76371.

7. Anderson KS, Sibani S, Wallstrom G, Qiu J, Mendoza EA, Raphael J, et al. Protein Microarray Signature of Autoantibody Biomarkers for the Early Detection of Breast Cancer. Journal of Proteome Research. 2011;10:85–96.

8. Miersch S, Bian X, Wallstrom G, Sibani S, Logvinenko T, Wasserfall CH, et al. Serological autoantibody profiling of type 1 diabetes by protein arrays. Journal of Proteomics. 2013;94:486–96.

9. Srivastava RM, Lee SC, Andrade Filho PA, Lord CA, Jie H-B, Davidson HC, et al. Cetuximab-Activated Natural Killer and Dendritic Cells Collaborate to Trigger Tumor Antigen-Specific T-cell Immunity in Head and Neck Cancer Patients. Clinical Cancer Research. 2013;19:1858–72.

10. Hulett TW. Coordinated responses to individual tumor antigens by IgG antibody and CD8+ T cells following cancer vaccination. 2018;14.

11. Gowen MF, Giles KM, Simpson D, Tchack J, Zhou H, Moran U, et al. Baseline antibody profiles predict toxicity in melanoma patients treated with immune checkpoint inhibitors. Journal of Translational Medicine [Internet]. 2018 [cited 2018 Nov 4];16. Available from: https://translational-medicine.biomedcentral.com/articles/10.1186/s12967-018-1452-4

12. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, Clark NR, Ma'ayan A. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;128(14).

13. Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, Koplev S, Jenkins SL, Jagodnik KM, Lachmann A, McDermott MG, Monteiro CD, Gundersen GW, Ma'ayan A. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Research. 2016; gkw377 .

14. Olsen JH, Friis S, Frederiksen K. Malignant Melanoma and Other Types of Cancer Preceding Parkinson Disease: Epidemiology. 2006;17:582–7.

15. Freedman DM, Curtis RE, Daugherty SE, Goedert JJ, Kuncl RW, Tucker MA. The association between cancer and amyotrophic lateral sclerosis. Cancer Causes & Control. 2013;24:55–60.

16. Roe CM, Fitzpatrick AL, Xiong C, Sieh W, Kuller L, Miller JP, et al. Cancer linked to Alzheimer disease but not vascular dementia. Neurology. 2010;74:106–12.

Ethics Approval

All data collected for this study was collected with approval of the NYU Institutional Review Board at the NYU Perlmutter Cancer Center with informed consent.

Consent

No sensitive or patient identifiable information is included in the data presented.

Correspondence: Tong Seng Lim (tongseng.lim@mbiomarkers.com)

Background

Tumor microenvironment contains a diverse array of cell types with heterogeneous genomic and molecular profiles. Averaging the characteristics of all cells in a cancerous tissue no doubt obscures important variations in biomarkers among minority, but critical, pathogenic cell populations. High resolution, single-cell analyses are thus needed in the clinic to precisely delineate the inherent heterogeneity of tumor microenvironment underlying oncogenesis in each patient. Biopsies derived from tumor microenvironment are routinely used for medical diagnosis or prognostications. The number of cells typically available from a biopsy is limited and heterogeneous. The ability to distinguish, select, and sort rare malignant or pathogenic immune cells from either tissue or liquid biopsies poses a unique challenge for single-cell based diagnosis and prognostication. Heterogeneous cell populations with non-target immunoreactive cells in pausicellular biopsies complicate conventional bulk-cell analysis, and could lead to disease misdiagnosis or prognostication.

Methods

We adopted a state-of-the-art multi-modal strategy including the real-time imaging-based DEPArray with downstream molecular and genomic assays [1], and quantitative multiplex immunofluorescent technique [2] in order to predict clinical outcome or direct therapy, based on single-cell based diagnostic or prognostic biomarkers derived from tumor microenvironment.

Results

We provided proof-of concepts that DEPArray technology enabled automated isolation and recovery of rare malignant or pathogenic immune cells from liquid or tissue biopsies in several malignancies including vitreoretinal lymphoma (VRL), hepatocellular carcinoma (HCC) and colorectal carcinoma (CRC). Rare target B lymphoma cells were distinguished and sorted from pausicellular ocular vitreous biopsies with high resolution and purity required for sensitive single-cell based MYD88 mutational profiling to aid VRL diagnosis [1]. Single cellular imaging revealed the presence of large (>10μm), irregular shaped of a novel population of HCC-infiltrating macrophages in association with improved prognosis after surgery. A unique signature regulatory T-cells (Tregs) population was identified in both blood circulation and cancerous tissues of CRC. These signature Tregs expressed phenotypically distinct surface markers in association with better disease-free and overall survival of CRC patients.

Conclusions

Using real-time imaging-based, digital sorting DEPArray, we could distinguish, select and sort different types of malignant or target immune cells including B-cells, T-cells and macrophages from heterogeneous tumor microenvironment or liquid biopsies with low cellularity. Comprehensive genomic and molecular characterizations at single cell resolution revealed crucial biomarkers associated with clinicopathological features that impact clinical outcome of patients. The single cell interrogation using DEPArray technology provides a novel precision medicine tool for diagnostics and prognostications of malignancies in future.

Acknowledgements

This study was supported by research funding from the research collaboration between A. Menarini Biomarkers Singapore Pte Ltd, Singapore Eye Research Institute and Singapore General Hospital. Some data presented here are part of patent filed on 21 August 2018 (#10201807097T).

References

1. Tan, W.J., et al., Single-cell-MYD88 sequencing of isolated B cells from vitreous biopsies aids vitreoretinal lymphoma diagnosis. Blood, 2019.

2. Lim, J.C.T., et al., An automated staining protocol for seven-colour immunofluorescence of human tissue sections for diagnostic and prognostic use. Pathology, 2018. 50(3): p. 333-341.

Ethics Approval

This study was approved by the SingHealth Institutional Review Board in accordance with the Singapore Guidelines for Good Clinical Practice and the Declaration of Helsinki, approval number #2009/907/B, 2012/104/F and #2017/2494

Correspondence: Tony Lim (Lim.Kiat.Hon@singhealth.com.sg); Jinmiao Chen (Chen_Jinmiao@immunol.a-star.edu.sg)

Background

In the advent of immuno-technology, newer single-cell flow cytometry techniques have greatly increased the capacity for the maximum number of immunological parameters measured. Notably, multiplex-immunofluorescence (mIF) can perform measurements for 7 markers, flow cytometry can handle 20, and imaging mass cytometry can process up to 37 biomarkers simultaneously. Hence, dimensionality reduction techniques such as t-SNE and UMAP are becoming increasingly important for tumour single-cell data analysis. Using human hepatocellular carcinoma (HCC) tissue samples, we aim to compare and evaluate the use of a new technique, UMAP, as an alternative to t-SNE in mIF derived single-cell data.

Methods

We adopted an unsupervised clustering approach using FlowSOM to identify 8 major cell types present in human HCC tissues by staining them with 7 markers, including immune-checkpoint molecules and one nuclear counterstain. Following that, UMAP and t-SNE were ran independently on the dataset to qualitatively compare the distribution of clustered cell types in both dimensionality reduction tools.

Results

The key advantage of UMAP is its superior runtime – it takes approximately one-fifth the time required to run t-SNE. Both techniques provide similar arrangements of cell clusters, with the key difference being UMAP’s extensive characteristic branching. Also, increasing perplexity values in t-SNE results in a t-SNE visualisation with certain degrees of branching like that of UMAP’s, albeit limited. When parameters such as standard deviation, minimum and maximum intensity from the mIF image cytometry data were included, a t-SNE plot with virtually the same morphology as the resulting UMAP plot can be visualised. Most interestingly, UMAP’s branching highlighted biological lineages, especially in identifying potential hybrid tumour cells (HTC). Survival analysis shows patients with higher proportion of HTC have a worse prognosis (p-value = 0.019).

Conclusions

We conclude that both techniques are similar in their visualisation capabilities, but UMAP has a clear advantage over t-SNE in runtime, making it highly plausible to employ UMAP as an alternative to t-SNE in single-cell data analysis.

Correspondence: Cynthia Sirard (csirard@leaptx.com)

Background

Dickkopf-1 (DKK1), a secreted modulator of Wnt signaling, contributes to an immune suppressive tumor microenvironment and promotes tumor growth, angiogenesis and metastasis. DKN-01, a DKK1 neutralizing antibody, has demonstrated clinical activity across multiple solid tumors as both a monotherapy and in combination with checkpoint inhibitors and chemotherapies. Nonclinical studies indicate that DKN-01 efficacy depends on a functioning immune system, notably natural killer (NK) cells. High tumor expression of DKK1 correlates with a worse clinical prognosis in many solid tumors. As such, we evaluated tumor levels of DKK1 and association with clinical outcomes for DKN-01 based therapies.

Methods

DKK1 mRNA expression in patient tumor biopsies was evaluated with a RNAscope in situ hybridization assay. Expression levels were semi-quantified with QuPath or manually scored. Data was pooled from three separate clinical trials, DKN-01 as monotherapy or in combination with paclitaxel or pembrolizumab in esophagogastric cancer (EGC) (NCT02013154), DKN-01 as monotherapy or in combination with paclitaxel in epithelial endometrial cancer (EEC) or epithelial ovarian cancer (EOC) (NCT03395080) and DKN-01 in combination with gemcitabine/cisplatin in biliary tract cancer (BTC) (NCT02375880). Survival analysis was performed by the Kaplan-Meier method and multi-variable Cox proportional-hazards and logistic regression models were used to study the association of DKK1 H-score cutoffs (tertiles and quartiles) with survival and clinical benefit (CR, PR or SD per RECIST v1.1) outcomes.

Results

A total of 120 patients (59 EGC, 28 EEC, 20 EOC and 13 BTC) had DKK1 tumor expression with response and survival outcomes. Patients who had an H-score ≥ upper-quartile (≥50) of DKK1 expression versus < upper-quartile had a higher-odds of having clinical benefit/response with an adjusted OR of 4.46 (95% CI: 1.78, 11.71) and a longer PFS with an adjusted HR of 0.49 (95% CI: 0.29, 0.81). Patients who had an H-score ≥ upper-tertile (≥35) versus < upper-tertile had a higher-odds of having clinical benefit/response with an adjusted OR of 2.82 (95% CI: 1.21, 6.67) and a longer PFS with an adjusted HR of 0.53 (95% CI: 0.33, 0.85).

Conclusions

Elevated DKK1 tumor expression was associated with a higher clinical benefit/response rate and longer PFS for DKN-01 based treatments across multiple solid cancers. Tumor expression of DKK1 may represent an important patient selection criterion for further development of DKN-01 based therapies in solid cancers. The contribution of high levels of tumor DKK1 expression to an immune suppressive tumor microenvironment and the role of NK cells is currently under investigation.

Ethics Approval

Studies were approved by the Institutional Review Boards of each participating institution.

Correspondence: Lukas Käsmann (lkaesmann@gmail.com)

Background

The prognostic role of the tumor immunity microenvironment (TIME) in multimodal treatment for locally advanced non-small cell lung cancer (LA-NSCLC) is unclear. Increasing evidence suggests treatment benefit depending on tumor cell PD-L1 expression. The purpose of this retrospective single-center study was to investigate the prognostic value of PD-L1 expression on tumor cells in combination with CD8+ tumor stroma-infiltrating lymphocytes (TILs) density in inoperable LA-NSCLC treated with concurrent chemoradiotherapy (CRT).

Methods

We collected retrospectively clinical characteristics and initial tumor biopsy samples of 31 inoperable LA-NSCLC patients treated with concurrent CRT. PD-L1 expression on tumor cells (0% versus ≥1%), CD8+ TILs density (0-40% vs. 41-100%) and TIME according to classification by Zhang et al. were evaluated for potential prognostic value in terms of local control, progression-free (PFS) and overall survival (OS) as well as correlations with clinic-pathological features investigated.

Results

Median OS was 14 months (range: 3-167 months). The OS rates at 1- and 2 years were 68% and 20%. Local control rates for the entire cohort at 1 and 2 years were 74% and 61%, respectively. Median PFS and PFS at 1 and 2 years were 13±1.4 months, 58% and 19%. PD-L1 expression <1% on tumor cells was associated with improved OS, PFS and local control in patients treated with concurrent CRT. Univariate analysis showed a trend for improved OS and local control in patients with low CD8+ TILs density. Evaluation of TIME appears to be an independent prognostic factor for local control, PFS and OS. The longest and shortest OS were achieved in patients with type I (PD-L1neg/CD8low) and type IV (PD-L1pos/CD8low) tumors (median OS: 57±37 vs. 10±5 months, p=0.05), respectively.

Conclusions

Assessment of the tumor immunity microenvironment (TIME) by PD-L1 expression on tumor cells and CD8+ TILs density is a predictive biomarker in patients treated with concurrent CRT for inoperable LA-NSCLC.

Acknowledgements

The study was funded by the German Center for Lung Research (DZL).

Ethics Approval

The study was approved by the University Ethics Board, approval number 493-16.

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

Correspondence: Sarita Agarwal (saritasgpgi@gmail.com)

Background

Loss of Y chromosome (LOY) is a well know established phenomenon associated with cancers and ageing. Recently, LOY in peripheral blood cells was suggested as a possible biomarker for different cancers in males. On the basis of previous findings, the present case-control study was conducted to evaluate the association of LOY in peripheral blood cells in prostate (PC) and colorectal cancers (CRC) in males [1-4].

Methods

30 CRC patients (mean age = 44.03±10.8), 36 PC patients (mean age = 60.8 ± 15.8 yrs) and 36 healthy control male cases (mean age = 54.6± 15.1 years) were recruited. DNA was extracted by using a standard phenol-chloroform method. Multiplex quantitative fluorescent (QF) PCR was used to co-amplify the homologous sequences present on the Y chromosome and other chromosome followed by their analysis on the genetic analyzer (ABI 3500) and finally the Y/X ratio was calculated on the basis of the peak height obtained from the electropherogram.

Results

The mean Y/X ratio was significantly lower in the whole group of cancer patients (0.709±0.02; p <0.0001) when compared to the controls (0.92±0.044). Also, the Y/X ratio when calculated separately was found to be lower in CRC (0.701±0.078; p <0.0001) and PC (0.717±0.044; p <0.0001) cases, when compared to controls (0.92±0.044). Multivariate logistic regression was performed by matching cancer and control subjects with age and the results suggest that LOY is not influenced by their age.

Conclusions

The results support the significant association of LOY in peripheral blood cells carcinogenesis in males. LOY can also serve as a non-invasive cancer biomarker to improve the early diagnosis and management of cancer patients in males.

Acknowledgements

We are highly grateful to Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, Uttar Pradesh, India for providing the infrastructure and lab facilities for research work. The authors also thank all the consultant and residents of SGPGIMS, who helped in carrying out the study. Dr Ambreen Asim is the first author, who collected data, carried out all the practical work and drafted this abstract. Prof. Sarita Agarwal is the corresponding and second author, who helped in finalizing, correcting and critical review of the work. Prof.Rakesh Kapoor and Prof. Neeraj Rastogi are the oncologist consultant who has provided prostate and colorectal cancer patients blood samples after taking informed consent for this study.

References

1. Noveski P, Madjunkova S, Stefanovska E. Loss of Y Chromosome in Peripheral Blood of Colorectal and Prostate Cancer Patients. Plos One. 2016;11(1).

2. Chang Y M, Perumal R, Keat P Y, Rita Y.Y. Yong, Daniel L.C. Kuehn, Leigh Burgoyne. A distinct Y-STR haplotype for Amelogenin negative males characterized by a large Yp11.2 (DYS458-MSY1-AMEL-Y) deletion. Forensic Sci Int. 2007; 166(2-3):115-20.

3. Donaghue C, Mann K, Docherty Z and Ogilvie C M. Detection of mosaicism for primary trisomies in prenatal samples by QF-PCR and karyotype analysis.Prenat Diagn .2005; 25: 65–72.

4. Plaseski T, Noveski P, Trivodalieva S, Georgi D. Quantitative Fluorescent-PCR Detection of Sex Chromosome Aneuploidies and AZF Deletions /Duplications. Genetic Teasting, 2008 : 12: 4.

Ethics Approval

This study was approved by Sanjay Gandhi Post Graduate Institute of Medical Sciences Ethics Board; approval number IEC CODE – 2018- 53-IMP-103 dated 18th June 2018.”

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

Correspondence: Kurt Schalper (kurt.schalper@yale.edu)

Background

Regulatory T cells (Tregs) mediate potent tolerogenic signals, are involved in adaptive anti-tumor immune responses and T-cell reinvigoration using immune checkpoint blockers. Despite their prominent immune suppressive role, the tissue distribution and contribution of Tregs to clinical outcomes in human lung cancer is not well understood.

Methods

The levels and tissue distribution of Tregs and major tumor infiltrating lymphocyte (TIL) subsets were measured using simultaneous detection of FOXP3, CD4, CD8, pancytokeratin and DAPI by multiplexed quantitative immunofluorescence in 619 formalin-fixed paraffin embedded (FFPE) NSCLCs from 4 independent cohorts represented in tissue microarrays (cohort #1 [Yale, n=210], cohort #2 [Greece, n=192]; cohort #3] [80 immunotherapy-treated NSCLCs]; cohort #4 [Yale, n=137, adenocarcinomas with mutation testing). Markers were measured in different tissue compartments and cell phenotypes were used for individual cell counts and machine-learning-based spatial analysis. We studied the association between T-cell populations, tissue distribution, clinicopathologic/molecular characteristics and outcomes.

Results

Tregs (DAPI+/CD4+/FOXP3+ cells) were predominantly located in the stromal compartment and represented 3-10% of the total T-cell population. The level of Tregs was positively associated with higher CD8+ T-cell infiltration across the cohorts. There was no consistent association between Treg levels and patient age, gender, smoking status, clinical stage or tumor histology. However, Tregs were significantly higher in KRAS mutated lung adenocarcinomas than in EGFR mutant or KRAS/EGFR wild-type cases. As a single marker, the level of Tregs was not significantly associated with survival. However, the Treg to CD8 signal ratio was associated with shorter 5-year overall survival across the cohorts. Reduced survival was also seen in cases with a higher 5-nearest neighbor (5NN) mean distance between CD4+/Tregs and CD8+/CD4+ cells. Notably, the survival effect of the Treg-associated metrics was numerically higher in patients treated with immune checkpoint blockers.

Conclusions

Tregs are prominently less abundant than other TIL subsets in NSCLC microenvironments and they are increased in T-cell inflamed tumors. Their positive association with CD8+ cytotoxic TILs suggests their upregulation upon adaptive anti-tumor immune pressure and could explain the inconsistent reported relationship between Tregs and prognosis. Elevated Treg to CD8 signal ratio and reduced spatial clustering between CD4-Tregs and CD8-CD4 are indicative of poor outcome preferentially in NSCLC patients treated with checkpoint blockade suggesting a biomarker role.

Ethics Approval

All tissues were used after approval from the Yale Human Investigation committee protocol #9505008219 which approved patient consent forms or waivers of consent.

Correspondence: Subham Basu (Subham.Basu@abcam.com)

Background

PD-L1 protein expression by immunohistochemistry (IHC) measurement is the only FDA-approved protein diagnostic biomarker for PD1/PD-L1 immunotherapies [1]. However, the tumor-immune interaction is complex: PD-L1 expression alone is not predictive of patient response [2]. This has led to the investigation of other PD-1 ligands such as PD-L2 [2]. PD-L2 even in the absence of PD-L1 has been associated with clinical response to PD-1 blockade in multiple tumor types [2]. PD-L2 status has also been investigated where immunotherapy based on PD-L1 has been less successful, such as prostate cancer, where PD-L1 expression is typically low [3]. Here, significantly higher levels of PD-L2 were associated with multiple survival and response measures [3]. Due to the diagnostic and therapeutic potential indicated by the presence of this key immune checkpoint ligand in patients irrespective of PD-L1 expression [2-4], dependable detection tools for investigating the presence and role of PD-L2 are crucial. To address this need, Abcam have developed and extensively characterized and validated a recombinant rabbit monoclonal antibody specific to PD-L2 (CAL28). For checkpoint inhibitors, Abcam already has research use only versions of three anti-PD-L1 RabMAb® antibodies employed in the clinical setting (73-10, 28-8 and SP142), co-developed with pharmaceutical and diagnostic companies.

Methods

A recombinant rabbit monoclonal antibody was generated using a direct B cell cloning process and characterized for IHC. The clone was tested using PD-L2-transfected and non-transfected HEK293 cells fixed in formaldehyde and processed into paraffin wax (FFPE) and further validated alongside In Situ Hybridization (ISH) for PD-L2 mRNA in FFPE commercial cell lines. Once specificity was determined, it was tested in positive and negative tissues and TMAs of Head & Neck Squamous Cell Carcinoma (HNSCC), Prostate Carcinoma (PC) and Renal Cell Carcinoma (RCC).

Results

CAL28 demonstrated positive IHC staining on PD-L2-overexpressed HEK293 cells processed in FFPE with a lack of staining in the parental line. Additionally, CAL28 demonstrated IHC staining in FFPE cell lines where PD-L2 expression was confirmed with ISH for PD-L2 mRNA. Expression in tumor tissue in TMAs from HNSCC, PC and RCC was evaluated with no non-specific background staining.

Conclusions

We have demonstrated sensitivity, specificity and reproducibility of a recombinant rabbit monoclonal antibody to PD-L2 in IHC (CAL28). The global, commercial availability of this recombinant clone to researchers, pathologists, clinicians and the biopharmaceutical industry will enable further progress to be made in understanding the clinical relevance and predictive value that PD-L2 promises for cancer immunotherapy.

References

1. Tsao MT, Kerr K, Yatabe Y, et al. PL 03.03 Blueprint 2: PD-L1 Immunohistochemistry Comparability Study in Real-Life, Clinical Samples. J Thor Oncol. 2017;12(Suppl 2):S1606

2. Yearley JH, Gibson C, Yu N, Moon C, Murphy E, Juco J, Lunceford J, Cheng J, Chow LQM, Seiwert TY, Handa M, Tomassini JE, McClanahan T. Clin Cancer Res. 2017;23:3158-3167

3. Zhao SG, Lehrer J, Chang SL, Das R, Erho N, Liu Y, Sjöström M, Den RB, Freedland SJ, Klein EA, Karnes RJ, Schaeffer EM, Xu M, Speers C, Nguyen PL, Ross AE, Chan JM, Cooperberg MR, Carroll PR, Davicioni E, Fong L, Spratt DE, Feng FY.The Immune Landscape of Prostate Cancer and Nomination of PD-L2 as a Potential Therapeutic Target. J Natl Cancer Inst. 2019;111:301-310

4. Takamori S, Takada K, Toyokawa G, Azuma K, Shimokawa M, Jogo T, Yamada Y, Hirai F, Tagawa T, Kawahara A, Akiba J, Okamoto I, Nakanishi Y, Oda Y, Hoshino T, Maehara Y. PD-L2 Expression as a Potential Predictive Biomarker for the Response to Anti-PD-1 Drugs in Patients with Non-small Cell Lung Cancer. Anticancer Res. 2018;38:5897-5901

Correspondence: Cara Mathews(cmathews@wihri.org)

Background

Multiple immunotherapies have been evaluated in patients with advanced or recurrent endometrial cancer (EC) using molecular biomarkers, including microsatellite instability-high (MSI-H) and stable (MSS) status. Clinical outcomes appear to be different in patients with MSI-H/mismatch repair (MMR)-deficient status versus MSS/MMR-proficient status when receiving anti-programmed cell death (ligand) 1 (PD-[L]1) therapies [1,2]. It is unclear if these differences are due to the therapies themselves or to differences inherent to the patient populations.

