You are here

  • Home
  • Archive
  • Volume 11, Issue 6
  • PEOPLE (NTC03447678), a phase II trial to test pembrolizumab as first-line treatment in patients with advanced NSCLC with PD-L1

Article Text

Article menu

This article has a correction. Please see:

Immunotherapy biomarkers
Original research
PEOPLE (NTC03447678), a phase II trial to test pembrolizumab as first-line treatment in patients with advanced NSCLC with PD-L1 <50%: a multiomics analysis
  1. Giuseppe Lo Russo1,
  2. Arsela Prelaj1,2,
  3. James Dolezal3,
  4. Teresa Beninato1,
  5. Luca Agnelli4,5,
  6. Tiziana Triulzi6,
  7. Alessandra Fabbri4,
  8. Daniele Lorenzini4,
  9. Roberto Ferrara1,7,
  10. Marta Brambilla1,
  11. Mario Occhipinti1,
  12. Laura Mazzeo1,
  13. Leonardo Provenzano1,
  14. Andrea Spagnoletti1,
  15. Giuseppe Viscardi8,
  16. Francesco Sgambelluri9,
  17. Silvia Brich4,
  18. Vanja Miskovic2,
  19. Alessandra Laura Giulia Pedrocchi2,
  20. Francesco Trovo'2,
  21. Sara Manglaviti1,
  22. Claudia Giani1,
  23. Paolo Ambrosini1,
  24. Rita Leporati1,
  25. Andrea Franza1,
  26. John McCulloch10,
  27. Tommaso Torelli4,
  28. Andrea Anichini9,
  29. Roberta Mortarini9,
  30. Giorgio Trinchieri11,
  31. Giancarlo Pruneri5,
  32. Valter Torri12,
  33. Filippo De Braud1,5,
  34. Claudia Proto1,
  35. Monica Ganzinelli1 and
  36. Marina Chiara Garassino1,3
  1. 1Medical Oncology, Fondazione IRCCS Istituto Nazionale dei Tumori, Milano, Lombardia, Italy
  2. 2Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, Milano, Lombardia, Italy
  3. 3Thoracic Oncology Program, Section of Hematology/Oncology, University of Chicago Department of Medicine, Chicago, Illinois, USA
  4. 4Department of Advanced Diagnostics, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy
  5. 5Department of Oncology and Hemato-Oncology, University of Milan, Milano, Lombardia, Italy
  6. 6Molecular Targeting Unit, Department of Research, Fondazione IRCCS Istituto Nazionale dei Tumori, Milano, Lombardia, Italy
  7. 7Medical Oncology, Università Vita Salute San Raffaele, Milano, Lombardia, Italy
  8. 8Medical Oncology, Department of Precision Medicine, Università degli Studi della Campania Luigi Vanvitelli, Caserta, Campania, Italy
  9. 9Department of Research, Human Tumors Immunobiology Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milano, Lombardia, Italy
  10. 10Genetics and Microbiome Core, Laboratory of Integrative Cancer Immunology, Center for Cancer Research, NCI, Bethesda, Maryland, USA
  11. 11Laboratory of Integrative Cancer Immunology, Center for Cancer Research, NIH, Bethesda, Maryland, USA
  12. 12Oncology department, Mario Negri Institute for Pharmacological Research (IRCCS), Milano, Lombardia, Italy
  1. Correspondence to Dr Arsela Prelaj; arsela.prelaj{at}istitutotumori.mi.it

Abstract

Background Chemoimmunotherapy represents the standard of care for patients with advanced non-small cell lung cancer (NSCLC) and programmed death-ligand 1 (PD-L1) <50%. Although single-agent pembrolizumab has also demonstrated some activity in this setting, no reliable biomarkers yet exist for selecting patients likely to respond to single-agent immunotherapy. The main purpose of the study was to identify potential new biomarkers associated with progression-free-survival (PFS) within a multiomics analysis.

Methods PEOPLE (NTC03447678) was a prospective phase II trial evaluating first-line pembrolizumab in patients with advanced EGFR and ALK wild type treatment-naïve NSCLC with PD-L1 <50%. Circulating immune profiling was performed by determination of absolute cell counts with multiparametric flow cytometry on freshly isolated whole blood samples at baseline and at first radiological evaluation. Gene expression profiling was performed using nCounter PanCancer IO 360 Panel (NanoString) on baseline tissue. Gut bacterial taxonomic abundance was obtained by shotgun metagenomic sequencing of stool samples at baseline. Omics data were analyzed with sequential univariate Cox proportional hazards regression predicting PFS, with Benjamini-Hochberg multiple comparisons correction. Biological features significant with univariate analysis were analyzed with multivariate least absolute shrinkage and selection operator (LASSO).

Results From May 2018 to October 2020, 65 patients were enrolled. Median follow-up and PFS were 26.4 and 2.9 months, respectively. LASSO integration analysis, with an optimal lambda of 0.28, showed that peripheral blood natural killer cells/CD56dimCD16+ (HR 0.56, 0.41–0.76, p=0.006) abundance at baseline and non-classical CD14dimCD16+monocytes (HR 0.52, 0.36–0.75, p=0.004), eosinophils (CD15+CD16−) (HR 0.62, 0.44–0.89, p=0.03) and lymphocytes (HR 0.32, 0.19–0.56, p=0.001) after first radiologic evaluation correlated with favorable PFS as well as high baseline expression levels of CD244 (HR 0.74, 0.62–0.87, p=0.05) protein tyrosine phosphatase receptor type C (HR 0.55, 0.38–0.81, p=0.098) and killer cell lectin like receptor B1 (HR 0.76, 0.66–0.89, p=0.05). Interferon-responsive factor 9 and cartilage oligomeric matrix protein genes correlated with unfavorable PFS (HR 3.03, 1.52–6.02, p 0.08 and HR 1.22, 1.08–1.37, p=0.06, corrected). No microbiome features were selected.

Conclusions This multiomics approach was able to identify immune cell subsets and expression levels of genes associated to PFS in patients with PD-L1 <50% NSCLC treated with first-line pembrolizumab. These preliminary data will be confirmed in the larger multicentric international I3LUNG trial (NCT05537922).

Trial registration number 2017-002841-31.

  • non-small cell lung cancer
  • immunotherapy
  • biomarkers, tumor

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-2023-006833

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

    • In advanced non-small cell lung cancer (NSCLC), the lack of driver molecular alterations and programmed death-ligand 1 (PD-L1) Tumor Proportion Score levels (≥ or <50%) represents the standard tool to candidate patients to first-line single-agent immunotherapy or to immunotherapy and chemotherapy combinations. A subgroup of patients with advanced NSCLC and PD-L1 <50% might benefit from first-line pembrolizumab single agent, though biomarkers able to identify these patients are still lacking.

