Purpose: The efficacy of PD-1 checkpoint blockade as adjuvant therapy in localized clear cell renal cell carcinoma (ccRCC) is currently unknown. The identification of tumor microenvironment (TME) prognostic biomarkers in this setting may help define which patients could benefit from checkpoint blockade and uncover new therapeutic targets.

Experimental Design: We performed multiparametric flow cytometric immunophenotypic analysis of T cells isolated from tumor tissue [tumor-infiltrating lymphocytes (TIL)], adjacent non-malignant renal tissue [renal-infiltrating lymphocytes (RIL)], and peripheral blood lymphocytes (PBL), in a cohort of patients (n = 40) with localized ccRCC. Immunophenotypic data were integrated with prognostic and histopathologic variables, T-cell receptor (TCR) repertoire analysis of sorted CD8+PD-1+ TILs, tumor mRNA expression, and digital quantitative immunohistochemistry.

Results: On the basis of TIL phenotypic characterization, we identified three dominant immune profiles in localized ccRCC: (i) immune-regulated, characterized by polyclonal/poorly cytotoxic CD8+PD-1+Tim-3+Lag-3+ TILs and CD4+ICOS+ cells with a Treg phenotype (CD25+CD127Foxp3+/Helios+GITR+), that developed in inflamed tumors with prominent infiltrations by dysfunctional dendritic cells and high PD-L1 expression; (ii) immune-activated, enriched in oligoclonal/cytotoxic CD8+PD-1+Tim-3+ TILs, that represented 22% of the tumors; and (iii) immune-silent, enriched in TILs exhibiting RIL-like phenotype, that represented 56% of patients in the cohort. Only immune-regulated tumors displayed aggressive histologic features, high risk of disease progression in the year following nephrectomy, and a CD8+PD-1+Tim-3+ and CD4+ICOS+ PBL phenotypic signature.

Conclusions: In localized ccRCC, the infiltration with CD8+PD-1+Tim-3+Lag-3+ exhausted TILs and ICOS+ Treg identifies the patients with deleterious prognosis who could benefit from adjuvant therapy with TME-modulating agents and checkpoint blockade. This work also provides PBL phenotypic markers that could allow their identification. Clin Cancer Res; 23(15); 4416–28. ©2017 AACR.

Translational Relevance

The identification of tumor immune subgroups is a crucial task to predict the patient's sensitivity to novel therapies, including checkpoint blockade. In this study, we analyzed the tumor-infiltrating lymphocyte (TIL) and peripheral blood lymphocyte (PBL) phenotypes in a cohort of patients with localized clear cell renal cell carcinoma and correlated it with major clinicopathologic characteristics and prognosis. With two 11-color flow cytometric surface panels, we identified a group of patients (22%) who presented a high risk of disease progression in the year following nephrectomy despite having a localized disease. A detailed characterization of their tumors revealed a highly inflammatory and suppressive microenvironment, rich in dysfunctional TIL. These patients also exhibited a PBL phenotypic signature that could allow their identification by noninvasive methods. This study unravels TIL and PBL phenotypic biomarkers that could be used in routine clinical practice to identify patients with early-stage ccRCC and deleterious prognosis who could benefit from adjuvant therapy with TME-modulating agents.

During the evolution of a normal cell toward a disseminated neoplastic cancer, host immunity plays an active role in controlling tumor growth and metastatic spreading (1, 2). Different types of infiltrating immune cells with diverse effects on tumor progression and patient's clinical outcome have been described (3, 4). A large amount of evidence suggests that increased infiltrations with CD8+ and Th1-CD4+ T cells are associated with good prognosis in the majority of tumors (3, 4). Nevertheless, several exceptions to this rule have recently emerged, including clear cell renal cell carcinoma (ccRCC; refs. 5, 6), Hodgkin lymphoma (7), and gastric cancer (8), where high densities of CD8+ tumor-Infiltrating lymphocytes (TIL) have been associated with poor prognosis.

The study of the tumor microenvironment (TME) immune infiltrates in ccRCC has revealed specific hallmarks that could explain this counterintuitive correlation. In fact, ccRCC enriched in CD8+ TILs and associated with poor clinical outcome (20%–30%) exhibit high expression of immune checkpoints and their ligands, abundant infiltration with dysfunctional dendritic cells (DC) and scarce tertiary lymphoid structures (TLS; ref. 5). Similar results have been reported in subgroups of patients with other pathologies, including melanoma (9, 10), non–small cell lung cancer (refs. 11, 12), and colorectal cancer (13).

The identification of biomarkers of this type of tumors will allow to detect patients at high risk of disease recurrence and potentially sensitive to checkpoint blockade (14). Currently, the expression of PD-L1 by immune or tumor cells, the extensive tumor infiltration by CD8+ TILs, and high tumor mutational burdens are the most sensitive and specific biomarkers of clinical response to checkpoint blockade (9, 10, 15, 16). Interestingly, the overexpression of PD-1 by TILs has limited accuracy in predicting clinical response to these therapies (9), and recent evidence supports that the mono-expression of this marker does not necessarily correlate with the loss of T-cell functions (17, 18). In fact, the simultaneous expression of other InR (e.g., Tim-3, Lag-3, and TIGIT) on this population is necessary to accurately identify functionally exhausted TILs (17, 18), and the expression of PD-1 ligands in the TME seems necessary to induce PD-1+ TIL inhibition (5).

Till date, most of the studies analyzing TME biomarkers of clinical response to checkpoint blockade in ccRCC have been performed in cohorts of patients with metastatic disease (15), and its clinical efficacy in localized tumors is currently unknown. Thus, a detailed study of novel prognostic and theranostic TME biomarkers in non-metastatic ccRCC is necessary to determine whether these patients could benefit from checkpoint blockade in the adjuvant setting. In addition, the difficulties in identifying exhausted TILs by the mono-expression of InR support the need to study the ccRCC-TIL phenotype with a multidimensional perspective. In this study, we performed a multiparametric immunophenotypic flow cytometric analysis of TIL in non-metastatic ccRCC and exhaustively characterized other TME characteristics by immunohistochemistry (IHC), next-generation T-cell receptor (TCR)-B sequencing, and gene expression analyses. We identify a group of patients whose tumors are enriched in polyclonal and poorly cytotoxic CD8+PD-1+Tim-3+Lag-3+ cells, CD4+ICOS+ T cells with a Treg phenotype (CD25+CD127Foxp3+), dysfunctional DC, M2-skewed macrophages, PD-L1+ immune and tumor cells and who have a very high risk of early disease progression (<12 months). Furthermore, we describe that this group of patients exhibited a CD4+ICOS+ and CD8+PD-1+Tim-3+ phenotypic signature in their peripheral blood lymphocytes (PBL). Our work provides a phenotypic signature of TIL and PBL associated with deleterious prognosis in early-stage ccRCC and, upon further validation in prospective cohorts, may define a group of patients who could benefit from adjuvant therapy with TME-modulating agents and checkpoint blockade.

Patients

A cohort of 40 primary ccRCC human tumors was collected between January 2014 and June 2016, from specimens of radical or partial nephrectomy operated at the hospital Institut Mutualiste Montsouris (Paris, France). We obtained preoperative peripheral blood samples (n = 33) and adjacent non-tumor kidney tissue (n = 34) in some of the patients. Also, we analyzed the peripheral blood from 5 healthy volunteers. This research was conducted according to the recommendations outlined in the Helsinki declaration and was approved by the medical ethics boards of all participating institutions (no. CEPAR-2014-001). All the included patients signed an informed consent prior to inclusion in the study. The demographic characteristics of the cohorts are depicted in Supplementary Table S1.

