immune and stromal classiﬁcation of colorectal ... › content › clincanres › 22 › 16 ›...

Personalized Medicine and Imaging

Immune and Stromal Classification of ColorectalCancer Is AssociatedwithMolecular Subtypes andRelevant for Precision ImmunotherapyEtienne Becht1,2,3, Aur�elien de Reyni�es4, Nicolas A. Giraldo1,2,3, Camilla Pilati2,5,B�en�edicte Buttard1,2,3, Laetitia Lacroix1,2,3, Janick Selves6,7, Catherine Saut�es-Fridman1,2,3,Pierre Laurent-Puig2,5, and Wolf Herman Fridman1,2,3

Abstract

Purpose: The tumor microenvironment is formed by manydistinct and interacting cell populations, and its compositionmaypredict patients' prognosis and response to therapies. Colorectalcancer is a heterogeneous disease inwhich immune classificationsand four consensus molecular subgroups (CMS) have beendescribed. Our aimwas to integrate the composition of the tumormicroenvironment with the consensusmolecular classification ofcolorectal cancer.

Experimental Design: We retrospectively analyzed the com-position and the functional orientation of the immune, fibro-blastic, and angiogenic microenvironment of 1,388 colorectalcancer tumors from three independent cohorts using transcrip-tomics. We validated our findings using immunohistochemistry.

Results: We report that colorectal cancer molecular subgroupsand microenvironmental signatures are highly correlated. Out ofthe four molecular subgroups, two highly express immune-spe-cific genes. The good-prognosis microsatellite instable–enriched

subgroup (CMS1) is characterized by overexpression of genesspecific to cytotoxic lymphocytes. In contrast, the poor-prognosismesenchymal subgroup (CMS4) expresses markers of lympho-cytes and of cells of monocytic origin. The mesenchymal sub-group also displays an angiogenic, inflammatory, and immuno-suppressive signature, a coordinated pattern that we also found inbreast (n¼ 254), ovarian (n¼ 97), lung (n¼ 80), and kidney (n¼143) cancers. Pathologic examination revealed that the mesen-chymal subtype is characterized by a high density of fibroblaststhat likely produce the chemokines and cytokines that favortumor-associated inflammation and support angiogenesis, result-ing in a poor prognosis. In contrast, the canonical (CMS2) andmetabolic (CMS3) subtypes with intermediate prognosis exhibitlow immune and inflammatory signatures.

Conclusions: The distinct immune orientations of the colo-rectal cancer molecular subtypes pave the way for tailored immu-notherapies. Clin Cancer Res; 22(16); 4057–66. �2016 AACR.

IntroductionCancers are generally classified according to their localization,

histology, and the genomic alterations of themalignant cells, suchas chromosomal rearrangements or DNA mutations. Prognosishas been based essentially on tumor extension categorized by theTNM staging method which incorporates local tumor spread (T)

and distant lymph node (N) and organ (M) metastases (1).Cancer therapies are proposed based on these classificationsincluding conventional chemotherapies in advanced cancers andpersonalized therapies targeting products of mutated genes orrearranged genes. Mutational analyses also unveiled unexpectedfindings, such as the resistance of patients with colorectalcancer exhibiting KRAS mutation to treatment with cetuximab,an anti-EGF receptor antibody (2). These classifications have beencomplemented by high-throughput transcriptome analyses thatidentified dominant oncogenic pathways and established prog-nostic subtypes, as in diffuse large B-cell lymphoma (3), breastcancer (4, 5) or clear-cell renal cell carcinoma (6).

In the last decade, immune classification of cancers has shednew light in patients' care providing prognostic (7) and predictivefactors for chemotherapies (8) and immunotherapies, such asimmune checkpoint inhibitors (9). Colorectal cancer has been aparadigmatic tumor for immune classifications. Our laboratoryhas demonstrated that patients whose tumors are highly infiltrat-ed bymemory T cells, particularly cytotoxic CD8þ T lymphocytes,had a longer progression-free survival (PFS) and overall survival(OS; refs. 10–14). We have hypothesized that tumor-associatedantigens could locally induce antitumor adaptive immuneresponses and have characterized tertiary lymphoid structures(TLS), adjacent to the tumor nests, that could be sites whereantitumor immunity is generated (15). Indeed, we found that