    WHAT THIS STUDY ADDS

    • Through a multiomics approach, this study identified immune circulating cell subsets (mainly natural killer cells at baseline) and tumor tissue expression levels of genes (interferon-responsive factor 9, cartilage oligomeric matrix protein, killer cell lectin like receptor B1, protein tyrosine phosphatase receptor type C and CD244) positively associated to progression-free-survival in PD-L1 <50% advanced NSCLC treated with first-line pembrolizumab.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • This study provides a promising strategy for identification of biomarkers useful for selecting patients with PD-L1 <50% advanced NSCLC for de-escalation of chemoimmunotherapy to single-agent immunotherapy. These preliminary data need confirmation in larger studies.

    Background

    Immunotherapy (IT) is a mainstay of treatment in patients with metastatic and locally advanced non-small cell lung cancer (aNSCLC) without driver alterations.1 Despite many limits, programmed death-ligand 1 (PD-L1) Tumor Proportion Score (TPS) still remains the only used and approved predictor of activity.2 Tumor mutational burden (TMB) was also prospectively assessed in several randomized controlled trials3 4; however, its value is still debatable in effectively predicting overall survival (OS). In clinical practice, the lack of driver molecular alterations and a PD-L1 TPS ≥50% represents the standard condition to candidate patients to first-line single-agent IT or to IT-chemotherapy combinations (although it depends on regulatory rules applied among different countries).5 In patients with PD-L1 TPS <50% IT monotherapy has shown minor activity, while IT-chemotherapy combinations based on platinum regimens proved to significantly prolong patients OS compared with chemotherapy alone, both for squamous and non-squamous histology.6 7 However, in some countries, IT single agent is also approved in patients with aNSCLC with PD-L1 TPS 1–49%, based on KEYNOTE-042 trial results. In this study pembrolizumab improved OS in patients with aNSCLC with PD-L1 TPS ≥1% compared with standard chemotherapy.8 Although in this trial the positive results were mostly driven by patients with PD-L1 TPS ≥50%, in the real-life clinical practice IT-chemotherapy combinations may lead to higher incidence of adverse events compared with IT single agent; therefore, pembrolizumab remains a feasible first-line choice also in patients with low PD-L1. Enhancing patient selection remains a crucial point. Unfortunately, all these treatment strategies were developed in all-comer populations (any PD-L1 expression), where either PD-L1 expression or TMB were considered only in subgroup analyses reducing the statistical significance.

    Immune biomarkers are fundamentally different from oncogene biomarkers since they are continuous rather than categorical (binary) parameters, spatially variable, and reliant on multiple complex interactions rather than a single, dominant determinant. Beside PD-L1 and TMB, many biomarkers have been explored to perform more personalized diagnosis and consequently more individualized treatment decisions for patients with aNSCLC,9 10 but results did not reach enough power to be practise changing. Different challenges remain and one of the most interesting to yet solve is surely the de-escalation of chemotherapy in patients with aNSCLC with PD-L1 TPS <50% subgroup who potentially are responsive to first-line IT single agent.

    The main purpose of PEOPLE trial was to assess new biomarkers on tumor tissue, immune circulating and microbiome associated with treatment efficacy in patients with aNSCLC with PD-L1 TPS <50% treated with first-line pembrolizumab single agent. Here we report the multiomics biomarker analysis.

    Methods

    Study design and patients

    PEOPLE (NTC03447678) is a prospective, monocentric, open-label phase II trial conducted at Fondazione IRCCS Istituto Nazionale dei Tumori in Milan (INT). The main inclusion criteria were diagnosis of stage IIIB/IV EGFR and ALK wild type aNSCLC, who had not received prior systemic chemotherapy for advanced disease, and with PD-L1 TPS expression <50%. The availability of tissue (archival or newly obtained), blood and stool samples was mandatory for the inclusion in the trial. Full inclusion and exclusion criteria have already been described elsewhere.11 The local Ethical Committee approved the trial protocol and all related amendments. The trial was conducted in accordance with the International Conference on Harmonization Guidelines on Good Clinical Practice and the Declaration of Helsinki. All patients provided their written informed consent before enrollment.

    Treatment and procedures

    Patients were enrolled from May 31, 2018, to October 07, 2020. Treatment with pembrolizumab at a flat dose of 200 mg was administered intravenously every 3 weeks (21±3 days) and continued until 2 years or 35 cycles (whichever occurred later) or until documented disease progression, unacceptable toxicity or withdrawal of the patient’s consent. Tumor response was assessed every 9 weeks (63±3 days) according to the Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1. PD-L1 expression was centrally assessed during the screening phase by immunohistochemistry with the anti-PD-L1 antibody (DAKO 22C3) on archival or newly obtained tumor-biopsy specimens (within 45 days prior treatment initiation) and only patients with a PD-L1 TPS expression <50% were eligible. Baseline tumor tissue was used for RNA immune gene-expression profiling. Blood samples were collected at baseline, every two cycles, and at the end of treatment for circulating immune profiling analysis. Stool samples were collected at baseline, after the first cycle and at the end of treatment.

    Objectives

    The primary objective of PEOPLE trial was to identify immune biomarkers associated with progression-free-survival (PFS). Secondary objectives were: to detect differences in immune biomarkers distribution between pre-study and post-study treatment; to estimate the activity of pembrolizumab first-line monotherapy in terms of objective response rate (ORR), duration of response (DoR) and disease control rate (DCR); to assess the efficacy of the treatment in terms of OS.

    Statistical analyses

    The median follow-up was estimated using the inverse Kaplan-Meier (KM) method. The survival curves were estimated by the KM method.

    To identify candidate biomarkers/features in the data sets obtained with each technique, a first step of sequential univariate Cox proportional hazard regression model was used for predicting PFS, taking into account the low number of patients. This procedure was applied singularly to all four set of features: circulating immune profiling at baseline(bCIP), circulating immune profiling post-immunotherapy (pCIP), gene expression profiling (GEP) and microbiome. To decrease the false discovery rate, the Benjamini-Hochberg procedure was used to correct multiple comparisons. To select the statistically significant features within each omics analysis, an alpha=0.05 was used for circulating immune profiling, while an alpha=0.1 for the microbiome and RNA expression. Correlation matrix was used to visualize the correlation between the features selected in univariate analysis and with the main relevant clinical ones. To find the optimal lambda for PFS prediction for the least absolute shrinkage and selection operator (LASSO) a 10-fold cross-validation on the previously selected features from all four modalities was performed. After this step a further feature selection using LASSO was made among the most important ones using the optimal lambda. All analyses were done with the R programming language and ‘glmnet’ package.