Tumor processing, surface staining, and cell sorting

Tumors were dilacerated and incubated for 1 hour at 4°C with Cell Recovery Solution (Fisher Scientific); mixtures were filtrated and TILs separated with Ficoll-Paque PLUS (GE Healthcare Life Science). Cells were then stained with the following monoclonal antibodies: CD3 AF700 (UCHT1, BD), CD4 BV605 (OKT4, Biolegend), CD8 BV650 (RPA-T8, Biolegend), CD45RA ECD (2H4, Beckman coulter), CCR7 PE-Cy7 (G043H7, Biolegend), CD69 PE (FN50, BD), CD38 PercpeF710 (HB7, eBioscience), CD40L APC-Cy7 (Biolegend), ICOS FITC (Isa-3, eBioscience), GITR APC (AITR, eBioscience), PD-1 APC-Cy7 or PerCP (EH12.2H7, Biolegend), TIM-3 BV421 (F38-2E2, Biolegend), CTLA-4 APC (L3D10, Biolegend), LAG-3 FITC (17B4, Enzo Life Sciences), CD127 PE-Vio615 (A019D5, Miltenyi), and CD25 APC-Cy7 (M-A251, Biolegend). Certain samples were stained after cellular fixation/permeabilization with the eBioscience Kit (Intracellular Fixation & Permeabilization Buffer Set) following the manufacturer's instruction with the next set of antibodies: Perforin AF647 (dG9, Biolegend), FoxP3 BV450 (259D/C7, BD), and HELIOS FITC (22F6, Miltenyi). Samples were acquired in a FACS Fortessa cytometer with FACSDiva software (BD Bioscience) and data analyzed with FlowJo 7.9.4 software (Tree Star, Inc.). The fraction of cells co-expressing multiple markers was calculated in SPICE 5.3033 (Exon), a data mining software application that normalizes and analyzes large FlowJo datasets (19). The gating and data analysis strategy is displayed in Supplementary Fig. S1.

For CD8+PD-1 (n = 7) and CD8+PD-1+ (n = 6) TIL sorting, the abovementioned isolation protocol was followed and cells were incubated with: CD45-BV510 (HI30, Biolegend), CD3-PE (UCTH1, eBioscience), CD4-AF700 (RPA-T4, eBioscience), CD8-APC-H7 (SK1, BD Bioscience), PD-1-PE (EH12.2H7, Biolegend), and DAPI (1 μg/mL). Samples were sorted using FACSAria cytometer. The achieved purity of viable CD45+CD3+CD4CD8+PD-1 and CD45+CD3+CD4CD8+PD-1+ cells was superior to 95%.

IHC staining and quantification

Serial 5-μm formalin-fixed, paraffin-embedded (FFPE) tissue sections from primary ccRCC were stained using autostainerPlus Link 48 (Dako). Antigen retrieval and deparaffinization were carried out on a PT-Link (Dako) using the EnVision FLEX Target Retrieval Solutions (Dako). The antibodies used in this study for IHC are listed in Supplementary Table S2. Peroxidase activity was detected using 3-amino-9-ethylcarbazole substrate (AEC) and alkaline phosphatase using alkaline phosphatase substrate III (Vector Laboratories). Stained slides were scanned with a Nanozoomer (Hamamatsu) and analyzed with Calopix software (Tribvn). We quantified the CD3+, CD8+, CD20+, CD163+, PD-1+, and Lag-3+ cells; in addition to DC-Lamp+ cells outside TLS [considered as dysfunctional (3, 5)] and semiquantified PD-L1 expression. The percentage of tumor and/or immune cells expressing PD-L1 was determined by 2 independent reviewers (L. Lacroix and N.A. Giraldo) and an expression superior to 5% was considered as positive.

Gene expression analysis

RNA was extracted from frozen n = 33 ccRCC (15 sections of 15 μm each), using the Maxwell 16 LEV simplyRNA Purification Kit (Promega) as described by the manufacturer's instructions. RNA quality and concentration was determined with a 2100 Bioanalyzer (Agilent). Reverse transcription PCR was conducted with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystem), and the quantitative gene expression analysis of 107-immune related genes was determined with the TaqMan Human Immune Array on an Applied Biosystems 7900HT Fast Real-Time PCR System. Expression levels of genes were determined using threshold cycle (Ct) values normalized to actin B (ΔCt). The list of analyzed genes is displayed in Supplementary Table S3.

TCR-B deep sequencing

CD8+PD-1+ and CD8+PD-1 cells were sorted in PBS 1× and 5% FBS, cell pellet was resuspended in 200 μL PBS 1×, and immediately frozen. The number of sequenced CD8+ TILs ranged from 10 × 103 to 10 × 104. Deep sequencing of the variable VDJ regions of TCR-B genes was performed on gDNA by Adaptive Biotechnologies. The coverage per sample was >10×. As measure of the clonality of each population, we used the Clonality Index calculated as the inverse of the normalized Shannon entropy of all productive clones in a sample [where values near 1 represent samples with one or a few predominant clones and values near 0 represent more polyclonal samples (ref. 20)] as well as the fraction of total sequenced clonotypes that represented the 15 most expanded clones (or top clones).

Statistical analysis

Data were reported as the mean ± SD, interquartile range (IQR), or SEM. We used the Kruskal–Wallis test with post-hoc Dunn test to compare the percentage of positive cells between PBL, renal-infiltrating lymphocytes (RIL), and TIL. Principal Component Analysis (PCA) was conducted by the FactoMineR R package. The unsupervised classifications of patients (TIL and PBL) were performed by hierarchical clustering using Euclidian distance and the Ward linkage criterion. We z-score transformed parameters before running the clustering algorithm. To determine the optimal number of clusters according to TIL phenotype, we used Gap Statistics (21). The association of patient's clusters and prognosis was assessed using the Kaplan–Meier method, univariate Cox regression models, and the log-rank test. Kruskal–Wallis test with Dunn post-hoc test or Fisher exact test were used to analyze the statistical differences between ccRCC clusters for numerical and categorical variables, respectively. To compare the relative frequencies of the 15 most recurrent unique TCRβ clonotypes in CD8+PD-1+ TIL between cluster 2 and 3 tumors, we used χ2 test for trend. For gene expression analyses, 2-sided comparison was carried using the Mann–Whitney U test, corrected using Bonferroni method for multiple testing. Correlation matrix and r values were determined using Spearman method, using the corrplot R package.

CD4+ and CD8+ ccRCC TILs display increased but highly heterogeneous expression of activation markers and inhibitory receptors

ccRCC is one of the tumors where increased PD-1+ TIL infiltrations have been associated with poor clinical outcome (5, 22). To characterize the simultaneous expression of PD-1 and other InR, as well as the activation and differentiation status of ccRCC TILs, we performed a multiparametric phenotypic analysis of 40 ccRCC TILs and compared it with autologous PBLs and RILs isolated from adjacent non-malignant kidney tissue. The demographic and histopathologic characteristics of this cohort are displayed in Supplementary Table S1. We measured cell surface expression of differentiation markers (CCR7 and CD45RA), InR (PD-1, Tim-3, Lag-3, and CTLA-4) typically expressed in chronically stimulated T cells, and activation markers (AM: CD69, CD38, CD40L, ICOS, and GITR) which can be unregulated upon TCR activation.

The expression of differentiation markers significantly varied between PBLs and TILs in the CD4+ and CD8+ T-cell compartments, but not between RIL and TIL (Supplementary Fig. S2). As compared with PBLs, CD4+ andCD8+ TILs displayed lower frequencies of cells with central memory (CM, CCR7+CD45RA) and naïve (CCR7CD45RA) phenotypes, but higher fractions of effector memory (EM, CCR7+CD45RA+) T cells (Supplementary Fig. S2). Overall, the expression of activation markers and InR was considerably higher in CD4+ and CD8+ TILs than in PBL, whereas only some markers were upregulated in TILs as compared with RILs (Fig. 1A and B). Indeed, CD4+ TILs exhibited an enhanced expression of all the analyzed activation markers (except CD38) and InR as compared with PBL (Fig. 1A and B); but, when compared with CD4+RIL, they only displayed increased percentages of GITR+, PD-1+, and Tim-3+ cells (Fig. 1A and B). Similarly, CD8+ TILs exhibited an enhanced expression of almost all the activation markers and InR (except CD40L and GITR) in comparison to PBL (Fig. 1A and B); but only increased fraction of CD38+, ICOS+, PD-1+, and Tim-3+ cells than RIL (Fig. 1A and B).

Figure 1.

The intertumor heterogeneity of TIL phenotype in ccRCC. A, Percentage of CD4+ (top) or CD8+ (bottom) PBLs (red dots), RILs (black dots), and TILs (blue dots) expressing activation or inhibitory receptors (mean ± SD) in ccRCC-bearing patients (n = 40). B, Heatmap of the scaled mean fraction of cells expressing activation markers or inhibitory receptors in PBLs, RILs, and TILs. C, PCA including the fractions cells expressing activation or inhibitory receptors in CD4+ (top) or CD8+ (bottom) in PBLs, RILs, and TIL. Squares in the top left corner in each panel display the dispersion of each compartment (red dots) in relation to all the samples in PCA. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by Kruskal–Wallis with Dunn post-hoc test.