1INSERMUMR_S 1138, Cancer, Immune Control and Escape, CordeliersResearch Centre, Paris, France. 2Universit�e Paris Descartes, Paris,France. 3Universit�e Pierre et Marie Curie, Paris, France. 4ProgrammeCartesd'Identit�edesTumeurs, LigueNationaleContre leCancer, Paris,France. 5INSERM, UMR_S1147, Paris, France. 6Centre de Recherche enCanc�erologie de Toulouse, Unit�e Mixte de Recherche, 1037 INSERM -Universit�e Toulouse III, Toulouse, France. 7Department of Pathology,Centre Hospitalier Universitaire de Toulouse, Toulouse, France.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

A. de Reyni�es and W.H. Fridman jointly directed this work with their respectiveexpertise.

Corresponding Author:W.H. Fridman, INSERM UMRS 1138, 15 rue de l0Ecole deM�edecine, 75006 Paris, France. Phone: þ33144279102; Fax: þ33140510420;E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-15-2879

�2016 American Association for Cancer Research.

ClinicalCancerResearch

www.aacrjournals.org 4057

on July 8, 2020. © 2016 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 18, 2016; DOI: 10.1158/1078-0432.CCR-15-2879

http://clincancerres.aacrjournals.org/

high T- and B-cell infiltration and a high expression of genescoding for lymphocyte-attracting chemokines, i.e., CX3CL1,CXCL9, and CXCL10 for T cells (13) and CXCL13 for B cells(14), as well as genes involved in a Th1 orientation (IFNG andTBX21) and cytotoxicity (GZMB andGNLY; ref. 10), are associatedwith favorable prognosis (10, 12, 14).MSI tumors, with their highmutational load and high leukocyte infiltration, fall perfectly inthis category. It has recently been reported that metastatic colo-rectal cancer tumors with this phenotype responded to treatmentswith PD-1 immune checkpoint–blocking antibodies that increasethe local immune reaction, potentially against tumor-associatedantigens (16, 17).

A very recent publication proposed a transcriptomic classi-fication of colorectal cancer into four consensus molecularsubtypes (CMS; ref. 18). CMS1, called MSI-like, contains mostmicrosatellite instable (MSI) tumors with mutations in genesencoding DNA mismatch-repair proteins, resulting in highmutational burden. The MSI-like subtype is also enriched fortumors with a CpG-island methylator phenotype (CIMP) andmutations in the BRAF oncogene. CMS2, called canonical, is asubtype with high chromosomal instability (CIN) as well asactivation of the Wnt and MYC pathways. CMS3, calledmetabolic, is enriched in tumors with KRAS mutations andshows a disruption of metabolic pathways. Finally, CMS4,called mesenchymal, has a mesenchymal phenotype and fre-quent CIMP phenotype. This classification stratifies colorectalcancer into intrinsic subtypes with different prognosis (18). Ithas been independently well established that the compositionof the microenvironment in which the malignant cells growand expand is essential for predicting patient's prognosis (10,7) and can be a target for cancer therapies (19).

In the era of targeted therapies, particularly immunothera-pies that are dependent on the composition of the tumormicroenvironment, we undertook to integrate molecular andimmune classifications of colorectal cancer, by addressing the

question of the immune, inflammatory, angiogenic, and fibro-blastic composition of colorectal cancer molecular subtypes.We thus quantified these components in a discovery cohort of458 CMS-classified colorectal cancer tumors (CIT cohort). Itwas validated in two independent validation cohorts of 404(CIT validation cohort) and 526 (PETACC3 cohort) tumors.For this purpose, we applied the MCP-counter algorithm, acomputational method able to infer the abundance of nineimmune and two other stromal cell populations from a tran-scriptomic sample. Using this method, we quantified immuneand stromal infiltration of the four CMS subtypes of colorectalcancer and found a significant correlation with CMS subtypes,validated the predicted infiltration profiles using immunohis-tochemistry, and discussed immunotherapeutic approachesthat could benefit each subtype.

Materials and MethodsPublic transcriptomic datasets

The complete lists of selected gene expression profiles (GEP),related type, and experimental conditions are given in Supple-mentary Tables S1, S2, and S3.