    Omics methodology

    Circulating immune profiling

    The methodology for circulating immune profiling has been already extensively described in a previous publication.11 Absolute cell counts of 36 immune subsets in peripheral blood were obtained using flow cytometry. Blood was collected at baseline and at day 63±3 days (first radiological evaluation) and stained with fluorescently labeled antibodies in Trucount Absolute Counting Tubes (Becton Dickinson). Samples were analyzed using a 10-color cytometer and data were analyzed with FlowJo software. Beads were gated out and peripheral blood mononuclear cells (PBMCs) were gated using side scatter (SSC) versus CD45 dot plots. Absolute cell count was calculated using the formula A=X/Y×N/V, where X=number of positive cells events, Y=number of bead events, N=number of beads in the test tube and V=test vol. Validation of flow cytometry cell counts data was obtained through correlation analysis with counts of lymphocytes, granulocytes and monocytes generated on the same samples by an automated hematology cell counter. Thirty-six distinct immune cell subsets were identified and counted using the gating strategy described in Lo Russo et al.11

    GEP

    Total RNA was extracted from formalin-fixed, paraffin-embedded (FFPE) tumor tissue at baseline, using the RNeasy FFPE Kit (Qiagen, Maryland, USA), according to manufacturer’s instructions. RNA 150 ng was used for gene expression analysis performed by means of the nCounter PanCancer Immune IO 360 Panel (NanoString). The 770-plex assay panel contains: 109 genes to cell surface markers capable of quantitating 24 different tumor infiltrating immune cell types and populations, 30 genes for commonly studied antigens, over 500 genes for measuring immune response with a special emphasis on checkpoint regulation/signaling and 40 PanCancer reference genes. The NanoStringNorm package for R software was used to assess quality and process the data; the geometric mean of the counts relative to each sample, the mean plus two SD and the total sum of counts options were used to correct the data for technical, background and batch effect issues, respectively. The expression counts of housekeeping genes and quantile normalization were used to account for inter sample variations with the panel.

    Microbiome

    Metagenomic analyses were performed as described in a previous publication.12 Briefly, DNA was extracted from frozen stool samples collected from patients before the treatment using the MOBIO PowerSoil DNA Isolation Kit (MOBIO Laboratories) and epMotion 5075 liquid handling robot (Eppendorf). DNA libraries were prepared using the Nextera DNA Flex Library Prep Kit, quantified using Qubit, and sequenced on the NovaSeq System (Illumina) using the 2×150 base pair (bp) paired-end protocol. Trimmomatic V.0.36 was used for quality trimming and adapter clipping of raw reads. The reads were then aligned against the human genome with Bowtie 2 V.2.3.2 and unaligned (non-host) reads were then assembled using MEGAHIT V.1.2.9. Assembly contigs smaller than 500 bp were discarded. Taxonomic classification of contigs was obtained by k-mer analysis using Kraken 2, with a custom 96 Gb Kraken 2 database built using draft and complete genomes of all bacteria, archaea, fungi, viruses, and protozoa available in the National Center for Biotechnology Information (NCBI) GenBank in January 2022, in addition to the human and mouse genomes. Taxonomy was expressed as the last known taxon (LKT), being the taxonomically lowest unambiguous classification determined for each sequence, using Kraken’s confidence scoring threshold of 5e-06 (using the confidence parameter). Reads were aligned back to contigs, and the relative abundance for each LKT within each sample was obtained by dividing the number of bp covering all contigs pertaining to that LKT by the total number of non-host base pairs sequenced for that sample. Relative abundances were expressed in parts per million (PPM).

    Results

    Patients’ characteristics

    This report is based on the final data analysis (database locked on July 22, 2021, data previously published.11 Briefly, from May 31, 2018, to October 07, 2020, 87 treatment naïve patients with aNSCLC diagnosis, were screened in INT. Among 87 patients screened, 17 were declared screening failure and 65 were enrolled and treated. The median age was 70 years (47–87), with 44 (68%) men and 21 (32%) women. Eastern Cooperative Oncology Group Performance Status (ECOG PS) was 0 in 23 (35.4%), 1 in 30 (46.2%), and 2 in 12 (18.5%) patients. Most patients (50, 77%) had an adenocarcinoma histology while 10 (15%) patients had a squamous histology and 5 (8%) other types of histology. PD-L1 TPS expression was 1–49% in 47 (73%) patients and 0 in 18 (27%) patients. Twenty-eight (43.1%) patients received antibiotics during study treatment.

    Biomarkers investigated beyond PD-L1

    Three different omics evaluations have been performed, generating four different set of features. As reported in a previous publication, scanty samples or inadequate sample quality were the main reasons for not evaluating all the patients within each data set.11

    Circulating immune profiling at baseline (bCIP): 36 immune subsets evaluated in 57 patients.

    Circulating immune profiling post pembrolizumab (pCIP): 36 immune subsets evaluated in 46 patients.

    Gene expression profiling (GEP) generated 778 features in 48 patients.

    Microbiome: 492 gut bacterial taxa were obtained in 54 patients.

    Survival analysis

    At the time of analysis, with a median follow-up of 26.4 months (mo), 51 (78%) patients experienced progression and 46 (70%) died. The median PFS was 2.9 mo (95% CI 1.8 to 5.8) and the median OS was 12.1 mo (95% CI 9.0 to 20.2). Seven patients experienced an early death (before the first radiological evaluation), while 58 patients underwent at least one radiological evaluation. The ORR was 24.1%, DCR was 53.4% and the median DoR was 14.5 mo (95% CI 8.4 to 24.9).

    Univariate feature analysis

    Table 1 reports the features in each of the four data sets significantly correlated with PFS in the univariate analysis.

    View this table:
    Table 1

    Multiomic features significantly associated with progression-free-survival in univariate analysis. P values are adjusted per Benjamini-Hochberg procedure

    In the bCIP data sets, 3 out of 36 immune subsets were significantly and positively correlated with PFS: CD56+, natural killer (NK) cells and NK cells/CD56dimCD16+, with the following HR of: 0.5 (0.32–0.78, p=0.0293), 0.54 (0.39–0.76, p=0.006) and 0.56 (0.41–0.76, p=0.006), respectively. In the pCIP, 17 immune subsets significantly associated with PFS. Among them, the most significant with a p value ranging from 0.0008 to 0.0061 included granulo-/HLA-DRdimCD14– (HR 0.28, 0.15–0.50, p=0.0008), CD3+ (H3 0.35, 0.20–0.60, p=0.0018), lymphocytes (CD3+ or CD19+ or CD56+) (HR 0.32, 0.19–0.56, p=0.001), non-classical CD14dimCD16+ (HR 0.52, 0.36–0.75, p=0.0039), non-classical CD14dimCD16+/HLA-DR++ (HR 0.56, 0.41–0.77, p=0.0036), NK cells (HR 0.55, 0.38–0.79, p=0.0061) and NK cells/CD56dimCD16+ (HR 0.61, 0.46–0.83, p=0.0061). Of note, granulo-/HLA-DRdimCD14− is a subset including T lymphocytes (predominantly) and also NK, from which activated T cells (HLA DR+T cells) and B lymphocytes (constitutively HLA-DR+) were excluded. In the GEP data sets, expression of 21 genes significantly associated with PFS, including CD48 (HR 0.59, 0.47–0.75, p=0.0146), CD84 (HR 0.59, 0.45–0.77, p=0.0486), CD244 (HR 0.74, 0.62–0.87, p value=0.0533), CD3D (HR 0.61, 0.47–0.80, p value=0.0533) and killer cell lectin like receptor B1 (KLRB1) (HR 0.76, 0.66–0.89, p value=0.0533). In the microbiome data set, the Eisenbergiella massiliensis species is the only taxon significantly associated with PFS (HR 13.9, 3.51–55.3, p=0.0891) in microbiome data sets.

