You are here

  • Home
  • Archive
  • Volume 10, Issue 10
  • Evaluation of tumor microenvironment and biomarkers of immune checkpoint inhibitor response in metastatic renal cell carcinoma

Article Text

Article menu
Immunotherapy biomarkers
Original research
Evaluation of tumor microenvironment and biomarkers of immune checkpoint inhibitor response in metastatic renal cell carcinoma
  1. Landon C Brown1,2,
  2. Jason Zhu1,2,
  3. Kunal Desai3,
  4. Emily Kinsey2,
  5. Chester Kao2,
  6. Yong Hee Lee4,
  7. Sarabjot Pabla4,
  8. Matthew K Labriola2,
  9. Jennifer Tran5,
  10. Konstantin H Dragnev6,
  11. Laura J Tafe6,
  12. Farshid Dayyani7,
  13. Rajan T Gupta2,
  14. Shannon McCall2,
  15. Daniel J George2,
  16. Sean T Glenn8,9,
  17. Mary K Nesline9,
  18. Saby George10,
  19. Matthew Zibelman11,
  20. Carl Morrison4,
  21. Moshe C Ornstein5 and
  22. Tian Zhang2,12
  1. 1Levine Cancer Institute, Charlotte, North Carolina, USA
  2. 2Department of Medicine, Duke Cancer Institute, Durham, North Carolina, USA
  3. 3Department of Internal Medicine, Cleveland Clinic, Cleveland, Ohio, USA
  4. 4Bioinformatics, OmniSeq, Buffalo, New York, USA
  5. 5Department of Hematology and Medical Oncology, Taussig Cancer Institute, Cleveland, Ohio, USA
  6. 6Department of Medicine, Dartmouth Cancer Center, Lebanon, Pennsylvania, USA
  7. 7Department of Medicine, University of California-Irvine Health, Orange, California, USA
  8. 8Center for Personalized Medicine, Roswell Park Cancer Institute, Buffalo, New York, USA
  9. 9OmniSeq, Inc, Buffalo, New York, USA
  10. 10Department of Medicine, Roswell Park Cancer Institute, Buffalo, New York, USA
  11. 11Department of Hematology/Oncology, Fox Chase Cancer Center, Philadelphia, Pennsylvania, USA
  12. 12Hematology and Oncology, The University of Texas Southwestern Medical Center, Dallas, Texas, USA
  1. Correspondence to Dr Tian Zhang; tian.zhang{at}utsouthwestern.edu

Abstract

Background Immunotherapy combinations including ipilimumab and nivolumab are now the standard of care for untreated metastatic renal cell carcinoma (mRCC). Biomarkers of response are lacking to predict patients who will have a favorable or unfavorable response to immunotherapy. This study aimed to use the OmniSeq transcriptome-based platform to develop biomarkers of response to immunotherapy.

Methods Two cohorts of patients were retrospectively collected. These included an investigational cohort of patients with mRCC treated with immune checkpoint inhibitor therapy from five institutions, and a subsequent validation cohort of patients with mRCC treated with combination ipilimumab and nivolumab from two institutions (Duke Cancer Institute and Cleveland Clinic Taussig Cancer Center). Tissue-based RNA sequencing was performed using the OmniSeq Immune Report Card on banked specimens to identify gene signatures and immune checkpoints associated with differential clinical outcomes. A 5-gene expression panel was developed based on the investigational cohort and was subsequently evaluated in the validation cohort. Clinical outcomes including progression-free survival (PFS) and overall survival (OS) were extracted by retrospective chart review. Objective response rate (ORR) was assessed by Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1.

Results The initial investigation cohort identified 86 patients with mRCC who received nivolumab (80%, 69/86), ipilimumab/nivolumab (14%, 12/86), or pembrolizumab (6%, 5/86). A gene expression score was created using the top five genes found in responders versus non-responders (FOXP3, CCR4, KLRK1, ITK, TIGIT). The ORR in patients with high gene expression (GEhigh) on the 5-gene panel was 29% (14/48), compared with low gene expression (GElow) 3% (1/38, χ2 p=0.001). The validation cohort was comprised of 62 patients who received ipilimumab/nivolumab. There was no difference between GEhigh and GElow in terms of ORR (44% vs 38.5%), PFS (HR 1.5, 95% CI 0.58 to 3.89), or OS (HR 0.96, 95% CI 0.51 to 1.83). Similarly, no differences in ORR, PFS or OS were observed when patients were stratified by tumor mutational burden (high=top 20%), PD-L1 (programmed death-ligand 1) expression by immunohistochemistry or RNA expression, or CTLA-4 (cytotoxic T-lymphocytes-associated protein 4) RNA expression. The International Metastatic RCC Database Consortium (IMDC) risk score was prognostic for OS but not PFS.

Conclusion A 5-gene panel that was associated with improved ORR in a predominantly nivolumab monotherapy population of patients with mRCC was not predictive for radiographic response, PFS, or OS among patients with mRCC treated with ipilimumab and nivolumab.

  • Immunotherapy
  • Kidney Neoplasms
  • Tumor Biomarkers
  • Gene Expression Profiling

Data availability statement

Data are available upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jitc-2022-005249

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • There are no validated genomic biomarkers used in clinical practice that can predict response or resistance to immune checkpoint inhibitors (ICIs) among patients with metastatic renal cell carcinoma (mRCC). RNA expression signatures of tumors and the tumor microenvironment could provide prognostic or predictive information in this population undergoing ICI-based therapy.

    WHAT THIS STUDY ADDS

    • This study sought to develop a gene expression signature associated with ICI response that was detected in a multi-institution cohort of patients primarily treated with single-agent anti-programmed cell death protein-1 checkpoint inhibitor with mRCC. This gene expression signature was not validated in a distinct multi-institution cohort of patients with mRCC treated with the combination of ipilimumab and nivolumab.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • RNA sequencing based predictive gene expression signatures may differ for individual treatment regimens among patients with mRCC and further research is indicated.