Figure 1.

The intertumor heterogeneity of TIL phenotype in ccRCC. A, Percentage of CD4+ (top) or CD8+ (bottom) PBLs (red dots), RILs (black dots), and TILs (blue dots) expressing activation or inhibitory receptors (mean ± SD) in ccRCC-bearing patients (n = 40). B, Heatmap of the scaled mean fraction of cells expressing activation markers or inhibitory receptors in PBLs, RILs, and TILs. C, PCA including the fractions cells expressing activation or inhibitory receptors in CD4+ (top) or CD8+ (bottom) in PBLs, RILs, and TIL. Squares in the top left corner in each panel display the dispersion of each compartment (red dots) in relation to all the samples in PCA. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by Kruskal–Wallis with Dunn post-hoc test.

Close modal

The frequency of expression of phenotypic markers on CD4+ and CD8+ TILs was highly heterogeneous throughout tumors, particularly for ICOS, GITR, PD-1, and Tim-3 (Fig. 1A). In view of this TIL phenotypic heterogeneity, we performed a PCA including the fractions of CD4+ or CD8+ T cells expressing differentiation markers, activation markers and InR in TILs, PBLs, and RILs (Fig. 1C). PCA confirmed that, contrary to PBLs and RILs, the CD4+ and CD8+ TIL phenotypes were highly heterogeneous throughout patients (Fig. 1C). Almost all the included markers (except CD45RA, CD38, CTLA-4 for the CD4+ T cells and GITR for the CD8+ T cells) contributed to the variance in the PCA modeling (Fig. 1C).

Unsupervised classification of TIL phenotype reveals 3 ccRCC major clusters, with different risk of early disease progression

In view of the heterogeneous expression of activation markers and InR in TILs across patients, our next aim was to evaluate the pertinence of a ccRCC classification model that categorized tumors according to TIL phenotype. We performed an unsupervised clustering of our cohort on the basis of the fractions of CD4+ and CD8+ TILs expressing activation markers and InR; we only included variables which displayed significant contribution to TILs versus RILs versus PBL variance (as weighed by the previous PCA analysis, Fig. 1C). In addition, as TILs were mainly composed by EM T cells (Supplementary Fig. S2), differentiation markers were also excluded from the analysis. Through gap statistic methods (21), we determined that the optimal number of clusters in our cohort was 3 (Supplementary Fig. S3). Accordingly, by hierarchical unsupervised clustering, we divided our cohort into 3 groups, named cluster 1, cluster 2, and cluster 3 (Fig. 2A). PCA supported this subdivision, as tumors from the 3 groups clustered separately (Supplementary Fig. S4).

Figure 2.

Unsupervised clustering of TIL phenotype defines 3 ccRCC clusters with different pathologic and prognostic features. A, Unsupervised hierarchical clustering and heatmap of the scaled fractions of TIL expressing phenotypic markers (on the right, n = 40). The dendogram was cut (dotted line) into 3 clusters (blue, green, and orange), according to gap statistic results. P values according to Kruskal–Wallis test are displayed on the left. B, Progression-free survival and fraction of tumors with nonaggressive (1–2, black) or aggressive (3–4, gray) nuclear grades according to TIL cluster. P values according to univariate Cox regression analysis and Fisher exact test are displayed, respectively. C, Percentage of CD4+ (left) and CD8+ (right) TILs expressing activation markers or InR (mean ± SD) according to TIL cluster (top). P values according to Kruskal–Wallis with Dunn post-hoc test are shown. Representative histograms of the expression of activation markers or InR according to TIL cluster are displayed (bottom). D, Spider-plots displaying the relative fraction of positive T cells for the mentioned markers according to the TIL clusters and RIL (black). Color code: Blue = Cluster 1, green = Cluster 2, orange = Cluster 3, black = RILs. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 2.

Unsupervised clustering of TIL phenotype defines 3 ccRCC clusters with different pathologic and prognostic features. A, Unsupervised hierarchical clustering and heatmap of the scaled fractions of TIL expressing phenotypic markers (on the right, n = 40). The dendogram was cut (dotted line) into 3 clusters (blue, green, and orange), according to gap statistic results. P values according to Kruskal–Wallis test are displayed on the left. B, Progression-free survival and fraction of tumors with nonaggressive (1–2, black) or aggressive (3–4, gray) nuclear grades according to TIL cluster. P values according to univariate Cox regression analysis and Fisher exact test are displayed, respectively. C, Percentage of CD4+ (left) and CD8+ (right) TILs expressing activation markers or InR (mean ± SD) according to TIL cluster (top). P values according to Kruskal–Wallis with Dunn post-hoc test are shown. Representative histograms of the expression of activation markers or InR according to TIL cluster are displayed (bottom). D, Spider-plots displaying the relative fraction of positive T cells for the mentioned markers according to the TIL clusters and RIL (black). Color code: Blue = Cluster 1, green = Cluster 2, orange = Cluster 3, black = RILs. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Patients whose tumors were classified as cluster 3 exhibited a shorter disease-free survival [DFS; median DFS = 7 months, P = 0.0005, HR, 20.6; 95% confidence interval (CI), 2.2–190] as compared with clusters 1 and 2 (median DFS not reached, Fig. 2B). In addition, cluster 3 ccRCC displayed more aggressive nuclear grades (Fuhrman grades 3–4 = 74%, P = 0.01, Fig. 2B) and a trend toward bigger tumors (Supplementary Fig. S5) than clusters 1 and2. Once we determined that this unsupervised clustering model of TIL phenotype could identify subgroups of tumors with different pathologic and prognostic features, we aimed to characterize in detail these 3 clusters.

A relevant fraction of non-metastatic ccRCC displays an immune-silent microenvironment

Cluster 1 (representing 56% of the tumors) was characterized by lower fractions of TILs expressing several activation markers and InR (Fig. 2C and Supplementary Figs. S6 and S7) compared with cluster 2; and by all the analyzed markers relatively to cluster 3 (except CD40L in CD4+ and GITR in CD8+ TILs; Fig. 2C and Supplementary Figs. S6 and S7). Interestingly, cluster 1 displayed the same T-cell phenotypic profile as autologous RIL (Fig. 2C and D) and they clustered together in PCA (Supplementary Fig. S4). In view of the relatively low expression of activation markers and InR in cluster1 ccRCC, we further named them Immune-silent tumors.

ccRCC cluster with the highest risk of early progression is enriched in CD4+ICOS+GITR+Helios+ cells with a Treg phenotype

Clusters 2 and 3 displayed similar percentages of several TIL populations (including CD4+PD-1+ and CD8+PD-1+; and Fig. 2C and Supplementary Figs. S6 and S7). What in fact differentiated these 2 subgroups was an expansion of the CD4+ICOS+ (P = 0.0002), CD4+Lag-3+ (P = 0.001), CD4+CTLA-4+ (P = 0.03), CD8+ICOS+ (P = 0.006), and CD8+Lag-3+ (P = 0.0003) TIL in cluster 3 (Fig. 2C and Supplementary Figs. S6 and S7). The overexpression of these 4 markers in the CD4+ TIL compartment suggested an expansion of a potential Treg population in cluster 3. To support this finding, we analyzed in detail the phenotype of CD4+ TIL population expressing ICOS, as this molecule has been described as a receptor expressed in Treg in cancer (23) and, among the abovementioned 4 markers, it had the highest representation in cluster 3 tumors (19.1% ± 6.5%; Fig. 3A). We measured the expression of GITR, PD-1, CD25, CD127, Helios, and Foxp3 in 8 of the 40 ccRCCs. Interestingly, CD4+ICOS+ TILs were enriched in cells displaying a Treg phenotype (CD25+CD127Foxp3+, 20.0% ± 12.8%) as compared with CD4+ICOS TILs (3.5% ± 2.2%, P = 0.008, Fig. 3B). In addition, an elevated fraction of CD4+ICOS+ TILs co-expressed GITR, Foxp3, and Helios (28.3% ± 21.9) as compared with CD4+ICOS TILs (5.4% ± 3.8%, P = 0.01, Fig. 3B). The CD4+ICOS+GITR+Foxp3+Helios+ TIL population represented approximately 8% of the CD4+ TILs in the cluster 3 tumors (and <1.2% in the other clusters).

Figure 3.