    Correlations among features

    A correlation matrix was created to illustrate correlation among features significantly associated with PFS in univariate analysis (figure 1). Clinical variables such as age, smoking habits, and PD-L1 generally lack a strong correlation with features identified in the univariate analysis, although sex was slightly correlated with the expression levels of most genes. High performance status by ECOG was negatively correlated with high expression of CD48, protein tyrosine phosphatase receptor type C (PTPRC), CD3D, CD45RA, CD69, as well as elevated levels of NK and CD56+ cells, both at baseline and post-pembrolizumab. Tumor histology was correlated with the expression of a half of genes significant by univariate analysis, as well as with NK cells/CD56dimCD16+ after two cycles of therapy.

    Figure 1
    Figure 1

    The correlation matrix visualizes the correlation between the features selected at univariate analysis and the most relevant clinical features. Green boxes indicate positive correlation, golden boxes indicate negative correlation, X indicates absence of correlation. bCIP, circulating immune profiling at baseline; COMP, cartilage oligomeric matrix protein; GEP, gene expression profiling; IRF9, interferon-responsive factor 9; KLRB1, killer cell lectin like receptor B1; LKT, last known taxon; NK, natural killer; pCIP, circulating immune profiling post-immunotherapy; PSMB5, proteasome 20S subunit beta 5; PTPRC, protein tyrosine phosphatase receptor type C; SCC, squamous cell carcinoma.

    The expression levels of all the genes, except interferon-responsive factor 9 (IRF9), were highly correlated with one another; most genes were positively correlated, apart from CD276, proteasome 20S subunit beta 5 and cartilage oligomeric matrix protein (COMP) which were negatively correlated with the remaining genes. Additionally, the expression levels of CD48, PTPRC, CD69, CD3D, KLRB1, CCL19, CD45RA, CD2, CD244 were positively correlated with lymphocytes subset and negatively correlated with granulocytes and neutrophils post-pembrolizumab subset.

    E. massiliensis, was strongly inversely correlated with post-immune subsets, mainly with lymphocytes and NK cells.

    Multiomic feature selection with LASSO

    All features significant in univariate analysis were pooled for a multiomic analysis, with feature selection performed with LASSO (figures 2 and 3). Optimal λ was determined through 10-fold cross-validation, and this value (0.28) was then used for the final analysis. Among circulating biomarkers, the multiomic LASSO analysis selected NK cells/CD56dimCD16+ at baseline and granulo-/HLA-DRdimCD14−, non-classical CD14dim CD16+ and eosinophils (CD 15+CD16−) after two pembrolizumab cycles, all of which were associated with a favorable PFS. Five genes were selected, with higher expression levels of CD244, PTPRC and KLRB1 genes associating with a favorable PFS and IRF9 and COMP gene expression associating with poorer PFS. No microbiome features were selected by LASSO in the final integration (figure 4).

    Figure 2
    Figure 2

    Feature selection using LASSO algorithm: LASSO coefficient profile for all features selected using univariant analysis. A vertical line was drawn at the optimal lambda value selected using the 10-fold cross-validation process. COMP, cartilage oligomeric matrix protein; IRF9, interferon-responsive factor 9; KLRB1, killer cell lectin like receptor B1; LASSO, least absolute shrinkage and selection operator; NK, natural killer; PSMB5, proteasome 20S subunit beta 5; PTPRC, protein tyrosine phosphatase receptor type C.

    Figure 3
    Figure 3

    Feature selection using least absolute shrinkage and selection operator algorithm: Zoom-in on optimal lambda and selected features. COMP, cartilage oligomeric matrix protein; KLRB1, killer cell lectin like receptor B1.

    Figure 4
    Figure 4

    The figure represents all features selected by LASSO for each set of data. The green or gold color code of the features indicates a positive or negative correlation with PFS. bCIP, circulating immune profiling at baseline; CIP, circulating immune profiling; COMP, cartilage oligomeric matrix protein; GEP, gene expression profiling; IRF9, interferon-responsive factor 9; KLRB1, killer cell lectin like receptor B1; LASSO, least absolute shrinkage and selection operator; NK, natural killer; pCIP, circulating immune profiling post-immunotherapy; PFS, progression-free-survival; PTPRC, protein tyrosine phosphatase receptor type C.

    Discussion

    To date, clinicians have a variety of first-line treatment options for patients with aNSCLC with PD-L1 TPS <50%. Nevertheless, nowadays treatment decisions cannot accurately be taken, due to the complexity of the immune system, its interaction with the tumor microenvironment (TME) and the lack of highly accurate biomarkers. This unmet need, led to the design of the PEOPLE study in 2017 and has now led to the carrying out of the present multiomic analysis.

    The characterization of 36 immune subpopulations was carried out both at baseline and during pembrolizumab treatment. The final analysis selected only one subset at baseline (NK cells/CD56dimCD16+) and three subsets post pembrolizumab treatment (granulo-/HLA-DRdimCD14−, non-classical CD14dim CD16+ and eosinophils CD15+CD16−). The levels of all these subsets were positively associated with PFS. These data confirmed our previously published results, validating and strengthening our previously reported evidence on the association between high frequency of T cells (programmed cell death protein-1 (PD-1)+, CD4+) and NK cells, with improved PFS in patients with aNSCLC treated with first-line pembrolizumab single agent.11 The potential predictive value of elevated NK cells for PFS challenges the widely held belief that immune checkpoint blockade depends primarily on T-cell activity. Indeed, several pieces of evidence from clinical and preclinical studies highlight the crucial role of innate immune cells in the context of PD-1 targeted therapy. According to Barry et al, NK cell frequency at the tumor site is correlated with stimulatory dendritic cells (DCs) and with higher OS and responsiveness to anti-PD-1 therapy in patients with melanoma.13 Interestingly, Dong et al found that patients with PD-L1 negative tumors can still respond to IT with anti-PD-L1, but that this response is conditioned by PD-L1 positive NK cells.14 Hsu et al showed in preclinical models that inhibition of PD-1 and PD-L1 promotes an NK cell response, which is necessary for IT efficacy.15 Granulo-/HLA-DRdimCD14− is a subset of immune cells including T lymphocytes (predominantly) and also NK, from which activated T cells (HLA DR+T cells) and B lymphocytes (constitutively HLA-DR+) were excluded. Its positive correlation with PFS confirmed our previous published results highlighting the importance of T and NK cells beyond B cells.11 Krieg et al, reported that CD14+CD16−HLA-DRhi monocytes were able to strongly predict PFS at baseline IT.16 Non-classical CD14dimCD16+, differently expressed in literature as ‘nonclassical patrolling monocytes’, seem to play an important role in patrolling the vascular endothelial during chronic inflammatory injury with high ability to remove damaged cells17 confirming their positive role after pembrolizumab administration. Finally, as reported in our final LASSO model, it is largely described in literature that an increased level of eosinophils cells after IT stimulation was associated with better disease control and survival outcomes.18–20