    Background

    Immune checkpoint inhibitors (ICIs) have revolutionized the treatment of metastatic renal cell carcinoma (mRCC). In the front-line setting, five immunotherapy-containing combinations are Food and Drug Administration (FDA) approved for mRCC: ipilimumab/nivolumab,1 pembrolizumab/axitinib,2 avelumab/axitinib,3 nivolumab/cabozantinib,4 and pembrolizumab/lenvatinib.5 While these therapies have improved overall survival for many patients with mRCC, biomarkers of treatment response or resistance are urgently needed. Unlike lung cancer where programmed death-ligand 1 (PD-L1) is an FDA-approved biomarker of response to ICI, there are no predictive immunotherapy biomarkers that have been shown to be clinically useful in RCC. While several biomarkers have been explored in mRCC, including tumor mutational burden (TMB), microsatellite instability status, intrinsic molecular subtypes, and composite immune biomarkers, none have been shown to strongly predict for ICI response or resistance.6

    Angiogenesis, T-effector/interferon-γ response, and myeloid inflammatory gene expression signatures have been retrospectively described based on data from IMmotion150 and validated from tumor samples in IMmotion151.1 2 IMmotion150 and IMmotion151 examined the efficacy of atezolizumab alone or in combination with bevacizumab, compared with sunitinib in patients with treatment naïve mRCC. In the exploratory biomarker analysis of IMmotion150, patients were categorized into T-effector gene signature high and low subgroups, as defined by the gene expression of CD8A, EOMES, PRF1, IFNG, and CD274. In patients receiving combination atezolizumab plus bevacizumab, those with T-effector high (TeffHigh) gene signature had improved progression-free survival (PFS) compared with T-effector low (TeffLow) signature (HR 0.50, 95% CI 0.30 to 0.86, p=0.011).1 This benefit was redemonstrated in the subsequent analysis of IMmotion151 (HR 0.76, 95% CI 0.59 to 0.99, p<0.05).3 While this gene signature was helpful for patients treated with combination therapy, PFS and the objective response rates were not significantly different for patients treated with single agent atezolizumab in IMmotion150. Separately, our group has shown that the combination of low cell proliferation as measured by RNA sequencing (RNA-seq) expression of 10 proliferation-related genes and PD-L1 immunohistochemistry (IHC) negativity is associated with lower response rates (6.5%) compared with moderately proliferative, PD-L1 IHC negative tumors (30%).4

    In this study, we aimed to define a new potential tumor microenvironment gene signature which could help identify patients with mRCC who have a higher likelihood of response to single-agent programmed cell death protein-1 (PD-1)/PD-L1 inhibition or combination PD-1/cytotoxic T-lymphocytes-associated protein 4 (CTLA-4) inhibition. In addition, we evaluated established gene signatures and immune checkpoints on the OmniSeq platform within a cohort of patients with mRCC treated with ICIs.

    Methods

    Patients and clinical data

    Two cohorts of patients were collected and analyzed for this study: (1) initial investigational cohort and (2) subsequent validation cohort. For the initial investigational cohort, patients from five institutions were included based on the following criteria: (1) history of advanced/metastatic RCC, (2) pre-ICI treatment archival formalin-fixed paraffin-embedded (FFPE) tissue for transcriptomic (RNA) sequencing, and; (3) availability of demographic, diagnosis, ICI treatment (line of therapy, single agent or combination), and follow-up data. All patients were evaluated for best response based on the Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1 criteria and were designated as having complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). The objective response rate (ORR) was defined as the rate of CR or PR.

    For the separate, subsequent validation cohort, only patients with mRCC treated with combination ipilimumab/nivolumab were included, given evolution in practice patterns away from single-agent PD-1 inhibition to combination therapy over time since the initial investigational cohort analysis. Patients from two institutions (Duke Cancer Institute and Cleveland Clinic Taussig Cancer Center) that met the following criteria were included: (1) history of advanced/metastatic RCC, (2) availability of adequate archival pre-ICI treatment FFPE tissue, (3) treated with combination ipilimumab and nivolumab, and (4) availability of demographic, diagnosis, and follow-up data. All patients were evaluated for best response based on RECIST V.1.1 criteria by each site and were designated as CR, PR, SD, or PD.

    Clinical outcomes

    Time to event clinical outcomes were extracted by retrospective chart review by the clinical authors for the validation cohort. Overall survival (OS) was defined as the time from initiation of ipilimumab/nivolumab until death. PFS was defined as the time from initiation of ipilimumab/nivolumab until the earliest occurrence of death, radiographic progression, or unequivocal clinical progression as indicated in the chart by the treating physician.

    Immunohistochemical studies

    PD-L1 expression was assessed in all cases using the Dako 22C3 pharmDx antibody (Agilent, Santa Clara, California, USA). PD-L1 levels were scored by a board-certified anatomic pathologist following FDA guidelines, with a tumor proportion score in neoplastic cells with a value of≥1% considered positive.5

    Transcriptomic analysis

    RNA was extracted from each sample and processed for targeted RNA-seq, as previously described.6 Gene expression was evaluated by amplicon sequencing of 394 immune transcripts on samples that met validated quality control thresholds. For the investigational cohort, the average gene expression of every immune transcript was compared between patients who were classified as responders (PR and CR) and non-responders (SD and PD) and the five genes which had the greatest differential in gene expression were used to generate a gene signature. The tumor sample was defined as gene expression high (GEhigh) if the average of gene expression in the tumor was higher than the corresponding score for all cases in the cohort (value of 46, the average gene expression of all five genes in the cohort) and otherwise, they were defined as gene expression low (GElow).

    For the validation cohort, RNA-seq was performed along with PD-L1 IHC. The five-gene panel defined by the investigational cohort was applied to the validation cohort. Additionally, previously described gene expression panels tumor immunogenic signature (TIGS)7 and cell proliferation (CP)8 were evaluated in this cohort. Lastly, PD-L1 by RNA expression (high defined as ≥75% rank), CTLA-4 by RNA expression (high defined as ≥75% rank) and TMB (high defined as >10 mut/Mb or >20% rank) were also evaluated.