We sought to evaluate the association between MSI-H/deficient MMR (dMMR) status and response among patients with advanced or recurrent EC.

Methods

We conducted a systematic review of the Embase, MEDLINE, and Cochrane Central Register of Controlled Trials databases from 2000 to present to identify publications (manuscripts and conference proceedings) on studies using chemotherapy, surgery, radiotherapy, hormonal therapy, or biological therapy (or any combination thereof) in adult patients (≥18 years) with stage III or IV advanced or recurrent EC, and where MMR or MSI status was identified (by any means). To better understand the prognostic value of MSI-H/MSS status, we excluded anti-PD-(L)1 therapies from the analysis, as recent evidence suggests that there is a positive predictive value for these agents in patients with MSI-H/dMMR status [2-4].

Results

Our systematic review of MSI/MMR status and recurrence-free survival (RFS), progression-free survival (PFS), and overall survival (OS) identified a total of 5 studies. One study reported dMMR status was associated with a reduction in RFS (hazard ratio, 2.02) [5], while another study found no significant effect [6]. A third study reported a trend towards a higher rate of recurrence among patients with advanced-stage EC with dMMR than among patients with MMR proficiency (P value not reported) [7]. Two studies reported no significant association between PFS and dMMR status [8,9]. Three studies found no statistically significant association between OS and dMMR status [5,6,8].

Conclusions

This review could not identify a consistent association between dMMR or MSI-H status and recurrence, RFS, PFS, or OS among patients with advanced or recurrent EC receiving therapy other than anti-PD-(L)1. For RFS, where differences were present, they trended towards worse outcomes for patients with MSI-H/dMMR status. Consequently, we have identified no evidence of a prognostic or predictive value of MSI-H or dMMR biomarker status for efficacy outcomes in patients with advanced or recurrent EC receiving non–anti-PD-(L)1 therapy. Further investigation into the prognostic or predictive value of MSI-H/dMMR status is warranted.

Acknowledgements

Clinical Trial Registration: N/A

References

1. Segal NH, Wainberg ZA, Overman MJ, et al. Safety and clinical activity of durvalumab monotherapy in patients with microsatellite instability–high (MSI-H) tumors [abstract]. J Clin Oncol. 2019; 37(Suppl 4):670.

2. Konstantinopoulos PA, Liu JF, Luo W, et al. Phase 2, two-group, two-stage study of avelumab in patients (pts) with microsatellite stable (MSS), microsatellite instable (MSI), and polymerase epsilon (POLE) mutated recurrent/persistent endometrial cancer (EC) [abstract]. J Clin Oncol. 2019; 37(Suppl 15):5502.

3. Oaknin A, Duska LR, Sullivan RJ, et al. Preliminary safety, efficacy, and pharmacokinetic/pharmacodynamic characterization from GARNET, a phase I/II clinical trial of the anti–PD-1 monoclonal antibody, TSR-042, in patients with recurrent or advanced MSI-H and MSS endometrial cancer. Gynecol Oncol. 2019; 154:17.

4. Antill YC, Kok PS, Robledo K, et al. Activity of durvalumab in advanced endometrial cancer (AEC) according to mismatch repair (MMR) status: The phase II PHAEDRA trial (ANZGOG1601). J Clin Oncol 37, 2019 (suppl; abstr 5501).

5. Djordjevic B, Bruegl A, Fellman B, et al. The prognostic effect of MLH1 loss in endometrial endometrioid adenocarcinoma. Lab Invest. 2014; 94:280A–281A.

6. Cohen JG, Goodman MT, Karlan BY, Walsh C. Genomic characterization of grade 3 endometrial carcinoma. Gynecol Oncol. 2014; 133:134–135.

7. Cosgrove CM, Cohn DE, Hampel H, et al. Epigenetic silencing of MLH1 in endometrial cancers is associated with larger tumor volume, increased rate of lymph node positivity and reduced recurrence-free survival. Gynecol Oncol. 2017; 146(3):588–595. doi:10.1016/j.ygyno.2017.07.003.

8. Kim SR, Pina A, Albert A, et al. Does MMR status in endometrial cancer influence response to adjuvant therapy? Gynecol Oncol. 2018; 151:76–81.

9. Aghajanian C, Filiaci V, Dizon DS, et al. A phase II study of frontline paclitaxel/carboplatin/bevacizumab, paclitaxel/carboplatin/temsirolimus, or ixabepilone/carboplatin/bevacizumab in advanced/recurrent endometrial cancer. Gynecol Oncol. 2018; 150:274–281.

Correspondence: Bernard Fox(foxb@foxlab.org)

Background

Head and neck squamous cell cancer (HNSCC) ranks as the 6th most common cancer afflicting humans and remains a significant unmet medical need. While interfering with the PD-1/PD-L1 axis improves outcomes, the majority of patients progress and die of their disease. To address this lack of efficacy our group has explored the immune makeup of HNSCC, hypothesizing that a better characterization of responders and non-responders will result in improved predictive biomarkers and insights into strategies to improve outcomes for the majority of patients.

Methods

Over the past 7 years we have collected and processed more than 350 HNSCC specimens. When sufficient tumor material was available, tumor-infiltrating lymphocytes (TIL) and primary tumor cultures were initiated and characterized for autologous tumor reactivity. Once established, tumor cell lines were characterized for phenotypic markers by flow cytometry. Flow cytometric analysis and RNASeq has been performed on some established cell lines and on FFPE tumor specimens.

Results

Consistent with previous reports increased expression of CD8 T cells was associated with improved outcome. In preliminary studies increased expression of GITR was also associated with improved outcome. Initial speculation was that GITR expression was coming from immune infiltrates. Subsequently a report suggested that GITR could be expressed by HNSCC. Using flow cytometry we detected low level GITR expression on 3 HNSCC cell lines and low to high level expression of GITR-L on a 8 HNSCC cell lines. Studies are continuing to expand on these preliminary observations.

Conclusions

Anti-GITR and GITR-L both have the potential to provide positive signals to immune cells. In addition to APC’s, GITR-L expression by some HNSCC cells may contribute to the make-up of the immune cells infiltrating these cancers.

Acknowledgements

Funding Support: The Harder Family, Robert and Elsie Franz, Wes and Nancy Lematta, Lynn and Jack Loacker, the Providence Portland Medical Foundation and the Oral and Maxillofacial Surgery Foundation, The Murdock Trust.

Ethics Approval

The study was approved by the institutional review board of the Providence Portland Medical Center (12-075A).

Correspondence: Kurt Schalper (kurt.schalper@yale.edu)

Background

Interleukin-8 (IL-8) is a chemokine expressed in multiple cancer types, including NSCLC. It exerts various functions in shaping cancer vascularization, cell dedifferentiation and inflammation/immunity. IL-8 was described as a chemotactic factor for neutrophils and it has been proposed to mediate recruitment of tolerogenic myeloid cells favoring a pro-tumorigenic microenvironment. Although clinical trials targeting IL-8 are ongoing, its expression and role in NSCLC is unclear.

Methods

We developed a multiplexed quantitative immunofluorescence (QIF) panel for simultaneous and localized measurement of IL-8, myeloperoxidase (MPO), CD15, cytokeratin (CK) and DAPI. We analyzed the expression of these markers and their association with PD-L1, CD4 and CD8-positive cells in 3 retrospective NSCLC immunotherapy-naive cohorts represented in tissue microarrays (cohort #1, n=262; #2, n=145; and #3, n=132); 1 cohort of NSCLC patients treated with immune checkpoint blockers (#4, n=59) and 1 collection of lung adenocarcinomas (LAC) analyzed for activating mutations in EGFR and KRAS (#5, n=121). We studied the level of the targets, their distribution and association with immune features, clinicopathological variables and survival.

Results

IL-8 protein signal was detected in ~85% of cases with cytoplasmic staining pattern and was higher in tumor than in stromal cells. Elevated tumor IL-8 was consistently associated with higher MPO+ neutrophils and CD15+ tumor-associated myeloid cells across the cohorts, but not with CD4+ and CD8+ T-cells. Increased IL-8 expression was not associated with major clinicopathologic variables. Elevated MPO+ and CD15+ cells was significantly higher in KRAS mutated than in EGFR mutated LACs. High MPO and CD15 signal was associated with shorter 5-year overall survival in all NSCLC cohorts. The negative prognostic effect of MPO and CD15 was comparable in both immunotherapy-naïve and immunotherapy-treated NSCLC collections.

Conclusions

IL-8 protein is frequently expressed in NSCLCs associated with increased tumor-associated myeloid cells but independent from intratumor T-cell responses. KRAS mutated LACs have prominent MPO+/CD15+ expression, supporting an immune suppressive role of myeloid cells in these malignancies. CD15 and MPO are prognostic markers in NSCLC and IL-8 blockade could mediate favorable immunomodulatory effects.

Ethics Approval

All tissues were used after approval from Yale Human Investigation committee protocol #9505008219 which approved the patient consent forms or waivers of consent

Correspondence: Gregory Vidal (gvidal@westclinic.com)

Background

Immune checkpoint inhibitors (PD-(L)1 inhibitors) have shown promising therapeutic outcomes and have been approved for multiple indications. However, widespread use of PD-(L)1 inhibitors has been limited by a low response rate and immune-related adverse events. Therefore, an improved method for predicting response to the immune checkpoint blockade would better identify patients misclassified by conventional testing. We have evaluated a proprietary algorithm which utilizes gene expression in solid tumors to assess the presence of an immunomodulatory (IM) signature intended to predict immunotherapy response. The purpose of this study was to evaluate the performance of the IM signature against progression-free survival (PFS) of patients treated with immune checkpoint inhibitors.

Methods

In this retrospective study, archival tumor tissue from metastatic lung cancer patients treated with one of three PD-(L)1 inhibitors (pembrolizumab, nivolumab, and atezolizumab) either as a single agent or in conjunction with standard chemotherapy, from whom response data was available, was tested for the IM signature. Patients were stratified into two groups based on IM signature classification as positive or negative, which was compared to immunohistochemistry PD-(L)1 testing with a primary endpoint of one-year progression-free survival. Additionally, the IM signature classification was compared with objective response by Spearman’s correlation as a continuous variable.

Results

A total of 71 metastatic lung cancer patients were included in the study with a median follow-up of 29 months. The one-year PFS hazard ratio for the IM positive group was 0.31 (95% CI 0.14 to 0.68; p=.004 - Figure 1). A total of 62 out of the 71 metastatic lung cancer patients had previous PD-L1 staining. Head-to-head analysis of PD-L1 and IM signature on these patients found the one-year PFS hazard ratio for the IM group to be 0.30 (95% CI 0.13 to 0.71; p=0.006) and the one-year PFS hazard ratio for PD-L1 positive staining to be 0.76 (95% CI 0.31 to 1.82; p=0.533). The mean IM correlation value with objective response for PD = -0.06; SD = -0.04; PR = 0.14; CR = 0.33; p

Conclusions

The IM signature was significantly associated with prolonged one-year progression-free survival among patients treated with PD-L1 inhibitors while PD-L1 staining failed to be significantly associated. Patients classified positive by the IM signature demonstrated a three-fold improved hazard ratio compared to those who were negative. Funding was provided by Insight Genetics working in cooperation with West Cancer Center and Research Institute.

Ethics Approval

This study was approved by the West Cancer Clinic Institutional Review Board.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P141).

See text for description

Correspondence: Parantu Shah (parantu.shah@emdserono.com); Matthew Salter (matthew.salter@oxfordbiodynamics.com)

Background

Development of baseline predictive classifiers for response to treatment can provide advantages for programs of targeted immunotherapies, development of successful combination therapies, and identification of responder populations to active therapies. Chromosome conformations represent strong systemic cellular network deregulations associated with differences in clinical phenotypes and outcomes [1].

Methods

Oxford Biodynamics, in collaboration with the EMD Serono, Inc., a business of Merck KGaA, Darmstadt, Germany/Pfizer alliance," has applied its proprietary technology EpiSwitchTM to monitor systemic epigenetic biomarkers for chromosome conformation signatures in baseline blood samples of patients with multiline anti–PD-L1 (avelumab) treatment of NSCLC. This application was based on the published methodology for validated predictive biomarkers for response to treatment [2], systemic blood-based monitoring of oncological conditions [3-5], and proprietary programs in collaboration with the Mayo Clinic for predictive and response biomarkers in melanoma patients treated with anti–PD-1 therapy (pembrolizumab).

Results

A 14-marker classifier was generated with 12 avelumab-treated patients in each response group; in this cohort, responders were defined as patients with complete or partial response, and non-responders were defined as patients with progressive disease. Validation of the developed predictive markers was performed on an independent cohort of 75 patients treated with avelumab as either first-line (1L) or 2L therapy. In the validation cohort, patients with stable disease were also considered as responders in addition to above. The classifier delivered stratifications for response vs nonresponse with 84% accuracy, 79% sensitivity, 92% specificity, 75% positive predictive value (PPV) and 95% negative predictive value (NPV). The associations of EpiSwitch™ response calls with overall survival (OS) and progressive free survival (PFS) in the independent cohort were significant (OS and PFS: log-rank p

Conclusions

The established EpiSwitchTM classifier contains strong binary markers of epigenetic deregulation with features normally attributed to genetic markers; the binary status of these classifying markers is statistically significant for survival. Altogether, these findings highlight the potential of the EpiSwitchTM approach for identifying responders and non-responders to immuno-oncology therapies.

Acknowledgements

The authors would like to thank patients enrolled in the EMR000070-001 JAVELIN Solid Tumor trial for agreeing for their samples to be used for research purposes. This work is funded by Merck KGaA, Darmstadt, Germany, as part of an alliance between Merck KGaA, Darmstadt, Germany and Pfizer Inc., New York, NY, USA.

References

1. Tordini F, Aldinucci M, Milanesi L, et al. The genome conformation as an integrator of multi-omic data: the example of damage spreading in cancer. Front Genet. 2016; 7:194.

2. Carini C, Hunter E; Scottish Early Rheumatoid Arthritis Inception Cohort investigators, et al. Chromosome conformation signatures define predictive markers of inadequate response to methotrexate in early rheumatoid arthritis. J Transl Med. 2018; 16:18.

3. Jakub JW, Grotz TE, Jordan P, et al. A pilot study of chromosomal aberrations and epigenetic changes in peripheral blood samples to identify patients with melanoma. Melanoma Res. 2015; 25:406-11.

4. Bastonini E, Jeznach M, Field M, et al. Chromatin barcodes as biomarkers for melanoma. Pigment Cell Melanoma Res. 2014; 27:788-800.

5. Yan H, Hunter E, Akoulitchev A, et al. Epigenetic chromatin conformation changes in peripheral blood can detect thyroid cancer. Surgery. 2019; 165:44-49.

Ethics Approval

The protocol was approved by the institutional review board or independent ethics committee at each center.

Correspondence: Parantu Shah (parantu.shah@emdserono.com)

Background

Development of baseline predictive classifiers for response to treatment can provide advantages for programs of targeted immunotherapies, development of successful combination therapies, and identification of responder populations to active therapies. Changes in chromosome conformations represent strong systemic cellular network deregulations associated with differences in clinical phenotypes and outcomes [1]. However, questions remain about the applicability of classifiers across treatment lines, indications, and drug combinations.

Methods

Oxford Biodynamics, in collaboration with the EMD Serono, Inc., a business of Merck KGaA, Darmstadt, Germany/Pfizer alliance, has applied its proprietary technology EpiSwitchTM to monitor systemic epigenetic biomarkers for chromosome conformation signatures at baseline in patients with multiline anti–PD-L1 (avelumab) treatment of non-small cell lung cancer (NSCLC). Additionally, epigenetic biomarkers to predict outcome and response in patients with melanoma treated with anti–PD-1 (pembrolizumab) and its combination with another agent were identified in collaboration with the Mayo Clinic.

Results

Three NSCLC classifiers predicting response to avelumab in first-line (1L), second-line (2L), and combined 1L + 2L cohorts were built and applied to test sets. Average accuracy, positive predictive value (PPV), and negative predictive value (NPV) for 10-fold cross-validation on data splits were reported. An NSCLC patient set treated with 2L pembrolizumab served as an independent test set. The 2L NSCLC classifier achieved high (defined hereafter as > 0.7) predictive power (PPV, NPV, and accuracy) in the 2L test set but not in the 1L test set. A reduced version of this classifier achieved a PPV of 0.71 in the 2L pembrolizumab population. The 1L classifier was not applicable in patients who received 2L treatment for NSCLC. The 1L + 2L composite classifier had high predictive power in both 1L and 2L cohorts and a high PPV for identifying responders in the 2L pembrolizumab population. A fourth classifier starting with preselected NSCLC markers had good predictive power for classifying responders in patients with melanoma treated with pembrolizumab. Finally, a 2L NSCLC classifier trained to classify response groups from pembrolizumab-treated patients also identified NSCLC responders with a high PPV from patients treated with pembrolizumab in combination with an epigenetic drug.

Conclusions

Collectively, these results suggest that a set of EpiSwitchTM biomarkers correlates with outcome on anti–PD-1/PD-L1 immunotherapies. Classifier signatures could be generated to work across treatment lines, indications, and combinations, and could be helpful for baseline patient stratification.

Acknowledgements

The authors would like to thank patients enrolled in the EMR000070-001 JAVELIN solid tumor trial for agreeing to consent usage of samples for research purposes. This work was funded by Merck KGaA, Darmstadt, Germany, as part of an alliance between Merck KGaA, Darmstadt, Germany and Pfizer Inc., New York, NY, USA.

References

1. Tordini F, Aldinucci M, Milanesi L, et al. The genome conformation as an integrator of multi-omic data: the example of damage spreading in cancer. Front Genet. 2016; 7:194.

Ethics Approval

The protocol was approved by the institutional review board or independent ethics committee at each enrolling center.

Correspondence: Yulei Wang (wang.yulei@gene.com)

Background

Understanding response to immunotherapies in relation to tumor-immune contexture requires a paradigm shift from a single-marker test towards multiplexed immunohistochemistry (IHC). Here we report the development, early proof of concept of two fully automated 5-plex fluorescent multiplex IHC assays and accompanying digital pathology algorithms.

Methods

Tyramide signal amplification detection was used to inform on the tumor/stroma/immune contexture (CD8, PanCK, FAP, MHC-I, CD31) and to characterize T-cell functions (PD1, CD3, PanCK, GZMB, and PD-L1). Whole slide digital pathology scoring algorithms were developed to identify all phenotypes represented by the markers and their specificity and sensitivity was verified against the results by expert observers.

Results

Assay performance including accuracy, precision, and sequential markers detection was validated on > 200 unique cases of Gastric, Pancreatic, Breast, Lung, Urothelial and Colorectal carcinomas. For an early proof-of-concept, we analyzed paired pre- vs. post treatment tumor biopsies from three pancreatic cancer patients treated with a combination of atezolizumab and chemotherapy. Digital pathology algorithms identified biologically and clinically relevant features: MHC-I is highly expressed in tumor cells of the primary lesions and very rare tumor cells express MHC-I in the liver met samples. A patient with partial response (PR) showed a significantly increased tumor MHC-I upon treatment, while two patients with stable disease (SD) did not show significant changes. The PR patient also showed an increased density of CD3+, CD8+, and GZMB+ T cells within the tumor post-treatment, indicating an increased tumoral T cell infiltration and activation by the treatment.

Conclusions

The automated 5-plex IHC assays and digital pathology algorithms developed in this study provide a robust tool for quantitative and spatially resolved whole-slide characterization of the tumor-immune contexture. Applying these tools in large-scale clinical investigations may provide better understanding of the response/resistance mechanisms to cancer immunotherapies.

Correspondence: Cecile Chartier (cecile.chartier@iovance.com)

Background

Iovance’s TIL products Lifileucel and LN-145 have demonstrated remarkable clinical activity in melanoma and cervical cancer utilizing Iovance’s proprietary 22-day manufacturing process and surgically resected tumor lesions ~ 1.5-cm diameter [1, 2]. Using a core needle biopsy procedure to obtain tumor samples could allow for greater convenience of collecting the tumor from patients [3]. We asked whether a streamlined manufacturing process could be implemented to produce therapeutically relevant TIL from multiple histologies starting with a core biopsy.

Methods

Core biopsies obtained from 4 melanoma and 3 pancreatic, 2 breast, 2 ovarian, and 1 lung tumors were processed in vitro, using a 22-day expansion method termed ‘Core process’. Core biopsy-derived TIL were assessed for expansion, phenotype (lineage, youth/differentiation, activation, and exhaustion markers), function (IFN-gamma and CD107a mobilization), and TCR repertoire.

Results

Iovance’s Core process successfully generated TIL products from all tested samples. One to 2 cores yielded more than 10e9 T cells for 10 of the 12 preparations. Phenotypic analyses revealed no significant differences in terms of T cell lineages and memory subsets, or expression of activation, differentiation, and exhaustion markers when compared to Iovance’s current products. Core-derived TIL products responded to PMA and to anti-CD3 stimulations by inducing levels of CD107a mobilization and IFN-gamma secretion like those produced by TIL derived from excisional biopsies. Preliminary TCR sequencing data suggest that high-diversity products can be also be obtained from small samples, similar to what is obtained from TIL expansion.

Conclusions

This work demonstrates that the Iovance 22-day Core manufacturing method is highly robust and that it is feasible to expand TIL to therapeutically relevant numbers from as little as 1 to 2 core biopsies from multiple histologies with this method. Resulting products were shown to be phenotypically comparable to, and as potent as, products generated with Iovance’s process from excisional biopsy. Iovance anticipates implementing this process in the clinic in the near future.

References

1. Jazaeri AA, Zsiros E, Amaria RN, Artz AS, Edwards RP, Robert Michael Wenham RM, et al. Safety and efficacy of adoptive cell transfer using autologous tumor infiltrating lymphocytes (LN-145) for treatment of recurrent, metastatic, or persistent cervical carcinoma. Clin Oncol. 2019;37:15:2538 (suppl).

2. Sarnaik A, Khushalani NI, Chesney JA, Kluger HM, Curti BD, et al. Safety and efficacy of cryopreserved autologous tumor infiltrating lymphocyte therapy (LN-144, lifileucel) in advanced metastatic melanoma patients who progressed on multiple prior therapies including anti-PD-1. J Clin Oncol. 2019;37:15:2518 (suppl).

3. Ullenhag GJ, Sadeghi AM, Carlsson B, Ahlström H, Mosavi F, Wagenius G, Tötterman TH, et al. Adoptive T-cell therapy for malignant melanoma patients with TILs obtained by ultrasound-guided needle biopsy. Cancer Immunol Immunother. 2012;61:725–732.