    Looking at GEP selected genes: COMP and IRF9 in our final LASSO model negatively correlate with PFS. In particular, IRF9 encodes for a transcription factor in the type I interferon (IFN) receptor signaling pathway.21 Interestingly, preclinical studies have demonstrated that IRF-9 dependent signaling may be involved in induction of PD-L1 expression in mouse and human lung cancer cells exposed to IFN-β22 and of PD-1 expression in murine T cells exposed to IFN-α.23 Recent studies have also recognized the involvement of IRF9 in vascular smooth muscle cells proliferation and migration through growth factors stimulation.24 25 Therefore, we can hypothesize that the unfavorable prognostic role reported in our analysis could be driven by these two mechanisms leading to immune-escape and angiogenesis.22–25 COMP encodes for a regulator of the extracellular matrix,26 highly represented in fibrotic scars, and involved in vascular wall remodeling. Similarly, to our findings, previous studies reported a correlation with poor survival outcomes which seem to be mediated by influence on cancer cell’s migration, invasion and metabolism.27–30 On the other hand, KLRB1, PTPRC (also known as CD45) and CD244 correlated positively with PFS in the final model. These genes encode for transmembrane receptors expressed on hematological cells. KLRB1 is expressed on NK cells, CD8+, CD4+, and other T-cell subgroups.31 Various studies support its favorable prognostic role in multiple tumors.31–34 Cheng et al reported on 33 tumor types that KLRB1 was positively correlated to tumor infiltrating lymphocytes, while negatively correlated to cancer-promoting myeloid cells.31 Moreover, KLRB1 expression reflected higher levels of immune checkpoint genes’ expression, suggesting a potential capacity to predict IT response, which would be consistent with our results.31 By the way, according to a meta-analysis performed by Braud and collaborators, KLRB1 expression correlates with better OS in non-small cell lung cancer (NSCLC) independently from the level of tumor infiltration by CD8+T and NK cells.34

    Elevated CD45 levels could be an index of higher tumor inflammation though they are not informative regarding the type and function of tumor infiltrate since CD45 is a pan-leukocyte protein expressed on almost all immune cells.35 CD244 has an immunomodulatory function and is expressed by hematopoietic cells, including NK cells, a subset of CD8+ αβ T cells, DCs and myeloid derived suppressor cells.36–39 Few conflicting data are available regarding the impact of CD244 expression on NSCLC outcomes. A negative prognostic role has been hypothesized by Vaes and coauthors, who reported a negative correlation of CD244 expression levels with PFS among 26 patients with stage I NSCLC treated with stereotactic body radiation therapy.40 However, based on this study results, high CD244 expression levels seem to confer a positive role on PFS. This hypothesis is also supported by an exploratory analysis conducted on 14 patients with NSCLC treated with atezolizumab within POPLAR (study comparing second line atezolizumab vs docetaxel).41

    The positive correlation between most of the immune-related genes assessed by NanoString and the peripheral immune profile suggests that a better clinical outcome to pembrolizumab depends on both peripheral and tumor-related factors. Although not surprizing, this also means that patients with the best clinical outcome need to have at the same time a systemic immune profile geared towards a functional T/NK compartment and, likely, also an inflamed/leukocyte-infiltrated TME (PTPRC, CD244, KLRB1) that is well poised to react/be functionally rescued by pembrolizumab. In other words, clinical response to pembrolizumab is a systemic process and only the integrated analysis on tumor tissue and periphery can provide such evidence.

    Finally, among microbiome features, no one was selected in the final LASSO model. However, it is now well accepted that the gut microbiota is causally implicated in the efficacy and toxicity of IT in several solid tumors including aNSCLC.42 43 In fact, fecal microbiota transplantation or dietary interventions have been clinically investigated to improve the success rate of IT.44 However, little concordance exists among species identified in different studies. In the present study only E. massiliensis was found associated with PFS. Whether it depends on negative or low PD-L1 expression or on the high microbiota heterogeneity which cannot be mitigated due to the small number of patients, remains to be determined. In addition, the use of antibiotics in a subgroup of patients, reportedly associated with dysbiosis and progression on IT treatment43 could have contributed to limit the power of the analysis. E. massiliensis is a gram-negative bacterium isolated from human stool that belongs to the Firmicutes phylum and Lachnospiraceae family.45 46 Even if not selected as significant variable in multivariate analysis, its amount in the gut at baseline was highly negatively correlated with post-treatment immune features that remained in the final model—mainly lymphocytes, reportedly determinant of IT efficacy47—supporting its inverse association with IT activity and its putative role as modulator of the immune system. This correlation may be clinically relevant, because this bacterium was characterized in stool collected at baseline, so it anticipates the immune features that would be evaluated later. The impact of this bacterium in immune modulation and IT efficacy has not been investigated yet, but it was recently found to be increased in the gut of mice by ketogenic diet and to be highly correlated with the main-ketogenic metabolite, three hydroxybutyric acid, in the blood of mice and humans.48 Although this metabolite, similar to butyrate, was found associated with antitumor immune modulation and IT efficacy in preclinical models,48 49 it was also described to have anti-inflammatory and tolerogenic roles mainly in inflammatory disease,50 supporting a context and dose dependent effects. Comprehensive metagenomic analysis and preclinical studies would shed light on the role of gut microbiota and of this bacterium in the response to anti-PD-1 therapy.

    The present study has some limitations, mainly the non-randomized design and the small sample size.

    Conversely, the strength of the study lies in its innovative multiomics approach, which allows the concomitant evaluation of multiple biologic sets to select the most relevant biomarkers.

    The strength of the present study mainly relates to its prospective phase II biomarker-driven design including innovative multiomics approach. Compared with other first-line IT single-agent studies, the present study population was frailer with a higher proportion of patients with ECOG PS 2. Also, the study included patients with negative PD-L1 expression levels, a population that has not been included in the previous trials with single-agent IT.

    Over the last years, omics technologies have experienced a sharp acceleration in their development and their costs have also been reduced.51 But there is still one thing missing: their integration based on the consideration of the individual as a whole. Since the human mind is unable to integrate them and—far less—to correlate them with response to IT, our main ambition is to create and prospectively validate an artificial intelligence (AI) tool that allows scientists to predict more precisely the efficacy of treatment while allowing us to integrate scientific knowledge. In this perspective, the INT-led I3LUNG Horizon Europe project (https://cordis.europa.eu/project/id/101057695) aims to expand this cohort with other multiomics data collected across Europe and beyond, with the objective to achieve peak performance in personalized medicine through an AI/machine learning method modeled on multimodal patients’ data.