Phenotypic characterization CD4+ICOS+ TILs in ccRCC. A, Representative flow cytometric density plots displaying the expression of ICOS (y-axis) and GITR (x-axis) among CD4+ T cells according to TIL cluster. The percentage of cells in each quadrant is displayed. B, Percentage (mean ± SD) of CD25+CD127Foxp3+ and GITR+HELIOS+Foxp3+ among CD4+ICOS and CD4+ICOS+ TIL (left). P values according to Mann–Whitney U test are displayed. Pie charts and arcs representing the median fraction of cells expressing GITR (green), HELIOS (yellow), and Foxp3 (red) among CD4+ICOS and CD4+ICOS+ TILs; in green, median fraction of GITR+HELIOS+Foxp3+ cells (middle). P values according to Fisher exact test are displayed. Representative histograms of the expression of GITR, Helios, and Foxp3 in CD4+ICOS and CD4+ICOS+ TILs (left). *, P < 0.05; **, P < 0.01.

Figure 3.

Phenotypic characterization CD4+ICOS+ TILs in ccRCC. A, Representative flow cytometric density plots displaying the expression of ICOS (y-axis) and GITR (x-axis) among CD4+ T cells according to TIL cluster. The percentage of cells in each quadrant is displayed. B, Percentage (mean ± SD) of CD25+CD127Foxp3+ and GITR+HELIOS+Foxp3+ among CD4+ICOS and CD4+ICOS+ TIL (left). P values according to Mann–Whitney U test are displayed. Pie charts and arcs representing the median fraction of cells expressing GITR (green), HELIOS (yellow), and Foxp3 (red) among CD4+ICOS and CD4+ICOS+ TILs; in green, median fraction of GITR+HELIOS+Foxp3+ cells (middle). P values according to Fisher exact test are displayed. Representative histograms of the expression of GITR, Helios, and Foxp3 in CD4+ICOS and CD4+ICOS+ TILs (left). *, P < 0.05; **, P < 0.01.

Close modal

ccRCC cluster enriched in CD4+ICOS+GITR+ TILs exhibit poorly cytotoxic and polyclonal CD8+PD-1+ TILs

Recent evidence supports that in melanoma and ovarian cancers, CD8+PD-1+ TILs are enriched in clonally expanded tumor-reactive cells (24, 25). We hypothesized that although cluster 2 and 3 tumors displayed similar fractions of CD8+PD-1+ TILs (Fig. 4A), their phenotype and function could significantly differ. Since exhausted T cells often co-express several InR (17, 18), we first assessed the simultaneous presence of Tim-3, Lag-3, and other InR on the CD8+PD-1+ cell surface, according to TIL clusters. Interestingly, although the fraction of CD8+PD-1+ TIL–expressing Tim-3 did not differ between cluster 2 and 3 tumors (P = 0.59, Fig. 4B and C), the latter displayed higher percentages of CD8+PD-1+Tim-3+Lag-3+ TILs (14.4% ± 9.6% vs. 5.0% ± 3.8%, P = 0.004, Fig. 4D). We also assessed the cytotoxic potential of CD8+PD-1+ TIL according to ccRCC clusters and found a significantly higher fraction of CD8+PD-1+ TILs expressing perforin in cluster 2 (65.8% ± 15.3), as compared with immune-silent (20.6% ± 4.4%, P = 0.004) and cluster 3 tumors (18.5% ± 6.0%, P = 0.009, Fig. 4E).

Figure 4.

Phenotypic and clonal characterization of CD8+PD-1+ T lymphocyte population among immune-silent, immune-activated, and immune-regulated ccRCC. A, Representative flow cytometric density plots displaying the expression of PD-1 (x-axis) and Tim-3 (y-axis) among CD8+ T cells according to TIL cluster. The percentage of cells in each quadrant is displayed. B, Pie charts representing the median fraction of CD8+PD-1+ cells expressing Tim-3 and/or Lag-3 in RIL (right) and TIL according to tumor's cluster. Percentage (mean ± SD) of CD8+PD-1+ cells expressing (C) Tim-3, (D) Tim-3/Lag-3 (n = 40), or (E) perforin (n = 15) in RILs and TILs according to tumor's cluster. P values according to Kruskal–Wallis with Dunn post-hoc test are shown. F, Relative frequencies of the first most recurrent (red), second most recurrent (orange), third to 15th most recurrent (yellow), and the rest (gray) of unique TCRβ clonotype in CD8+PD-1+ TILs according to tumor's cluster is displayed. The same information for CD8+PD-1 TILs (n = 6, not divided by TIL cluster) is displayed on the left. *, P < 0.05; ***, P < 0.001.

Figure 4.

Phenotypic and clonal characterization of CD8+PD-1+ T lymphocyte population among immune-silent, immune-activated, and immune-regulated ccRCC. A, Representative flow cytometric density plots displaying the expression of PD-1 (x-axis) and Tim-3 (y-axis) among CD8+ T cells according to TIL cluster. The percentage of cells in each quadrant is displayed. B, Pie charts representing the median fraction of CD8+PD-1+ cells expressing Tim-3 and/or Lag-3 in RIL (right) and TIL according to tumor's cluster. Percentage (mean ± SD) of CD8+PD-1+ cells expressing (C) Tim-3, (D) Tim-3/Lag-3 (n = 40), or (E) perforin (n = 15) in RILs and TILs according to tumor's cluster. P values according to Kruskal–Wallis with Dunn post-hoc test are shown. F, Relative frequencies of the first most recurrent (red), second most recurrent (orange), third to 15th most recurrent (yellow), and the rest (gray) of unique TCRβ clonotype in CD8+PD-1+ TILs according to tumor's cluster is displayed. The same information for CD8+PD-1 TILs (n = 6, not divided by TIL cluster) is displayed on the left. *, P < 0.05; ***, P < 0.001.

Close modal

In view of the phenotypic heterogeneity of CD8+PD-1+ between ccRCC clusters, we hypothesized they could also differ in their clonality. We sorted CD8+PD-1+ TILs from each cluster and sequenced the variable VDJ regions of TCR-B. CD8+PD-1+ TILs from cluster 2 tumors demonstrated a trend toward higher clonality [clonality index (CI) = 0.5 ± 0.1, n = 3] than CD8+PD-1+ TILs from cluster 3 tumors (CI = 0.2 ± 0.1, n = 3, P = 0.09; Supplementary Fig. S8). The fraction of CD8+PD-1+ TILs exhibiting one of the top 15 clonotypes in each tumor was significantly higher in cluster 2 than in cluster 3 tumors (χ2 for trend: P < 0.0001; Fig. 4F and Supplementary Fig. S8). For instance, the first and second top clones in cluster 2 represented 23.2% (±12.0%) and 14.3% (±6.9%) of the total amount of CD8+PD-1+ TILs, respectively, and only 7.9% (±4.1%) and 3.1% (±0.4%) in cluster 3 tumors (Fig. 4F and Supplementary Fig. S8). As compared with CD8+PD-1 TILs, only CD8+PD-1+ TILs from cluster 2 tumors displayed a clonal expansion (Fig. 4F and Supplementary Fig. S8).

On the basis of the above-described differential features, we further named cluster 2 tumors immune-activated and cluster 3 tumors immune-regulated.

Immune-regulated tumors exhibit an inflammatory, M2-enriched, and poorly cytotoxic microenvironment

To better understand the immune contexture accompanying the poor cytotoxic, polyclonal PD-1+Lag-3+ and Treg-enriched TME in immune-regulated ccRCC, we analyzed the expression of 107 immune-related genes (Supplementary Table S2) in whole tumor tissues by low-density microarray in 33 of the 40 tumors (immune-silent, n = 18; immune-activated, n = 6; and immune-regulated, n = 9).

Immune-activated (cluster 2) and immune-regulated (cluster 3) tumors were both relatively enriched in CD8A, CD8B, PDCD1, CD79B, and CD14 transcripts as compared with immune-silent tumors (Fig. 5A and B and Supplementary Fig. S9). Interestingly, immune-activated tumors displayed increased expression of cytotoxic genes (e.g., PRF1 and GNLY, Fig. 5A and B and Supplementary Fig. S9) as compared with the other 2 clusters. T-cell attracting chemokine genes (e.g., CXCL9 and CXCL10) were also lower in immune-silent tumors (Fig. 5A and Supplementary Fig. S9). Immune-regulated ccRCC demonstrated an elevated expression of CXCL2, IL8, and CCL20 genes as compared with the other 2 clusters (Fig. 5A and B and Supplementary Fig. S9). Interestingly, most of the genes associated with blood vessel signature (e.g., MMRN2 and VWF) or lymphangiogenesis (e.g., VEFGC) were underexpressed in immune-regulated tumors (Fig. 5A and Supplementary Fig. S9).