Correspondence: Martin Pule (m.pule@autolus.com)

Background

Neuroblastoma is the most common extracranial solid cancer in children with poor long-term survival in those with high-risk disease. A currently ongoing phase I clinical study of GD2-targeted CART for refractory/relapsed neuroblastoma (NCT02761915) shows activity against disseminated disease without inducing on target/off tumor toxicity. However, CART persistence was limited and clinical activity transient and incomplete.

Building on the GD2 CAR used in this study, we have developed a next generation T-cell product candidate termed AUTO6NG. The AUTO6NG product consists of 3 distinct populations of GD2-targeted CAR T-cells, produced by dual transduction of T-cells with two separate retroviral vectors. The first vector directs the expression of a GD2-targeting CAR, co-expressed with a constitutively signalling IL7 cytokine receptor (IL7R_CCR) (product A), while the second vector is a tri-cistronic retroviral vector encoding the same GD2 CAR, co-expressed with dominant negative TGFbRII (dnTGFbRII) and truncated SHP2 (dSHP2) (product B). dSHP2 confers resistance to inhibitory signals such as those from PD1.

Methods

Human T-cells were either dual transduced with both vectors yielding a mix of product A/B/A+B (AUTO6NG) or single transduced with each vector individually giving raise to product A or B. Both single and dual transduced CAR T-cells were extensively evaluated in vitro for redirected lysis, cytokine secretion, T-cell proliferation and survival and resistance to immunosuppressive pathways (including TGFb and PD1/PDL1 inhibition) in co-culture assays with GD2-positive and negative tumour cell lines. Additionally, anti-tumour activity of AUTO6NG was evaluated in vivo by intravenous administration in an established neuroblastoma xenograft model in NSG mice.

Results

AUTO6NG T-cells (product A/B/A+B) were highly potent in cytotoxicity assays against GD2 positive tumour cell lines with no differences observed compared with single transduced CAR T-cells (product A or B). Expression of the IL7R_CCR in both AUTO6NG and product A conferred exogenous-cytokine-independent viability and homeostatic proliferation of modified T-cells, without causing autonomous T-cell growth. Furthermore, AUOTO6NG T-cells and product B but not product A proved resistant to both TGFb- and PD1/PDL1-mediated immunosuppression in vitro due to the presence of dnTGFbRII and dSHP2 in those genetically engineered CAR T-cells. Finally, intravenous delivery of AUTO6NG exhibited potent anti-tumour activity and extended survival in NSG mice with established tumour burden.

Conclusions

These results demonstrate the feasibility, safety, and efficacy of AUTO6NG T-cells. The addition of IL7R_CCR, dnTGFbRII and dSHP2 modules to the AUTO6NG product augment its functions by extending T-cell persistence and rendering modified T-cells resistant to TGFb- and PD1/PDL1-driven immune inhibition.

Correspondence: Frederic Lehmann (flehmann@celyad.com)

Background

Autologous and allogeneic Chimeric Antigen Receptor (CAR) T-cells are under thorough investigation to translate their success in B-cell malignancies to other types of cancer. Previous studies associated the anti-tumour effect of CAR T-cells to their long-term persistence. Most studies use cyclophosphamide and fludarabine (CyFlu) preconditioning chemotherapy to facilitate CAR T-cell persistence. However, the effect of CyFlu preconditioning was rarely compared to other chemotherapies or to CAR T-cells alone. The THINK, SHRINK and ALLOSHRINK trials evaluate the safety and clinical activity of NKG2D receptor-based CAR T-cells in metastatic colorectal cancer (mCRC) patients. THINK and SHRINK utilize autologous CAR T-cells, whereas ALLOSHRINK utilizes allogeneic CAR T-cells. In THINK, CAR T-cells are injected without preconditioning chemotherapy or after CyFlu. In SHRINK and ALLOSHRINK, FOLFOX chemotherapy is given before CAR T-cell injections. Herein we present cellular kinetics results from these three trials.

Methods

Whole blood samples were drawn at various timepoints from patients receiving at least one injection of CAR T-cells. Peripheral blood mononuclear cells (PBMCs) were isolated by ficoll gradient centrifugation at a central laboratory designated by the Sponsor. Genomic DNA was isolated using a commercially available kit. Engraftment of CAR T-cells was measured by digital droplet polymerase chain reaction (ddPCR) using transgene-specific primers and reported as transgene copies per microgram of genomic DNA. Long-term persistence of CAR T-cells was measured by calculating the area under the curve (AUC) using the linear trapezoidal rule.

Results

35 mCRC patients have been treated in THINK (14), SHRINK (9) and ALLOSHRINK (12). Preliminary results are available for 29 subjects. Cell kinetics for subjects having received one injection of autologous CAR T-cells show a seven-fold increase in mean peak levels of T-cell engraftment with CyFlu compared to FOLFOX. Mean AUC is four times higher with CyFlu compared to FOLFOX. Peak levels of engraftment and persistence observed with FOLFOX and without previous chemotherapy are similar. Additionally, allogeneic CAR T-cells exhibit a five-fold increase in mean AUC and a ten-fold increase in mean peak levels compared to autologous cells with the same prior chemotherapy regimen. Additional analyses will be presented during the congress.

Conclusions

Analyses of the initial 29 patients receiving either autologous or allogeneic NKG2D-based CAR T-cells demonstrate that CyFlu enhances peak levels and persistence of adoptively transferred cells. FOLFOX does not appear to influence engraftment or persistence of CAR T-cells. Allogeneic CAR T-cells show higher peaks and time-averaged persistence compared to autologous cells. Analysis of the results is ongoing.

Ethics Approval

The studies referred to in this abstract were approved by all relevant ethical committees and authorities.

Correspondence: Clint Allen (clint.allen@nih.gov)

Background

A significant portion of head and neck cancers (HNCs) harbor genomic alterations that render them insensitive to T cell detection. For these patients, natural killer (NK) cellular therapy may be an effective complementary treatment approach. We studied the anti-tumor activity of a novel, off the shelf, NK cellular therapy consisting of high affinity NK cells engineered to express a chimeric antigen receptor (CAR) targeting PD-L1 (PD-L1 t-haNKs).

Methods

Irradiated (15 Gy) PD-L1 t-haNK cells were assessed for direct cytotoxicity of five human and two murine HNC cell lines by real-time impedance analysis. PD-L1 knockout by CRISPR/Cas9 gene editing was performed in select cells to assess PD-L1-specific killing. Co-culture assays with PD-L1 t-haNKs and murine or human peripheral and tumor infiltrating leukocytes were performed to determine selective elimination of cells. Wild-type C57BL/6 (B6) or NSG mice were engrafted with parental or PD-L1 knockout murine or human tumors and assessed for tumor growth inhibition (TGI) following PD-L1 t-haNK treatment.

Results

PD-L1 CAR expression on PD-L1 t-haNKs was verified. PD-L1 t-haNKs killed all human and murine HNC cell lines at low effector:target ratios. Killing of cells was significantly enhanced with increased PD-L1 expression following IFN-γ pre-treatment. Baseline killing was partially reversed and IFN-γ -enhanced killing was completely abrogated in PD-L1 knockout cells. Ex vivo co-culture of PD-L1 t-haNKs with peripheral and tumor infiltrating leukocytes from tumor bearing mice or with peripheral leukocytes from HNC patients revealed selective elimination of PD-L1 high macrophages and myeloid derived suppressor cells (MDSC) but not lymphocyte subsets. Treatment of B6 mice bearing murine oral cancers with PD-L1 t-haNKs in vivo resulted in ≥50% reduction in PD-L1 high macrophages and MDSC but no reduction in lymphocytes. Treatment of NSG mice bearing parental human HNC or B6 mice bearing parental murine oral cancer resulted in significant TGI after PD-L1 t-haNK treatment. TGI was completely abrogated in mice bearing PD-L1 knockout tumors.

Conclusions

PD-L1 t-haNKs mediated potent PD-L1-specific cytotoxicity against HNC cells and selectively eliminate immunosuppressive macrophages and MDSC expressing high levels of PD-L1 from the periphery and tumor microenvironment. PD-L1 t-haNK monotherapy resulted in PD-L1-specific TGI in xenograft and syngeneic models. These data provide the pre-clinical rationale for the clinical study of PD-L1 t-haNKs in solid tumors. Evidence that PD-L1 t-haNKs selectively eliminate immunosuppressive macrophages and MDSC support the clinical study of PD-L1 t-haNKs as a monotherapy or in combination with treatments designed to activate T cell immunity.

Ethics Approval

The study was approved by the NIH Animal care and Use Committee, approval number 1464-18.

Correspondence: Cecile Chartier (cecile.chartier@iovance.com)

Background

Adoptive T-cell transfer with tumor infiltrating lymphocytes (TIL) is an investigational immunotherapy for advanced solid cancers. Ongoing Phase II clinical trials of Iovance’s lifileucel and LN-145 TIL products have demonstrated efficacy with ORRs of 38% and 44% in patients with melanoma and cervical cancer, respectively [1,2]. Anti-PD-1 therapy has been widely used as a first-line therapy in several types of cancer. TIL infusion products from the patients previously treated with anti-PD-1 therapy still sustain PD-1 expression, especially the subset of tumor antigen-specific TIL [3]. Building on the therapeutic efficacy of PD-1 blockade, we reasoned that intrinsic silencing of PD-1 in our TIL products, may provide similar benefits to systemic administration of anti-PD-1 therapy, while decreasing the side effects associated with systemic anti-PD-1 [3]. Self-delivering small interfering RNA (sd-rxRNA) is a chemically modified siRNA molecule, which has ability to penetrate cell types with high knockdown efficiency of specific target genes [4]. Furthermore, a knockdown approach yields a transient effect, which may prove a more favorable approach when compared with permanent genetic modification. Here, we tested the silencing efficiency of a PD-1-targeted sd-rxRNA, termed PH-762, in TIL and its effect on TIL phenotype and function.

Methods

TIL from melanoma, breast cancer, lung cancer, H&N cancer, and sarcoma were expanded ex vivo with Iovance’s proprietary 22-day process in the presence of PH-762. Resulting TIL products were assessed for PD-1 knockdown, cell expansion and viability, phenotype (T-cell lineage, differentiation, activation, and exhaustion), and effector functions (IFN-gamma induction).

Results

Average silencing of the PD-1 levels was 85%. Sixteen of the 19 tumors tested demonstrated >80% silencing at the surface of PH-762-treated TIL relative to control sd-rxRNA-treated TIL. The remaining 3 samples had ~70% silencing efficiency. Expression of T-cell activation markers including 4-1BB and OX40 was significantly increased in TIL expanded with PH-762. Importantly, other inhibitory and exhaustion molecules remained unaffected, suggesting that compensatory mechanisms were not triggered by PD-1 silencing. Functionally, PD-1 knockdown TIL displayed elevated IFN-gamma secretion when co-cultured with autologous tumor cells, indicating improved effector function upon specific T-cell re-stimulation.

Conclusions

sd-rxRNA-mediated silencing of PD-1 with PH-762 in TIL was highly efficient and generated TIL products with elevated effector function, providing a strong rationale for clinical testing.

Acknowledgements

PH-762 was kindly provided by Phio Pharmaceuticals.

References

1. Sarnaik A. et al. Safety and efficacy of cryopreserved autologous tumor infiltrating lymphocyte therapy (LN-144, lifileucel) in advanced metastatic melanoma patients who progressed on multiple prior therapies including anti-PD-1. J Clin Oncol. 2019;37:2518-2518.

2. Jazaeri A A, et al. Safety and efficacy of adoptive cell transfer using autologous tumor infiltrating lymphocytes (LN-145) for treatment of recurrent, metastatic, or persistent cervical carcinoma. J ClinOncol. 2019;37:2538-2538.

3. Gros A, et al. PD-1 identifies the patient-specific CD8(+) tumor-reactive repertoire infiltrating human tumors. J Clin Invest. 2014;124:2246-2259.

4. Ligtenberg M A, et al. Self-Delivering RNAi Targeting PD-1 Improves Tumor-Specific T Cell Functionality for Adoptive Cell Therapy of Malignant Melanoma. Mol Ther. 2018;26:1482-1493.

Correspondence: Cynthia Bamdad (cbamdad@minervabio.com)

Background

Minerva will open a 1st-in-human CAR T clinical trial for metastatic breast cancers at the Fred Hutchinson Center September, 2019. huMNC2-CAR44 targets a novel form of MUC1; no therapeutic that targets this form has ever been tested in humans. All previous, failed attempts to therapeutically target MUC1 have targeted the tandem repeat domains, which are cleaved and shed from the surface of cancer cells. Cleavage and shedding of the tandem repeat domain increases as tumor stage increases. huMNC2-CAR44 targets the truncated extra cellular domain of MUC1* (muk 1 star), also known as MUC1-C, which is the transmembrane cleavage product that remains after MUC1 is cleaved and the tandem repeat domain is shed from the cancer cells. The MNC2 antibody, which is the targeting head of the CAR, cannot bind to full-length MUC1. It binds to an ectopic epitope that is only unmasked by cleavage and release of the MUC1 tandem repeat domain. MUC1* growth factor receptor is activated when onco-embryonic growth factor NME7AB dimerizes its truncated extracellular domain. NME7AB and the huMNC2 antibody both compete for the same binding site, which is masked in full-length MUC1.

Methods

Monoclonal antibody MNC2 was selected because it recognizes a conformational epitope within MUC1* that is created by cleavage by MMP9, which is overexpressed in breast cancers and is an indicator of poor prognosis. The luminal edge of some normal tissues express a cleaved MUC1*-like form; however, on normal tissues, MUC1 is cleaved by a different cleavage enzyme, which alters the conformation of the truncated extra cellular domain and it is not recognized by the MNC2 antibody.

Results

huMNC2-scFv recognizes 95% of breast cancers, across all subtypes, wherein the average percent staining for each tissue specimen is ~80%. Despite this robust staining of cancerous tissues, huMNC2-scFv showed almost no binding to normal tissues and no staining of critical organs. In vitro, huMNC2-CAR44 T cells killed cancer cells, but not non-cancer cells even if they expressed MUC1 or a cleaved MUC1. In NSG mice (n>300), huMNC2-CAR44 T cells eliminated MUC1* positive tumors from implanted naturally occurring breast cancer cells. A single CAR T cell injection eliminated tumors for 100 days; control animals had to be sacrificed at Day 20. Further, huMNC2-CAR44 T cell mediated killing increased as MUC1* density increased (Figure 1).

Conclusions

If successful, huMNC2-CAR44 could treat a wide variety of solid tumors. huMNC2-scFv binds to 95% of breast, 83% ovarian, 78% pancreatic and 71% of lung cancers.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P150).

huMNC2-CAR44 T cells kill MUC1* positive tumors

Correspondence: Aaron Foster (afoster@bellicum.com); Joseph Bayle (jhbayle@bellicum.com)

Background

The potent, innate anti-tumor cytotoxicity of natural killer (NK) cells combined with their low risk of inducing graft-versus-host disease have made NK cells an emerging platform for allogeneic, off-the-shelf CAR-based cell therapies. However, adoptive transfers of NK cells have shown limited expansion and persistence which may impact their ability to induce durable anti-tumor responses. Here, we demonstrate that constitutive expression of a novel chimeric costimulatory protein, comprised of the signaling domains from MyD88 and CD40 (MC) and secreted IL-15 dramatically improves the proliferation and anti-tumor efficacy of HER2 CAR-redirected NK cells.

Methods

Human CD56-positive cells were enriched from PBMCs derived from healthy donors and activated with irradiated K562 cells in the presence of IL-15. NK cells were subsequently transduced with retroviral vector encoding inducible Caspase-9 (iC9), a HER2-specific CAR (HER2.ζ), MyD88/CD40 (MC) [1] and IL-15. Gene-modified NK cells were evaluated for expansion, cytotoxicity, cell phenotype and cytokine production, in vitro and in an HER2 positive OE-19 NSG mouse xenograft model.

Results

NK cells were efficiently transduced (>50%) and demonstrated robust ex vivo expansion (150 fold, 14 days post-activation) in culture relative to transduced NK cells. In coculture assays with HER2-expressing OE19 and SKOV3 tumor cells, MC-enhanced CAR-NK cells showed potent cytotoxicity with elevated expression of pro-inflammatory cytokines and chemokines including MIP1α, IFN-γ, and GM-CSF. In addition, NK cells expressing the iC9 safety switch could be rapidly ablated by treatment with 1 nM rimiducid to initiate apoptosis. In animals engrafted with OE-19 tumor cells, iC9-CAR.ζ-MC-IL15 modified NK cells demonstrated significantly improved control of tumor expansion compared with control NK cells.

Conclusions

MyD88/CD40 and IL-15 enhance the proliferation and anti-tumor potency of CAR-modified NK cells. Further, inclusion of the iC9 safety switch can be used to mitigate potential toxicities. These technologies have the potential to provide a potent, off-the-shelf allogeneic cell therapy to treat solid tumors.

Reference

1. Collinson-Pautz MR, Chang WC, Lu A, Khalil M, Crisostomo JW, Lin PY, Mahendravada A, Shinners NP, Brandt ME, Zhang M, Duong M, Bayle JH, Slawin KM, Spencer DM, Foster AE. Constitutively active MyD88CD40 costimulation enhances expansion and efficacy of chimeric antigen receptor T cells targeting hematological malignancies. Leukemia. 2019; 33:2195-2207.

Ethics Approval

This study was approved by Bellicum's IACUC and performed in its AAALAC approved vivarium.

Correspondence: Matthew Booty (matt.booty@sqzbiotech.com)

Background

The presentation of sufficient antigen on major histocompatibility complex class I (MHC-I) is a potential barrier to generating potent cancer immunizations. We use microfluidics-based squeezing to deliver antigen directly to the cytosol of target antigen presenting cells (APCs) – resulting in the enhanced presentation of antigen on MHC-I. In addition to facilitating potent CD8+ T cell priming by professional APCs, this approach can make unfractionated peripheral blood mononuclear cells (PBMCs) effective, unorthodox APCs capable of priming CD8+ T cell responses in mouse and human systems.

Methods

Protein and peptide antigens were delivered to the cytosol of murine splenocytes or human PBMCs by microfluidic squeezing. The response to in vivo immunization was assessed by flow cytometry in a series of experiments in mice. Tumor experiments were conducted with the TC-1 cell line, which expresses the viral antigens E6 and E7 from human papilloma virus type 16 (HPV16).

Human PBMCs were loaded with synthetic long peptides (SLPs) containing MHC-I restricted epitopes from cytomegalovirus (CMV) or HPV16. These PBMCs were co-cultured with epitope-reactive human responder CD8+ T cells, and interferon gamma production was quantified to assess antigen-specific responses in vitro.

Results

In mice, we demonstrate that microfluidic squeezing enables delivery to all cell subsets within the spleen and that delivered protein antigen is rapidly processed and presented on MHC-I. In vivo immunization using splenocytes squeezed with a HPV16-derived E7 SLP primes E7-specific responses. Prophylactic immunization of mice implanted with TC-1 resulted in complete protection and these responses were durable, as mice were protected upon TC-1 re-challenge. Therapeutic immunization following TC-1 implantation reduced tumor growth and extended survival compared to unimmunized mice (25 days vs 50 days). Following therapeutic immunization, 85% of tumor infiltrating CD8+ T cells were found to be E7-specific compared to 3% in unimmunized mice.

In human cells, we demonstrate that squeezing of primary PBMCs enables delivery to all cell subsets. Delivery of CMV and HPV16 SLPs leads to presentation on MHC-I, as demonstrated by in vitro responses of both CD8+ T cell clones and patient-derived memory populations. Delivery of CMV antigens at the manufacturing scale (~1 x 10^9 cells) also results in presentation and activation of CD8+ T cells.

Conclusions

Through the direct cytosolic delivery of antigen, we engineered unfractionated PBMCs to function as potent APCs. This strategy has demonstrated significant potential to generate CD8+ T cell responses in both mouse and human systems and has been scaled for clinical implementation.

Ethics Approval

Human samples were supplied by an approved vendor and animal studies were conducted in accordance with SQZ Biotech's Animal Care Program and IACUC which operate according to principles set forth in PHS Policy and the Guide for the Care and Use of Laboratory Animals - 8th edition.

Correspondence: Clifford Cho (cliffcho@med.umich.edu)

Background

The ability to suppress immune reactivity is a defining hallmark of cancer [1-3]. Both the administration and disinhibition of CD8+ T cells, through adoptive immunotherapy and checkpoint inhibition respectively, have yielded unprecedented responses in patients with advanced melanoma [4-8]. However, a majority of patients remain stubbornly unresponsive to T cell-based therapy [9,10]. A better knowledge of cancer-induced T cell suppression is needed improve efficacy. Memory CD8+ T cells (Tmem) are more effective than effector CD8+ T cells (Teff) at controlling melanoma growth after adoptive cell transfer (ACT) in a murine melanoma model [11,12]. Melanoma cancer stem cells (CSC) are primarily responsible for tumor growth and metastasis [13,14]. We hypothesized that Tmem are both more resistant to CSC suppression and more effective at targeting CSC after ACT.

Methods

The B16F10 melanoma cell line was stably transfected to express low levels of lymphocytic choriomeningitis virus (LCMV) peptide antigen GP33 (B16GP33). Ly5.1+/C57BL/6 mice were infected with LCMV to isolate Teff and Tmem on post-infection days 8 or > 30, respectively. Ly5.2+/C57BL/6 mice were inoculated with subcutaneous B16GP33 tumors followed by either no treatment or ACT with Teff or Tmem on days 1 or 7. On day 18-20, tumors were harvested for flow cytometric analysis (FACS) to characterize tumor-infiltrating lymphocytes (TIL) and composition of melanoma CSC versus non-CSC (NCSC) based on expression of the CSC-specific marker aldehyde dehydrogenase (ALDH).

Results

Tumor inhibition was observed after ACT, with greatest treatment effect found after Tmem ACT (Figure 1). FACS analysis of CD8+ TIL showed a predominant exhausted and non-activated phenotype after Teff ACT; in contrast, CD8+ TIL exhibited a highly activated phenotype as well as superior endogenous CD8+ T cell recruitment (Figure 2) after Tmem ACT. FACS analysis of tumor cells after ACT demonstrated that ALDHhigh CSC fractions were markedly expanded after Teff ACT, but diminished after Tmem ACT (Figure 3).

Conclusions

Tmem-based ACT resulted in optimal tumor growth suppression, a more activated TIL phenotype with superior CD8+ T cell recruitment, and substantially stronger clearance of CSC compared to Teff ACT and controls. These observations suggest that use of Tmem may enable cellular therapies to more effectively evade the suppressive effects of melanoma while selectively targeting CSC.

References

1. Marincola F, Wang E, Herlyn M, et al. Tumors as elusive targets of T-cell-based active immunotherapy. Trends Immunol. 2003; 24:335-342.

2. Dunn GP, Old LJ, et al. The three E’s of cancer immunoediting. Ann Rev Immunol. 2004; 22:329-360.

3. Rabinovich GA, Gabrilovich D, Sotomayor EM. Immunosuppressive strategies that are mediated by tumor cells. Ann Rev Immunol. 2007; 25:267-295.