Colorectal tumors samples and subtypes annotationsThe GEP from 1,750 colorectal tumor samples were collected.

The GSE39582 dataset (fresh-frozen samples; Affymetrix HG-U133Plus2.0; n ¼ 566) was used as a discovery cohort (hereintermed CIT discovery). Samples from series GSE13067 (n ¼ 74),GSE13294 (n¼ 155), GSE17536 (n¼ 177), and GSE33113 (n¼90) were aggregated as a validation meta-series (herein termedCIT validation; fresh-frozen samples; Affymetrix HG-U133Plus2.0; n ¼ 496). Samples from the PETACC3 (ArrayEx-press:E-MTAB-990) series (n ¼ 688, formalin-fixed, paraffin-embedded samples, custom Affymetrix microarrays) were usedto validate the nondependency of the results on microarraytechnology and sample processing. The CMS subtype annotationof all tumors analyzed was provided by the Colorectal CancerSubtyping Consortium (CRCSC). CMS-unclassified samplesreduced the numbers of samples analyzed to 458 for the CITdiscovery cohort (81% classified), 404 for the CIT validationcohort (81% classified), and 526 for the PETACC3 cohort(76% classified). The total number of colorectal cancer tumorsanalyzed was therefore 1,388.

Multi-cancers datasetThe GEP of breast (n¼ 254), colorectal (n¼ 173), kidney (n¼

144), ovarian (n ¼ 97), lung (n ¼ 80), and endometrial (n ¼ 69)were retrieved from expO dataset (GEO:GSE2109).

Microenvironment-purified cellsWe screened the GEO database (20) for GEP of purified

samples of human immune cells, fibroblasts, and endothelialcells hybridized on Affymetrix HG-U133Plus2.0 microarrays. Wecollected 1,194 GEP from 80 series, including 1,114 immune, 36endothelial, and 50 fibroblast samples.

Colorectal tumor cell linesThe Affymetrix HG-U133Plus2.0 GEP from the 55 colorectal

tumor cell lines from the Cancer Cell Lines Encyclopedia (21)series were selected as tumor controls.

Translational Relevance

Targeted therapies have highly improved the treatment ofcolorectal cancer, such as cetuximab in patients with KRASwild-type tumors. Recently, immunotherapy using an antic-heckpoint antibody, anti–PD-1, has shown strikingly positiveeffects in patients with microsatellite instable (MSI) tumors.To guide future targeted immunotherapies, it is thereforeessential to integrate molecular and immune classificationsof colorectal cancer. We analyze herein the immune, inflam-matory, angiogenic, and fibroblastic landscape of molecularlydefined colorectal cancer subtypes and identify, in addition tothe MSI-like subgroup with a Th1/cytotoxic orientation,tumors with high lymphocyte and stromal infiltration, sug-gesting that the corresponding patients could be treated by acombination of antiangiogenic, anti-inflammatory, and anti-checkpoint agents. In addition, we show that canonical andmetabolic colorectal cancer subtypes are immune down, sug-gesting the use of adoptive T-cell therapies. Our data thereforeprovide immune and molecular subtype integration withprognostic impact and pave the way for personalized colorec-tal cancer immunotherapy.

Becht et al.

Clin Cancer Res; 22(16) August 15, 2016 Clinical Cancer Research4058




ImmunohistochemistrySerial 5-mm formalin-fixed paraffin-embedded tissue sections

from colorectal cancer were stained using autostainerPlus Link 48(Dako). Antigen retrieval and deparaffinization were carried outon a PT-Link (Dako) using the EnVision FLEX Target RetrievalSolutions (Dako). The antibodies used are listed in Supplemen-tary Table S4. Peroxidase activity was detected using diamino-benzidine substrate (Dako). Slides stained with anti-CD8 andanti-CD68 were digitalized with a NanoZoomer scanner (Hama-matsu) and digitally quantified with Calopix software (Tribvn).The degree of smoothmuscle actin (SMA) expression in the tumorstroma was quantified according to the following grading system:(1) scarce fibroblasts; (2) continuous layer of fibroblast betweentumor nests with overall thickness inferior to three cells; (3)continuous layer of fibroblast between tumor nests with overallthickness superior to three cells andfibroblast area<50%of tumorarea; and (4) continuous layer of fibroblast between tumor nestswith overall thickness superior to three cells and fibroblastarea >50% of tumor area.