Figure 5.

Immune contexture and ccRCC TIL phenotype clusters. A, Heatmap of the median relative expression of immune-related genes in ccRCC tumors according to TIL cluster (n = 33). B, Mean (±SEM) of the relative expression of cytotoxic-associated (top) and inflammatory (bottom) genes in whole tumor tissue according to TIL cluster. C, Mean (±SEM) of immune cell densities by IHC according to TIL cluster. D, Fractions of ccRCC in each cluster displaying PD-L1+ tumor or immune-infiltrating cells. In red, fraction of tumors considered as PD-L1+ (>5% of expression). P value according to Fisher exact test is displayed. Representative photomicrographs of CD8 and PD-L1 staining (in red) by IHC according to TIL cluster. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

Immune contexture and ccRCC TIL phenotype clusters. A, Heatmap of the median relative expression of immune-related genes in ccRCC tumors according to TIL cluster (n = 33). B, Mean (±SEM) of the relative expression of cytotoxic-associated (top) and inflammatory (bottom) genes in whole tumor tissue according to TIL cluster. C, Mean (±SEM) of immune cell densities by IHC according to TIL cluster. D, Fractions of ccRCC in each cluster displaying PD-L1+ tumor or immune-infiltrating cells. In red, fraction of tumors considered as PD-L1+ (>5% of expression). P value according to Fisher exact test is displayed. Representative photomicrographs of CD8 and PD-L1 staining (in red) by IHC according to TIL cluster. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

To further characterize the immune infiltrates across ccRCC clusters, we quantified by IHC the densities of several immune populations, including CD3+ cells, CD8+ cells, B cells (CD20+), M2 macrophages (CD163+), and potentially dysfunctional DCs [DC-Lamp+ cells outside TLS (3, 5)], PD-1+, Lag-3+, and ICOS+ cells. Consistently with gene expression studies, we found that immune-silent tumors displayed lower densities of CD8+ cells as compared with the 2 other clusters (Fig. 5C). No differences in the CD8+ cell densities were found between immune-regulated and immune-activated tumors. In addition, immune-regulated tumors displayed higher densities of M2 macrophages than the immune-regulated ones (P = 0.03), as well as higher numbers of dysfunctional DC (P = 0.04), PD-1+ (P = 0.004), and Lag-3+ cells (P = 0.005) as compared with immune-silent ccRCC (Fig. 5C). Although not significant, a trend toward lower frequencies of CD20+ cells (Fig. 5C) and higher of ICOS+ cells (data not shown) in immune-regulated tumors was found. We also performed a semiquantification of the immune and tumor expression of PD-L1. Interestingly, we found that a higher fraction of immune-regulated tumors expressed PD-L1 (immune or tumor cell scores) as compared with immune-silent or immune-activated tumors (Fig. 5D). Using a cutoff expression level of 5% (5), we determined that 44% of the immune-regulated ccRCC displayed PD-L1+ tumor cells, versus 0% in immune-silent and immune-activated tumors (Fig. 5D). Similarly, we found that 55% of the immune-regulated ccRCC displayed PD-L1+ immune cells versus 0% in immune-silent and 22% in immune-activated tumors (Fig. 5D).

Interestingly, in immune-activated tumors, the expression of TLS-related genes (e.g., CCL19 and CCL21) was highly correlated with the expression of CD8 T-cell genes (CD8A and CD8B), as well as the infiltration with CD8+ T cells and mature DC by IHC (Supplementary Fig. S10). On the contrary, in immune-regulated tumors, the expression of TLS-related genes and the infiltration with mature DCs was not correlated with the CD8 T-cell genes, nor the densities of CD8+ TILs (Supplementary Fig. S10). Instead, the last 2 features were highly correlated with the tumor infiltration by dysfunctional DC (Supplementary Fig. S10).

Unsupervised classification of PBL phenotype identifies ccRCC-patients with poor prognosis and immune-regulated tumors

In parallel to the analysis of TIL, we performed a phenotypic characterization of autologous PBLs in 33 patients with ccRCC (immune-silent, n = 20; immune-activated, n = 7; and immune-regulated, n = 6). We questioned whether we could also define subgroups of patients based on their PBL phenotype, without previous knowledge of the patient's TIL group.

Although PBLs from patients with ccRCCs displayed lower expression of activation markers and InR as compared with TILs (Fig. 1), we first wondered whether they phenotypically differed from PBLs from healthy controls (HC, n = 5). Patients with ccRCC displayed higher fractions of PBLs expressing several activation markers and InR as compared with healthy controls, both in the CD4+ (CD4+CD69+P = 0.005, CD4+GITR+P = 0.007, CD4+CTLA-4+P = 0.0004, CD4+PD-1+P = 0.006 and CD4+Tim-3+P = 0.0006) and CD8+ (CD8+ICOS+P = 0.03, CD8+GITR+P = 0.008, CD8+PD-1+P = 0.008 and CD8+Tim-3+P = 0.0004) T-cell compartments (Supplementary Figs. S11 and S12). In contrast to healthy controls, the expression of activation markers and InR in PBLs from patients with ccRCC was heterogeneous across patients (Supplementary Figs. S11 and S12). PCA confirmed this heterogeneity, as PBL from healthy controls clustered together whereas those from patients with ccRCC showed higher dispersion (Fig. 6A).

Figure 6.

Unsupervised clustering of ccRCC-PBL phenotype defines 2 patient's clusters with different histopathologic and prognostic features. A, PCA including the fractions of cells expressing activation or inhibitory receptors in CD4+ or CD8+ PBL in healthy controls (HC; red dots, n = 5) or patients with ccRCC (black dots, n = 33). B, Unsupervised hierarchical clustering and heatmap of the scaled fractions of PBL expressing different phenotypic markers (right). The dendogram was cut (dotted line) into 2 clusters (blue and green). P values according to Mann–Whitney U test are displayed on the left. C, DFS and fraction of tumors with early (1–2, black) or advanced (3–4, gray) nuclear grades according to PBL cluster. P value according to univariate Cox regression analysis and Fisher exact test are displayed, respectively. D, Percentage of patients displaying PBL-A (blue) or PBL-B (green) profile according to TIL cluster. P value according to Fisher exact test is displayed. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 6.

Unsupervised clustering of ccRCC-PBL phenotype defines 2 patient's clusters with different histopathologic and prognostic features. A, PCA including the fractions of cells expressing activation or inhibitory receptors in CD4+ or CD8+ PBL in healthy controls (HC; red dots, n = 5) or patients with ccRCC (black dots, n = 33). B, Unsupervised hierarchical clustering and heatmap of the scaled fractions of PBL expressing different phenotypic markers (right). The dendogram was cut (dotted line) into 2 clusters (blue and green). P values according to Mann–Whitney U test are displayed on the left. C, DFS and fraction of tumors with early (1–2, black) or advanced (3–4, gray) nuclear grades according to PBL cluster. P value according to univariate Cox regression analysis and Fisher exact test are displayed, respectively. D, Percentage of patients displaying PBL-A (blue) or PBL-B (green) profile according to TIL cluster. P value according to Fisher exact test is displayed. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

We hypothesized that this interpatient PBL phenotype heterogeneity could be related to the tumor immune contexture and therefore that we could define patient's subgroups based on the PBL phenotype. We thus performed an unsupervised clustering of the patients with ccRCC on the basis of the fractions of CD4+ and CD8+ PBLs expressing activation markers or InR. For this classification, we only included markers significantly upregulated in patients with ccRCC compared with healthy controls. Two major groups were defined by this method and named PBL-A and PBL-B (Fig. 6B). Interestingly, patients classified as PBL-B displayed a shorted DFS (median DFS: 7 months, P = 0.0002, Fig. 6C) and higher tumor nuclear grades (FG3–4 = 67%, P = 0.0004, Fig. 6C) as compared with PBL-A (median DFS: not reached, FG3–4 = 28%).