4. Rosenberg SA, Yang JC, Sherry RM, et al. Durable complete responses in heavily pretreated patients with metastatic melanoma using T-cell transfer immunotherapy. Clin Cancer Res. 2011; 17:4550-4557.

5. Dudley MS, Wunderlich JR, Yang JC, et al. Adoptive cell transfer therapy folloing non-myeloablative but lymphodepleting chemotehrapy for the treatemnt of patients with refractory metastatic melanoma. J Clin Oncol. 2005; 23:2346-2357.

6. Hodi FS, O’Day SJ, McDermott DF, et al. Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med. 2010; 363:711-723.

7. Larkin J, Chiarion-Sileni V, Gonzalez R, et al. Combined nivolumab and ipilimumab or monotherapy in untreated melanoma. N Engl J Med. 2015; 373:23-34.

8. Wolchok JD, Kluger H, Callahan MK, et al. Nivolumab plus ipilimumab in advanced melanoma. N Engl J Med. 2013; 369:122-33.

9. Brahmer JR, Tykodi SS, Chow LQM, et al. Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. N Engl J Med. 2012; 366:2455-2465.

10. Royal RE, Levy C, Turner K, et al. Phase 2 trial of single agent ipilimumab (anti-CTLA-4) for locally advanced or metastatic pancreatic adenocarcinoma. J Immunother. 2010; 33:828-33.

11. Contreras A, Sen S, Tatar AJ, et al. Enhanced local and systemic anti-melanoma CD8+ T cell responses after memory T cell-based adoptive immunotherapy in mice. Cancer Immunol Immunother. 2016; 65:601-611.

12. Contreras A, Beems MV, Tatar AJ, et al. Co-transfer of tumor-specific effector and memory CD8+ T cells enhances the efficacy of adoptive melanoma immunotherapy in a mouse model. J Immunother Cancer. 2018; 6:41.

13. Maccalli C, DeMaria R. Cancer stem cells: perspectives for therapeutic targeting. Cancer Immunol Immunother. 2015; 64:91-97.

14. Pan Q, Li Q, Liu S, Ning N, Zhang X, Yu Y, Chang AE, Wicha MS. Concise review: targeting cancer stem cells using immunological approaches. Stem Cells. 2015; 33:2085-2092.

Ethics Approval

This study was approved by University of Michigan’s ethics board (IACUC), approval #1608-004.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P153).

Memory CD8+ T cells vs effector CD8+ T cells in melanoma

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P153).

CD8+ TIL exhibit a more activated phenotype after memory ACT

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P153).

Memory CD8+ T cells target melanoma CSC

Correspondence: Tammie Wahaus (twahaus@eliasah.com)

Background

Canine osteosarcoma (OSA) is an aggressively metastatic primary bone malignancy with a 90% mortality rate. Many naturally occurring canine cancers are genetically and biologically similar to their human counterparts. All cancers express neoantigens and therefore are potentially susceptible to vaccine-enhanced adoptive T cell therapy. Syngeneic rodent studies demonstrated that metastases could be permanently eliminated with vaccine-enhanced adoptive T cell therapy. Canine OSA studies provide an excellent translational bridge between experimental metastatic rodent cancer studies and metastatic human cancer clinical trials. We hypothesized that dogs with OSA could be safely treated at diagnosis with surgery, autologous cancer cell/P. acnes vaccination, adoptive T cell transfer (ACT) of ex vivo-activated T cells, and low dose human interleukin-2 (IL-2) resulting in improved survival compared to carboplatin. We further hypothesized that significant efficacy would be achieved by treating dogs with intact immune systems and minimal residual disease [1,2].

Methods

14 client-owned cancer bearing dogs were enrolled in a one-arm prospective trial. Dogs were staged with bloodwork, limb and thoracic radiographs, histopathology, and bone scans prior to amputation to remove the primary bone tumor. Autologous cancer cell/P. acnes vaccinations were administered intradermally weekly for three weeks. Dogs underwent leukapheresis. Mononuclear white blood cell products were stimulated ex vivo with a T cell-specific superantigen. Dogs received ACT of ex vivo-activated T cells followed by five subcutaneous IL-2 injections. Dogs were monitored for development of metastases via thoracic radiographs every three months.

Results

All 14 patients received autologous vaccinations. Due to early metastasis, 11 dogs received ACT. One dog did not receive adjuvant IL-2; ten dogs completed the entire protocol. Toxicity was minimal after pre-medicants (NSAID, antihistamine, and antiemetic) were instituted prior to ACT. With premedication, all toxicities were VCOG grade I/II. Median disease-free interval for all dogs was 213 days. Median survival time (MST) for all dogs was 415 days. Five dogs have survived for over 621 days and are disease-free (Figure 1.) In addition, the results included at least one dog with complete regression of distant macroscopic metastasis.

Conclusions

This immunotherapy protocol is safe and tolerable. Compared to MST for historical amputation alone with or without adjuvant chemotherapy (MST of 307(1) and 134(2) days, respectively), a significant survival benefit is noted in this group of patients. Further prospective studies are warranted to gain additional immunologic insight to the protocol, further improve disease response and survival, and evaluate the translational impact this treatment could have in advancing human medicine.

References

1. Phillips B, Powers BE, Dernell WS, Straw RC, Khanna C, Hogge GS, Vail DM. Use of single-agent carboplatin as adjuvant or neoadjuvant therapy in conjunction with amputation for appendicular osteosarcoma in dogs. J Am Anim Hosp Assoc. 2009; 45:33-8.

2. Spodnick GJ, Berg J, Rand WM, Schelling SH, Couto G, Harvey HJ, Henderson RA, MacEwen G, Mauldin N, McCaw DL, Moore AS, Morrison W, Norris AM, O’Bradovich, J, O’Keefe DA, Page R, Ruslander D, Klausner J, Straw R, Thompson JP, Withrow SJ. Prognosis for dogs with appendicular osteosarcoma treated by amputation alone: 162 cases (1978-1988). J Am Vet Med Assoc. 1992; 200(7):995-9.

Ethics Approval

The study was approved by University of Missouri’s IACUC, approval number 8280.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P155).

Survival Analysis of Dogs Completing Protocol

Correspondence: Rachit Ohri (rachit@enablelifesciences.com)

Background

Natural Killer (NK) cells hold great promise in immunotherapy, particularly for lung cancer [1]. However, there is a paucity of literature which organizes the susceptibility of various lung cancer subtypes to NK cells. We evaluated the cytotoxicity (necrosis and apoptosis) of the NK cell line KHYG-1 (Effector) against cell-lines of 3 different subtypes of lung cancer i.e. H1975, H1703, A549 (Target). We also determined the levels of biomarkers relevant to NK cell activation and function [2] i.e. the cell-surface biomarker CD107a and 5 soluble biomarkers [Perforin, Granzyme-A, Granzyme-B, IFN-gamma and TNF-alpha].

Methods

Three lung cancer cell lines (H1975, H1703, A549) (ATCC, Virginia) were plated at 100% confluency in 96-well plates (1.25 cells/cm^2, 3.2*10^5 cells/mL), while K562 cells (ATCC, Virginia), used as a positive control, were suspended in the plate wells at 3.2*10^5 cells/mL. KHYG-1 cells (JCRB, Japan) were added at a density of 6.4*10^6 cells/mL, at a 20:1 Effector:Target (E:T) ratio. Cells incubated for 5 hours at 5% CO2, 37 °C. Subsequently:

1. Cytotoxicity (necrosis and apoptosis) in target cells were quantified using flow cytometry with PerCP-Cy™5.5 Annexin V and Propidium Iodide (PI).

2. CD107a expression on KHYG-1 cells was evaluated using flow cytometry with PE-labeled anti-human CD107a Ab.

3. Expression levels of 5 soluble biomarkers (Perforin, Granzyme-A, Granzyme-B, IFN-gamma and TNF-alpha) were determined in the cell supernatants using Luminex.

Results

The cytotoxicity data suggests that for necrosis, all target cell-lines were significantly different (all p values < 0.0001) from each other in terms of susceptibility to the effector cells (A549 > H1703 > H1975 > K562) (Figure 1). K562 cells are significantly higher in late apoptosis than all three lung cancer cell lines (Figure 2). Amongst the 3 lung cancer cell-lines, the H1703 cell line is significantly higher in late apoptosis than H1975 and A549 cells (p-value < 0.05) (Figure 3). Although differences were seen in the necrotic and late apoptotic profiles of target cells, the CD107a expression on the KHYG-1 effector cells was similar across all co-cultures (Figure 4). The soluble biomarker data (Luminex) is being collected.

Conclusions

In conclusion, cell-lines corresponding to different lung cancer subtypes, i.e. A549 (Carcinoma), H1703 (Squamous Cell) and H1975 (Adenocarcinoma) exhibit significant differences in both their necrosis and late apoptosis susceptibility when co-cultured with NK cells. Such insight could be used to better guide NK cell based immunotherapy development.

References

1. Aktaş O, Öztürk A, Erman B, Erus S, Tanju S, Dilege S. Role of Natural Killer Cells in Lung Cancer. J Clin Cancer Res Clin Oncol. 2018; 144:997-1003.

2. Hodge G, Barnawi J, Jurisevic C, Moffat D, Holmes M, Reynolds PN, Jersmann H, Hodge S. Lung cancer is associated with decreased expression of perforin, granzyme B and interferon(IFN)-γ by infiltrating lung tissue T cells, natural killer (NK) T-like and NK cells. Clin Exp Immunol. 2014; 178:79–85.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P156).

Necrosis

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P156).

Late Apoptosis

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P156).

Late Apoptosis minus K562

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4 (abstract P156).

CD107a Expression

Correspondence: Thomas Andresen (tandresen@torquetx.com)

Background

Adoptive cell transfer (ACT) of tumor-targeted T cells has demonstrated encouraging clinical efficacy in some hematological cancers. However, in solid tumors, targeting a single antigen (e.g., CAR-T and TCR therapies) can lead to antigen escape and development of resistance. Furthermore, although support provided by lymphodepletion or cytokine administration can enhance responses to ACT, these systemic treatments are often associated with significant toxicities.

Torque’s Slipstream™ T cell manufacturing platform is a high-efficiency process for generating Deep-Primed™ T cells: polyclonal non-genetically engineered T cells that (1) are targeted against multiple tumor-specific antigens and (2) carry immunomodulating cytokine payloads to provide prolonged and locally directed immune support without systemic toxicities. The Slipstream™ process is designed to resolve the manufacturing challenge of generating high yields of early memory phenotype tumor-reactive T cells, which are associated with clinical benefit. Here, we show that the Slipstream™ process drives robust expansion while preserving favorable memory characteristics of natural tumor-reactive T cells, and we demonstrate that Deep-Priming™ T cells with Deep IL-15 or Deep IL-12 improves function.

Methods

Multi-targeted T cells (MTC) were comparatively generated from donors via either a first-generation process or the new Slipstream™ process that leverages ex vivo expansion conditions optimized for MTC production. T cell reactivity against tumor-associated antigens, memory, polyfunctionality, cytotoxicity, and response to Deep IL-15 and Deep IL-12 were measured. The modularity of Slipstream™ was tested by training MTC against antigen cassettes including cancer or viral antigens and measuring reactivity against antigen subsets.

Results

Compared to a first-generation process, MTC generated with Slipstream™ exhibited >20-fold improvement in antigen-specific reactivity and a substantial improvement in the yield of memory-phenotype antigen-specific T cells including a 10-fold increase in Tcf1-positive cells. Furthermore, the Slipstream™ process yielded MTC with increased polyfunctionality and specificity as measured by cytokine production and TCR sequencing, respectively. Notably, T cells expanded using the Slipstream™ process showed potent cytotoxicity against human cancer cells as well as responsiveness to Deep IL-15 and Deep IL-12. The Slipstream™ process can also be adapted for simultaneous training of MTC against different antigens including virus-associated tumor antigens.

Conclusions

The Slipstream™ process is optimized to produce Deep-Primed™ MTC with substantive increases in characteristics associated with clinical efficacy: antigen reactivity, memory phenotype, and polyfunctionality. Modularity of the Slipstream™ process has been demonstrated by simultaneously training T cell clones reactive to cancer and virus-associated antigens, and Deep-Primed™ MTC with cell-associated Deep IL-15 or Deep IL-12 drives enhanced T cell function in vitro.

Correspondence: Shikhar Mehrotra (mehrotr@musc.edu)

Background

Endoplasmic reticulum (ER) stress induced by external or internal stimuli activates a number of well-orchestrated cellular signaling processes aimed to promote either cell apoptosis or to restore cellular function and resolve the stress. In tumor microenvironment, induction of ER stress is known to dampen the antitumor activity of T cells by reducing their mitochondrial function. However, if magnitude of ER stress governs the T cell fate and function is unknown.

Methods

We performed our study on B16 murine melanoma model and used standard immunological techniques like flow cytometry, immunoblot analysis, microscopy, real time PCR

Results

Using melanoma antigen gp100 reactive T cells, we found that low level of ER stress enhances T cell stemness and promotes mitochondrial biogenesis, whereas high level of ER stress triggers T cell death. Moreover, upon adoptive transfer, T cells treated with low dose ER stress inducer are able to form long-lived memory in vivo, express reduced level of co-inhibitory molecule, and demonstrate superior anti-tumor immunity by increasing overall survival of B16 murine melanoma bearing mouse. Mechanistically, we discovered that, upon ER insult at suboptimal level, a protective autophagy pathway is induced to promote cell survival and maintain stemness through the protein kinase R-like endoplasmic reticulum kinase (PERK)/ activating transcription factor-4 (ATF4)-dependent manner. Conversely, knockdown of PERK abrogates autophagy activation, hampers mitochondrial biogenesis in response to suboptimal ER stress, which in-turn compromises the antitumor function of melanoma antigen specific T cells. Furthermore, we demonstrated that blocking autophagy in T cells hampers T cell anti-tumor activity. Lastly, T cells which initiates autophagic process due to suboptimal ER stress show better potential to control tumor compared to those, that do not enter into the process

Conclusions

Overall, these preclinical data highlights that, low level of ER stress response is important for healthy cellular function and therapeutically, ER stress pathways can be manipulated in T cells in order to regulate their antitumor potential.

Ethics Approval

The study was approved by Medical University of South Carolina‘s Ethics Board, approval number 2018-00628.

Correspondence: John Castle (john.castle@agenusbio.com); Marc Van Dijk (marc.vandijk@agentustherapeutics.com)

Background

Adoptive T cell therapy using patient T cells redirected to recognize tumor-specific antigens by expressing genetically engineered high-affinity T-cell receptors (TCRs) has therapeutic potential for melanoma and other solid tumors. Clinical trials implementing genetically modified TCRs in melanoma patients have raised concerns regarding off-target toxicities resulting in lethal destruction of healthy tissue, highlighting the urgency of assessing which off-target peptides can be recognized by a TCR.

Methods

As a model system we used the clinically efficacious NY-ESO-1-specific TCR C259, which recognizes the peptide epitope SLLMWITQC presented by HLA-A*02:01. We investigated which amino acids at each position enable a TCR interaction by sequentially replacing every amino acid position outside of anchor positions 2 and 9 with all 19 possible alternative amino acids, resulting in 134 peptides (133 altered peptides plus epitope peptide). Each peptide was individually evaluated using three different in vitro assays: binding of the NY-ESO C259 TCR to the peptide, peptide-dependent activation of TCR-expressing cells, and killing of peptide-presenting target cells. To represent the TCR recognition kernel, we defined Position Weight Matrices (PWMs) for each assay by assigning normalized measurements to each of the 20 amino acids in each position. To predict potential off-target peptides, we applied a novel algorithm projecting the PWM-defined kernel into the human proteome, scoring NY-ESOc259 TCR recognition of 336,921 predicted human HLA-A*02:01 binding 9-mer peptides.

Results

Of the 12 peptides with high predicted score, we confirmed 7 (including NY-ESO-1 antigen SLLMWITQC) strongly activate human primary NY-ESO C259-expressing T cells. These off-target peptides include peptides with up to 7 amino acid changes (of 9 possible), which could not be predicted using the recognition motif as determined by alanine scans.

Conclusions

Thus, this replacement scan assay determines the “TCR fingerprint” and, when coupled with the algorithm applied to the database of human 9-mer peptides binding to HLA-A*02:01, enables identification of potential off-target antigens and the tissues where they are expressed. This platform enables both screening of multiple TCRs to identify the best candidate for clinical development and identification of TCR-specific cross-reactive peptide recognition and constitutes an improved methodology for the identification of potential off-target peptides presented on MHC class I molecules. We used this platform and demonstrate screening of multiple TCRs targeting tumor antigens.

Correspondence: Sandro Matosevic (sandro@purdue.edu)

Background

NK cells are powerful effectors in cancer immunotherapy and have potential to treat various cancers; however significant challenges remain in the treatment of solid tumors. Energy availability is compromised surrounding solid tumors and NK cell metabolic reprogramming can occur to inhibit NK effector functions [1]. Accumulation of adenosine in the tumor microenvironment (TME) from the activity of ectoenzymes CD39 and CD73 on cancer cells is one mechanism that leads to impaired NK cell function. Our previously published data has established that the effects of TME adenosine on NK cells cause specific reorganization of the cells’ metabolism and effector signatures to suppress NK cell function, and the cytokine combination of IL-12/15 was hyperresponsive to adenosine [2]. One way to combat immunosuppression induced by cancer-produced adenosine is to engineer NK cells to overcome this inhibition. To that end, we engineered NK cells to directly target CD73 by imparting NK-specific signaling to enhance anti-tumor activity against CD73+ lung carcinoma.

Methods

Peripheral blood-derived NK cells were isolated from healthy human donors and expanded using feeder cells. NK cells were electroporated using mRNA or transduced with lentivirus expressing the CD73-targeting construct which bears signaling domains derived from FcγRIIIa. Engineered NK cells expressing the construct were tested for their killing ability against lung carcinoma A549 cells. The engineered NK cells were then adoptively transferred into a CD73+ lung cancer xenograft into NSG mice. Circulating CD73-CAR NK cells were quantified for their expression of activating markers NKG2D, DNAM, and NKp30 and visualized using immunohistochemistry to determine infiltration into tumors, and mice were assessed for tumor growth.

Results

We showed NK cells can be efficiently redirected against CD73 to block the generation of immunosuppressive adenosine and rescue impaired NK cell anti-tumor immunity. Specifically, primary human NK cells were successfully engineered to express the synthetic CD73-FCyRIIIa construct. Retargeted NK cells showed enhanced anti-tumor functions in vitro against CD73-expressing A549 cells. Engineered primary NK cells also showed promise in stunting CD73+ lung cancer tumor growth for up to 3 weeks in vivo. Current and future studies include evaluation of off target effects and local injection to further evaluate infiltration of the NK cells in vivo.

Conclusions

The microenvironment of solid tumors is highly immunosuppressive and adenosine has been shown to impair NK cell anti-tumor immunity. A novel anti-CD73 targeting construct using NK cell signaling components has been developed and shown to prevent tumor growth of CD73+ lung carcinoma.

References

1. Chambers A *, Lupo K*, Matosevic S. Tumor-microenvironment-induced immunometabolic reprogramming of natural killer cells. Front Immunol. 2018; 9:2517.

2. Chambers A, Wang J, Lupo K, and Matosevic S. Adenosinergic signaling alters natural killer cell functional responses. Front Immunol, 2018; 9:2533.

Ethics Approval

The study was approved by Purdue University Institution's Review Board, approval number 1804020540.

Correspondence: Jian Chen (jchen@celetrix.com)

Background

CAR-T cells are currently manufactured for clinical use by infection of human T cells with viral vectors containing the CAR gene. The current viral CAR-T manufacturing process is lengthy and costly and electroporation has emerged as a promising alternative. However, clinical use of electroporation technology in CAR-T has been difficult and several clinical trials have met significant problems due to the low transfection efficiency and/or high cell mortality.

Methods

Our novel understanding of the electroporation mechanism revealed that the current widely-used electroporation methods have significant mistakes in the physical design as well as electroporation buffer design. The first problem is the electroporation sample container design. It is well known that electrochemical reaction generates gas bubbles that are harmful to the cells and there was no good solution to the problem. Here we used a novel pressurization approach to largely eliminate the effect.

Results

Combined with other improvements including electroporation buffer design and post-electroporation cell culture strategy, we have been able to achieve over 80% plasmid transfection efficiency in unstimulated T cells and over 90% plasmid transfection efficiency in stimulated T cells. The viability in survived cells is over 95% measured by live/dead staining and the true survival rate measured by survived cell number is over 66%. The new electroporation method can achieve over 90% in gene editing and the method is also widely applicable in electroporation of NK cells, DC cells and monocytes.

Conclusions

The new method is also scalable as billions of cells can be processed in the large volume electroporation setting. Our method can potentially eliminate the need for expensive cell expansion and virus production altogether, therefore cutting the huge economic burden of CAR-T therapy.

Correspondence: Zhi-jie Cheng (jey.cheng@promega.com)

Background

Adoptive cancer antigen-specific T cell therapy currently comprised of chimeric antigen receptor (CAR-) and T cell receptor (TCR) engineered T cells. Clinical results from CAR-T cells have demonstrated promising results in treating leukemia, while TCR-engineered T cells which have the advantage of recognizing intracellular tumor antigens is still in very early development.

Methods

Here, we report the development of two CD4+ or CD8+ TCRαβ-KO reporter T cell lines for the screening and characterization of transgenic TCRs. A TCRαβ-KO reporter T cell line was first developed by knocking out the endogenous TCR α and β chains in the reporter T cell line using CRISPR/Cas9 and the successful knockout is confirmed by phenotypic assays and TCR v chain locus sequencing.

Results

We demonstrated that re-introduction of HA peptide-specific HA1.7 TCR α and β chains into TCRαβ-KO reporter T cell lines results in HA peptide-dependent TCR activation and luciferase reporter expression when HA peptide is presented by a MHCII+ cell line. Furthermore, the select expression of CD4 or CD8 variants in the TCRαβ-KO reporter T cell line could enable the development of TCRs for both MHCI- and MHCII-restricted tumor antigen targets.

Conclusions

The CD4+ and CD8+ TCRαβ-deficient reporter T cells can serve as valuable tools for screening and characterization of neoantigen-specific TCRs

Correspondence: Robin Parihar (rxpariha@texaschildrens.org)

Background

Immunotherapy using antigen-redirected lymphocytes such as chimeric antigen receptor (CAR)-T or -NK cells in patients with solid tumors has shown poor efficacy. Cell therapies are hindered by immunosuppressive cells such as inhibitory macrophages (M2s) and myeloid-derived suppressor cells (MDSCs) that contribute to a highly suppressive tumor microenvironment (TME) [1]. Researchers have armed redirected lymphocytes with the ability to secrete cytokines in hopes of promoting their proliferation and function in suppressive TMEs [2,3]. However, the effect of these cytokines on other immune cells within the TME, such as MDSCs and M2s, is unknown.