Microarrays analysisGEP normalization. The GEP from microenvironment-purifiedcells and pan-cancers cell lines were normalized independentlyfor each series, using the frozen RobustMultiarray Average (RMA)method (21a) on each independent series ("fRMA" R package).The RMA-normalized GEP from the CIT colorectal cancer discov-ery series were downloaded directly from GEO. The GEP fromPETACC3 colorectal cancer series were normalized in batch usingthe RMAmethod ('affy' R package). Each GEP series from the CITcolorectal cancer validationmeta-series was normalized indepen-dently using frozen RMA method; then the corresponding matri-ces were combined into one matrix, further normalized withCombat method (22), using series' identifiers as batch variablesand no covariates. The GEP from RNA mixture models werenormalized using the RMA method. When mapping probesetsto HUGO Gene Symbols, the mean across probesets was chosento represent the gene's expression level.

Use of the MCP-counter algorithm. MCP-counter is an algorithmthat aims to estimate samples' infiltration by various immune andother stromal cell populations using transcriptomic data. Itsoutput can be used to compare cellularly heterogeneous samplesfor their relative infiltration by eleven cell populations. MCP-counter relies on the identification of so-called "Transcriptomicmarkers," which are transcripts specifically expressed by a givencell population and not by the others. These transcriptomicmarkers have been identified on a discovery series of 1,939curated gene expression profiles representing cell populationspresent in the tumor microenvironment. The specificity of theirexpression for the corresponding cell populationwas validated ontwo independent series of 1,596 and3,208 samples.MCP-counterscores summarize the expression of the transcriptomic markersspecific for a given population, and have been validated for theirability to correlate with the fraction ofmRNAoriginating from thecorresponding cell population and to cell infiltration estimatedbyimmunohistochemistry (Becht et al.; submitted for publication).

We applied the R (version 3.1.3) package "MCPcounter" ver-sion 0.1.0 on each normalized tumor GEP, using probesets asidentifiers for the CIT, CIT validation and multi-cancer cohorts,and gene symbols for the PETACC3 cohort. The MCP-counterdesign and workflow are illustrated in Supplementary Fig. S1.

MCPcounter version 0.1.0 is available at http://cit.ligue-cancer.net/?page_id¼1243&lang¼en.

Supervised tests of differential expression.ANOVA testswere used toassess the dependency of genes or MCP-counter scores to themolecular subgroups classification. Student t tests were used toinvestigate differential expression of genes between subgroups orcell line phenotypes. To test for differential level of the MCP-counter scores in a given molecular subgroup, Student t testsagainst the cohort's medianMCP-counter score were used. To testfor differential level of MCP-counter scores between molecularsubgroups, pairwise one-tailed t tests with Bonferroni correctionwere used (Supplementary Table S5).

List of immune-related genes. We curated a list of genes known toencode proteins with immunomodulatory functions (Supple-mentary Table S6). Representatives of the chemokines, chemo-kine receptors, interleukins, interleukins receptors, TNF and TNFreceptors, growth factors, interferons and interferon receptors,inhibitory receptors and their ligands, TLR, and class I MHC genefamilies were included.

Results1-colorectal cancer molecular subgroups show distinctexpression patterns of immune and stromal signatures

We used the MCP-counter algorithm to obtain estimates oftumor infiltration by immune and other stromal cell populations.A brief presentation of MCP-counter design and workflow isprovided in Supplementary Fig. S1. In the CIT and CIT validationcohorts, this algorithm revealed that the molecular subgroupsshowed consistent and distinct patterns for the immune, endo-thelial, and fibroblastic cell populations' abundance estimates(Fig 1A). Tumors of theMSI-like andmesenchymal subtypes had ahigh expression of lymphoid (Fig 1A and B) as well as myeloidcell–specific genes (Fig 1A and C), thus exhibiting a strongimmune and inflammatory contexture, whereas tumors of thecanonical and metabolic subtypes had low expression of thelymphocytic and myeloid signatures. Tumors of the MSI-like andmesenchymal subtypes differed in that MSI-like samples exhib-ited a higher cytotoxic-cells abundance estimate, reflecting highinfiltration by activated CD8þ and natural killer (NK) cells.Granulocyte-specific transcripts were poorly discriminative (Fig1A and C). In addition, mesenchymal samples exhibited a highexpression of the fibroblastic and endothelial cell abundanceestimates, compatible with highly vascularized and inflammatorytumors that have a high density of cancer-associated fibroblasts(CAF) in their microenvironments (Fig 1A and D).