    Data analysis

    In the investigational cohort, statistical testing of differences in ORR was performed using χ2 test without Yates’ continuity correction for individual variables. For the validation cohort, ORR was assessed by the proportion test to evaluate differences in percentages and logistic regression to evaluate OR. For survival outcomes (OS and PFS) the Kaplan-Meier method was used to calculate median survival and Cox regression were used to calculate HRs. Cox regression models were controlled using the covariates sex, stage at diagnosis, ECOG (Eastern Cooperative Oncology Group) performance status at ipilimumab/nivolumab treatment, and time from diagnosis to ipilimumab/nivolumab of greater or less than 1 year. Analysis was performed in SAS V.9.4.

    Results

    Investigational cohort

    For the initial investigational cohort, a total of 104 tissue samples from five institutions were collected for analysis from 2015 to 2017, as previously reported.4 Among them, 83% (86/104) of samples were sufficient for RNA analysis (demographics shown in table 1). Subtypes of RCC evaluated included clear cell (87%, 75/86), papillary (6%, 5/86), mixed subtypes not otherwise specified (NOS) (6%, 5/86), clear cell with sarcomatoid features (2%, 2/86), and spindle cell (1%, 1/86). Among the patients evaluated, 80% (69/86) were Caucasian, 5% (4/86) were African American, 3% (3/86) were Asian, and race was unknown in 12% (10/86), with a median age of all patients of 60 years (range 37–80). The majority of the patients received nivolumab 80% (69/86), followed by ipilimumab/nivolumab at 14% (12/86), and pembrolizumab at 6% (5/86). By RECIST V.1.1, three patients experienced CR (3%, 3/86), 12 patients had a PR (14%, 12/86), 32 patients had SD (37%, 32/86), and 38 patients had PD (44%, 38/86).

    View this table:
    Table 1

    Demographics for patients in investigational cohort and validation cohort

    Evaluation of tumor PD-L1 (Dako 22C3) status and TMB is summarized in table 2. The majority of patients (70%, 60/86) had PD-L1 status of 0% and almost all patients had a TMB <10 mut/Mb (95%, 82/86). A summary of the gene expression data is provided in online supplemental table 1. For the responders, the 10 most highly expressed genes were TLR3, VEGFA, ITGB1, EGFR, NRP1, ITGA1, CD68, CSF1R, AXL, and LRP1, and the 10 genes with the lowest expression were MLANA, MAGEC2, MAGEA12, MAGEA3, BAGE, IL17A, IL17F, KIR2DL1, MAGEA1, and MAGEA4. For non-responders, the 10 most highly expressed genes were NRP1, TLR3, VEGFA, ID2, VCAM1, ITGA1, EGFR, AXL, TGFB1, ICOSLG and the 10 least expressed were IL17F, SSX2, IL17A, KIR2DL1, MAGEA3, MAGEA12, MAGEA4, MAGEC2, BAGE, and MAGEA1. The top 10 genes which were more highly expressed in responders compared with non-responders were FOXP3, CCR4, KLRK1, ITK, TIGIT, CD27, MYC, TP63, MS4A1, and IFITM1. The top 10 genes which were more highly expressed in non-responders compared with responders were ITGB2, SDHA, FUT4, LST1, HLA-DQB2, KLRF1, CEACAM1, POLR2A, HLA-DRB1, and GADD45GIP1.

    Supplemental material

    View this table:
    Table 2

    Clinical outcomes of patients in validation cohort treated with ipilimumab/nivolumab are evaluated by different gene signatures and by immune checkpoint status

    A gene expression score was created using the top five genes which were most highly expressed in responders compared with non-responders (FOXP3, CCR4, KLRK1, ITK, TIGIT) (figure 1A). Using a cut-off point of 46% (the average gene expression of these five genes), 48 patients were classified as GEhigh, and 38 patients were classified as GElow. Of the patients who were classified as GEhigh, the ORR was 29% (14/48), which was significantly different compared with ORR 3% (1/38) of GElow patients (χ2, p=0.001). For the continuum of 5-gene expression, RECIST responses were plotted by gene expression score (online supplemental figure 1). When using median gene expression for this panel, ORR in the GElow group was 4.7% (2/43) and in the GEhigh group was 30.2% (13/43). In the upper quartile of the combined expression, there were no patients who had progressive disease as their best response. PD-L1 expression also correlated with ORR in this cohort. Of the eight patients who were PD-L1 positive, the ORR was 50%. For patients who were PD-L1 negative, the ORR was 14% (table 1). TMB high status (≥10 mut/Mb), CD8 expression high (defined as >75% normalized RNA expression)and Teff gene expression (from IMMotion150/151 analysis) did not correlate with ORR in this cohort. Figure 1B graphically represents patients stratified by radiographic response, PD-L1, TMB, CD8, the 5-gene expression panel, and the Teff gene expression. Of the 48 patients categorized as GEhigh, 36 (75%) were categorized as TeffHigh and 28 (58%) were CD8 high. Of the 38 patients categorized as GElow, 31 (82%) were categorized as TeffLow and 36 (95%) were categorized as CD8 low.

    Figure 1
    Figure 1

    (A) Objective response rates (ORR) by tumor microenvironment factors evaluated in the investigation cohort of patients with metastatic renal cell carcinoma (mRCC) predominantly treated with single-agent programmed cell death protein-1 inhibitors. (B) Individual patient level data showing heat maps of the 5-gene panel and other tumor microenvironment factors. Every horizontal line across represents one patient. The 5-gene high and low cut-off point set as mean of 5-gene panel expression (46%). Green is higher gene expression and red is lower gene expression. CR, complete response; PR, partial response; PD, progressive disease; PD-L1, programmed death-ligand 1; SD, stable disease; Teff, T-effector; TMB, tumor mutational burden.