Methods

To determine how the human common gamma-chain cytokines, interleukin(IL)-2, IL-7, IL-15, and IL-21 affect human MDSCs and M2s, we exposed ex vivo enriched M2s and MDSCs to each cytokine separately and assessed changes in MDSC and M2 phenotype and ability to dampen T-cell activation and proliferation. To further define cytokine-induced changes in MDSC/M2 function in a more clinically relevant system, we tested the ability of cytokine-exposed MDSCs/M2s to impair CAR-T cell proliferation and anti-tumor activity in a TME co-culture. As a clinical correlate, we assessed common gamma-chain cytokine receptor expression on MDSCs and M2s within neuroblastoma and sarcoma patient tumors and tested the effects of cytokine exposure on their suppressive capacity.

Results

Subsets of ex vivo enriched M2s and MDSCs expressed common gamma-chain cytokine receptors. MDSCs expressed receptors for IL-2 (22%, avg. MFI=67, n=3), IL-7 (43%, avg. MFI=375, n=3), IL-15 (23%, avg. MFI=310, n=3), and IL-21 (65%, avg. MFI=124, n=4); whereas M2s expressed receptors for IL-2 (17%, avg. MFI=59), IL-7 (98%, avg. MFI=543), IL-15 (36%, avg. MFI=619), and IL-21 (91%, avg. MFI=296). Exposure of human MDSCs or M2s to IL-2, IL-7, IL-15, or IL-21 did not alter their cell-surface phenotype. Exposure of these suppressive myeloid cells to IL-2, IL-7, and IL-15 did not change their ability to suppress T-cell proliferation. In contrast, exposure of M2s and MDSCs to IL-21 increased their ability to suppress T-cell proliferation and activation (98% suppression by IL-21 exposed MDSCs vs. 72% suppression by control MDSCs; 98% suppression by IL-21 exposed M2 vs. 79% suppression by control M2 at a 2:1 T cell:MDSC/M2 ratio).

Conclusions

These results suggest that IL-21 increases the suppressive capacity of human MDSCs and M2s. Ongoing experiments will define the mechanisms by which IL-21 alters MDSC and M2 suppression and further define the effect of IL-21 exposed MDSCs and M2s on tumor growth and CAR-T cell therapeutic efficacy in vivo.

References

1. Martinez M, Moon EK. CAR T cells for solid tumors: New strategies for finding, infiltrating, and surviving in the tumor microenvironment. Front. Immunol. 2019; 10:128.

2. Yeku OO, Purdon TJ, Koneru M, Spriggs D, Brentjens RJ. Armored CAR T cells enhance antitumor efficacy and overcome the tumor microenvironment. Sci Rep. 2017; 7:10541.

3. Liu D, Song L, Wei J, Courtney AN, Gao X, Marinova E, Guo L, Heczey A, Asgharzadeh S, Kim E, et al. IL-15 protects NKT cells from inhibition by tumor-associated macrophages and enhances antimetastatic activity. J Clin Invest. 2012; 122:2221-2233.

Ethics Approval

Tumor tissue use was approved by Baylor College of Medicine IRB study protocol #26691, and samples were de-identified prior to laboratory evaluation.

Correspondence: Deborah Altomare (deborah.altomare@ucf.edu)

Background

PD-1 axis blockade therapies have shown success but responses are limited to ~15% of cancer patients. These responses correlate with presence of lymphocyte infiltrated, PD-L1 positive tumors. Strategies that increase PD-L1 expression may improve outcomes of PD-1 axis blockade. PD-L1 on tumor cells is induced by IFNγ, secreted by NK cells. We developed a method for producing therapeutic quantities (>1,000 fold expansion within two weeks) of hyper-activated NK cells with high anti-tumor cytotoxicity and enhanced IFNγ secretion. The utilizes particles from Plasma Membrane of K562 cells expressing membrane bound IL21 (PM21-particles). Herein, the ability of PM21-particle expanded NK cells to induce PD-L1 expression on various tumors was tested in vitro and in vivo in ovarian cancer model. Furthermore, the effect of anti-PD-L1 on NK cell anti-tumor activity was tested in vitro and in vivo.

Methods

NK cells were expanded with PM21-particles as described. For in vivo experiments, NSG mice were implanted with 1x10^6 SKOV-3 cells i.p.. Mice were treated with 10^7 PM21-NK cells (n=6) or with vehicle control (n=6) on days 8 and 13. Mice were sacrificed on day 20 to collect tumors. Tumors were perfused and retrieved tumor cells were analyzed for PD-L1 expression while infiltrating immune cells were phenotyped.

Results

PM21-NK treatment induced PD-L1 on >30% of tumor cells across multiple cell lines. PM21-NK cells are negative for PD-1 and addition of anti-PD-L1 had no effect on their cytotoxicity or cytokine production. In in vivo experiment, PM21-NK cell treated mice had increased PD-L1+ tumors vs. the untreated group (29.7% vs 14.5%, p<0.0001). Despite T-cell depletion, T-cells made up ~22% of hCD45+ events in perfused tumors, 83% of which were Tregs. PM21-NK cells are PD-1-, but are inhibited by Tregs. Untreated and anti-PD-L1 alone mice had median survival of 24 days. Treatment with PM21-NK cells improved survival over untreated (p=0.0003) and PD-L1 alone (p=0.0002) groups having median survival of 40 days. Combination of PM21-NK cells with anti-PD-L1 further improved of survival over the PM21-NK cells alone group (48 days, p=0.042) with 25% of mice still remaining in good health at day 58.

Conclusions

These data support the use of anti-PD-L1 in NK cell therapy, regardless of initial tumor PD-L1 status. PM21-NK cells can be used for tumor treatment and to prime tumors to express PD-L1. The PD-L1 induced upon NK cell treatment can serve as “universal targetable ligand” if used with humanized anti-PD-L1 antibodies to cause tumor killing by ADCC.

Correspondence: Devon Shedlock (dshedlock@poseida.com)

Background

Autologous Chimeric Antigen Receptor (CAR) T cell therapy for relapsed/refractory Multiple Myeloma (MM), such as Poseida’s anti-B cell maturation antigen (BCMA) product candidate, P-BCMA-101, have shown significant efficacy in the clinic. P-BCMA-101 is comprised of a high percentage of stem cell memory T cells (TSCM), resulting in a product that is much safer and potentially more durable than other anti-BCMA autologous product candidates. However, individualized products have expensive and time-consuming manufacturing and significant variability in input patient T cells characteristics. We are developing P-BCMA-ALLO1, an off-the-shelf anti-BCMA allogeneic (allo) CAR-T product candidate manufactured from serial healthy donor material that circumvents many of the downsides of an individualized CAR-T product.

Methods

P-BCMA-ALLO1 is produced using two key platform technologies: the nonviral piggyBac® (PB) DNA Modification System and the high-fidelity Cas-CLOVER™ (CC) Site-Specific Gene Editing System. The PB transposase mRNA and DNA encoding the PB-based transgene are electroporated along with the components of the CC system needed to knockout (KO) the T Cell Receptor (TCR) and beta-2 microglobulin, thereby eliminating expression of Major Histocompatibility Complex (MHC) class I. The T cells are then expanded using our proprietary "booster molecule.” The resulting product demonstrates expression of the transgene in nearly all cells, and after a purification step, have eliminated all TCR expression and most MHC class I expression.

Results

We have produced P-BCMA-ALLO1 at both research and near-clinical scale from >35 donors with >97% manufacturing success. Efficiencies of TCR-KO ranged from ~50-90%, with final product always demonstrating >99% TCR-KO. T cell expansion varied from ~0.5-20 fold. At clinical production scale, this translates to up to 250 doses of CAR-T per manufacturing run at a dose of 150x10e6 cells/patient. P-BCMA-ALLO1 demonstrated a high-percentage of TSCM cells (CD45RA+CD62L+CD45RO-). Furthermore, P-BCMA-ALLO1 generated from multiple donors demonstrated potent efficacy in the RPMI-8226 xenograft model in NSG mice, thus establishing the feasibility of using serial individual donors in our manufacturing process.

Conclusions

In summary, these data demonstrate a robust, reproducible and highly scalable manufacturing process. Moreover, this production process can be expanded for use with additional targets for treatment of other heme or solid tumors.

Correspondence: Hong Liu (hong.liu@eurekainc.com)

Background

The use of engineered T cells for the treatment of solid cancers remains challenging. Recently, we developed a novel antibody-T cell receptor (AbTCR) ARTEMIS™ T-cell platform, which combines antibody-based target recognition with gamma/delta TCR-based cellular activation [1]. In contrast to chimeric antigen receptors (CARs), the AbTCR forms a natural multimeric receptor with the endogenous CD3 complex, which feeds into a network of signaling pathways that regulate T-cell activation. In addition, the inclusion of gamma/delta TCR chains within the AbTCR avoids the formation of mispaired receptors with unknown cross-reactivity, which is a potential risk associated with current alpha/beta TCR-based therapies.

Using the core design of the AbTCR ARTEMIS™ T-cell platform, we developed ET140202 for the treatment of hepatocellular carcinoma (HCC). ET140202 features an AbTCR targeting alpha-fetoprotein (AFP)peptide/MHC complexes (specifically AFP158-166/HLA-A2) expressed on HCC cancer cells. To optimize T-cell activation and expansion, the AbTCR is co-expressed with a CD28-based co-stimulatory molecule engineered to target Glypican 3 (GPC3) expressed on HCC cancer cells.

Methods

To test the specificity and potency of ET140202 T cells in vitro, ET140202 T cells were co-incubated with either target-positive or target-negative cells. Lactate Dehydrogenase release was used to quantify target cell lysis. CFSE assay was used to measure cell proliferation. Expression of differentiation and exhaustion markers were determined by flow cytometry. The in vivo anti-tumor activity of ET140202 T cells was tested in an AFP+/HLA-A2+ Hep G2 liver cancer xenograft model. We also engineered the same anti-AFP158-166/HLA-A2 binding moiety onto a CD28-based CAR (AFP-CAR) and compared AFP-CAR-T cells to ET140202 T cells in various assays.

Results

ET140202 T cells specifically lysed AFP-positive tumor cells. Compared to AFP-CAR-T cells, ET140202 T cells displayed enhanced in vitro cell killing and proliferation even after repetitive antigen stimulations. ET140202 T cells also display a less exhausted surface phenotype (e.g. lower PD-1 expression) and a higher percentage of central memory T cells (CCR7+ CD45RA-) after antigen stimulation. In vivo, both intravenous and intratumoral single administration of ET140202 T cells led to significant tumor growth inhibition.

Conclusions

ET140202 is built upon our novel AbTCR ARTEMIS™ T-cell platform, which was designed to harness the natural biology of T cells to fight cancer. Both in vitro cellular assays and in vivo mouse studies support the safety and efficacy of ET140202 T cells. Whether these pre-clinical findings for AbTCR-based ET140202 T-cell therapy translate into the clinical setting is currently being tested (clinicaltrial.gov, NCT0399803).

Reference

1. Xu Y, Yang Z, Horan LH, Zhang P, Liu L, Zimdahl B, Green S, Lu J, Morales JF, Barrett DM et al. A novel antibody-TCR (AbTCR) platform combines Fab-based antigen recognition with gamma/delta-TCR signaling to facilitate T-cell cytotoxicity with low cytokine release. Cell Discovery 2018, 4(1):62.

Ethics Approval

All animal experiments were conducted according to protocols approved by their Institutional Animal Care and Use Committee (IACUC) and in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, National Academy Press, Washington, DC, 1996) and the Policy on Humane Care and Use of Laboratory Animals (Department of Health and Human Services, Bethesda, MD).

Correspondence: Gus ZeinerGus Zeiner (gus@chimera.bio)

Background

Current chimeric antigen receptor (CAR) T cell therapies are clinically efficacious against several B cell malignancies, but are less effective at eliminating solid tumors. A key contributor to this observed lack of efficacy is the tumor microenvironment (TME) that is erected by solid tumors to impose immunosuppressive physical and chemical barriers to T cell function and survival. To break TME-driven immunosuppression, attempts have been made to arm CAR-T cells with the ability to produce immunomodulatory payloads that target features of the TME. By their nature, these immunomodulators (e.g. Interleukin 12) are frequently toxic when systemically delivered. Due to a limited repertoire of existing programmable gene regulators, constitutive expression and systemic distribution of immunomodulators by armed CAR-T cells is common.

Chimera Bioengineering has characterized a novel, post-transcriptional gene regulatory node that strictly governs effector outputs in human T cells. This gene regulatory node, termed Gold, has been used to create enhanced CAR-T therapeutics that produce tightly controlled immunomodulatory outputs only upon tumor engagement.

Methods

Methods include: design and characterization of a proprietary bicistronic GoldCAR lentiviral vector to simultaneously deliver both CAR and IL12 transgenes; production of lentivirus and infection of primary human T cells; in vitro characterization of GoldCAR-T cell function utilizing cell based assays, immunoassays and flow cytometry; implantation of subcutaneous xenograft tumors in NSG mice with tumor growth monitored by caliper and bioluminescent imaging (BLI) to assess in vivo efficacy; and ex vivo analysis of blood, tumor and lymphatic organs to characterize safety profile.

Results

CD19 targeted GoldCAR-T cells delivering IL12 demonstrate improved efficacy over standard CD19 CAR-T cells in a subcutaneous Daudi B cell xenograft mouse model. These GoldCAR-T cells exhibit comparable efficacy to CAR-T cells with constitutive IL12 expression, but levels of pro-inflammatory cytokines in the peripheral blood are lower in the mice treated with GoldCAR-T cells.

Conclusions

GoldCAR-T cells with tumor proximal IL12 delivery demonstrate enhanced efficacy over standard CAR-T cells and an improved safety profile compared to CAR-T cells with constitutive IL12 expression. The implications of this study point to an amplified CAR-T cell response in an immunosuppressive tumor microenvironment.

Trial Registration

Not applicable

Ethics Approval

The animal study was conducted under the Institutional Animal Care and Use Committee (IACUC) of LumiGenics, LLC, 750 Alfred Nobel Drive, Suite 103, Hercules CA 94547

Correspondence: Sandro Matosevic (smatosev@purdue.edu)

Background

Natural killer (NK) cells are part of the innate immune system, but are capable of participating in both innate and adaptive immune responses due to their wide range of cytolytic activities, from degranulation, secretion of cytokines to antibody-dependent cell-mediated cytotoxicity. These are possible due to the cells’ ability to recognize self and non-self-entities via the net signal generated from their activating and inhibitory receptors upon engagement. One such receptor is TIM-3, which is expressed on various lymphocytes. In T-cells, TIM-3 is an exhaustion marker [1], but on NK cells, results are conflicting in regards to its function as the receptor exhibits both activating and inhibitory effects depending on disease type and activation status [2-6].

Methods

NK cells were isolated from peripheral blood of healthy donors. After expansion, they were co-cultured for 4 hours with glioblastoma (U87) at effector:target (E:T) ratios of 2.5:1 and 10:1, and various receptors were screened by flow cytometry, including PD-1, NKG2A, LAG-3, CD158b, CEAMCAM-1 and TIM-3. Then, expression of TIM-3 was measured when in the presence of patient-derived primary glioblastoma cells (GBM43) and prostate cancer (PC3) for 4 hours. To determine the effect of TIM-3 expression, killing assay are being carried out by blocking TIM-3 on NK cells. Further investigation is being performed by blocking one of TIM-3’s primary ligands, Galectin-9, on cancer cells to determine its impact on NK cell cytotoxicity. Statistical analyses are completed in SAS JMP Pro 14.

Results

We found that TIM-3 is significantly downregulated on primary human NK cells, in both frequency and surface density, when exposed to solid tumor cells such as U87, GBM43 and PC3 at multiple E:T ratios. Unlike other inhibitory NK receptors, this downregulation was unique to TIM-3. However, it is not known why the downregulation occurs with solid tumors, and whether this change in expression affects NK killing capacity. Here, we report the role of TIM-3 on NK cell cytotoxicity against solid tumor cell lines and the role of Galectin-9 in mediating NK cell activity.

Conclusions

We found that Tim-3 was significantly downregulated on NK cells in response to solid tumor cells. Understanding the complex roles of Tim-3 expression on NK cells allows us to better understand the nuanced immunomodulatory role of Tim-3 on NK cell anti-tumor responses, and provide a basis for the development of immunotherapies targeting impaired NK cell function in solid tumors

References

1. Sánchez-Fueyo A, Tian J, Picarella D, Domenig C, Zheng XX, Sabatos CA, Manlongat N, Bender O, Kamradt T, Kuchroo VK, Gutiérrez-Ramos JC, Coyle AJ, Strom TB. Tim-3 inhibits T helper type 1-mediated auto- and alloimmune responses and promotes immunological tolerance. Nat Immunol. 2003; 4:1093-101.

2. Ndhlovu LC, Lopez-Vergès S, Barbour JD, Jones RB, Jha AR, Long BR, Schoeffler EC, Fujita T, Nixon DF, Lanier LL. Tim-3 marks human natural killer cell maturation and suppresses cell-mediated cytotoxicity. Blood. 2012; 119:3734-43.

3. Jost S, Moreno-Nieves UY, Garcia-Beltran WF, Rands K, Reardon J, Toth I, Piechocka-Trocha A, Altfeld M, Addo MM. Dysregulated Tim-3 expression on natural killer cells is associated with increased Galectin-9 levels in HIV-1 infection. Retrovirology. 2013; 10:74.

4. Gleason MK1, Lenvik TR, McCullar V, Felices M, O'Brien MS, Cooley SA, Verneris MR, Cichocki F, Holman CJ, Panoskaltsis-Mortari A, Niki T, Hirashima M, Blazar BR, Miller JS. Tim-3 is an inducible human natural killer cell receptor that enhances interferon gamma production in response to galectin-9. Blood. 2012; 119:3064-72.

5. Ju Y, Hou N, Meng J, Wang X, Zhang X, Zhao D, Liu Y, Zhu F, Zhang L, Sun W, Liang X, Gao L, Ma C. T cell immunoglobulin- and mucin-domain-containing molecule-3 (Tim-3) mediates natural killer cell suppression in chronic hepatitis B. J Heptatol. 2010; 52:322-9.

6. da Silva IP, Gallois A, Jimenez-Baranda S, Khan S, Anderson AC, Kuchroo VK, Osman I, Bhardwaj N. Reversal of NK-cell exhaustion in advanced melanoma by Tim-3 blockade. Cancer Immunol Res. 2014; 2:410-22.

Ethics Approval

This study was approved by Purdue Intuition’s Ethics Board, approval number 1804020540.

Correspondence: Danielle Dillard (danielle.dillard@jhmi.edu)

Background

Human papillomavirus positive (HPV+) oropharyngeal squamous cell carcinoma (HPV-OPSCC) accounts for ~80% of OPSCCs in the United States. Although HPV-positivity confers improved survival relative to HPV(-) OPSCC, outcomes for metastatic HPV-OPSCC remain dismal. High tumor infiltrating lymphocytes (TILs) are associated with better outcomes in HPV-OPSCC and have the promise of a therapeutic role based upon their intrinsic enhanced tumor specificity. However, not all tumors possess TILs. To date, therapeutic expansions involve surgical excision of the tumor, expansion of TILs with high-dose IL-2 for up to 4 weeks and clinical product success rate between 40-60%. Marrow infiltrating lymphocytes (MILs) represent a novel and distinct T cell population obtained from the bone marrow (BM) of patients that possess significant tumor-specificity over peripheral blood lymphocytes (PBLs). Although the early work was done in myeloma, the unique nature of the BM microenvironment makes it a reservoir of antigen-experienced memory T cells in numerous solid tumors. As such, we sought to determine whether HPV-specific MILs could be identified and expanded ex vivo.

Methods

We obtained bone marrow and peripheral blood from patients with localized, HPV-OPSCC. A modified version of the MANAFEST assay was used to evaluate proliferation of peripheral and bone marrow-derived CD8+ T cells in response to HPV early 6 (E6) and early 7 (E7) peptides (HPVFEST) in a HPV-OPSCC patient.

Results

T cell receptor sequencing and bioinformatic analysis of each peptide-stimulated culture revealed a markedly increased frequency of HPV-specific T cells in MILs compared to peripheral blood. HPV-specific MILs showed a higher average TCR clone frequency relative to PBLs (p value = 0.0037). Additionally, of the 5 highest expanded TCR clones for both compartments HPV-specific MILs possessed an increased average representative clone frequency (p value = 0.0001). To investigate general responsiveness to shared tumor antigens, we incubated autologous BM with lysate from an HPV+ OPSCC and then added ex-vivo activated MILs and PBLs. Tumor specificity was defined as IFN훾 production in CD3 cells. MILs possessed an average of 50.73% IFN훾 production as compared to 0.11% in PBLs to HPV-OPSCC lysate.

Conclusions

Collectively, these data indicate that HPV-specific T cells exist in the BM of patients with localized disease and possess a greater clonoytpic frequency and functional tumor recognition compared to PBLs. These data provide the rationale for developing this novel adoptive T cell approach using MILs in this patient population.

Ethics Approval

This study was approved by Johns Hopkins University Institution Review Board, approval number IRB00128334 and NA_00028682.

Correspondence: Yonghong Zhang (yhzhang58@126.com)

Background

Currently, the prognosis of children with relapsed refractory(r/r) Burkitt lymphoma(BL) remains dismal . New therapies are exlored to achieve a higher remission rate such as immunotherapy for these patients .We have successfully treated a case by adopting sequential autologous chimeric antigen receptor T cell(CAR-T) therapies, targeting antigen CD19,CD22,and CD20.

Methods

An 8-year-old boy was studied,who presented with a mass on the right side of the neck and was diagnosed with BL by pathology. The child was treated with standard chemotherapy but suffered from relapse. Subsequently, anti-CD19, anti-CD22, and anti-CD20 autologous CAR-T cell treatments were sequentially administered. We observed the clinical manifestations and response to the three cycles of CAR-T treatments, values of peripheral CAR-T cells were also monitored and side effects were assessed.

Results

The patient displayed no response to anti-CD19-CART treatment .After CD-22 directed CART the patient got partial remission (PR),but relapse occurred quickly. Finally, after the use of anti-CD20 CAR-T cell therapy, the child achieved complete remission (CR) and has currently achieved a 6-month event-free survival (EFS). During the CD19 and CD20 CAR-T cell treatments, only mild cytokine release syndrome (CRS) were observed in the patient (grade 1) while he developed a grade 3 CRS during CD22 CAR-T therapy, the symptoms included fever and hypoxemia.

Conclusions

Autologous CAR-T cell therapies targeting multplei tumor antigens could be novel and safe treatments for children with r/r BL.

Consent

Written informed consent was obtained from the patient for publication of this abstract and any accompanying images. A copy of the written consent is available for review by the Editor of this journal.

Correspondence: Christopher Klebanoff (klebanoc@mskcc.org); Smita Chandran (chandrs1@mskcc.org)

Background

“Public” neoantigens represent immunogenic epitopes encompassing hotspot mutations in driver oncogenes that are also restricted by common HLA alleles. In contrast with patient-specific “private” neoantigens, public neoantigens are conceptually attractive because they are tumor-specific, clonally conserved, and shared across patients. Whether PIK3CA, the most common driver oncogene in breast and endometrial cancer, can yield public neoepitopes that may be exploited for cancer immunotherapy is unknown.