The immune infiltrations in the four subtypes predicted byMCP-counter were confirmed using immunohistochemical anal-yses in a subset of 38 randomly selected tumors from the CITdiscovery cohort. CD8þ T cells and CD68þ macrophages werequantified within the tumor center. These analyses showed asignificant correlation between the density of CD8þ cells in thetumor and the cytotoxic abundance estimate from transcriptomicanalyses (P¼ 2.10�5; r¼ 0.67) andbetweenCD68þmacrophagesand the monocytic-lineage abundance estimates (P¼ 1.10�5; r¼0.68). We confirmed that the MSI-like and the mesenchymal-likesubgroups had higher densities of CD8 T cells and CD68 macro-phages than the canonical andmetabolic subtypes, validating thetranscriptomic predictions (Fig 2A and B). In addition, we

Distinct Immune Phenotypes of Colorectal Cancer Molecular Subtypes

www.aacrjournals.org Clin Cancer Res; 22(16) August 15, 2016 4059




performed SMA immunohistochemical labeling. The mesenchy-mal subtype had the highest SMAgrading, supporting the fact thatthe transcriptomic fibroblastic signature was reflecting a highpresence of CAF (Fig 2C and D).

Having analyzed patterns related to microenvironment cellpopulations, we focused on features related to immune cellsfunction andmigration.We thus analyzed the expression of genesencodingmolecules involved in T-cell chemotaxis, activation andinhibition, inflammation and complement components, angio-genesis as well as MHC1 molecules (Fig 3; Supplementary TableS6). The four consensus molecular subgroups again showedstrikingly reproducible data across the two independent cohorts.The MSI-like subtype exhibited a high expression of genes codingfor T-cell–attracting chemokines [CXCL9 (13), CXCL10 (13), andCXCL16] or involved in the formation of tumor-adjacent tertiarylympho€�d structures (CXCL13; refs. 23, 24), as well as the Th1cytokines IFNG and IL15, all of which have been shown tocorrelate with good prognosis in colorectal cancer (10, 12, 13,14). In contrast, the mesenchymal subtype exhibited a highexpression of the myeloid chemokine CCL2, complement com-ponents (C1QA, C1QB, C1QC, C1R, C1S, C3, C3AR1, C5AR1,

C7, CFD, CFH, and CFI), angiogenic factors (VEGFB, VEGFC, andPDGFC), and immunosuppressive molecules [TGFB1, TGFB3,LGALS1 (25), and CXCL12]. CD274 and PDCD1LG2, the genesencoding the PD-1 ligands, were highly expressed in MSI-liketumors but also in some tumors of the mesenchymal group.Strikingly, MHC1 genes, whose products present peptides toCD8þ T cells, were poorly expressed in the poorly infiltratedcanonical subtype.

We were able to reproduce these results on an independentcohort, called "PETACC3," of 526 colorectal cancer samples,whose RNA was extracted from paraffin-embedded tissues andhybridized on another microarray platform, indicating strongreproducibility (Fig. 4).

2-mesenchymal cells induce an inflammatory and angiogenictumor microenvironment

The poor-prognostic CMS4 colorectal cancer subgroup is char-acterized by a high fibroblastic MCP-counter score, an estimate ofcellular abundance, as well as a high expression of the myeloidand endothelial cells scores. We found that the fibroblastic scorehighly correlated with the endothelial (P < 10�15 on the three

Figure 1.Immune and stromal signatures of the four molecular subgroups of colorectal cancer. A, heatmap showing the level of the of the nine immune and two otherstromal MCP-counter abundance estimates among two transcriptomic cohorts of colorectal cancer patients that were classified in four molecularly definedcolorectal cancer subgroups. Distributions of the (B) lymphocytic, (C) myeloid, and (D) stromal abundance estimates across subgroups in the two cohorts.� , P < 0.05; �� , P < 0.001; and �� , P < 0.0001 compared with the cohort's median using a Student t test.