    Validation cohort

    For the validation cohort, 69 patients with archival tumor specimens were initially identified and 62 passed quality control for marker testing among patients treated with ipilimumab and nivolumab at Duke Cancer Institute (n=39) and Cleveland Clinic Taussig Cancer Center (n=23). Demographics are shown in table 1. The median age at diagnosis was 59 (range 34–83) and International Metastatic RCC Database Consortium (IMDC) risk groups were favorable, intermediate, or poor risk in 15 (24%), 40 (65%), and 7 (11%), respectively. Clear cell histology was documented in 53 (85%) patients. The majority of patients were treated in the first-line setting (76%), and 26% of patients had bone metastases. A complete course of four doses of ipilimumab was administered in 39 (62.9%) patients. Median follow-up time was 581 days (range 51–1101 days) with 42 (67.7%) patients alive at the time of data cut-off. The median time on ipilimumab/nivolumab, including maintenance nivolumab, was 126 days (range 21–792). Transcriptomic sequencing was performed using the CLIA-certified OmniSeq Immune Report Card in addition to IHC for PD-L1.

    5-gene panel

    We sought to validate the findings found in the investigational cohort that high expression of a 5-gene panel consisting of FOXP3, CCR4, KLRK1, ITK, and TIGIT (GEhigh) was associated with a higher ORR compared with those with GElow (figure 2A, table 2). In the validation cohort, 23 patients were found to be GEhigh and 39 patients were GElow. There was no difference in ORR between these two groups as patients with GEhigh were found to have ORR 44% compared with ORR 38.5% in patients with GElow (OR 1.4, 95% CI 0.41 to 5.04). PFS and OS were also not statistically different between the GEhigh and GElow groups (OS HR 1.50, 95% CI 0.58 to 3.89; PFS HR 0.96, 95% CI 0.51 to 1.83). Receiver operating characteristic curve was estimated from logistic regression to evaluate the performance of 5-gene panel as a classifier of ORR. Based on our result, area under the curve of 0.52 suggests no ability to predict treatment response (ORR) based on 5-gene panel.

    Figure 2
    Figure 2

    Validation cohort objective response rate (ORR), left, and Kaplan-Meier curves, right, for overall survival (OS) and progression free survival (PFS) for (A) 5-gene panel as developed in investigational cohort (GEhigh vs GElow); (B) CP: high, moderate, and poor; (C) Tumor immunogenicity signature (TIGS): strong, moderate, and weak; and, (D) International Metastatic RCC Database Consortium (IMDC) risk groups. Full OR and HR data including CIs can be found in table 2. GEhigh, gene expression high; GElow, gene expression low.

    Predictors of immune response to ipilimumab and nivolumab

    CP is defined based on expression of 10 proliferation associated genes and was divided into one of three levels (high, moderate, and poor) as previously reported.8 The majority of patients were categorized as having poor CP (44, 71.0%) with 15 patients categorized as moderate (24.1%) and 3 patients as high CP (4.8%). ORR across these three groups were 66.7%, 46.7%, and 36.4% for high, moderate, and poor CP, respectively. Kaplan-Meier (KM) analysis for OS and PFS based on CP score are shown in figure 2B. Notably, poor CP was associated with the best OS compared with moderate CP (HR 2.9, 95% CI 1.05 to 7.99). The high CP group only included three patients, and while it demonstrated a worse OS compared with the poor CP group (HR 6.36, 95% CI 0.95 to 42.84) interpretation is limited. Interestingly the ORR and OS outcomes are opposed with the poor CP group having a lower ORR but the best OS; while moderate and high CP had higher ORR, but inferior OS.

    A three-level predefined TIGS score was assessed by RNA-seq and gene expression as previously defined,7 with strong, moderate, or weak immunogenicity identified in 27 (43.5%), 19 (30.6%), and 16 (25.8%) tumors, respectively. ORR across these levels was 40.7%, 31.6%, and 50.0% for strong, moderate, and weak immunogenicity. PFS across these three levels was similar (see figure 2C, table 2). OS was shorter in the moderate immunogenicity group as compared with weak immunogenicity (HR 3.28, 95% CI 0.9 to 1.96) and OS was longer in the strong immunogenicity group as compared with the moderate immunogenicity group (HR 0.30, 95% CI 0.1 to 0.92).

    IMDC risk group was analyzed as a predictor of clinical outcomes. ORR for favorable (n=15), intermediate (n=40), and poor risk (n=7) patients were 53%, 37.5%, and 28.6%, though the difference was not statistically significant. IMDC risk group demonstrated a statistically significant difference in OS by the log-rank test (p=0.02) but not PFS (p=0.6) with median OS of NR, 31.3, and 5.9 months and median PFS of 9.6, 9.7, and 3.5 months, respectively, for favorable, intermediate, and poor risk groups. KM curves for OS and PFS by IMDC risk group are shown in figure 2D

    In the validation cohort, other predictors of immunotherapy response were analyzed including high PD-L1 expression. High PD-L1 expression was detected by IHC (≥1%) and RNA expression (defined as ≥75% rank) in 13 (20.0%) and 21 (33.9%) patients, respectively. High PD-L1 by either IHC or RNA expression was not associated with a significantly different ORR, OS, or PFS (figure 3A–B, table 2). Notably high CTLA-4 expression by RNA-seq (defined as ≥75% rank) was detected in 13 (20.0%) patients and was associated with a numerically, but not statistically significantly higher ORR (64.5% vs 34.7%, OR 1.92, 95% CI 0.43 to 8.62, figure 3C, table 2). No statistical difference was detected in PFS or OS between CTLA-4 high versus low group, though PFS was numerically better in the CTLA-4 high group (PFS HR 0.66, 95% CI 0.31 to 1.43).