Methods

We have developed a high-throughput, single-cell functional assay for the discovery and retrieval of TCR alpha/beta gene sequences that confer specific recognition of endogenously processed and presented public neoantigens and not the corresponding wild type (WT) sequence. In this approach, donor-derived T cells are sensitized with autologous antigen presenting cells (APCs) electroporated with RNA encoding PIK3CA hotspot mutations. Expanded T cells are subsequently divided into paired daughter wells for short-term co-culture with APCs electroporated with minigenes containing either mutant or WT PIK3CA sequences. Acutely re-stimulated T cells from paired wells are subject to single-cell alpha/beta TCR VDJ and RNA sequencing. TCR alpha/beta gene sequences associated with selective upregulation of TCR signaling transcripts to mutant but not WT PIK3CA stimulation are subsequently cloned into retroviral vectors to confirm reactivity.

Results

Using this method, we have retrieved multiple TCRs that confer specific recognition of a public neoepitope derived from a PIK3CA hotspot mutation (H1047L). These TCRs are restricted by HLA-A*03:01, an allele present in 20.5% of the North American population. Immune-precipitation/tandem mass spectrometry analysis determined that the endogenously processed and presented public neoepitope is a 9 amino acid sequence containing a His to Leu substitution at position 2. To understand the mechanistic basis for the immunogenicity of this public neoepitope, we generated x-ray crystallography structures of mutant and WT epitopes bound to HLA-A*03:01 at ~2Å resolution. These studies revealed significant topologic overlap in the bound peptides. By contrast, the thermal and kinetic stability of the mutant peptide/HLA-A*03:01 complex was significantly enhanced relative to the WT complex, as measured by differential scanning fluorimetry and fluorescence anisotropy assays. Peripheral blood T cells genetically engineered with PIK3CA public neoantigen-specific TCRs cytolytically cleared target cells in a HLA/mutation-specific manner, leaving HLA-mismatched or WT target cells unperturbed.

Conclusions

These findings reveal for the first time the existence of an endogenously processed and presented public neoantigen derived from a PIK3CA hotspot mutation. These results open the possibility of targeting this common driver oncogene using adoptively transferred and genetically redirected T cells.

Ethics Approval

The study was approved by Smita Chandran's Institution‘s Ethics Board, approval number IRB 17-250.

Correspondence: Jordan Green (green@jhu.edu)

Background

Particulate delivery of artificial antigen presenting cells (aAPCs) is a promising cell-free strategy to initiate selective T cell stimulation for immunotherapy in vivo. While 2D aAPC strategies aim to optimize T cell proliferation and selection in vitro for subsequent cell therapy, it would be advantageous to deliver “off-the-shelf” biodegradable aAPCs directly as an in vivo therapeutic. 3D aAPC effectiveness has been limited due to inefficiencies in stimulating T cells as well as rapid clearance of delivered particles. To improve bioavailability as well as increase particulate aAPC effectiveness, we have developed a soft, biodegradable, microparticle aAPC (Figure 1). To create a new platform technology for immunoengineering, material and shape are investigated as parameters for improving T cell stimulation.

Methods

One micron size particles were synthesized using a poly(ethylene glycol) diacrylate (PEGDA) or using a poly(lactic-co-glycolic acid) emulsion method [1,2]. Anti-CD3 and anti-CD28 conjugated particles were incubated with primary mouse T cells and proliferation was quantified at 3 and 7 days. For macrophage uptake studies, particles were incubated with macrophages at 37 °C to analyze particle uptake and 4 °C to evaluate binding of particles to cells. To synthesize soft, ellipsoidal aAPCs, a novel thin-film stretching technique was developed where emulsified PEGDA droplets were frozen then cast into films and stretched [3].

Results

Protein conjugation efficiency and T cell proliferation were 10-fold higher for PEGDA particles than PLGA particles (Figure 2a-b). Uptake studies indicate a ~20-fold decrease in binding and uptake by the PEGDA particles (Figure 2c). Ellipsoidal aAPCs stimulate T cells 3 times more effectively than spherical particles (Figure 3b,d). Uptake studies indicate a ~10-fold decrease in nonspecific uptake of ellipsoidal particles (Figure 3c).

Conclusions

Particle material and shape are significant factors in designing particulates as biomimetic aAPCs for in vitro T cell stimulation. Ongoing work on further decoupling these parameters and optimizing in vivo efficacy has the potential to unleash a promising biomimetic platform technology for immunoengineering of T cells.

References

1. Anselmo AC, Mitragotri S. Impact of Particle Elasticity on Particle-Based Drug Delivery Systems. Adv. Drug Deliv. Rev. 2017; 108:51–67.

2. Meyer RA, Sunshine JC. Biodegradable Nanoellipsoidal Artificial Antigen Presenting Cells for T Cell Activation. Small. 2016; 11:1519–1525.

3. Meyer RA, Meyer RS, Green JJ. An Automated Multidimensional Thin Film Stretching Device for the Generation of Anisotropic Polymeric Micro- and Nanoparticles. J. Biomed. Mater. Res. A. 2015; 103:2747–2757.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P172).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P172).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P172).

See text for description

Correspondence: Indu Ramachandran (indu.ramachandran@adaptimmune.com)

Background

ADP-A2M4 is a genetically engineered autologous affinity-enhanced receptor immunotherapy (SPEAR T-cells) directed towards a MAGE-A4 peptide expressed in the context of HLA-A*02 on tumor cells. Clinical responses with ADP-A2M4 have been reported in patients with advanced MAGE-A4+ synovial sarcoma (SS) tumors. Here, we describe intra-tumoral and peripheral correlates associated with clinical response and resistance in two patients with SS.

Methods

Transduced T-cell persistence was determined by qPCR in PBMCs. Serum cytokines were measured via a multiplexed electrochemiluminescence-based immunoassay (MSD). Immunohistochemistry for antigen and immune markers was performed on FFPE tumor biopsies collected from patients prior to and following ADP-A2M4 transfer. A digital PCR-based assay was performed on FFPE tumor biopsies to detect the presence of SPEAR T-cells in the tumor. T-cell cytotoxicity assays were performed in vitro using the IncuCyte® platform. Clinical responses were assessed by RECIST v1.1.

Results

In the first patient, the best overall response (BOR) following ADP-A2M4 treatment was a partial response. Multiple correlates previously shown to be associated with response were observed. The patient’s pre- and post-infusion tumor biopsies expressed high levels of MAGE-A4 protein. Post-infusion, high levels of persisting transduced cells were observed in peripheral blood. Additionally, the patient had a grade 2 CRS event associated with a high level of serum IFN-g and IL-15 induction. In the post-infusion tumor sample, a notable increase in CD3+ T-cell infiltration, including SPEAR T-cells, was observed along with PD-L1 induction.

In the second patient, the BOR was stable disease; then the disease progressed. MAGE-A4 protein expression was lower prior to ADP-A2M4 infusion, compared to the 1st patient. Minimal peripheral induction of IFN-g and IL-15 was observed post-infusion along with a lower level of transduced T-cell persistence, compared with the 1st patient. No CRS was reported in this patient. Both patients’ manufactured products contained transduced CD8+ T-cells capable of killing antigen-expressing targets in vitro. In the responding patient, effective target killing was observed in transduced CD8+ T-cells isolated from the tumor site post-infusion. Eight additional SS patients have been treated, and we continue to analyze biomarkers in these patients.

Conclusions

Based on these two cases, we have identified some factors that may contribute to the anti-tumor activity of ADP-A2M4. High antigen expression levels, IL-15 and IFN-g cytokine induction, good engraftment, tumor site trafficking, and cytolytic function of SPEAR T-cells may be associated with favorable responses in SS patients treated with ADP-A2M4.

Trial Registration

NCT03132922

Correspondence: Anja Feldmann (a.feldmann@hzdr.de)

Background

Although T-cells genetically modified to express chimeric antigen receptors (CARs) are successfully used to treat hematological malignancies, patients still suffer from several drawbacks of conventional CAR (cCAR) therapy. CAR-T-cells can cause severe to life-threatening adverse reactions like on-target, off-tumor toxicities which cannot be controlled in patients. Moreover, cCAR therapy often fails to successfully affect solid tumors and bears the risk to encourage tumor escape variants upon targeting of only one single tumor-associated antigen (TAA). In order to overcome these problems, we have established a novel on/off-switchable RevCAR system facilitating combinatorial targeting strategies.

Methods

For combinatorial targeting one T-cell has to be modified with two separate CARs recognizing different TAAs. The first CAR mediates the activation and the second CAR the costimulatory signal. In case of ‘AND’ gate targeting, dual-CAR-T-cells have to recognize both TAAs on the surface of the target cells to get activated. However, such combinatorial targeting strategies are struggling with several challenges including the adjustment of signal strength and affinity of both split CARs as well as the CAR size limiting the number of transduced specificities. In order to overcome these obstacles, our idea was to construct small RevCARs comprising only a small peptide epitope as extracellular domain. By removing the extracellular single-chain variable fragment (scFv) of cCARs, RevCARs avoid tonic signaling induced by scFv dimerization. As RevCARs do not have an extracellular antigen binding moiety, they cannot bind to any antigen per se. Thus, actually they are switched off. Only in the presence of a bispecific target module (RevTM), RevCAR-T-cells can be redirected to tumor cells and switched on. Finally, short-living RevTMs allow a repeatedly on/off-switch and controllability of RevCAR-T-cells and furthermore a flexible redirection of RevCAR-T-cells to any target.

Results

For proof of concept two small peptide epitopes were selected to construct the respective RevCARs. Additionally, a series of different RevTMs was generated recognizing one of the two peptide epitopes and simultaneously any potential TAA. RevTMs were able to efficiently redirect RevCAR-T-cells specifically against different tumor targets. Moreover, we show that combinatorial targeting can be achieved using our RevCAR system. Here, dual-RevCAR-T-cells were efficiently activated only after engagement by two RevTMs targeting the activating or costimulatory RevCAR and different TAAs.

Conclusions

Taken together, we developed a switchable RevCAR platform showing high effectiveness, increased specificity, improved safety, easy controllability, and small size facilitating combinatorial tumor targeting.

Ethics Approval

The study was approved by local authorities and the Ethics Board.

Correspondence: Antonella Vitiello (avitiello@persimmune.com)

Background

Immune checkpoint inhibitors (ICIs) are being tested in myelodysplastic syndromes (MDS) based on pre-clinical data suggesting that the relevant targets are expressed on tumor and immune cells. Here we study both tumor cells and T cells from patients with higher-risk MDS to assess the role of PD-L1.

Methods

Patients’ CD3+ control cells, CD34+ stem cells, and their autologous MDS cell lines (MCLs) were analyzed by DNA and RNA sequencing to identify somatic variants present in the tumor cells and absent from the control cells. From all somatic variants identified, we generated and tested neopeptides in vitro for their ability to induce tumor-specific T cell responses. A T cell killing assay was performed to assess which neopeptide-specific T cells were capable of mediating tumor cell lysis (Figure 1A). In parallel, tumor PD-L1 expression levels were measured by flow cytometry before and after 24-hour incubation with tumor-specific autologous T cells. As a control for background tumor cell lysis and PD-L1 expression, tumor cells were also incubated with CEF-specific T cells (CEF: CMV, EBV, and flu peptides).

Results

Patients’ tumor cells did not express PD-L1 at baseline. Remarkably, after co-culture, PD-L1 expression on the tumor cells ranged from 20% to 70% (Figure 1B). Tumor cells, when incubated with CEF-specific T cells, did not upregulate PD-L1, suggesting that PD-L1 expression may be linked to target recognition by neoantigen-specific T cells (Figure 2). Interestingly, tumor cell lysis was independent of PD-L1 expression on tumor cells (Figure 1C). Additionally, IFNg neutralization did not affect PD-L1 expression nor ability to lyse tumor cells (Figure 3). These data show that when tumor cells are incubated with autologous tumor-specific T cells, the tumor cells upregulate PD-L1 expression yet do not escape lysis by T cells.

Since lysis of tumor cells may occur prior to their upregulating PD-L1, we pre-incubated tumor cells with soluble IFNg prior to co-culture with T cells. Tumor cells were 96% PD-L1+, and were lysed by tumor-specific T cells at the same level of target cells that had not been treated with IFNg (Figure 4).

Conclusions

Collectively, these data lend support to the notion that, in this system, PD-L1 is not a main player in MDS tumor immune evasion, suggesting that tumor immune evasion might function in a PD-L1-independent way. They also suggest that cognate recognition of tumor cells by neoantigen-specific T cells can cause the upregulation PD-L1 on target cells via a yet to be identified mechanism.

Ethics Approval

The study was approved by the University of California, San Diego's Institutional Review Board, HRPP #161345.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P175).

Ferrari, V SITC 2019 abstract slide 1

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P175).

Ferrari, V SITC 2019 abstract slide 2

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P175).

Ferrari, V SITC 2019 abstract slide 2

Correspondence: Chantale Bernatchez (CBernatchez@mdanderson.org)

Background

MD Anderson Cancer Center’s tumor-infiltrating lymphocyte (TIL) program has expanded TIL from tumor fragments of cutaneous metastatic melanoma using high dose IL-2 from over 900 patients with a growth success averaging 62% [1]. Surprisingly, this growth success plunges to 45% for metastatic uveal melanoma tumor fragments [2] and less than 20% from primary uveal tumors. The reason for this drop is unclear as uveal melanomas have an infiltration of CD8+TIL comparable to cutaneous melanoma [2], which in theory makes it an attractive candidate for immunotherapy. However, limited success observed with checkpoint therapy prompted us to explore ex vivo manipulation of the TIL. A previous report demonstrated the feasibility of TIL adoptive cell therapy for metastatic uveal patients using a TIL product expanded from metastases [3].

Since the primary and metastatic sites of uveal melanoma display preserved gene mutations, indicating a potential shared antigen landscape, one could propose generating a TIL product from the primary tumor when the patient undergoes enucleation and to utilize this product for treatment at time of recurrence.

Methods

Given the challenge of propagating TIL from a primary uveal tumor in high dose of IL-2 only, we hypothesized that our new TIL3.0 method to propagate TIL from tumor fragments (1st phase of expansion), based on the 3-signals required for optimal activation of a T-cell (TCR engagement, costimulation and cytokine exposure) would enable TIL growth from a higher percentage of primary uveal tumors given the success obtained with metastatic sites [1]. This product can be banked and accessed later for treatment at recurrence.

Results

The TIL3.0 expansion platform was shown to be optimal for T-cell propagation allowing for successful expansion of TIL from primary uveal melanoma tumors in >90% of the cases (n=20). This expansion was rapid (less than 3 weeks) and consistently composed of CD8+CD3+TIL. This later observation is attributed to the use of the agonistic anti-CD137/4-1BB, Urelumab, as of costimulation signal in our TIL3.0 method.

The TIL3.0 method applied to primary tumors could be scaled and adapted for GMP. This process, followed by a rapid expansion protocol, was applied to treat the first metastatic uveal patient with a TIL product generated from the primary tumor. The patient was infused with a total of 14.4 billion TIL with a viability of 99%.

Conclusions

This study demonstrates the feasibility of generating a TIL product from a primary uveal tumor to be used for treatment at recurrence.

Trial Registration

NCT00338377

References

1. Tavera RJ, Forget MA, et al. Utilizing T-cell Activation Signals 1, 2, and 3 for Tumor-infiltrating Lymphocytes (TIL) Expansion: The Advantage Over the Sole Use of Interleukin-2 in Cutaneous and Uveal Melanoma. J Immunother. 2018; 41:399-405.

2. Qin Y, de Macedo MP, Reuben A, et al. Parallel profiling of immune infiltrate subsets in uveal melanoma versus cutaneous melanoma unveils similarities and differences: A pilot study. OncoImmunology. 2017; 6:e1321187.

3. Chandran SS, et al. Treatment of metastatic uveal melanoma with adoptive transfer of tumour-infiltrating lymphocytes: a single-centre, two-stage, single-arm, phase 2 study. Lancet Oncol. 2017; 18:792-802.

Ethics Approval

Institutional review board (IRB)-approved protocol# 2004-0069

Correspondence: Alan Friedman (afriedm2@jhmi.edu)

Background

NF-kB p50 binds DNA but, unlike p65, lacks a trans-activation domain and recruits co-repressors. Macrophages and dendritic cells lacking NF-kB p50 are skewed towards a pro-inflammatory phenotype, with increased cytokine expression and enhanced T cell activation; additionally, murine melanoma, fibrosarcoma, colon carcinoma, and glioblastoma grow slower in p50-/- mice. Given these data, we evaluated efficacy of p50-deficient immature myeloid cells (p50-IMC) adoptively transferred into tumor-bearing hosts. Immature cells were utilized to maximize tumor localization, and pretreatment with 5-fluorouracil (5FU) was examined due to its potential to impair marrow production of myeloid cells, to target tumor myeloid cells, and to potentially release tumor neoantigens.

Methods

WT-IMC or p50-IMC were generated by culturing lineage-negative marrow cells from WT or p50-/- mice in media containing TPO, SCF, and FL for six days followed by M-CSF for one day on ultra-low attachment plates. Mice inoculated with Hi-Myc prostate cancer (PCa) or K-Ras(G12D) pancreatic ductal carcinoma (PDC)-luciferase cells received 5FU followed five days later by three doses of 1E7 IMC every three to four days. Some groups also received four doses of anti-PD-1 antibody twice weekly alone or with p50-IMC.

Results

PCa grew slower in p50-/- mice, and absence of host p50 led to prolonged survival of mice inoculated orthotopically with PDC. 5FU followed by p50-IMC slowed PCa and PDC tumor growth ~3-fold in contrast to 5FU followed by WT-IMC, 5FU alone, or p50-IMC alone. Slowed tumor growth was evident for 93% of PCa tumors but only 53% of PDC tumors. In PCa, p50-IMC predominantly generated tumor and draining lymph node F4/80+ macrophages, but also CD11b+F4/80-CD11c+ conventional dendritic cells. A subset of tumor and nodal macrophages co-expressed Ly6C and MHCII and had reduced MR compared to host macrophages, collectively indicating a pro-inflammatory phenotype. p50-IMC also produced a 5-fold increase in activated PCa tumor CD8 T cells, and antibody-mediated CD8 T cell depletion obviated slower tumor growth induced by 5FU followed by p50-IMC. Anti-PD-1 markedly slowed PCa growth but had little efficacy against PDC, whereas anti-PD-1 combined with p50-IMC slowed PDC tumor growth to prolong survival more effectively than either alone in an initial experiment.

Conclusions

5FU followed by p50-IMC slows the growth of murine prostate and pancreatic ductal carcinoma and depends upon CD8 T cell activation. Deletion of p50 in patient-derived marrow CD34+ cells and subsequent production of IMC for adoptive transfer may contribute to the therapy of these and additional cancers, alone or with additional immunotherapies.

Ethics Approval

The study was approved by the Johns Hopkins University Animal Care and Use Committee, protocol MO19M10.

Correspondence: Edmund Waller (ewaller@emory.edu)

Background

For chimeric antigen receptor T-cell (CAR T) therapy to treat chronic lymphocytic leukemia (CLL), recent work associates remissions with infusion of sufficient non-exhausted memory CAR T, capable of oxidative phosphorylation [1]. Our work aims to modulate metabolic pathways during ex vivo expansion of T-cells for translation of these findings to clinical adoptive therapies. Class I catalytic PI3K enzymes, such as PI3Kdelta and PI3Kgamma, regulate T-cell differentiation, regulatory T cell formation, and TCR signaling [2]. In this study, we hypothesized that pharmacological inhibition of these pathways during ex vivo culture would increase populations of early memory CAR T with enhanced metabolic and survival potential.

Methods

Healthy- and CLL- donor peripheral blood mononuclear cells were isolated and cryopreserved. Thawed cells were sorted for CD3 expression prior to culture (or transduction) and expanded in G-Rex plates using anti-CD3/CD28 beads, 30 U/mL interleukin-2, and investigational drugs: idelalisib, duvelisib and ibrutinib for 9 days.

Results

Class I catalytic enzymes in T-cells, PI3Kdelta and PI3Kgamma, exhibit domain homology (Figure-1A). Accordingly, maximum cell yields for both idelalisib (Figure 1B) and duvelisib (Figure 1C) occurred upon inhibition of both isoforms. Comparing doses of duvelisib, idelalisib, and ibrutinib that yielded optimal T-cell expansion, we showed potent PI3Kdelta/gamma dual antagonism maximizes live T-cell yields (Figure 1D) out-performing interleukin-2-inducible kinase inhibition. PI3K antagonists increased frequencies of cells expressing co-stimulatory molecules (Figure 2A-B). A dose-dependent increase in cells expressing FAS/FAS-L (Figure 2C) and an increased expression of pro-survival BCL-2 (Figure 2D) after anti-CD3/28 stimulation suggests the increase in live cells with a TSCM phenotype is likely due to enhanced cell survival.

Next, we confirmed the positive effect of dual PI3Kdelta/gamma inhibition in expansion of T cells from CLL donors (Figure 3A-B). PI3K antagonists increased frequencies of CD8 cells (Figure 3C) and co-stimulatory molecule expressing cells (Figure 3D-E). Interestingly, addition of idelalisib to T cell cultures resulted in a dose-dependent decrease in immune checkpoint molecules LAG-3, Tim-3, and PD-1 (Figure 3F-J). Given the importance of T cells with the stem cell memory (TSCM) phenotype in adoptive T-cell therapy, T-cell differentiation was studied. PI3K antagonists increase the frequency of early, T-memory, and effector memory cells (Figure 4A). Notably, the frequency of TSCM doubled (Figure 4B). Lastly, PI3K antagonists significantly increased the mitochondrial mass (Figure 4C) within total CD3 (Figure 4D) and CD8 (Figure 4E) subsets.

Conclusions

Dual inhibition of PI3Kdelta and PI3Kgamma modulates aspects of T-cell biology relevant to CAR T remissions. PI3Kdelta/gamma antagonism enhances CD27 expression and mitochondrial mass, decreases immune checkpoint expression, and enriches the TSCM phenotype.

Acknowledgements

The authors thank the patients and healthy volunteers for their blood donations. CR Funk is supported by a Howard Hughes Medical Research Fellowship. The authors thank Emory’s Clinical Lymphoma Research Team for requesting patient consent and aiding in sample collection.

References

1. Fraietta JA, et al. Determinants of response and resistance to CD19 chimeric antigen receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat. Med. 2018; 24:563-571.

[2]2. Fruman DA, Chiu H, Hopkins BD, Bagrodia S, Cantley LC, and Abraham RT. Leading Edge The PI3K Pathway in Human Disease. Cell. 2017; 170:605-635.

Ethics Approval

The study was approved by Emory University's Institutional Review Committee, approval number 00057236.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P178).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2 (abstract P178).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3 (abstract P178).

See text for description

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4 (abstract P178).