Becht et al.





cohorts, Pearson r ¼ 0.84, 0.84, 0.82 for CIT, CIT validation, andPETACC3, respectively) andmyeloid ones (P < 10�15 on the threecohorts, Pearson r ¼ 0.6, 0.46, 0.46 for CIT, CIT validation, andPETACC3, respectively; Fig 5A). In contrast, there was no corre-lation between the fibroblastic and cytotoxic cells abundanceestimates (Fig. 5A). Correlations between the fibroblastic scoreand both the endothelial and myeloid cell scores were alsoobserved in breast, lung, and ovary cancers, and confirmed incolorectal cancer (Fig 5B), suggesting that the immune contexturefound in mesenchymal colorectal cancer tumors also exists inthese cancers. In kidney cancer, the correlation between thefibroblast and the myeloid scores was weaker, and it was absentin endometrium cancer (Fig 5B).

The coordination of these cell population abundance estimatesacross human tumors led us to hypothesize that fibroblasts

promote angiogenesis and inflammatory cells recruitment in themesenchymal colorectal cancer tumors' microenvironment.Because tumor samples correspond to a mixture of tumor cellsand microenvironment cells, transcriptomic samples of pure cellpopulations were used to investigate the cellular origin of theinflammatory and angiogenic signatures of the mesenchymalmolecular subgroup (Supplementary Tables S1 and S3). We firstidentified the genes upregulated in the mesenchymal subtypecompared with each of the other subtypes (Student t tests againsteachof the other three subtypes, allP<0.05; Supplementary TableS6). We then investigated the expression of these genes byimmune, stromal, and malignant cells (Fig. 6). B, T, and NKlymphocytes, as well as colorectal cancer cell lines, each over-expressed only a small subset of these genes. Fibroblasts had thehighest expression for the proangiogenic factors VEGFB, VEGFC,

Figure 2.Immunohistochemical characterization of the four colorectal cancer subgroups. A, distributions of the densities of tumor-infiltrating CD8þ T cells in the foursubgroups. B, distributions of the densities of tumor-infiltrating CD68þ macrophages in the four subgroups. P values were assessed using the Kruskal–Wallistest.C, representative tumor areas of each SMA-staining grades. SMA-positive areas are labeled in brown.D, distributions of each SMA grades in the four subgroups.P values were assessed using the Fisher exact test.






and PDGFC, the immunosuppressive factors LGALS1, CXCL12,PTGS1, and TGFB3, and the complement components C1S, C1R,CFH, C7, and CFHR2, and can thus promote angiogenesis andimmunosuppression. Endothelial cells had the highest expressionof the myeloid chemoattractant CCL2, the angiogenic factorPDGFB, and immunosuppressive molecules TGFB1 and TGFB2.Finally, monocytic cells expressed complement components(C1QA, C1QC, C3, C3AR1, and C5AR1) and chemokines attract-ing macrophages (CCL19 and CCL23). These results suggest thatthese three cell populations could foster inflammation, angio-genesis, and immunosuppression in mesenchymal colorectalcancer tumors.

DiscussionIn the last decade, the interplay between tumors and the

immune system has emerged as a critical aspect of tumor biologyand is strongly associated with the host ability to control tumorgrowth and to respond to therapies. Incorporating preciseimmune-related information in descriptive cancer-classification

studies or in prospective clinical trials is therefore critical. In thepresent work, we apply the MCP-counter method to study theheterogeneity of the immune, inflammatory, angiogenic, andfibroblastic tumormicroenvironment of the consensusmolecularclassification of colorectal cancer.

The expression of the immune cell abundance estimates,enriched by the analysis of a large array of functionally relevantgenes, in three colorectal cancer cohorts stratified using a recentlypublished molecular classification revealed a strong associationbetween the tumor cell phenotype and both the composition andthe functional orientation of its immune microenvironment.Notably, we demonstrate that mesenchymal tumors are associ-ated with a proinflammatory, proangiogenic, and immunosup-pressive microenvironment.