    Figure 3
    Figure 3

    Validation cohort objective response rate (ORR), left, and Kaplan-Meier curves, right, for overall survival (OS) and progression free survival (PFS) as stratified by checkpoint expression: (A) programmed death-ligand 1 (PD-L1) high versus low by immunohistochemistry (IHC), (B) PD-L1 high versus low by RNA expression (high defined as ≥75% rank)), and (C) cytotoxic T-lymphocytes-associated protein 4 (CTLA-4) by RNA expression (high defined as ≥75% rank). Full OR and HR data including CIs can be found in table 2. TMB, tumor mutational burden.

    In the TMB analysis, two patients were found to have a high TMB defined as ≥10 mut/Mb, and neither patient experienced an objective response (online supplemental figure 2). When high TMB was defined as the top 20% (cut-off 7.0 mut/Mb, figure 3D), those with high TMB had a lower ORR compared with those with low TMB (30.8% vs 42.9%) and similar OS (HR 1.01, 95% CI 0.28 to 3.62) and PFS (HR 1.39, 95% CI 0.65 to 3.0).

    Discussion

    At this time, there are no FDA-approved biomarkers for mRCC to select patients who may benefit from front-line combination immunotherapy (ipilimumab/nivolumab) versus ICI/tyrosine kinase inhibitor (TKI) or single-agent TKI therapy. Our study identified a gene expression panel consisting of five genes (FOXP3, CCR4, KLRK1, ITK, and TIGIT) which, when high, was found to be associated with a higher ORR among a cohort of patients primarily treated with single-agent nivolumab in the second-line or later mRCC setting. However, when evaluated in a subsequent validation cohort of patients treated predominantly in the front-line setting with combination ipilimumab/nivolumab, there was no difference in ORR, PFS, or OS observed between patients with high gene expression compared with those with low gene expression.

    The 5-gene panel described above can be seen as a surrogate marker of infiltrating regulatory T-cells (Treg) and natural killer (NK) cells. These genes include FOXP3 and CCR4 which are important for regulatory T-cells, and KLRK1, ITK, and TIGIT, which are highly expressed for NK cells. FOXP3, also known as scurfin, is a Treg protein, involved in the establishment and maintenance of the Treg.9 CCR4, also known as CD194, is a protein receptor predominantly expressed by T helper 2 (Th2) cells, whose primary task is to help orchestrate cell migration and homing of leukocytes during development, angiogenesis, and cancer metastases.10 Killer cell lectin like receptor K1 (KLRK1), also known as NKG2D, belongs to a family of C-type lectin-like receptors which are expressed by all NK cells.11 This activating receptor and its ligands are critical for the T-cell mediated immune response to tumors and are frequently detected on the surface of tumor cell lines and tumor tissues.11 ITK is a critical member of the TEC-kinase family, which controls proximal T-cell receptor signaling and supports resting lymphocyte kinase (RLX).12 ITK inhibition can diminish Th2 immunity and potentiate Th1 based immune responses.12 The T-cell immunoreceptor with immunoglobulin and ITIM domains (TIGIT), also known as WUCAM, is a member of the PVR/nectin.13

    Our investigational cohort consisted of patients primarily treated with single-agent PD-1 inhibitors. PD-1 inhibitors primarily act by blocking the inhibitory signaling between PD-1 and PD-L1 that occur between tumor cells and cytotoxic T-cells in the tumor microenvironment, but do not increase the immune infiltration. Thus, we hypothesize that patients who have gene expression profiles associated with higher rates of immune infiltration, possibly indicated by a high score on our 5-gene panel, may be more primed to respond to PD-1 inhibitors. However, the mechanism of CTLA-4 inhibitors occurs at the inhibitory interface between antigen presenting cells and T-cells, thus augmenting the immune cell infiltrate. CTLA-4 inhibition with ipilimumab may result in a more robust tumor inflammatory response, which could negate the predictive value of the above panel due to CTLA-4 modulated inflammatory infiltration. Additionally, our findings in the investigational cohort may not translate to the cohort of patients in the validation cohort who were primarily treated in the first-line setting as opposed to the later line setting.

    A key secondary endpoint of this study was also to assess the utility of checkpoint expression and its correlation with clinical outcomes in ICI treated mRCC. High PD-L1 when measured by IHC or RNA expression was consistently not associated with a difference in ORR, OS, or PFS in both our investigational cohort and validation cohort. Notably, CTLA-4 expression in the ipilimumab/nivolumab treated validation cohort was associated with higher ORR, though statistical significance was not reached, and no differences in PFS and OS were observed. No data exists on CTLA-4 expression as a biomarker in mRCC but interestingly, prior studies in stage I–III non-small cell lung cancer have correlated high tumor CTLA-4 expression in tumor involved lymph nodes with worse clinical outcomes14 and decreased CTLA-4 expression by RNA-seq has been associated with higher response rates among patients with melanoma.15 Additional secondary goals were to evaluate previously defined gene panels developed by OmniSeq, including the TIGS score and CP score in our validation cohort. Though numbers are low and results are hypothesis generating, patients in the validation cohort treated with ipilimumab/nivolumab with poor CP experienced longer median OS, but poor CP was not predictive for median PFS when compared with those with moderate or high cell proliferation. We hypothesize that CP may be prognostic as a marker of more indolent disease biology rather than predictive of response to immunotherapy. This hypothesis is also supported by the poor CP group having lower ORR than either the moderate or high CP groups despite longer OS. Results evaluating the TIGS in the validation cohort demonstrate that the moderate immunogenicity group experienced worse OS and lower ORR, but similar PFS when compared with the weak and strong immunogenicity groups. These results warrant future study in additional validation cohorts.