See text for description

Correspondence: Anderson Gaweco (anderson.gaweco@apeiron-biologics.com)

Background

The intracellular master checkpoint Cbl-b, an E3 ubiquitin ligase, negatively regulates the innate and adaptive anti-tumor immune responses. Selective cell-based targeting of Cbl-b not only induces anti-tumor activity in vivo but also overrides immune regulation by the PD-L1/PD-1 pathway in vitro [1]. Human APN401 is an autologous adoptive cellular therapy of ex vivo Cbl-b-silenced human PBMCs currently in a clinical Phase 1b multiple dose study demonstrating early clinical safety and tolerability in patients with advanced solid tumors. Herein the preclinical Proof of Concept efficacy of murine APN401 immunotherapy is established in the syngeneic mouse hepatocellular carcinoma Hepa1-6 tumor rechallenge study.

Methods

Hepa1-6-C57/BL6-tumor bearing mice were treated for 19 days with murine APN401, ex vivo silenced immune cells with Cbl-b specific siRNA, as monotherapy or in combination with anti-PD1 (clone RMP1-14) versus control siRNA on D7 post-inoculation. Mice were later rechallenged on D29 with Hepa1-6 s.c. on the contralateral flank in the absence of any further APN401 treatment.

Results

Significant tumor growth inhibition (TGI) was observed following 19 days of treatment with APN401 alone or in combination with anti-PD1 starting on D7 after s.c. inoculation (p<0.0001). Following rechallenge, significant TGI of 86% (p<0.001) was observed in prior APN401-treated mice versus control siRNA-treated mice. Mice that received APN401 in combination with anti-PD1 prior to rechallenge demonstrate profound synergistic anti-tumor efficacy (p<0.001) versus anti-PD1 alone with control siRNA. APN401 was well tolerated and APN401-treated mice were unremarkable with optimal body conditions.

Conclusions

APN401 monotherapy demonstrates striking anti-tumor efficacy in the murine hepatocellular carcinoma Hepa1-6 model. The preclinical synergistic effects of APN401 with anti-PD1 support its therapeutic utility as a combination therapy with immune checkpoint anti-PD1 treatment. The significant TGI observed following tumor rechallenge indicate that prior selective cell-based Cbl-b-silencing with APN401 alone or in combination with anti-PD1 induced systemic and durable anti-tumor immune memory responses. These findings highlight the potential promise of a selective adoptive cell-based Cbl-b silencing by APN401 as a novel immunotherapy for cancer.

Reference

1. Fujiwara M, Anstadt EJ, Clark RB. Cbl-b Deficiency Mediates Resistance to Programmed Death-Ligand 1/Programmed Death-1 Regulation. Front Immunol. 2017 Jan 26;8:42.

Correspondence: Gary Lee (gary.lee@sentibio.com)

Background

More effective therapies for disseminated peritoneal carcinomatosis, including high-grade serous ovarian cancer, remain a major medical need [1]. Although several treatments offer initial responses to localized disease, patients with disseminated peritoneal tumors face poor overall survival [2].

SENTI-101 is a novel therapeutic agent comprising allogeneic mesenchymal stromal cells (MSCs) genetically modified to express a potent combination of immunomodulatory cytokines: IL12 and IL21. Upon administration, SENTI-101 innately homes to peritoneal tumors, secretes IL12 and IL21 in a localized and sustained fashion, and induces a robust anti-tumor immune response.

Methods

Two syngeneic pre-clinical models of disseminated peritoneal carcinomatosis with distinct immune phenotypes were established by implanting cells in the peritoneal cavities of mice (CT26-fLUC = immune-inflamed; B16-F10-fLUC = immune-excluded) [3]. A library of over 50 murine MSC lines engineered to express immune effectors (cytokines, chemokines, growth factors), either individually or in combination, was administered intraperitoneally and evaluated for anti-tumor activity via bioluminescence and tumor weight measurements. Immune phenotype was characterized by flow-cytometry and multiplexed immunohistochemistry.

Results

MSCs expressing the combination of IL12 and IL21 (SENTI-101) were selected based on significant tumor-burden reduction and immune profile changes in both syngeneic models. Notably, the combination outperformed each individual cytokine in extending survival (p=0.02).

Intraperitoneal administration of SENTI-101 into tumor-bearing mice led to preferential co-localization with tumors (>10-fold higher vs. normal tissues, p=0.001). Local concentrations of IL12 and IL21 were ~100-fold greater in the peritoneal space vs. serum (p=0.002). SENTI-101 treatment reduced tumor-burden more than 200-fold (p50% of the mice were tumor-free after 90 days, while control groups and groups treated with anti-PD1 antibody had a median survival of 21 to 30 days. Surviving mice were able to reject newly implanted tumor cells, demonstrating anti-tumor immune memory.

Anti-tumor effects of SENTI-101 are mediated by a multi-modal immune response. The frequency of antigen-presenting cells in peritoneal tumor-draining lymph nodes was more than doubled vs. controls (p=0.01). This correlated with increased T-cell and B-cell tumor infiltrates forming tertiary-lymphoid structures, which are associated with improved prognosis in cancer [4]. T-cell activation markers (CD38, IFNg, GranzymeB) were significantly increased locally.

Conclusions

SENTI-101 induces localized immune-modulation, regulates multiple steps of the cancer immunity cycle, and results in durable anti-tumor responses. These data warrant further development of SENTI-101 for the loco-regional treatment of advanced solid tumors.

References

1. Desai JP, Moustarah F. Cancer, Peritoneal Metastasis. StatPearls Publishing. 2019 [Updated 2019 Jun 30]

2. Lengyel E. Ovarian Cancer Development and Metastasis. Am J Pathol. 2010; 177(3): 1053–1064

3. Mosely SI, Prime JE, Sainson RC, et al. Rational Selection of Syngeneic Preclinical Tumor Models for Immunotherapeutic Drug Discovery. Cancer Immunol Res. 2017; 5(1):29-41

4. Sautès-Fridman C, Petitprez F, Calderaro J, Fridman WH. Tertiary lymphoid structures in the era of cancer immunotherapy. Nat Rev Cancer. 2019;19(6):307-325

Correspondence: Nicholas Restifo (restifo@nih.gov)

Background

Adoptive T cell transfer immunotherapy (ACT) using tumor-infiltrating lymphocytes (TIL) and gene-modified T cells can induce complete and durable regression of metastatic human malignancies that are otherwise refractory to treatment. While successful T cell-based treatments of patients with widely metastatic melanoma, synovial sarcoma, cholangiocarcinoma and cancers of the breast, colon, and cervix have been reported in recent years, most patients with common epithelial cancers fail to respond to treatment. While several factors can contribute to the efficacy of ACT, a major inherent limitation is the induction of terminally differentiated phenotype coupled with the loss of proliferative capacity in TIL during current ex vivo expansion protocols. Individual gene knockout approaches for enhancing T cell-based cancer immunotherapies are low-throughput and can improve one desired function (T cell memory) at the expense of another equally important function (expansion). Thus, there is significant interest in identifying T-cell intrinsic negative regulatory circuits that limit their ability to expand robustly ex vivo, while dampening their terminal effector differentiation along with the reduction of oxidative stress and genomic damage.

Methods

To identify the T cell intrinsic negative regulatory circuits, we developed a multi-phenotype genetic screen to systematically target 29 major kinases screen to concurrently measure the impacts of individual gene knockouts on T cell expansion, differentiation, oxidative stress and genomic stress. Using CRISPR-Cas9-based gene perturbation combined with high-throughput flow cytometry, we developed and validated a multi-phenotype screen, which identified Mapk14/p38 kinase as a target that improved all four phenotypes in CD8+ T cells. We used murine and human ex vivo T cell expansion models to validate the results from our genetic screen.

Results

Results from our genetic screen identified p38 kinase as a unique multi-phenotypic regulator of cellular differentiation, oxidative, and genomic stress while achieving improved cellular expansion. Furthermore, pharmacological inhibition of p38 kinase in murine and human ex vivo T cell expansion models validated the results from our genetic screen. Cells cultured in the presence of a p38 inhibitor had increased capacity for cytokine production, specifically interferon-γ and demonstrated improved in vivo persistence. Additionally, cells cultured in the presence of the p38 inhibitor demonstrated enhanced in vivo cell-expansion, tumor infiltration, and anti-tumor efficacy in an immunocompetent tumor mouse model.

Conclusions

This study establishes p38 inhibition in T cells as a potentially important strategy for improving ACT immunotherapy for cancer patients.

Ethics Approval

All human samples were isolated in accordance with approved clinical protocols and in accordance with NIH institutional review board approval and informed consent from patients and healthy donors.

Correspondence: Yui Harada (rkfraile@med.kyushu-u.ac.jp)

Background

[Background] Cancer immunotherapy has been established as a new therapeutic category since the recent success of immune checkpoint inhibitors and a type of adoptive immunotherapy, namely chimeric antigen receptor-modified T cells (CAR-T). Although CAR-T demonstrated impressive clinical results, serious adverse effects (cytokine storm and on-target off-tumor toxicity) and undefined efficacy on solid tumors are important issues to be solved. We’ve developed a cutting-edge, simple, and feeder-free method to generate highly activated and expanded human NK cells from peripheral blood (US9404083, PCT/JP2018/018236, PCT/JP2019/012744), and have been conducting further investigation why our new type of NK cells, named as GAIA-102, are so effective to kill malignant cells.

Methods

[Materials and Methods] Cryopreserved PBMCs purchased from HemaCare Corporation were mixed and processed by using LOVO and CliniMACS® Prodigy (automated/closed systems). CD3+ and CD34+ cells were depleted, and the cells were cultured at a concentration of 1 x 106 cells/ml with high concentration of hIL-2 and 5% UltraGRO® for 14 days in our original closed system. Then, we confirmed the expression of surface markers, CD107a mobilization and cell-mediated cytotoxicity against various tumor cells and normal cells with or without monoclonal antibody drugs in vitro and antitumor effects against peritoneal dissemination model using SKOV3 in vivo.

Results

[Results and Discussion] Importantly, we’ve found that our GAIA-102 exhibited CD3-/CD56bright/CD57- immature phenotype that could kill various tumor cells efficiently from various origins, including Raji cells that was highly resistant to NK cell killing. More importantly, massive accumulation, retention, infiltration and sphere destruction by GAIA-102 were affected neither by myeloid-derived suppressor cells nor regulatory T-lymphocytes. GAIA-102 was also effective in vivo to murine model of peritoneal dissemination of human ovarian cancer.

Conclusions

Thus, these findings indicate that GAIA-102 has a potential to be an ‘upward compatible’ modality over CAR-T strategy, and would be a new and promising candidate for adoptive immunotherapy against solid tumors. We now just started GMP/GCTP production of this new and powerful NK cells and first-in-human clinical trials in use of GAIA-102 will be initiated on 2020.

Ethics Approval

[Ethics Approval] Written informed consent was obtained from all healthy volunteers, in accordance with the Declaration of Helsinki. Upon the approval of the institutional ethical committee (approval no. 29-315) of Kyushu University, peripheral blood samples were collected from healthy volunteers. The animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Kyushu University (approval nos. A30-234-0 and A30-359-0).

Correspondence: Yukai He (yhe@augusta.edu)

Background

Immunotherapy has a great potential for hepatocellular carcinoma (HCC). Several human glypican 3 (hGPC3)-specific chimeric antigen receptor T cells (CARTs) are being tested for HCC. But, most, if not all, are constructed from one monoclonal antibody (mAb). It is unknown whether targeting different epitopes of hGPC3 will create more effective CARTs. Here, we aim to develop novel CARTs that target different regions of hGPC3.

Methods

BalB/C mice were immunized with hGPC3 protein. Hybridomas and mAbs were generated and characterized. Then, CARTs were built from the novel mAbs and their antitumor effect was studied.

Results

Twenty-two hGPC3-specific mAbs were identified by ELISA. Out of them, 14 bound HepG2 cells. Five mAbs were further characterized by immunohistochemical staining. Three of them (6G11, 8F8, and 12D7) were found to specifically stain HCC tumor but not adjacent normal tissues. The 3 mAb’s affinity were in the nanomolar range. 6G11 and 8F8 bound to hGPC3 epitope aa25-39 and aa463-496, respectively. No specific epitope was identified for 12D7 though it bound to the N-fragment (25-358aa). CARTs built from the 3 mAbs underwent expansion in response to HepG2 cell stimulation. However, their effector function was significantly different. 8F8 CARTs possessed the strongest effector function. 6G11 CARTs generated the greatest expansion, but with slightly weaker function. In contrast, 12D7 CART had the weakest effector function. Soluble hGPC3 did not activate CARTs, nor blocked CART activation by tumor cells. Adoptive transfer of 8F8 and 6G11, but not 12D7, CARTs generated potent antitumor effects with complete regression of HCC xenografts, which correlated to their expansion in vivo.

Conclusions

The three novel CARTs that target different hGPC3 regions possess significantly different effector function and antitumor effects. Adoptive transfer of CARTs targeting the hGPC3 N- or C-epitope results in complete eradication of HCC xenografts.

Correspondence: Ivan Borrello (iborrell@jhmi.edu)

Background

Marrow infiltrating lymphocytes (MILs) are a promising candidate for adoptive cell therapy (ACT) due to their broader anti-tumor specificity and persistence. These characteristics are due to intrinsic properties of the bone marrow (BM); known to be a reservoir for long-lived memory T cells. It has also been established that naïve and memory T cells are metabolically quiescent, favoring oxidative phosphorylation (OXPHOS) over glycolysis, while effector T cells favor glycolysis to fuel their rapid proliferation.

Methods

We examined how activation and expansion of MILs in hypoxia could be used to better understand the inherent properties of the BM, to exploit these properties, and enhance the efficacy of MILs in ACT, especially when compared to that of peripheral blood lymphocytes (PBLs). By activating MILs in hypoxia, we can select for and/or alter the cells best suited to mount an effective anti-tumor response.

Results

Activation under hypoxic conditions alters MILs in several unique ways. MILs show greater overall expansion, enhanced tumor-specificity, and a unique metabolic profile—upregulating both OXPHOS and glycolytic machinery. This metabolic profile suggests that hypoxia-activated MILs possess properties of both effector and memory cells. PBLs grown under the same conditions fail to expand significantly and show no metabolic differences or specificity. Following activation in hypoxia we have found that MILs have upregulated metabolism-related genes such as CPT1A and GLUT1 and 3, as well as anti-apoptotic factors such as BCL2 and BCL2L1 at an RNA level. The post-expansion MILs product, using intracellular staining and FACs analysis, shows an increase in mitochondrial proteins, including TOMM20, CPT1a, and SDHa, increased mTOR signaling, and increased glycolytic machinery—HK2 and GLUT1. Additionally, while cell-cycle analysis with Ki67 and PI shows that MILs are more resting at baseline and arrested in G0, upon activation in hypoxia they become more proliferative—moving into S and G2/M phase—than normoxic MILs or PBLs and maintain this phenotype over time. This finding is supported by our RNAseq data showing a lack of transcriptional activity at baseline but a significant increase by day 3 of activation. Gene expression analysis has also shown that there is a substantial increase in expression of IL2 and IL15Ra with a significant decrease in the expression of IL10 transcripts.

Conclusions

These findings suggest that hypoxia contributes to the unique properties of MILs through modification of their metabolic profile and is being uniquely employed to generate more effective MILs, and not PBLs, for adoptive cell therapy.

Correspondence: Thomas Long (thomas.long@junotherapeutics.com)

Background

Preclinical safety evaluation of chimeric antigen receptor (CAR) T cells presents a number of unique challenges. One of those challenges is the development of a Cynomolgus macaque toxicology model as a tool to understand potential on-target CAR T cell toxicity. This is important for CAR T cell programs where the lead binder is cyno cross-reactive and the target has known normal tissue expression conserved across species. Unlike many mouse models, non-human primates lack target-expressing tumors that can drive CAR T cell activation and expansion; it would be advantageous to recapitulate that expansion in the context of a toxicity model.

Methods

T cell antigen presenting cells (tAPCs) have been reported to drive measurable CAR T cell expansion in Rhesus macaques [1]. The advantages of this strategy include the ease of manufacturing tAPCs in parallel to CAR T cells and co-engraftment of tAPCs with CAR T cells in hematological niches to ensure antigen availability. The optimal tAPC dosing strategy to drive a strong and persistent CAR T cell activity has not been determined. Here, we generated human anti-CD19 CAR T cells expressing firefly luciferase and human T cells expressing a truncated human CD19 (CD19t) lacking the intracellular domain as tAPCs. We then dosed mice either with CAR and tAPC concurrently, or with CAR first then followed by tAPCs three days later.

Results

(Figure 1) shows bioluminescent imaging (BLI) measurements that indicate concurrent and delayed tAPC dosing have similar CAR T cell expansion kinetics; however, the delayed tAPC dosing exhibits a greater magnitude of CAR T cell expansion. In both cases, the CAR T cell expansion occurs in a tAPC dose-dependent manner. Imaging shows that concurrent dosing leads to CAR T cell proliferation primary in the lungs, whereas delayed tAPC dosing leads to more systemic CAR T cell expansion. In addition, flow cytometry data show a significant depletion of tAPCs in the peripheral blood between day 7 and day 14.

Conclusions

Follow-up studies delivering additional tAPC doses three, six, and nine days after CAR T cells show that repeated tAPC administration can significantly increase the CAR T cell exposure over time compared to a single tAPC dose. Overall, these data demonstrate that tAPCs can be used to induce CAR T cell expansion in vivo in the absence of a tumor and will enable us to design a tAPC strategy for use in a Cynomolgus macaque model to evaluate the safety of CAR T cell candidates.

Reference

1. Berger C, Sommermeyer D, Hudecek M, Berger M, Balakrishnan A, Paszkiewicz PJ, Kosasih PL, Rader C, Riddell SR. Safety of targeting ROR1 in primates with chimeric antigen receptor-modified T cells. Cancer Immunol Res. 2015;3(2):206-16.

Ethics Approval

All animal studies were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1 (abstract P185).

See text for description

Correspondence: Liselotte Brix (lb@immudex.com)

Background

Identification of disease-specific T-cell epitopes is key to developing novel cancer vaccines and immunotherapies. Profiling disease-specific T cells, emerging during an induced cellular immune response is important for understanding anti-tumor immunity and guide personalized therapy. The MHC dCODE™Dextramer® technology enables simultaneous screening of high numbers of T-cell specificities in the same sample using MHC multimer-specific DNA barcodes and next generation sequencing as readout. Combining this technology with 10x Genomics Chromium single cell assay further enables the simultaneous analysis of antigen specific T-cells, sequencing of the cognate T-cell receptors and single cell gene expression profiling.

Methods

Panels of up to 50 dCODE™Dextramer® reagents were used for screening antigen-specific T-cells in human blood samples. Single cell sequencing was performed on the isolated T-cell subpopulations, and their gene expression profile analyzed in combination with cell phenotype and TCR sequences.

Results

The experiment generated a large dataset and we show one example of how such a dataset can be analyzed to generate useful information. By combining the gene expression profile, cellular phenotypes and Dextramer specificity we identified expanded populations of antigen-specific T cells in the memory T cell compartment and characterized their individual TCR clonotypes based on TCR sequence. pMHC-specific T-cells were also detected in the naïve T cell compartment, showing a more diverse TCR sequence profile.

Conclusions

This experiment demonstrates a novel method of exploring antigen-specific T-cell responses. Linking TCR sequences with pMHC specificity, cellular phenotypes and gene expression at this scale and resolution provide a more comprehensive analysis of the antigen-specific T cell response than previously available in a single workflow. Novel biomarkers and improved strategies of T cell based immunotherapeutic development will result from T cell analysis at this scale and resolution.

Correspondence: Demetrios Kalaitzidis (d.kalaitzidis@crisprtx.com)

Background

Autologous chimeric antigen receptor T (CAR-T) cell therapies have shown remarkable success in treating relapsed/refractory B-cell malignancies. However, even in indications with high complete response rates, not all patients respond or have durable responses after CAR-T treatment. Furthermore, autologous CAR-T treatments have not yielded the same impressive outcomes in solid malignancies to date. A major limitation of autologous CAR-T therapy may be the dysfunctional state of a patient’s T cell populations used for manufacturing of a drug product. Allogeneic therapeutics can bypass this limitation by enabling the use of healthy donor starting material. Moreover, healthy donor material that exhibits specific T cell attributes can be selected for drug product manufacturing.

Methods

To identify attributes that can be associated with improved performance of CAR-T cells we have characterized T cells from healthy donors as well as cancer patients, in particular from chronic lymphocytic leukemia (CLL) patients as these have been described previously to be dysfunctional.

Results

We show impaired function of cancer patient-derived CAR-T cells when compared to healthy donor-derived cells utilizing both in vitro and in vivo assays. We have performed single-cell RNA sequencing (scRNA seq) on both starting material T cells and CAR-T cells from multiple healthy and CLL donors used in functional assays to uncover both gene expression and population differences associated with CAR-T cell performance. scRNA seq analysis revealed marked heterogeneity among starting populations as well as CAR-T lots from the cancer patient-derived T cells.

Conclusions

Our analysis has allowed us to associate distinct cellular subpopulation and gene expression profiles with preclinical functional outputs.

Correspondence: Xiaokui Zhang (xiaokui.zhang@celularity.com)

Background

Celularity, Inc. is developing human placental CD34+ derived, off-the-shelf, and allogeneic natural killer (PNK) cells for various hematologic malignancies and solid tumors. ACY-241 is an orally bioavailable and selective histone deacetylase (HDAC) 6 inhibitor in MM clinical development. ACY-241 has been shown to sensitize MM cells to endogenous NK cell killing [1]. Here, we investigated the potential augmentation of PNK mediated anti-MM activity by ACY-241 treatment.

Methods

Placental CD34+ cells were cultivated in the presence of cytokines including thrombopoietin, SCF, Flt3 ligand, IL-7, IL-15 and IL-2 for 35 days to generate PNK cells. MM cell lines were treated with different doses (0, 0.1, 0.3, 1, 3, 10 and 30μM) of ACY-241 over 24h, 48h, or 72h. Cytotoxicity of PNK against different doses of ACY-241 pretreated MM cell lines was assessed by a PKH26/TO-PRO-3 FACS based assay. Ligands to NK activating receptors of ACY-241 treated MM cells were evaluated by flow cytometry. The RPMI8226 subcutaneously(SubQ) xenografted NOD scid gamma (NSG) mouse model was used for in vivo efficacy study.

Results

ACY-241 treatment of MM cell lines for >48h resulted in significant inhibition of cancer cell growth and decreased cell viability at doses >10μM. In a dose dependent manner, ACY-241 further enhanced cytotoxicity of PNK against MM cells. In a 4h cytotoxicity assay at effector to target (E:T) ratio of 10:1, relative to vehicle control, PNK (n=3 donors) showed increased cytotoxicity to ACY-241 pretreated MM cell lines: RPMI8226 (16.3% to 34.9%), MM.1S (15.1% to 26.0%), OPM2 (12.7% to 37.2%), and U266 (0% to 9.1%). Increased expression of ligands to activating NK receptors, MIC A/B, CD56, CD54, and CD155 was detected from ACY-241 treated MM cells, suggesting that engagement of NKG2D, CD11a or DNAM-1 of NK cells leads to enhancement of the anti-MM effect.