In the three cohorts, two subgroups were characterized by highexpression of immune signatures: the expected MSI-rich CMS1group and the unexpected mesenchymal CMS4 group. Strikingly,although the MSI-like group correlated with favorable patient'sprognostic in terms of RFS (18), the mesenchymal subgroup ofpatients had the worst prognosis of the four subgroups (18). We

Figure 3.Expression of functionally relevant immune genes among the four subgroups in the two cohorts. The heatmaps on the left represent the level of expressionof the genes. Rows were centered and scaled. Red denotes a higher expression, and blue a lower expression. The heatmaps on the right represent the P value of aStudent t test against the cohort median, for each gene.

Becht et al.





describe for the first time a group of colorectal cancer tumors withhigh lymphoid gene expression associated with poor prognosisfor the patients. This subgroup is characterized by an extensivetumor infiltration by CAF (Fig. 1A and D and Fig. 2C and D),

correlating with high angiogenesis and myeloid cells infiltration(Fig. 1A and D, and Fig. 5A). Recent studies reported an extensiveinfiltration by stromal fibroblasts in mesenchymal colorectalcancer tumors (26, 27). These studies additionally showed that

Figure 4.Results are reproducible on the independent PETACC3 colorectal cancer cohort (526 CMS-classified samples). A, heatmap showing the level of theabundance estimates of the immune and stromal signatures in the PETACC3 colorectal cancer transcriptomic cohort that was classified according to the fourmolecularly defined colorectal cancer subgroups. Distributions of the (B) lymphocytic, (C) myeloid, and (D) stromal abundance estimates across subgroupsin the two cohorts. �, P < 0.05; ��, P < 0.001; and �� , P < 0.0001 compared with the cohort's median using a Student t test. E, the heatmap on the left representsthe level of expression of the genes. Rows were centered and scaled. Red denotes a higher expression, and blue a lower expression. The heatmap on theright represents the P value of a Student t test against the cohort median, for each gene.

Figure 5.The fibroblast abundance estimate correlates with endothelial and myeloid cells abundance estimates in colorectal cancer and other cancers. Scatterplotsrepresenting the relationshipsbetween the endothelial cells,myeloid cells, and cytotoxic cellsMCP-counter scores (cellular abundance estimates) comparedwith thefibroblast MCP-counter score (A) in the two colorectal cancer cohorts (B) across six cancers, including colorectal cancer, in the expO dataset.






the presence of stromal fibroblasts and TGFß signaling enhancedtumors' metastatic capacities and abilities to form xenografts inmice. Our results suggest that in addition to this prometastaticeffect, CAF promote inflammation and angiogenesis in mesen-chymal colorectal cancer tumors. We hypothesize that this stronginflammatory component hampers the positive value of the Th1/CD8þ T cells in these tumors, by repressing the antitumor activityof cytotoxic T cells while fueling tumor growth, angiogenesis, andstroma remodeling. These findings are reminiscent of our recentobservations in clear-cell renal cell carcinoma, a paradigmatictumor for high angiogenesis (28), in which we identified a poor-prognosis subgroup with high CD8þ T cells infiltration with highinflammatory and immunosuppressive contexture (6, 29).