    Our results highlight that biomarkers may be specific to a treatment (such as single-agent PD-1 inhibition) and are not generalizable to other combinations (such as combination PD-1/CTLA-4 inhibition). In particular, transcriptome signatures in other phase 3 trials in mRCC have also demonstrated this principle, as the effector T-cell enriched and myeloid-enriched gene signatures found from the IMmotion150 trial did not predict for PFS to avelumab/axitinib-treated patients in the JAVELIN Renal 101 trial.16 Most applicable to our study, RNA-seq data from a subset of patients treated with ipilimumab and nivolumab in Checkmate 214 was recently published did not show discrimination in outcomes when stratified by previously published gene expression signatures such as tumor inflammation, angiogenesis, T-effector, and the JAVELIN Renal 101 signature.17 Multiple other clinical trials have been analyzed for predictive gene signatures relating to the tumor microenvironment18 and angiogenic factors: IMmotion151,19 JAVELIN Renal 100, JAVELIN Renal 101,10 and Keynote 427.20 Other predictive biomarkers that are promising include the baseline neutrophil to eosinophil ratio21 and HLA typing.22 Despite these results, predictive biomarkers in mRCC continue to have limited role in clinical practice and need further prospective testing for widespread clinical utility.

    There are several limitations to our data. All data was assessed retrospectively, and clinical variables were abstracted based on chart review. Our targeted RNA-seq approach with 394 immune transcripts likely limits our scope. There was significant patient heterogeneity within our cohorts regarding line of therapy, and baseline variables that were not adjusted for. A limitation of our investigational cohort was the lack of OS and PFS, which were included in our validation cohort. Both the investigational and validation cohorts were composed of a relatively small number of patients, which also limits the interpretation of results. Additionally, our two cohorts were treated exclusively with ICIs, and these results likely do not apply to patients treated with ICI/TKI combinations. Lastly, our two cohorts were indeed different patient groups given the investigational cohort was primarily treated in the second and later line whereas the validation cohort was primarily treated in the front-line setting, though this is a reflection of the real-world evolution in practice patterns based on available clinical evidence. A strength of our analysis is the presence of two independent cohorts to evaluate our gene expression panel. An additional strength is our inclusion of both an investigational cohort in which we identified a biomarker of interest in the 5-gene panel as well as a validation cohort to prospectively assess this biomarker. Finally, further prognostic markers such as cell proliferation and immunogenicity subgroups were further evaluated across the assessed cohorts. Together these data show use of molecular testing in mRCC and its application to patients treated with immunotherapies.

    Conclusion

    We identified a 5-gene expression panel that was associated with treatment response among patients with mRCC treated with single-agent PD-1 inhibitors. However, these results were not verified when this 5-gene panel was applied to a distinct cohort of patients treated with ipilimumab and nivolumab. Other results demonstrated that moderate tumor inflammation and high/moderate cell proliferation were negatively prognostic for OS. Predictive markers remain elusive for mRCC, and gene expression profiles that predict for response may vary by immunotherapy treatment regimen.

    Data availability statement

    Data are available upon reasonable request.