    In conclusion, to our knowledge this is the first prospective drug-interventional trial investigating the association of multiomics biomarkers and efficacy of IT within PD-L1 TPS <50% aNSCLC.

    The results suggested that mainly NK cells subset evaluated at baseline IT may be useful in identifying those patients who may benefit from pembrolizumab, avoiding the additional toxicity expected to be induced by chemotherapy. Other relevant immune subsets, evaluated in the post pembrolizumab setting, included lymphocytes, monocytes and eosinophils subsets. In addition, the evaluation of some selected genes on tumor tissue such as of IRF9, COMP, KLRB1, PTPRC (CD45) and CD244 could be implemented within clinical practice to individualize selection among single agent and IT-chemotherapy combination in this subset of patients. Comprehensive metagenomic studies are required to clarify the influence of E. massiliensis and other metagenomic biomarkers on IT outcomes.

    Supplemental material

    Data availability statement

    Data are available upon reasonable request.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and was approved by Fondazione IRCCS Istituto Nazionale dei Tumori Ethics Committee, ID of the study: INT 178/17. Participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    We would like to acknowledge the pharmacy, Dr Vito Ladisa and his staff, for supporting us in managing the drug. We also would like to acknowledge our nurses for their contribution in assisting the patients and our Trial Center for their support to this study and Giovanni Scoazec for his contribution in English editing. We also would like to acknowledge donors for the Excalibur project in memory of Giorgiana Marchesi Bianchini.