A detailed analysis of the PBL phenotype in the PBL-B cluster demonstrated an expansion of the CD4+ICOS+ (P = 0.0005), CD4+GITR+ (P = 0.009), CD4+CTLA-4+ (P = 0.003), CD4+PD-1+ (P = 0.03), CD8+ICOS+ (P < 0.0001), CD8+PD-1+ (P = 0.002), and CD8+Tim-3+ (P = 0.0004) populations as compared with PBL-A (Supplementary Figs. S13 and S14). As many of these markers were also expanded in the TIL from immune-regulated tumors, we hypothesized that PBL and TIL clusters could overlap. Interestingly, we found that most of the patients with immune-regulated tumors exhibited a PBL-B cluster (83%, Fig. 6D). On the contrary, the vast majority of immune-silent and immune-activated patients belonged to PBL-A cluster (85% and 86%, respectively, Fisher exact test P = 0.004, Fig. 6D). Interestingly, all the patients with immune-regulated tumors who progressed displayed a PBL-B cluster (4 of 4). In addition, the sole patient with an immune-silent tumor who progressed also displayed a PBL-B cluster.

The exhaustive analysis of the tumor immune microenvironment in retrospective cohorts of human cancers has revealed that the density, phenotype, and localization of TILs can predict both patient's prognosis and clinical response to diverse treatments (4, 3, 5, 26, 27). In this study, we analyzed the TIL and PBL phenotypes in a cohort of 40 patients with localized ccRCC and correlated it with major clinicopathologic characteristics and the risk of early disease progression.

Consistently with previous studies, we found that ccRCC TILs overexpress activation markers and InR as compared with autologous PBL (28, 29). In addition, we report that the TIL phenotypic profile in more than half of the tumors (56%) highly resembles that of autologous RIL, characterized by discreet features of an active immune response. Interestingly, this group of tumors exhibited nonaggressive histologic characteristics and low risk of early progression. ccRCC subgroups with similar immune-silent microenvironments have been described in retrospective cohorts of tumors and have been associated with intermediate patient's prognosis (5, 6, 30). Consistently, some studies support that a fraction of ccRCC represents poorly aggressive neoplasias (31) that may share genetic and phenotypic features with proximal tubule cells (32).

Contrary to immune-silent tumors, the second half of ccRCC exhibited an expansion of TIL-expressing activation markers and InR. Unsupervised clustering methods identified 2 major types of CD8+PD-1+ populations in these tumors and allowed to define 2 subgroups of patients. The first population was characterized by an enhanced expression of Tim-3 (but not Lag-3) and perforin, in addition to an oligoclonal expansion. ccRCC enriched in these cells (roughly 22% of the tumors, here named immune-activated) displayed an increased expression of cytotoxic-associated genes (e.g., PRF1 and GNLY), nonaggressive nuclear grades and low risk of early disease progression. These results suggest that, in this particular scenario, the expression of PD-1 by CD8+ TILs might be induced by tumor–antigen TCR-dependent activation (33), as reported in other cancers (24, 25, 34).

Moreover, we also found a second population of CD8+PD-1+ TILs, which develops in 22% of non-metastatic ccRCC. These cells were characterized by the co-expression of Tim-3 and Lag-3 and exhibited T-cell exhaustion features (35), including diminished cytotoxic potential, polyclonality, and co-expression of at least 3 InR (18). Interestingly, ccRCC enriched in this population (called immune-regulated tumors) exhibited increased expression of inflammatory genes and aggressive nuclear grades and were associated with high risk of early disease progression. All these features suggest that PD-1+ TILs in this subgroup of tumors could be exhausted (35, 36) and provided a biologic basis to the correlation between increased densities of PD-1+ cells and poor prognosis in ccRCC (5, 22), in contrast to other tumors (37, 38). Previous studies using gene expression analyses have described a similar correlation between a poor tumor cell differentiation and the overexpression of immunomodulatory molecules in ccRCCs (13, 30). Interestingly, in our study, we found 2 patients with a nonaggressive nuclear grade (1 or 2) who exhibited an immune-regulated TIL phenotype and high risk of early disease progression, suggesting that the tumor histopathologic features could be insufficient to identify patients with deleterious prognosis.

Our results reveal mechanisms potentially implicated in the development of exhausted CD8+PD-1+Tim-3+Lag-3+ TILs in ccRCC and support that the patients with immune-regulated tumors could benefit from checkpoint blockade in combination with TME-modulating adjuvant treatments and closer clinical follow-up (5, 10, 39). Immune-regulated tumors are enriched in CD4+ TILs exhibiting a Treg phenotype (CD25+CD127Foxp3+), suggesting that this population could be related to the presence of poorly cytotoxic and polyclonal TILs in ccRCC. Consistently with our results, previous studies have described that ccRCC are enriched in Tregs, and their increased densities are associated with poor prognosis (40–42). In addition, we report that a relevant fraction of these cells in ccRCC co-expressed Helios and ICOS, markers that have been related with a highly suppressive potential in mouse models (43, 44) and human cancer (23). As targeted therapies reduce the densities of intratumor CD4+ Tregs (45) and boost the TIL clonal expansion in metastatic ccRCC (46), our data support that this therapeutic approach could be beneficial to patients with immune-regulated tumors. Although with did not demonstrate the suppressive capacity of the ICOS+ TIL population in immune-regulated tumors (due to technical limitations derived from the small size of localized tumors), our data provide a basis for future functional explorations in ccRCC.

A second feature of the ccRCC TME potentially implicated in the development of exhausted TIL is the tumor enrichment in dysfunctional DC and M2 macrophages (35). Here, we describe that immune-regulated ccRCC exhibit high densities of both populations. Consistently, previous reports have found that the infiltration with dysfunctional DC and M2 macrophages in ccRCC is associated with poor clinical outcome (5, 29, 47, 48) and correlates with PD-1+ TIL infiltration (5, 29). Interestingly, it has been reported that dysfunctional DC in ccRCC can by induced by an environment rich in IL8 (overexpressed in immune-regulated tumors; ref. 48). One may hypothesize that these dysfunctional antigen-presenting cells in immune-regulated tumors may express other immunoregulatory molecules (e.g., IL-10 or PD-L1) to promote T-cell exhaustion and/or anergy (49). It is technically challenging to perform multiparametric flow cytometric staining on TILs in the routine clinical practice, and many studies have previously attempted to identify PBL phenotypic biomarkers to identify patients with aggressive histopathologic tumor characteristics in ccRCC (50). To our knowledge, this is the first report correlating the multiparametric expression of InR in PBLs with prognosis in a cohort of ccRCC-bearing patients, highlighting its potential utility to detect patients with clinically aggressive tumors. These results are highly relevant, as the unsupervised classification of patients according to TIL or PBL phenotype independently identified the same group of progressor patients, supporting that both type of analyses could be used as prognostic and theranostic markers.

ccRCC is a highly vascularized neoplasia. Interestingly, in this study, we found that the tumors with the best prognosis were characterized by higher expression of endothelial-associated genes. It would be highly relevant to explore this unexpected correlation that has been previously reported in the literature (13, 30).

In this study, the testing of a statistical hypothesis was not prespecified, and therefore these results must be interpreted with caution and need to be validated in prospective cohorts. However, despite the limitations inherent to this multiparametric flow cytometric study, including the relatively small number of tumors and the short patients' follow-up, the ccRCC TILs, and PBL immune clusters described here provide a basis for future explorations.

To summarize, by unsupervised clustering methods, we identified a group of non-metastatic ccRCC where a highly inflammatory and suppressive TME probably induce dysfunctional in situ T-cell responses and is associated with very high risk of early disease progression after nephrectomy. In view of the diverse phenotypes and functions PD-1+ TIL in ccRCC, we support that other markers, either in the TME (e.g., PD-L1) or in the TIL (e.g., Lag-3, perforin or Treg markers), should be routinely measured to identify patients with deleterious prognosis who probably will benefit from TME-modulating adjuvant treatments and closer clinical follow-up.

W.H. Fridman is a consultant/advisory board member for Adaptimmune, Bristol-Myers Squibb, Curetech, Efranat, Novartis, PFM, and Servier. No potential conflicts of interest were disclosed by the other authors.