    Ethics statements

    Patient consent for publication

    References

    1. ↵
      1. McDermott DF,
      2. Huseni MA,
      3. Atkins MB, et al
      . Clinical activity and molecular correlates of response to atezolizumab alone or in combination with bevacizumab versus sunitinib in renal cell carcinoma. Nat Med 2018;24:749–57.doi:10.1038/s41591-018-0053-3pmid:http://www.ncbi.nlm.nih.gov/pubmed/29867230
      OpenUrlCrossRefPubMed
    2. ↵
      1. Motzer RJ,
      2. Banchereau R,
      3. Hamidi H, et al
      . Molecular subsets in renal cancer determine outcome to checkpoint and angiogenesis blockade. Cancer Cell 2020;38:803–17.doi:10.1016/j.ccell.2020.10.011pmid:http://www.ncbi.nlm.nih.gov/pubmed/33157048
      OpenUrlCrossRefPubMed
    3. ↵
      1. Rini B,
      2. Huseni M,
      3. Atkins M
      . Molecular correlates differentiate response to atezolizumab (atezo) plus bevacizumab (bev) vs sunitinib (sun): results from a phase III study (IMmotion151) in untreated metastatic renal cell carcinoma (mRCC). Annals Of Oncology, Oxford Univ Press Great Clarendon St, Oxford Ox2 6dp, England 2018:724–5.
    4. ↵
      1. Zhang T,
      2. Pabla S,
      3. Lenzo FL, et al
      . Proliferative potential and response to nivolumab in clear cell renal cell carcinoma patients. Oncoimmunology 2020;9:1773200.doi:10.1080/2162402X.2020.1773200pmid:http://www.ncbi.nlm.nih.gov/pubmed/32923131
      OpenUrlPubMed
    5. ↵
      1. Büttner R,
      2. Gosney JR,
      3. Skov BG, et al
      . Programmed death-ligand 1 immunohistochemistry testing: a review of analytical assays and clinical implementation in non-small-cell lung cancer. J Clin Oncol 2017;35:3867–76.doi:10.1200/JCO.2017.74.7642pmid:http://www.ncbi.nlm.nih.gov/pubmed/29053400
      OpenUrlCrossRefPubMed
    6. ↵
      1. Conroy JM,
      2. Pabla S,
      3. Glenn ST, et al
      . Analytical validation of a next-generation sequencing assay to monitor immune responses in solid tumors. J Mol Diagn 2018;20:95–109.doi:10.1016/j.jmoldx.2017.10.001pmid:http://www.ncbi.nlm.nih.gov/pubmed/29061374
      OpenUrlPubMed
    7. ↵
      1. Pabla S,
      2. Seager RJ,
      3. Van Roey E, et al
      . Integration of tumor inflammation, cell proliferation, and traditional biomarkers improves prediction of immunotherapy resistance and response. Biomark Res 2021;9:56.doi:10.1186/s40364-021-00308-6pmid:http://www.ncbi.nlm.nih.gov/pubmed/34233760
      OpenUrlPubMed
    8. ↵
      1. Pabla S,
      2. Conroy JM,
      3. Nesline MK, et al
      . Proliferative potential and resistance to immune checkpoint blockade in lung cancer patients. J Immunother Cancer 2019;7:27.doi:10.1186/s40425-019-0506-3pmid:http://www.ncbi.nlm.nih.gov/pubmed/30709424
      OpenUrlAbstract/FREE Full Text
    9. ↵
      1. Lu L,
      2. Barbi J,
      3. Pan F
      . The regulation of immune tolerance by FOXP3. Nat Rev Immunol 2017;17:703–17.doi:10.1038/nri.2017.75pmid:http://www.ncbi.nlm.nih.gov/pubmed/28757603
      OpenUrlPubMed
    10. ↵
      1. Choueiri TK,
      2. Albiges L,
      3. Haanen JB
      . Biomarker analyses from javelin renal 101: Avelumab+ axitinib (A+ Ax) versus sunitinib (S) in advanced renal cell carcinoma (aRCC). American Society of Clinical Oncology 2019.
    11. ↵
      1. Nausch N,
      2. Cerwenka A
      . NKG2D ligands in tumor immunity. Oncogene 2008;27:5944–58.doi:10.1038/onc.2008.272pmid:http://www.ncbi.nlm.nih.gov/pubmed/18836475
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Dubovsky JA,
      2. Beckwith KA,
      3. Natarajan G, et al
      . Ibrutinib is an irreversible molecular inhibitor of ITK driving a Th1-selective pressure in T lymphocytes. Blood 2013;122:2539–49.doi:10.1182/blood-2013-06-507947pmid:http://www.ncbi.nlm.nih.gov/pubmed/23886836
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Manieri NA,
      2. Chiang EY,
      3. Grogan JL
      . TIGIT: a key inhibitor of the cancer immunity cycle. Trends Immunol 2017;38:20–8.doi:10.1016/j.it.2016.10.002pmid:http://www.ncbi.nlm.nih.gov/pubmed/27793572
      OpenUrlCrossRefPubMed
    14. ↵
      1. Paulsen E-E,
      2. Kilvaer TK,
      3. Rakaee M, et al
      . CTLA-4 expression in the non-small cell lung cancer patient tumor microenvironment: diverging prognostic impact in primary tumors and lymph node metastases. Cancer Immunol Immunother 2017;66:1449–61.doi:10.1007/s00262-017-2039-2pmid:http://www.ncbi.nlm.nih.gov/pubmed/28707078
      OpenUrlCrossRefPubMed
    15. ↵
      1. Nesline M,
      2. Puzanov I,
      3. Ernstoff MS, et al
      . Effect of CTLA-4 overexpression on response to ipilimumab in melanoma. Journal of Clinical Oncology 2018;36:190.doi:10.1200/JCO.2018.36.5_suppl.190
    16. ↵
      1. Motzer RJ,
      2. Robbins PB,
      3. Powles T, et al
      . Avelumab plus axitinib versus sunitinib in advanced renal cell carcinoma: biomarker analysis of the phase 3 JAVELIN renal 101 trial. Nat Med 2020;26:1733–41.doi:10.1038/s41591-020-1044-8pmid:http://www.ncbi.nlm.nih.gov/pubmed/32895571
      OpenUrlPubMed
    17. ↵
      1. Motzer RJ,
      2. Choueiri TK,
      3. McDermott DF, et al
      . Biomarker analysis from CheckMate 214: nivolumab plus ipilimumab versus sunitinib in renal cell carcinoma. J Immunother Cancer 2022;10:e004316.doi:10.1136/jitc-2021-004316pmid:http://www.ncbi.nlm.nih.gov/pubmed/35304405
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Paik S,
      2. Shak S,
      3. Tang G, et al
      . A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004;351:2817–26.doi:10.1056/NEJMoa041588pmid:http://www.ncbi.nlm.nih.gov/pubmed/15591335
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Rini B,
      2. Huseni M,
      3. Atkins M
      . LBA31 Molecular correlates differentiate response to atezolizumab (atezo)+ bevacizumab (bev) vs sunitinib (sun): Results from a phase III study (IMmotion151) in untreated metastatic renal cell carcinoma (mRCC). Annals of Oncology 2018;037:mdy424.
      OpenUrl
    20. ↵
      1. Martini J-F,
      2. Plimack ER,
      3. Choueiri TK, et al
      . Angiogenic and immune-related biomarkers and outcomes following Axitinib/Pembrolizumab treatment in patients with advanced renal cell carcinoma. Clin Cancer Res 2020;26:5598–608.doi:10.1158/1078-0432.CCR-20-1408pmid:http://www.ncbi.nlm.nih.gov/pubmed/32816890
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Tucker MD,
      2. Brown LC,
      3. Chen Y-W, et al
      . Association of baseline neutrophil-to-eosinophil ratio with response to nivolumab plus ipilimumab in patients with metastatic renal cell carcinoma. Biomark Res 2021;9:80.doi:10.1186/s40364-021-00334-4pmid:http://www.ncbi.nlm.nih.gov/pubmed/34732251
      OpenUrlPubMed
    22. ↵
      1. Naranbhai V,
      2. Viard M,
      3. Dean M, et al
      . HLA-A*03 and response to immune checkpoint blockade in cancer: an epidemiological biomarker study. Lancet Oncol 2022;23:172–84.doi:10.1016/S1470-2045(21)00582-9pmid:http://www.ncbi.nlm.nih.gov/pubmed/34895481
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Twitter @LandonBrownMD, @MattLabs831, @mattzibelman

    • Contributors LCB: Writing—original draft. JZ: Investigation, writing—original draft. KD: Investigation. EK: Investigation. CK: Investigation. YHL: Formal analysis. SP: Investigation. MKL: Investigation. JT: Investigation. KHD: Investigation. LJT: Investigation. FD: Investigation. RTG: Formal analysis. SM: Formal analysis. DJG: Supervision, writing—review and editing. STG: Investigation. MKN: Investigation. SG: Investigation. MZ: Investigation. CM: Investigation, supervision, writing—review and editing. MCO: Investigation, supervision, writing—review and editing. TZ: Guarantor; Conceptualization, supervision, writing—review and editing.

    • Funding This research received external funding from Kidney Cancer Association Young Investigator Award. The Duke University BioRepository & Precision Pathology Center receives Cancer Center Support Grant funding from P30 CA014236.