    References

    1. ↵
      1. Proto C,
      2. Ferrara R,
      3. Signorelli D, et al
      . Choosing wisely first line immunotherapy in non-small cell lung cancer (NSCLC): what to add and what to leave out. Cancer Treat Rev 2019;75:39–51. doi:10.1016/j.ctrv.2019.03.004
      OpenUrlCrossRefPubMed
    2. ↵
      1. Paver EC,
      2. Cooper WA,
      3. Colebatch AJ, et al
      . Programmed death ligand-1 (PD-L1) as a predictive marker for Immunotherapy in solid tumours: a guide to immunohistochemistry implementation and interpretation. Pathology 2021;53:141–56. doi:10.1016/j.pathol.2020.10.007
      OpenUrlPubMed
    3. ↵
      1. Gandara DR,
      2. Paul SM,
      3. Kowanetz M, et al
      . Blood-based tumor mutational burden as a predictor of clinical benefit in non-small-cell lung cancer patients treated with atezolizumab. Nat Med 2018;24:1441–8. doi:10.1038/s41591-018-0134-3
      OpenUrlCrossRefPubMed
    4. ↵
      1. Kim ES,
      2. Velcheti V,
      3. Mekhail T, et al
      . Blood-based tumor mutational burden as a biomarker for atezolizumab in non-small cell lung cancer: the phase 2 B-F1Rst trial. Nat Med 2022;28:939–45. doi:10.1038/s41591-022-01754-x
      OpenUrlPubMed
    5. ↵
      1. Reck M,
      2. Rodríguez-Abreu D,
      3. Robinson AG, et al
      . Pembrolizumab versus chemotherapy for PD-L1–positive non–small-cell lung cancer. N Engl J Med 2016;375:1823–33. doi:10.1056/NEJMoa1606774
      OpenUrlCrossRefPubMed
    6. ↵
      1. Paz-Ares L,
      2. Luft A,
      3. Vicente D, et al
      . Pembrolizumab plus chemotherapy for squamous non–small-cell lung cancer. N Engl J Med 2018;379:2040–51. doi:10.1056/NEJMoa1810865
      OpenUrlCrossRefPubMed
    7. ↵
      1. Gandhi L,
      2. Rodríguez-Abreu D,
      3. Gadgeel S, et al
      . Pembrolizumab plus chemotherapy in metastatic non–small-cell lung cancer. N Engl J Med 2018;378:2078–92. doi:10.1056/NEJMoa1801005
      OpenUrlCrossRefPubMed
    8. ↵
      1. Mok TSK,
      2. Wu Y-L,
      3. Kudaba I, et al
      . Pembrolizumab versus chemotherapy for previously untreated, PD-L1-expressing, locally advanced or metastatic non-small-cell lung cancer (KEYNOTE-042): a randomised, open-label, controlled, phase 3 trial. The Lancet 2019;393:1819–30. doi:10.1016/S0140-6736(18)32409-7
      OpenUrl
    9. ↵
      1. An HJ,
      2. Chon HJ,
      3. Kim C
      . Peripheral blood-based biomarkers for immune checkpoint inhibitors. IJMS 2021;22:9414. doi:10.3390/ijms22179414
      OpenUrl
    10. ↵
      1. Brueckl WM,
      2. Ficker JH,
      3. Zeitler G
      . Clinically relevant Prognostic and predictive markers for immune-checkpoint-inhibitor (ICI) therapy in non-small cell lung cancer (NSCLC). BMC Cancer 2020;20:1185. doi:10.1186/s12885-020-07690-8
    11. ↵
      1. Lo Russo G,
      2. Sgambelluri F,
      3. Prelaj A, et al
      . PEOPLE (Nct03447678), a first-line phase II pembrolizumab trial, in negative and low PD-L1 advanced NSCLC: clinical outcomes and association with circulating immune biomarkers. ESMO Open 2022;7:100645. doi:10.1016/j.esmoop.2022.100645
      OpenUrl
    12. ↵
      1. McCulloch JA,
      2. Davar D,
      3. Rodrigues RR, et al
      . Intestinal microbiota signatures of clinical response and immune-related adverse events in melanoma patients treated with anti-PD-1. Nat Med 2022;28:545–56. doi:10.1038/s41591-022-01698-2
      OpenUrlPubMed
    13. ↵
      1. Barry KC,
      2. Hsu J,
      3. Broz ML, et al
      . A natural killer–dendritic cell axis defines checkpoint therapy–responsive tumor microenvironments. Nat Med 2018;24:1178–91. doi:10.1038/s41591-018-0085-8
      OpenUrlCrossRefPubMed
    14. ↵
      1. Dong W,
      2. Wu X,
      3. Ma S, et al
      . The mechanism of anti–PD-L1 antibody efficacy against PD-L1–negative tumors identifies NK cells expressing PD-L1 as a cytolytic effector. Cancer Discov 2019;9:1422–37. doi:10.1158/2159-8290.CD-18-1259
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Hsu J,
      2. Hodgins JJ,
      3. Marathe M, et al
      . Contribution of NK cells to Immunotherapy mediated by PD-1/PD-L1 blockade. J Clin Invest 2018;128:4654–68.:99317. doi:10.1172/JCI99317
      OpenUrlCrossRefPubMed
    16. ↵
      1. Krieg C,
      2. Nowicka M,
      3. Guglietta S, et al
      . High-dimensional single-cell analysis predicts response to anti-PD-1 immunotherapy. Nat Med 2018;24:1773–5. doi:10.1038/s41591-018-0094-7
      OpenUrlPubMed
    17. ↵
      1. Kapellos TS,
      2. Bonaguro L,
      3. Gemünd I, et al
      . Human monocyte Subsets and phenotypes in major chronic inflammatory diseases. Front Immunol 2019;10:2035. doi:10.3389/fimmu.2019.02035
    18. ↵
      1. Delyon J,
      2. Mateus C,
      3. Lefeuvre D, et al
      . Experience in daily practice with Ipilimumab for the treatment of patients with metastatic melanoma: an early increase in lymphocyte and eosinophil counts is associated with improved survival. Ann Oncol 2013;24:1697–703. doi:10.1093/annonc/mdt027
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Alves A,
      2. Dias M,
      3. Campainha S, et al
      . Peripheral blood eosinophilia may be a Prognostic biomarker in non-small cell lung cancer patients treated with Immunotherapy. J Thorac Dis 2021;13:2716–27. doi:10.21037/jtd-20-3525
      OpenUrl
    20. ↵
      1. Gaba L,
      2. Victoria I,
      3. Pineda E, et al
      . Changes in blood eosinophilia during anti-Pd1 therapy as a Predictor of long term disease control in metastatic Melanoma. JCO 2015;33:9069. doi:10.1200/jco.2015.33.15_suppl.9069
      OpenUrl
    21. ↵
      1. Platanias LC
      . Mechanisms of type-I- and type-II-interferon-mediated signalling. Nat Rev Immunol 2005;5:375–86. doi:10.1038/nri1604
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Morimoto Y,
      2. Kishida T,
      3. Kotani S-I, et al
      . Interferon-Β signal may up-regulate PD-L1 expression through Irf9-dependent and independent pathways in lung cancer cells. Biochem Biophys Res Commun 2018;507:330–6. doi:10.1016/j.bbrc.2018.11.035
      OpenUrlCrossRefPubMed
    23. ↵
      1. Terawaki S,
      2. Chikuma S,
      3. Shibayama S, et al
      . IFN-Α directly promotes programmed cell death-1 transcription and limits the duration of T cell-mediated immunity. J Immunol 2011;186:2772–9. doi:10.4049/jimmunol.1003208
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Wan W,
      2. Ding Y,
      3. Xie Z, et al
      . PDGFR-Β modulates vascular smooth muscle cell phenotype via IRF-9/SIRT-1/NF-ΚB pathway in subarachnoid hemorrhage rats. J Cereb Blood Flow Metab 2019;39:1369–80. doi:10.1177/0271678X18760954
      OpenUrl
    25. ↵
      1. Zhang S-M,
      2. Zhu L-H,
      3. Chen H-Z, et al
      . Interferon regulatory factor 9 is critical for neointima formation following vascular injury. Nat Commun 2014;5. doi:10.1038/ncomms6160
    26. ↵
      1. Posey KL,
      2. Coustry F,
      3. Hecht JT
      . Cartilage oligomeric matrix protein: compopathies and beyond. Matrix Biology 2018;71–72:161–73. doi:10.1016/j.matbio.2018.02.023
      OpenUrl
    27. ↵
      1. Englund E,
      2. Bartoschek M,
      3. Reitsma B, et al
      . Cartilage oligomeric matrix protein contributes to the development and metastasis of breast cancer. Oncogene 2016;35:5585–96. doi:10.1038/onc.2016.98
      OpenUrlCrossRef
    28. ↵
      1. Englund E,
      2. Canesin G,
      3. Papadakos KS, et al
      . Cartilage oligomeric matrix protein promotes prostate cancer progression by enhancing invasion and disrupting intracellular calcium homeostasis. Oncotarget 2017;8:98298–311. doi:10.18632/oncotarget.21176
      OpenUrlCrossRef
    29. ↵
      1. Papadakos KS,
      2. Hagerling C,
      3. Rydén L, et al
      . High levels of expression of cartilage oligomeric matrix protein in lymph node metastases in breast cancer are associated with reduced survival. Cancers (Basel) 2021;13:5876. doi:10.3390/cancers13235876
    30. ↵
      1. Lu Y,
      2. Kong X,
      3. Zhong W, et al
      . Diagnostic, therapeutic, and prognostic value of the thrombospondin family in gastric cancer. Front Mol Biosci 2021;8:647095. doi:10.3389/fmolb.2021.647095
      OpenUrl
    31. ↵
      1. Cheng X,
      2. Cao Y,
      3. Wang X, et al
      . Systematic pan-cancer analysis of KLRB1 with prognostic value and immunological activity across human tumors. A Cura Di Xue-Li Zhang. J Immunol Res 2022;2022:5254911. doi:10.1155/2022/5254911
    32. ↵
      1. Gentles AJ,
      2. Newman AM,
      3. Liu CL, et al
      . The prognostic landscape of genes and infiltrating immune cells across human cancers. Nat Med 2015;21:938–45. doi:10.1038/nm.3909
      OpenUrlCrossRefPubMed
    33. ↵
      1. Ma C,
      2. Luo H,
      3. Cao J, et al
      . Identification of a novel tumor microenvironment–associated eight-gene signature for prognosis prediction in lung adenocarcinoma. Front Mol Biosci 2020;7:571641. doi:10.3389/fmolb.2020.571641
      OpenUrl
    34. ↵
      1. Braud VM,
      2. Biton J,
      3. Becht E, et al
      . Expression of LLT1 and its receptor CD161 in lung cancer is associated with better clinical outcome. Oncoimmunology 2018;7:e1423184. doi:10.1080/2162402X.2017.1423184
    35. ↵
      1. Al Barashdi MA,
      2. Ali A,
      3. McMullin MF, et al
      . Protein tyrosine phosphatase receptor type C (PTPRC or Cd45). J Clin Pathol 2021;74:548–52. doi:10.1136/jclinpath-2020-206927
      OpenUrlAbstract/FREE Full Text
    36. ↵
      1. Buller CW,
      2. Mathew PA,
      3. Mathew SO
      . Roles of NK cell receptors 2B4 (CD244), CS1 (CD319), and LLT1 (CLEC2D) in cancer. Cancers (Basel) 2020;12:1755. doi:10.3390/cancers12071755
    37. ↵
      1. Boles KS,
      2. Nakajima H,
      3. Colonna M, et al
      . Molecular characterization of a novel human natural killer cell receptor homologous to mouse 2B4. Tissue Antigens 1999;54:27–34. doi:10.1034/j.1399-0039.1999.540103.x
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Kubin MZ,
      2. Parshley DL,
      3. Din W, et al
      . Molecular cloning and biological characterization of NK cell activation-inducing ligand, a counterstructure for Cd48. Eur J Immunol 1999;29:3466–77. doi:10.1002/(SICI)1521-4141(199911)29:11<3466::AID-IMMU3466>3.0.CO;2-9
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Wang N,
      2. Calpe S,
      3. Westcott J, et al
      . Cutting edge: the adapters EAT-2A and -2B are positive regulators of CD244- and CD84-dependent NK cell functions in the C57Bl/6 mouse. J Immunol 2010;185:5683–7. doi:10.4049/jimmunol.1001974
      OpenUrlAbstract/FREE Full Text
    40. ↵
      1. Vaes RDW,
      2. Reynders K,
      3. Sprooten J, et al
      . Identification of potential prognostic and predictive immunological biomarkers in patients with stage I and stage III non-small cell lung cancer (NSCLC): a prospective exploratory study. Cancers (Basel) 2021;13:6259. doi:10.3390/cancers13246259
    41. ↵
      1. Fehlings M,
      2. Jhunjhunwala S,
      3. Kowanetz M, et al
      . Late-differentiated effector neoantigen-specific Cd8+ T cells are enriched in peripheral blood of non-small cell lung carcinoma patients responding to atezolizumab treatment. J Immunotherapy Cancer 2019;7. doi:10.1186/s40425-019-0695-9
    42. ↵
      1. Zitvogel L,
      2. Ma Y,
      3. Raoult D, et al
      . The microbiome in cancer Immunotherapy: diagnostic tools and therapeutic strategies. Science 2018;359:1366–70. doi:10.1126/science.aar6918
      OpenUrlAbstract/FREE Full Text
    43. ↵
      1. Bredin P,
      2. Naidoo J
      . The gut microbiome, immune check point inhibition and immune-related adverse events in non-small cell lung cancer. Cancer Metastasis Rev 2022;41:347–66. doi:10.1007/s10555-022-10039-1
      OpenUrl
    44. ↵
      1. Fernandes MR,
      2. Aggarwal P,
      3. Costa RGF, et al
      . Targeting the gut microbiota for cancer therapy. Nat Rev Cancer 2022;22:703–22. doi:10.1038/s41568-022-00513-x
      OpenUrlCrossRef
    45. ↵
      1. Togo AH,
      2. Khelaifia S,
      3. Bittar F, et al
      . 'Eisenbergiella Massiliensis', a new species isolated from human stool collected after bariatric surgery. New Microbes New Infect 2016;13:15–6. doi:10.1016/j.nmni.2016.05.015
      OpenUrl
    46. ↵
      1. Togo AH,
      2. Diop A,
      3. Million M, et al
      . Draft genome and description of eisenbergiella Massiliensis strain AT11T: a new species isolated from human Feces after bariatric surgery. Curr Microbiol 2018;75:1274–81. doi:10.1007/s00284-018-1520-2
      OpenUrl
    47. ↵
      1. Waldman AD,
      2. Fritz JM,
      3. Lenardo MJ
      . A guide to cancer Immunotherapy: from T cell basic science to clinical practice. Nat Rev Immunol 2020;20:651–68. doi:10.1038/s41577-020-0306-5
      OpenUrlCrossRefPubMed
    48. ↵
      1. Ferrere G,
      2. Tidjani Alou M,
      3. Liu P, et al
      . Ketogenic diet and ketone bodies enhance the anticancer effects of PD-1 blockade. JCI Insight 2021;6:e145207. doi:10.1172/jci.insight.145207
    49. ↵
      1. Dai X,
      2. Bu X,
      3. Gao Y, et al
      . Energy status dictates PD-L1 protein abundance and anti-tumor immunity to enable checkpoint blockade. Mol Cell 2021;81:2317–31. doi:10.1016/j.molcel.2021.03.037
      OpenUrl
    50. ↵
      1. Corrêa RO,
      2. Castro PR,
      3. Moser R, et al
      . Butyrate: connecting the gut-lung axis to the management of pulmonary disorders. Front Nutr 2022;9:1011732. doi:10.3389/fnut.2022.1011732
    51. ↵
      1. Vanguri RS,
      2. Luo J,
      3. Aukerman AT, et al
      . Multimodal integration of radiology, pathology and genomics for prediction of response to PD-(L)1 blockade in patients with non-small cell lung cancer. Nat Cancer 2022;3:1151–64. doi:10.1038/s43018-022-00416-8
      OpenUrl