Conception and design: N.A. Giraldo, P. Validire, X. Cathelineau, W.H. Fridman, C. Sautès-Fridman

Development of methodology: N.A. Giraldo, L. Lacroix, C. Germain

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N.A. Giraldo, Y. Vano, P. Validire, R. Sanchez-Salas, A. Ingels, A. Moatti, B. Buttard, S. Bourass, X. Cathelineau

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N.A. Giraldo, E. Becht, Y. Vano, F. Petitprez, C. Sautès-Fridman

Writing, review, and/or revision of the manuscript: N.A. Giraldo, E. Becht, Y. Vano, F. Petitprez, R. Sanchez-Salas, S. Oudard, X. Cathelineau, W.H. Fridman, C. Sautès-Fridman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N.A. Giraldo, L. Lacroix, S. Bourass, X. Cathelineau, P. Validire, W.H. Fridman

Study supervision: W.H. Fridman, C. Sautès-Fridman

We would like to thank all members of the teams 10, 13, and 15 in the Cordeliers Research Center (CRC) for their valuable discussions; Estelle Devevre and Helene Fohrer-Ting from the “Centre d'imagerie cellulaire et cytométrie” (CRC) for their technical support; Isabelle Sauret and Simon Lefranc from the “Centre de ressources biologiques”; and the Pathology Department at the Institut Mutualiste Montsouris for their help in the sample storing and collection.