    • Competing interests LCB declares the following relationships: Consulting with Seattle Genetics. JZ declares the following relationships: Consulting with Tempus. KD declares the following relationship: Consulting with Tyra BioSciences. EK declares that she has no conflicts of interest that might be relevant to the contents of this manuscript. CK declares that he has no conflicts of interest that might be relevant to the contents of this manuscript. YHL is/was an employee at OmniSeq. SP is/was an employee at OmniSeq. MKL declares that he has no conflicts of interest that might be relevant to the contents of this manuscript. JT declares that she has no conflicts of interest that might be relevant to the contents of this manuscript. KHD declares that he has no conflicts of interest that might be relevant to the contents of this manuscript. LJT declares that she has no conflicts of interest that might be relevant to the contents of this manuscript. FD declares that he has received research grants (to the institution) from AstraZeneca, Bristol Myers Squibb, Merck, Genentech/Roche, Taiho, Exelixis, Trishula, Leap Therapeutics, has received a speaker honorarium from Amgen, Eisai, Ipsen, Exelixis, Sirtex, Deciphera, Ipsen, Natera, and has received a consultancy honorarium from Natera, QED, Eisai, Exelixis, Genentech/Roche. RTG declares the following relationships: Bayer Pharma AG—consultant & speakers bureau, Invivo—consultant, Bard—consultant, DxTx Medical—consultant, Quibim—consultant, Bracco Diagnostics—consultant. SM declares the following relationships: AstraZeneca—consultant; Takeda Pharmaceuticals—speaker, honorarium. DJG declares the following relationships: Acerta Pharmaceuticals—research; American Association for Cancer Research—senior editor; Astellas—consultant, research, advisory board; AstraZeneca—consultant, advisory board; Axess Oncology—independent contractor; Bayer H/C Pharmaceuticals—research, consultant, speaker, honorarium, travel accommodations; SC, BMS—consultant, research, steering committee; Calithera—research; Capio Biosciences—scientific advisory board; EMD Serono—honorarium; Exelixis—research, consultant, speaker, honorarium, travel accommodations; Flatiron—consultant; Ipsen—honorarium; Janssen Pharmaceuticals—research, consultant, Independent Data Monitoring Committee; Leidos Biomedical Research—consultant; Merck Sharp & Dohme—consultant; Michael J Hennessey Associates—honorarium, consultant; Millennium Medical Publishing, Clinical Advances in Hematology & Oncology—co-editor-in-chief; Modra Pharmaceuticals B.V.—advisory board; Myovant Sciences—consultant; Nektar Therapeutics—steering committee; Novartis—research; Physician Education Resource LLC—consultant; Pfizer—research, consultant, steering committee, honorarium; Sanofi—research, consultant, speaker, honorarium, travel accommodations; UroGPO—honorarium; UroToday—honorarium, travel accommodations; Vizuri Health Sciences, LLC—consultant; NCI—steering committee. YHL is/was an employee at OmniSeq. STG is/was an employee at OmniSeq. MKN is/was an employee at OmniSeq. SG declares the following relationships: Aveo—advisory board, personal, advisor/consultant; Bayer—advisory board, personal, advisor/consultant; BMS—advisory board, personal, advisor/consultant; Corvus—advisory board, personal, advisor/consultant; Eisai—advisory board, personal, advisor/consultant; EMD Serono—advisory board, personal, advisor/consultant; Exelixis—advisory board, personal, advisor/consultant; Merck—advisory board, personal, advisor/consultant; Pfizer—advisory board, personal, advisor/consultant; QED Therapeutics—advisory board, personal, advisor/consultant; Sanofi/Genzyme—advisory board, personal, advisor/consultant; Seattle Genetics—advisory board, personal, advisor/consultant; Agensys—local PI, institutional, no financial interest; Aravive—local PI, institutional, no financial interest; Aveo—local PI, institutional, no financial interest; Bayer—local PI, institutional, no financial interest; BMS—local PI, institutional, no financial interest; Calithera—local PI, institutional, no financial interest; Corvus—local PI, institutional, no financial interest; Eisai—local PI, institutional, no financial interest; Exelixis—local PI, institutional, no financial interest; Gilead—local PI, institutional, no financial interest; Merck—local PI, institutional, no financial interest; Novartis—local PI, institutional, no financial interest; Pfizer—local PI, institutional, no financial interest; Seattle Genetics—local PI, institutional, no financial interest; Surface Oncology—local PI, institutional, no financial interest. MZ declares the following relationships: Consulting or advisory roles for: Exelixis, Pfizer, Aveo, EMD Serono, Janssen, Blue Earth; Institutional research funding from Bristol Myers Squibb, Exelixis. CM is/was an employee at OmniSeq. MCO declares the following relationships: Consulting or advisory roles for Eisai, Exelixis, Pfizer, Aveo, Merck, and Bristol Myers Squibb; speakers bureaus for Bristol Myers Squibb and Exelixis; institutional research funding from Bristol Myers Squibb, Pfizer, Merck, AstraZeneca, Astellas; reimbursement for travel and accommodations expenses from Bristol Myers Squibb, Pfizer, and Exelixis. TZ declares the following relationships: Research funding (to Duke University) from Pfizer, Janssen, Acerta, AbbVie, Novartis, Merrimack, OmniSeq, PGDx, Merck, Mirati, Astellas, and Genentech; Speaking with Genomic Health and Sanofi Aventis; Consulting/Advisory board with AstraZeneca, Bayer, Pfizer, Janssen, Merck, Amgen, MJH Associates, BMS, Dendreon, Calithera, QED therapeutics, Aveo Therapeutics, Eli Lilly, SeaGen, Sanofi-Aventis, Pharmacyclics, Aptitude Health, PeerView, Clinical Care Options. Stock ownership/employment (spouse) from Capio Biosciences, Archimmune Therapeutics and Nanorobotics.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.