    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 @andrea_franza

    • GLR, AP and JD contributed equally.

    • CP, MG and MCG contributed equally.

    • Correction notice This article has been corrected since it was first published online. Affiliation 12 has been updated to: Oncology Department, Mario Negri Institute for Pharmacological Research (IRCCS), Milano, Lombardia, Italy

    • Contributors GLR is responsible for overall content as guarantor. GLR, AP, CP, MG and MCG substantially contributed to the conception or design of the work, the acquisition, or interpretation of data for the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published. JD substantially contributed to the conception or design of the work, analysis, or interpretation of data for the work; drafting the work or revising it critically for important intellectual content; final approval of the version to be published. TB, LA, TTr, AFa, DL, RF, MB, MO, LM, LP, AS, GV, FS, SB, VM, ALGP, FT, SM, CG, PA, RL, AFr, JMC, TTo, AA, RM, GT, GP, VT and FDB contributed to drafting the work or revising it critically for important intellectual content; final approval of the version to be published.

    • Funding This trial was supported by Merck Sharp & Dohme (grant number IIS52278).

    • Competing interests GLR provided consultation, attended advisory boards and/or provided lectures for the following organizations, from whom received honoraria or education grants: Merck Sharp and Dohme, Takeda, Amgen, Eli Lilly, BMS, Roche, Italfarmaco, Novartis, Sanofi, Pfizer and AstraZeneca. AP declares personal fees from AstraZeneca, Italfarmaco, Roche, BMS. RF declares advisory role from Merck Sharp and Dohme. FDB provided consultation, attended advisory boards and/or provided lectures for the following organizations, from whom received honoraria or education grants: Amgen, AstraZeneca, Boehringer-Ingelheim, BMS, Eli Lilly, F. Hoffmann-La Roche, Ignyta, Merck Sharp and Dohme, Merck Serono, Novartis, Pfizer. CP declares personal fees from Italfarmaco, AstraZeneca, BMS and Merck Sharp and Dohme. MCG declares personal financial interests with the following organizations: AstraZeneca, MSD International GmbH, BMS, Boehringer Ingelheim Italia S.p.A, Celgene, Eli Lilly, Ignyta, Incyte, Inivata, MedImmune, Novartis, Pfizer, Roche, Takeda, Seattle Genetics, Mirati, Daiichi Sankyo, Regeneron, Merck, Ose Immuno Therapeutics, Blueprint, Jansenn, Sanofi; she also declares Institutional financial interests with the following organizations: Eli Lilly, MSD, Pfizer (MISP); AstraZeneca, MSD International GmbH, BMS, Boehringer Ingelheim Italia S.p.A, Celgene, Eli Lilly, Ignyta, Incyte, MedImmune, Novartis, Pfizer, Roche, Takeda, Tiziana, Foundation Medicine, Glaxo Smith Kline GSK, Spectrum pharmaceuticals.

    • 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.