This work was supported by the Institut National de la Santé et de la Recherche Médicale (INSERMU1138T13), the University Paris-Descartes (UPDINSERMU1138T13), the University Pierre et Marie Curie (UPMCINSERMU1138T13), the Institut National du Cancer (2011-1-PLBIO-06-INSERM 6-1, PLBIO09-088-IDF-KROEMER), and CARPEM (T8) and Labex Immuno-Oncology (LAXE62_9UMRS972 FRIDMAN).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Hanahan
D
,
Weinberg
RA
. 
Hallmarks of cancer: the next generation
.
Cell
2011
;
144
:
646
74
.
2.
Giraldo
NA
,
Becht
E
,
Vano
Y
,
Sautès-Fridman
C
,
Fridman
WH
. 
The immune response in cancer: from immunology to pathology to immunotherapy
.
Virchows Arch Int J Pathol
2015
;
467
:
127
35
.
3.
Becht
E
,
Giraldo
NA
,
Germain
C
,
de Reyniès
A
,
Laurent-Puig
P
,
Zucman-Rossi
J
, et al
Immune contexture, immunoscore, and malignant cell molecular subgroups for prognostic and theranostic classifications of cancers
.
Adv Immunol
2016
;
130
:
95
190
.
4.
Fridman
WH
,
Pagès
F
,
Sautès-Fridman
C
,
Galon
J
. 
The immune contexture in human tumours: impact on clinical outcome
.
Nat Rev Cancer
2012
;
12
:
298
306
.
5.
Giraldo
NA
,
Becht
E
,
Pagès
F
,
Skliris
G
,
Verkarre
V
,
Vano
Y
, et al
Orchestration and prognostic significance of immune checkpoints in the microenvironment of primary and metastatic renal cell cancer
.
Clin Cancer Res
2015
;
21
:
3031
40
.
6.
Nakano
O
,
Sato
M
,
Naito
Y
,
Suzuki
K
,
Orikasa
S
,
Aizawa
M
, et al
Proliferative activity of intratumoral CD8(+) T-lymphocytes as a prognostic factor in human renal cell carcinoma: clinicopathologic demonstration of antitumor immunity
.
Cancer Res
2001
;
61
:
5132
6
.
7.
Scott
DW
,
Chan
FC
,
Hong
F
,
Rogic
S
,
Tan
KL
,
Meissner
B
, et al
Gene expression-based model using formalin-fixed paraffin-embedded biopsies predicts overall survival in advanced-stage classical Hodgkin lymphoma
.
J Clin Oncol
2013
;
31
:
692
700
.
8.
Thompson
ED
,
Zahurak
M
,
Murphy
A
,
Cornish
T
,
Cuka
N
,
Abdelfatah
E
, et al
Patterns of PD-L1 expression and CD8 T cell infiltration in gastric adenocarcinomas and associated immune stroma
.
Gut.
2016
Jan 22. [Epub ahead of print].
9.
Taube
JM
,
Klein
A
,
Brahmer
JR
,
Xu
H
,
Pan
X
,
Kim
JH
, et al
Association of PD-1, PD-1 ligands, and other features of the tumor immune microenvironment with response to anti-PD-1 therapy
.
Clin Cancer Res
2014
;
20
:
5064
74
.
10.
Teng
MWL
,
Ngiow
SF
,
Ribas
A
,
Smyth
MJ
. 
Classifying cancers based on T-cell infiltration and PD-L1
.
Cancer Res
2015
;
75
:
2139
45
.
11.
Goc
J
,
Germain
C
,
Vo-Bourgais
TKD
,
Lupo
A
,
Klein
C
,
Knockaert
S
, et al
Dendritic cells in tumor-associated tertiary lymphoid structures signal a Th1 cytotoxic immune contexture and license the positive prognostic value of infiltrating CD8+ T cells
.
Cancer Res
2014
;
74
:
705
15
.
12.
Tokito
T
,
Azuma
K
,
Kawahara
A
,
Ishii
H
,
Yamada
K
,
Matsuo
N
, et al
Predictive relevance of PD-L1 expression combined with CD8+ TIL density in stage III non-small cell lung cancer patients receiving concurrent chemoradiotherapy
.
Eur J Cancer
2016
;
55
:
7
14
.
13.
Becht
E
,
de Reyniès
A
,
Giraldo
NA
,
Pilati
C
,
Buttard
B
,
Lacroix
L
, et al
Immune and stromal classification of colorectal cancer is associated with molecular subtypes and relevant for precision immunotherapy
.
Clin Cancer Res
2016
;
22
:
4057
66
.
14.
Hugo
W
,
Zaretsky
JM
,
Sun
L
,
Song
C
,
Moreno
BH
,
Hu-Lieskovan
S
, et al
Genomic and transcriptomic features of response to anti-PD-1 therapy in metastatic melanoma
.
Cell
2016
;
165
:
35
44
.
15.
Lipson
EJ
,
Forde
PM
,
Hammers
H-J
,
Emens
LA
,
Taube
JM
,
Topalian
SL
. 
Antagonists of PD-1 and PD-L1 in cancer treatment
.
Semin Oncol
2015
;
42
:
587
600
.
16.
Rizvi
NA
,
Hellmann
MD
,
Snyder
A
,
Kvistborg
P
,
Makarov
V
,
Havel
JJ
, et al
Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer
.
Science
2015
;
348
:
124
8
.
17.
Duraiswamy
J
,
Freeman
GJ
,
Coukos
G
. 
Dual blockade of PD-1 and CTLA-4 combined with tumor vaccine effectively restores T-cell rejection function in tumors–response
.
Cancer Res
2014
;
74
:
633
4
;
discussion 635
.
18.
Anderson
AC
,
Joller
N
,
Kuchroo
VK
. 
Lag-3, Tim-3, and TIGIT: co-inhibitory receptors with specialized functions in immune regulation
.
Immunity
2016
;
44
:
989
1004
.
19.
Roederer
M
,
Nozzi
JL
,
Nason
MC
. 
SPICE: exploration and analysis of post-cytometric complex multivariate datasets
.
Cytometry A
2011
;
79A
:
167
74
.
20.
Stewart
JJ
,
Lee
CY
,
Ibrahim
S
,
Watts
P
,
Shlomchik
M
,
Weigert
M
, et al
A Shannon entropy analysis of immunoglobulin and T cell receptor
.
Mol Immunol
1997
;
34
:
1067
82
.
21.
Tibshirani
R
,
Walther
G
,
Hastie
T
. 
Estimating the number of clusters in a data set via the gap statistic
.
J R Stat Soc Ser B Stat Methodol
2001
;
63
:
411
23
.
22.
Thompson
RH
,
Dong
H
,
Lohse
CM
,
Leibovich
BC
,
Blute
ML
,
Cheville
JC
, et al
PD-1 is expressed by tumor-infiltrating immune cells and is associated with poor outcome for patients with renal cell carcinoma
.
Clin Cancer Res
2007
;
13
:
1757
61
.
23.
Faget
J
,
Bendriss-Vermare
N
,
Gobert
M
,
Durand
I
,
Olive
D
,
Biota
C
, et al
ICOS-ligand expression on plasmacytoid dendritic cells supports breast cancer progression by promoting the accumulation of immunosuppressive CD4+ T cells
.
Cancer Res
2012
;
72
:
6130
41
.
24.
Ye
Q
,
Song
D-G
,
Poussin
M
,
Yamamoto
T
,
Best
A
,
Li
C
, et al
CD137 accurately identifies and enriches for naturally occurring tumor-reactive T cells in tumor
.
Clin Cancer Res
2014
;
20
:
44
55
.
25.
Gros
A
,
Robbins
PF
,
Yao
X
,
Li
YF
,
Turcotte
S
,
Tran
E
, et al
PD-1 identifies the patient-specific CD8+ tumor-reactive repertoire infiltrating human tumors
.
J Clin Invest
2014
;
124
:
2246
59
.
26.
Taube
JM
,
Anders
RA
,
Young
GD
,
Xu
H
,
Sharma
R
,
McMiller
TL
, et al
Colocalization of inflammatory response with B7-h1 expression in human melanocytic lesions supports an adaptive resistance mechanism of immune escape
.
Sci Transl Med
2012
;
4
:
127ra37
.
27.
Mlecnik
B
,
Bindea
G
,
Angell
HK
,
Maby
P
,
Angelova
M
,
Tougeron
D
, et al
Integrative analyses of colorectal cancer show immunoscore is a stronger predictor of patient survival than microsatellite instability
.
Immunity
2016
;
44
:
698
711
.
28.
Sittig
SP
,
Køllgaard
T
,
Grønbæk
K
,
Idorn
M
,
Hennenlotter
J
,
Stenzl
A
, et al
Clonal expansion of renal cell carcinoma-infiltrating T lymphocytes
.
Oncoimmunology
2013
;
2
:
e26014
.
29.
Dannenmann
SR
,
Thielicke
J
,
Stöckli
M
,
Matter
C
,
von Boehmer
L
,
Cecconi
V
, et al
Tumor-associated macrophages subvert T-cell function and correlate with reduced survival in clear cell renal cell carcinoma
.
Oncoimmunology
2013
;
2
:
e23562
.
30.
Beuselinck
B
,
Job
S
,
Becht
E
,
Karadimou
A
,
Verkarre
V
,
Couchy
G
, et al
Molecular subtypes of clear cell renal cell carcinoma are associated with sunitinib response in the metastatic setting
.
Clin Cancer Res
2015
;
21
:
1329
39
.
31.
Pierorazio
PM
,
Johnson
MH
,
Ball
MW
,
Gorin
MA
,
Trock
BJ
,
Chang
P
, et al
Five-year analysis of a multi-institutional prospective clinical trial of delayed intervention and surveillance for small renal masses: the DISSRM registry
.
Eur Urol
2015
;
68
:
408
15
.
32.
Skinnider
BF
,
Folpe
AL
,
Hennigar
RA
,
Lim
SD
,
Cohen
C
,
Tamboli
P
, et al
Distribution of cytokeratins and vimentin in adult renal neoplasms and normal renal tissue: potential utility of a cytokeratin antibody panel in the differential diagnosis of renal tumors
.
Am J Surg Pathol
2005
;
29
:
747
54
.
33.
Legat
A
,
Speiser
DE
,
Pircher
H
,
Zehn
D
,
Fuertes Marraco
SA
. 
Inhibitory receptor expression depends more dominantly on differentiation and activation than “Exhaustion” of human CD8 T cells
.
Front Immunol
2013
;
4
:
455
.
34.
Inozume
T
,
Hanada
K-I
,
Wang
QJ
,
Ahmadzadeh
M
,
Wunderlich
JR
,
Rosenberg
SA
, et al
Selection of CD8+PD-1+ lymphocytes in fresh human melanomas enriches for tumor-reactive T cells
.
J Immunother
2010
;
33
:
956
64
.
35.
Wherry
EJ
,
Kurachi
M
. 
Molecular and cellular insights into T cell exhaustion
.
Nat Rev Immunol
2015
;
15
:
486
99
.
36.
Ahmadzadeh
M
,
Johnson
LA
,
Heemskerk
B
,
Wunderlich
JR
,
Dudley
ME
,
White
DE
, et al
Tumor antigen-specific CD8 T cells infiltrating the tumor express high levels of PD-1 and are functionally impaired
.
Blood
2009
;
114
:
1537
44
.
37.
Badoual
C
,
Hans
S
,
Merillon
N
,
Van Ryswick
C
,
Ravel
P
,
Benhamouda
N
, et al
PD-1-expressing tumor-infiltrating T cells are a favorable prognostic biomarker in HPV-associated head and neck cancer
.
Cancer Res
2013
;
73
:
128
38
.
38.
Webb
JR
,
Milne
K
,
Nelson
BH
. 
PD-1 and CD103 are widely coexpressed on prognostically favorable intraepithelial CD8 T cells in human ovarian cancer
.
Cancer Immunol Res
2015
;
3
:
926
35
.
39.
Becht
E
,
Giraldo
NA
,
Dieu-Nosjean
M-C
,
Sautès-Fridman
C
,
Fridman
WH
. 
Cancer immune contexture and immunotherapy
.
Curr Opin Immunol
2016
;
39
:
7
13
.
40.
Griffiths
RW
,
Elkord
E
,
Gilham
DE
,
Ramani
V
,
Clarke
N
,
Stern
PL
, et al
Frequency of regulatory T cells in renal cell carcinoma patients and investigation of correlation with survival
.
Cancer Immunol Immunother CII
2007
;
56
:
1743
53
.
41.
Li
JF
,
Chu
YW
,
Wang
GM
,
Zhu
TY
,
Rong
RM
,
Hou
J
, et al
The prognostic value of peritumoral regulatory T cells and its correlation with intratumoral cyclooxygenase-2 expression in clear cell renal cell carcinoma
.
BJU Int
2009
;
103
:
399
405
.
42.
Kang
MJ
,
Kim
KM
,
Bae
JS
,
Park
HS
,
Lee
H
,
Chung
MJ
, et al
Tumor-infiltrating PD1-positive lymphocytes and FoxP3-positive regulatory T cells predict distant metastatic relapse and survival of clear cell renal cell carcinoma
.
Transl Oncol
2013
;
6
:
282
9
.
43.
Kim
H-J
,
Barnitz
RA
,
Kreslavsky
T
,
Brown
FD
,
Moffett
H
,
Lemieux
ME
, et al
Stable inhibitory activity of regulatory T cells requires the transcription factor Helios
.
Science
2015
;
350
:
334
9
.
44.
Nakagawa
H
,
Sido
JM
,
Reyes
EE
,
Kiers
V
,
Cantor
H
,
Kim
H-J
. 
Instability of Helios-deficient Tregs is associated with conversion to a T-effector phenotype and enhanced antitumor immunity
.
Proc Natl Acad Sci U S A
2016
;
113
:
6248
53
.
45.
Desar
IME
,
Jacobs
JHFM
,
Hulsbergen-vandeKaa
CA
,
Oyen
WJG
,
Mulders
PFA
,
van der Graaf
WTA
, et al
Sorafenib reduces the percentage of tumour infiltrating regulatory T cells in renal cell carcinoma patients
.
Int J Cancer
2011
;
129
:
507
12
.
46.
Gerlinger
M
,
Quezada
SA
,
Peggs
KS
,
Furness
AJ
,
Fisher
R
,
Marafioti
T
, et al
Ultra-deep T cell receptor sequencing reveals the complexity and intratumour heterogeneity of T cell clones in renal cell carcinomas: ultra-deep sequencing of T cell repertoires in renal cancer
.
J Pathol
2013
;
231
:
424
32
.
47.
Gigante
M
,
Blasi
A
,
Loverre
A
,
Mancini
V
,
Battaglia
M
,
Selvaggi
FP
, et al
Dysfunctional DC subsets in RCC patients: ex vivo correction to yield an effective anti-cancer vaccine
.
Mol Immunol
2009
;
46
:
893
901
.
48.
Figel
A-M
,
Brech
D
,
Prinz
PU
,
Lettenmeyer
UK
,
Eckl
J
,
Turqueti-Neves
A
, et al
Human renal cell carcinoma induces a dendritic cell subset that uses T-cell crosstalk for tumor-permissive milieu alterations
.
Am J Pathol
2011
;
179
:
436
51
.
49.
Palucka
K
,
Banchereau
J
. 
Cancer immunotherapy via dendritic cells
.
Nat Rev Cancer
2012
;
12
:
265
77
.
50.
Kim
C-S
,
Kim
Y
,
Kwon
T
,
Yoon
JH
,
Kim
KH
,
You
D
, et al
Regulatory T cells and TGF-β1 in clinically localized renal cell carcinoma: comparison with age-matched healthy controls
.
Urol Oncol
2015
;
33
:
113
.
e19
25
.

Supplementary data