In vivo anti-MM activity of PNK in combination with ACY-241 was assessed in a RPMI8226 SubQ xenograft NSG model. Single intravenous dosing of 1.0E7 PNK in combination with ACY-241 significantly reduced the tumor growth compared to vehicle control (P<0.001).

Conclusions

Our data demonstrated that enhanced in vitro anti-MM activity of PNK in combination with ACY-241. In vivo efficacy of PNK in combination with ACY-241 was further demonstrated in a RPMI8226 SubQ xenograft NSG model. Taken together, our results demonstrate the synergistic effects of combining an HDAC inhibitor with an NK cell therapy for anti-MM enhancement. Further development of a combinatorial PNK and ACY-241 therapy for MM treatment is warranted.

Reference

1. Ray A, Das DS, Song Y, Hideshima T, Tai YT, Chauhan D, Anderson KC. Combination of a novel HDAC6 inhibitor ACY-241 and anti-PD-L1 antibody enhances anti-tumor immunity and cytotoxicity in multiple myeloma. Leukemia. 2018 Mar;32(3):843-846.

Correspondence: Moshe Giladi (mgiladi@novocure.com)

Background

Tumor Treating Fields (TTFields) are a clinically applied anti-neoplastic treatment modality delivered via noninvasive application of low-intensity, intermediate-frequency, alternating electric fields. In this study we evaluated whether TTFields-induced cell death can be immunogenic and therefore suitable for combination with anti-PD-1 therapy.

Methods

Cancer cells were treated with TTFields using the inovitro(TM) system. Immunogenic cell death (ICD) was characterized by the exposure of calreticulin on the cell surface, secretion of ATP, and release of HMGB1. For detection of ER stress, phosphorylation of eIF2α was assessed. TTFields effect on autophagy was evaluated using electron microscopy, and evaluation of LC3. Bone marrow derived dendritic cells (DCs) were co-incubated with TTFields treated cells and phagocytosis by DCs and DCs maturation were evaluated. The combination of TTFields and anti-PD-1 was evaluated in short duration treatment protocol in orthotopic lung cancer model and long duration treatment protocol in subcutaneous colon cancer model. Analysis of infiltrating cells was performed using flow cytometry.

Results

We demonstrate that cancer cells that die during TTFields application exhibit ER stress leading to calreticulin translocation to the cell surface, as well as release of damage-associated molecular patterns including HMGB1 and ATP. Moreover, we show that TTFields treated cells promote phagocytosis by DCs, DCs maturation in vitro, and promote immune cells recruitment in vivo. We also show that the combined treatment of TTFields plus anti-PD-1 led to a significant decrease in tumor volume and significant increases in CD45+ tumor infiltrating cells in both tumor models. In the lung tumors, these infiltrating cells, specifically macrophages and DCs, demonstrated upregulation of surface PD-L1 expression following short treatment duration. Correspondingly, cytotoxic T-cells isolated from these tumors have shown higher levels of IFN-γ production relative to untreated mice. In the colon cancer tumors, significant increases in T-cell infiltration was observed following long treatment duration with TTFields plus anti-PD-1.

Conclusions

Our results demonstrate the potential of TTFields therapy to induce ICD. We also demonstrate robust efficacy of concurrent application of TTFields and anti PD-1 therapy in mouse models of cancer. These data suggest that combining TTFields with anti-PD-1 might achieve tumor control by further enhancing antitumor immunity.

Acknowledgements

The authors would like to thank Dr. Kenneth Swanson from Beth Israel Deaconess Medical Center, and Dr. Ilan Volovitz from Tel Aviv Sourasky Medical Center for their helpful and constructive comments.

Ethics Approval

This study was approved by Novocure’s Ethics Board and by the Israel National Ethics Board; approval numbers 160816, 21015, IL-17-3-131 and IL-19-1-38.

Correspondence: Dhana Chinnasamy (dhanac@maxcyte.com)

Background

MaxCyte has developed a rapid and potent cell therapy that utilizes mRNA in transient Flow Electroporation (FEP) to produce gene-modified cell products, termed CARMA™. This proprietary CARMA platform modifies peripheral blood mononuclear cells (PBMC) from apheresis to generate a cryopreserved drug product in a single-day manufacturing process using the cGMP-compliant, closed MaxCyte GT® Transfection System, dramatically reducing the labor, facilities investment, and cost of raw materials typically required for such products. The CARMA one-day manufacturing process using cGMP grade mRNA has the potential to revolutionize cell therapy strategies by significantly reducing the wait time for patients receiving treatment.

Methods

We report here the implementation of the CARMA platform to manufacture MCY-M11, a PBMC cell therapy product expressing an anti-mesothelin chimeric antigen receptor (Meso-CAR) designed to target mesothelin-expressing solid malignancies. MCY-M11 expresses the Meso-CAR in all cells in the PBMC preparation, which are processed and cryopreserved without the need for prior activation or selective expansion. MCY-M11 for clinical application is manufactured under the appropriate cGMP quality systems and controls by MaxCyte at HCATS, a Contract Development Manufacturing Organization (CDMO). Manufacturing release specifications are preliminarily assigned, with multiple For Information Only (FIO) data points being accumulated during clinical production, while sufficient clinical data is being generated to establish meaningful release criteria.

Results

A total of 20 CARMA product development and engineering runs were performed during the technology transfer campaign, with analytical test methods and supply chain established. The viable cell yield from pre-FEP to post-FEP samples averaged around 92%. Meso-CAR expression in T and NK cell subsets in MCY-M11 ranged from 42-83% (average 73%) and 28-75% (average 59%), respectively. The anti-tumor bioactivity and target specificity of MCY-M11 was successfully established in vitro by demonstrating antigen-specific cytotoxicity and inflammatory cytokine release in co-culture assays with various mesothelin-expressing human tumor cell lines. Increased survival and efficacy were also demonstrated in vivo using a human mesothelin expressing ovarian syngeneic mouse tumor model.

Conclusions

The CARMA one-day manufacturing process using cGMP grade mRNA has the potential to revolutionize cell therapy strategies by significantly reducing the wait time for patients receiving treatment. MCY-M11 is currently being tested in a first-in-human clinical trial for advanced epithelial ovarian cancer and peritoneal mesothelioma (ClinicalTrials.gov Identifier: NCT03608618).

Trial Registration

NCT03608618

Correspondence: Hannah Knochelmann (knochelm@musc.edu)

Background

Accessibility of T cell transfer therapies for most patients is hindered by cost and time required for product development. Our lab has shown that shortening ex vivo expansion of Th17 cells licenses a proinflammatory cell product which induces cytokine storm with high levels of systemic IL-6 in tumor-bearing hosts. Despite potential toxicity, briefly expanded Th17 cells eradicate large established tumors in low doses and generate durable memory against tumor rechallenge, suggesting a therapeutic benefit to the inflammatory state. Prior reports show that IL-6 promotes functional CD4+ T cell memory formation. Given that IL-6 is blocked clinically to manage cytokine release syndrome, we addressed the physiologic impact of IL-6 on efficacy and durability of Th17 cell therapy.

Methods

Th17 cells were expanded ex vivo using the TRP-1 transgenic mouse model in which CD4+ T cells express a TCR that recognizes tyrosinase-related protein 1 on melanoma. Naïve CD4+ T cells were polarized to the Th17 phenotype and infused into mice with B16F10 melanoma after a nonmyeloablative total body irradiation (5 Gy) preparative regimen. Serum cytokine levels were obtained by multiplex array and IL-6 signaling was inhibited with antibodies targeting the IL-6R and neutralizing IL-6 cytokine.

Results

Acute IL-6 blockade post Th17 cell transfer did not impact the primary response against melanoma nor the engraftment of Th17 cells. However, blocking IL-6 abrogated long-term responses increasing the frequency of tumor relapse upon secondary challenge and reduced survival. Mechanistically, IL-6 blockade reduced phosphorylation of STAT3 in transferred T cells associating with diminished Bcl-2 expression. The CD4+ compartment was reshaped by IL-6 blockade via promoting a greater frequency of FoxP3+ Treg cells in the peripheral blood, tumor and draining lymph nodes. Given the plasticity of Th17 and Treg cells, we assessed FoxP3 expression within the cell product 10 days post transfer and found that the frequency of FoxP3+ transferred cells was significantly heightened through IL-6 blockade.

Conclusions

IL-6 induced by Th17 cell therapy promotes an inflammatory over regulatory phenotype in vivo permitting durable memory against tumors. The expansion of tumor-specific regulatory cells from the transferred product is enhanced in the absence of IL-6 signaling. This work implies that the universal strategy of IL-6 inhibition for cytokine release syndrome may come at the expense of long-term efficacy for cell therapy approaches.

Ethics Approval

All animal studies were approved by MUSC's IACUC committee, approval number 0488.

Correspondence: Alan Epstein (aepstein@usc.edu)

Background

HLA-G is highly expressed on human placenta during pregnancy and has been found to suppress the NK response to cells that lose their HLA and/or beta2-microglobulin expression. [1-2]. In addition, except for pregnancy, HLA-G is rarely expressed in normal adult tissues. Moreover, roughly 50% of human solid tumors lose their HLA expression to avoid detection by the human immune system. [3] HLA-G is therefore an outstanding target for CAR T-cells since, like the placenta, it is up-regulated in HLA-negative tumors to suppress NK destruction. [4] We have successfully generated anti-HLA-G CAR T-cells to treat solid tumors that express HLA-G.

Methods

An anti-HLA-G CAR construct was generated by fusing anti HLA-G scFv to a second generation CAR containing the CD8α leader sequence, 4-1BB co-stimulation sequence, and CD3ζ signaling domain. The CAR vector was then fused with a lentivirus vector in-frame with the CAR backbone, and was used to transduce primary human CD3 positive T-cells. After transduction, expanded CAR-T cells were characterized for their ability to bind HLA-G antigen and HLA-G positive SKOV-3 cells (human ovarian cancer model) using flow cytometry.

Results

Expanded CAR-T cells were able to bind successfully both the HLA-G antigen and SKOV-3 cells in vitro. Expanded CAR-T cells were then co-cultured with SKOV3-Luc cells and studied for their epitope-driven cytotoxicity. Anti HLA-G CAR T-cells displayed dose-dependent cytotoxicity when co-cultured with tumor cells. We have recently developed an in vivo model of ovarian cancer that can be used for testing the efficacy of our CAR-T cells. In this model, NSG mice are injected with 2 million SKOV3-Luc cells intraperitoneally (ip). Seven-10 days after injection, tumors are visible when observed by bioluminescence imaging, at which time the treatment group will receive an ip injection of anti HLA-G CAR-T cells. Since ovarian cancer rapidly metastasizes to the peritoneum, the aforementioned model should provide relevant clinical data that can be translated to patients, and like hematopoietic cancers, will present antigen quickly after injection of CAR T-cell to keep them stimulated and functional.

Conclusions

We are currently testing the efficacy of anti-HLA-G CAR-T cells in vivo using this ip model, and plan to show that HLA-G as a pan tumor target will provide selective and specific cell based therapy which may in the near future be clinically relevant for chemotherapy resistant ovarian cancer and other tumors.

Acknowledgements

This work is supported by Cell Biotherapy, Inc., Los Angeles, CA.

References

1. Apps R, Gardner L, Moffett A. A critical look at HLA-G. Trends Immunol. 2008; 29:313–321.

2. Jurisicova A, Casper RF, MacLusky NJ, Mills GB, Librach CL. HLA-G expression during preimplantation human embryo development. Proc. Natl. Acad. Sci. U. S. A. 1996; 93:161–5.

3. de Kruijf EM et al. HLA-E and HLA-G expression in classical HLA class I-negative tumors is of prognostic value for clinical outcome of early breast cancer patients. J. Immunol. 2010; 185:7452–9.

4. Lin, A. & Yan, W.-H. Human Leukocyte Antigen-G (HLA-G) Expression in Cancers: Roles in Immune Evasion, Metastasis and Target for Therapy. Mol. Med. 2015; 21:782–791.

Ethics Approval

This study was approved by the IRB of the University of Southern California protocol #HS-16-00029 on 2-29-16.

Correspondence: Robert Hawkins (r.hawkins@immetacyte.com)

Background

The efficacy of TIL therapy is limited in some patients due to the failure of the cells to respond to tumour sufficiently or persist long enough to have a necessary anti-tumour effect. We have addressed this issue by developing Co-stimulatory receptors (CoStARs) that provide enhanced signaling to tumour-specific T-cells upon encountering tumour associated antigens.

Since tumour reactivity is determined by natural TCRs that have undergone thymic selection, this approach does not bear with it the risks of other therapies targeting tumour antigens expressed on the cell surface

Methods

In order to identify optimal signalling domain for Co-StAR molecules, several iterations of our prototype receptor were synthesised. The ability of each to enhance T-cell activation, proliferation, secretion of cytokines and increase resistance to apoptosis were assessed.

To explore if this approach has the potential of wide applicability, we went on to assess targeting of two additional ovarian cancer tumour associated antigens. To achieve this, the antigen binding moiety was exchanged and signaling domain kept constant.

Results

We show that colorectal cancer specific Co-StAR significantly enhances the number of T-cells expressing IL2, TNFα, 41BB, CD107a and bcl-xL by factors ranging from 2-4 fold in model systems. This shows an increase in activation, effector function and resistance to apoptosis. We also identified an optimal signaling domain that caused the greatest magnitude of enhancement for the above factors.

Observations of Co-StAR enhancement were mirrored in model systems for ovarian cancer, targeting two separate ovarian cancer tumour associated antigens.

In addition, stimulation assays showed that Co-StAR with optimal signaling domain increased T-cell proliferation over 3 weeks in comparison to prototype Co-StAR, Co-StAR that binds an irrelevant target, or indeed, mock transduced T-cells.

Conclusions

Our optimal Co-StAR provides a means to effectively deliver “signal 2” to T-cells. Enhancing activation, effector functions and resistance to apoptosis upon contact with target tumour cells.

Since T-cell activation primarily requires “signal 1”, application to of Co-StAR to enhance TIL therapy, which works through natural, thymically selected TCRs, provides a means to increase the activity of tumour-reactive TIL without risking severe off-tumour side effects.

Immetacyte is assessing this approach in our current Phase I/II clinical trial of TIL in ovarian cancer patients (EudraCT–2019-000106-30).

Acknowledgements

This research was supported by IUK projects 133299 and 104468

Ethics Approval

This study was approved by the South Central Research Ethics Committee : 19/SC/0355

Correspondence: James Trager (jtrager@nkartatx.com)

Background

NK cells have been expanded on K562 stimulatory cells expressing membrane-bound (mb) IL-15 and 41BBL for clinical use, and can be genetically modified to express activating chimeric receptors [1,2,3]. Engineered NK cells targeting CD19 or ligands of NKG2D show in vitro and in vivo cytotoxicity against relevant tumor targets that can overcome endogenous resistance to NK cells. NK cells activated in the presence of IL-12, IL-15 and IL-18 develop cytokine induced memory-like phenotype and function; these cells have shown clinical promise [4]. Here we describe NK cell function and phenotype achieved by combining the robust driven by K562-mbIL15-41BBL with the induction of a cytokine-induced memory phenotype achieved after exposure to IL-12 and IL-18.

Methods

Healthy donor PBMC NK were expanded on K562-mbIL15-41BBL stimulatory cells with IL-2 alone or with IL-2 plus IL-12 and IL-18 (12-18). We compared NK cell expansion, cytokine secretion, cytotoxicity against tumor lines at various time points, and persistence in culture over 4 weeks. The expanded NK were transduced with CD19 and NKG2D CAR constructs, and the resulting cells evaluated for CAR expression, cytotoxicity and in vivo efficacy against relevant cell lines.

Results

Addition of 12-18 to the K562-mbIL15-41BBL stimulatory cells improves NK expansion 2-3 fold [If we have a p value, it’s better to give it:<br>‘…significantly improves NK expansion 2-3 fold (p<x) relative to that…’][Will put together later for the poster]relative to that achieved using the stimulatory cell line alone, while NK cell cytotoxicity is unchanged. IFNγ and TNFα production and transduction efficiency are also improved in this setting. Over 3 weeks of culture following brief exposure to 12-18, NK cell phenotype changes, with an increased percentage of CD62L+ and NKG2C+ cells, and increased NKG2C expression per NK cell. While the NKG2D-CAR driven cytotoxic activity is unchanged by 12-18 at 14 days post-exposure, cytotoxic activity increases in these cells by day 21. In addition, this improved cytotoxicity at day 21 is reflected by improved CD19-CAR driven in vivo activity against the CD19+ tumor target [Raji or Nalm6?]Nalm6 [Oops, Nalm6, thx.] with an increased presence of circulating NK cells over 4 weeks in the mice.

Conclusions

The data demonstrates that the addition of IL-12 and IL-18 to K562-mbIL15-41BBL stimulatory cells during NK expansion maintains in vitro cytotoxicity and improves expansion, transduction, persistence, and in vivo efficacy. Further, IL-12 and IL-18 drive development of a memory phenotype and more sustained NK cell function over the course of several weeks. This activation system may allow for development of more robust and potent engineered NK cells for clinical use.

References

1. Lapteva N, Durett AG, Sun J, Rollins LA, Huye LL, Fang J, Dandekar V, Mei Z, Jackson K, Vera J, Ando J, Ngo MC, Coustan-Smith E, Campana D, Szmania S, Garg T, Moreno-Bost A, Vanrhee F, Gee AP, Rooney CM. Large-scale ex vivo expansion and characterization of natural killer cells for clinical applications. Cytotherapy. 2012;14(9):1131-1143

2. Chihaya I, Iwamoto S, Campana D. Genetic modification of primary natural killer cells overcomes inhibitory signals and induces specific killing of leukemic cells. Blood. 2005; 106:376-383.

3. Yang Y, Connolly J, Shimasaki N, Mimura K, Kono K, Campana D. A Chimeric Receptor with NKG2D Specificity Enhances Natural Killer Cell Activation and Killing of Tumor Cells. Cancer Res. 2013;73(6):1777-1786

4. Romee R, Rosario M, Berrien-Elliott MM, Wagner JA, Jewell BA, Schappe T, Leong JW, Abdel-Latif S, Schneider SE, Willey S, Neal CC, Yu L, Oh ST, Lee YS, Mulder A, Claas F, Cooper MA, Fehniger TA. Cytokine-induced memory-like natural killer cells exhibit enhanced responses against myeloid leukemia. Sci Trans Med. 2016;8(357): 357ra123

Ethics Approval

Animal studies were conducted and approved by the Explora IACUC committee.

Correspondence: Caroline Le Poole (caroline.lepoole@northwestern.edu)

Background

Benign tumors can arise from bi-allelic mutations in a single gene. In tuberous sclerosis complex (TSC) and lymphangioleiomyomatosis (LAM), tumors do not acquire additional mutations, and patients are not eligible for therapeutics that rely on neoantigen formation. However, the affected gene is responsible for several predictable phenotypic changes. As mTOR hyperactivity resulting from mutations in TSC1 or TSC2 is associated with overexpression of some melanoma-associated antigens, de novo expression of ganglioside D3 expression may render the resulting, benign tumors eligible for immunotherapy.

Methods

We probed the expression of GD3 in human TSC lesions of the lungs, kidneys, skin and brain by immunostaining and monitored anti-GD3 titers in serum by ELISA. Infiltration by NK cells and NKT was measured to look for natural responses to the cell surface antigen. We next isolated tumors cells from TSC2 heterozygote mice and confirmed loss of heterozygosity by genotyping before challenging groups of 10 SCID/beige mice in 2 repeat experiments, and subjecting them to adoptive transfer by GD3-CART cells and measured tumor sizes over time. Similarly we treated groups of 8 ageing TSC mice >16 months of age by adoptive GD3 CART-cell transfer, and measured surface tumor growth on internal organs.

Results

We found consistent overexpression of GD3 in tissues from TSC patients compared to healthy controls. GD3 overexpression was not accompanied by an influx of NK(T) cells, and anti-GD3 titers were reduced rather than elevated in patients, supporting the concept that slow growing tumors in TSC patients are not immunogenic. However, CART cells responsive to GD3, supplemented by IL-2, mediated prolonged and significant anti-tumor responses in both immunodeficient and immune competent hosts. The majority of TSC2 heterozygote mice treated by CAR T cells displayed no tumors at end point, versus all mice treated with untransduced T cells.

Conclusions

These promising results infer that adoptive transfer of transgenic T cells can offer an effective strategy to not only prevent further tumor growth as rapamycin therapy does, but also to treat and even eliminate arising tumors. This strategy might offer a cure for patients with LAM, a disease that hits women in the prime of their lives.

Acknowledgements

Studies supported by a DoD Tuberous Sclerosis Complex Research Program Clinical Translational Research Award to CLP.

Ethics Approval

All animal experiments were approved by the Animal Care and Use Committee of Northwestern University and followed the institutional guidelines; protocol number IS00008259.

Correspondence: Okjae Lim (blubelle@mogam.re.kr)

Background

Chimeric antigen receptor (CAR) T cell therapy is the revolutionary treatment of choice for hematologic malignancies. Currently approved CAR T therapies require patients’ own immune cells, and this autologous T cell manufacturing process involves certain limitations primarily derived from the nature of individualized therapy. Thus, engineering allogeneic donor cells to evade host immune rejection is required for a broader clinical application of the therapy.

Methods

In this study, we attempted to inhibit expression of both HLA I and II through the CRISPR/Cas9 gene editing system to reduce allo-reactive immune rejection response. First, we screened 60 gRNAs targeting B2M and 60 gRNAs each targeting alpha chains of HLA-II molecules (DP, DQ and DR, respectively) to find gRNA sequences efficiently ablate expression of HLA molecules on T cell surface. Next, we investigated whether the absence of HLA-I/II expression in donor T cells could alleviate immune response from allogeneic responders using in vitro mixed lymphocyte reaction (MLR) assays.

Results

We have identified gRNA sequences highly efficient in targeting B2M and alpha chains of HLA-II molecules without carrying off-target effects. Selected gRNA sequences for HLA-II ablation covered the vast majority of each HLA-II alpha chain allele. HLA-I/II double negative T cells generated by simultaneous quadruple genome editing with the selected gRNAs maintained their phenotypes and cytotoxicity upon TCR stimulations compared to the control cells treated with non-target gRNA. Furthermore, the MLR assays showed that IFN- and TNF-α production in allo-responder T cells was significantly decreased in the absence of donor HLA-I alone and was further diminished in response to HLA-I/II double negative donor T cells compared with the control cells, implicating prolonged survival of the adoptively transferred immune cells.

Conclusions

In conclusion, we have identified novel gRNA sequences ablating expression of HLA molecules on donor T cell surfaces to dramatically reduce donor-derived allo-responses, establishing an essential cornerstone towards the universal T cell therapy.

Ethics Approval

Human PBMCs were obtained from healthy volunteers by leukapheresis from the Samsung Medical Center (SMC) under IRB approval (SMC IRB no.2018-01-089).