The microenvironment of mesenchymal colorectal cancertumors, characterized by high fibroblastic, endothelial, andmyeloid densities, extends to other cancers than colorectal cancer(Fig. 5A). It is thus tempting to postulate that similar immune,inflammatory, and immunosuppressive microenvironmentsmight also be found in these tumors, indicating that similartherapies aimed at modifying the tumor microenvironmentcould be applied to cohorts of cancers of different origins andlocations exhibiting a mesenchymal phenotype. In particular,antiangiogenic treatments and/or inhibitors of LGALS1-encodedprotein (31) should be tested in mesenchymal colorectal cancerand the mesenchymal-like tumors. The mesenchymal subgroupalso exhibits an angiogenic and inflammatory signature which isprobably the consequence of their high fibroblastic infiltration.Angiogenesis and inflammation are intertwined pathways,which both fuel tumor growth through the production of sur-vival and proliferative signals and by favoring blood supply (31).Yet, because the mesenchymal subtype is highly infiltrated byCD8þ T cells, one could expect it to be associated with favorableoutcome (7). However, an extensive number of studies haveshown that inflammatory and angiogenic microenvironmentswere associated with the inhibition of antitumor cytotoxic T-cellimmune responses, notably through the inhibition of the mat-uration of dendritic cells (31). Immature dendritic cells deliverinhibitory secondary signals to T cells upon antigen presenta-tion, inhibiting their activation. MSI-like is the other "immune-high subgroup" of colorectal cancer. This group contains mostpatients harboring MSI tumors, and is known to be associatedwith a good prognosis, and to feature a strong CD8þ T-cellinfiltration. Strikingly, MSI-like is the group featuring the highestexpression of class I MHC genes (Fig. 3; Supplementary TableS6), as well as genes specific for cytotoxic lymphocytes (Fig. 1A;Supplementary Table S5) or attractingmemory T cells (CXCL9 orCXCL10), activating T cells (IFNG), supporting proliferation of Tand NK cells (IL15) and helping in the formation of TLS(CXCL13), where antitumor adaptive immune responses arelikely shaped (ref. 32; Fig. 3; Supplementary Table S6). Highexpression of these genes has been reported to be associated with

Figure 6.Inflammatory, angiogenic, and suppressive molecules overexpressed inmesenchymal tumors are highly expressed by fibroblastic, endothelial, andmonocytic cells. Expression of the genes specifically upregulated inmesenchymal tumors and related to inflammation, angiogenesis,immunosuppression, and immune cell functional orientations, inhomogeneous samples of immune, stromal, or colorectal cancer cell lines(Supplementary Table S1). Black frames indicate that the corresponding cellpopulation has the highest expression of the gene.

Becht et al.





good prognosis in colorectal cancer (7, 10, 13, 14). The cellularcomposition and functional orientation of the MSI-like sub-group thus suggest that it corresponds to tumors with a highImmunoscore (33). CXCL13 and IL15 have been shown to beproduced by the tumor cells (12, 14), whereas IFNG is clearlyproduced by the infiltrating cells. MSI-like is also characterizedby a lower expression of the myeloid and endothelial cellssignatures (Fig. 1A, C, and D) as well as angiogenesis-inducinggenes (Fig. 3; Supplementary Table S6). It is therefore likely thatthe MSI-like subgroup contains highly immunogenic tumors, inthe context of mild inflammation and angiogenesis, whichresults in the generation of antitumor adaptive immuneresponses educated in tumor-adjacent TLS (34). Effector mem-ory CD8 Tþ cells (34) and B cells (35) would then control thegrowth and metastasis in this subgroup (36), as exemplified innon-small cell lung cancer (NSCLC) (37). IFNG produced byinfiltrating T cells is known to induce a phenomenon called"adaptive resistance" by increasing the expression of the inhib-itory checkpoint molecule PD-1 on T cells (38) and of its ligandsCD274 (PD-L1; ref. 38) and PDCD1LG2 (PD-L2; ref. 39) on thetumor cells, which may result in inefficient antitumor T-cellreaction (40). It is striking that MSI-like tumors also show thehighest expression of PD-L1 and PD-L2 genes, followed bymesenchymal tumors (Fig. 3; Supplementary Table S6). Theseresults prompt to treat colorectal cancer MSI-like patients withagents blocking the PD-1/PD-L1 pathway, such as anti–PD-1 andanti–PD-L1. Recent evidence using in-situ immunohistochemicalstaining of immune checkpoints molecules supports the use ofanticheckpoint immunotherapies in patients with MSI tumors(16), and a recent clinical trial showed that patients with MSItumors responded to PD-1 blockade (17). Because MSI-likesubtype is highly enriched for patients with MSI tumors butalso includes a group of microsatellite stable (MSS) patients(41), the use of molecular classifications might help identifyresponders to PD-1 blockade therapies amongMSS patients, andnonresponders among patients with MSI tumors, and we there-fore advocate to investigate themolecular subgroups of anti–PD-1 and anti–PD-L1-treated colorectal cancer patients.

Tumors of the canonical and metabolic subgroups were char-acterized by poor infiltration by immune cells and low class IMHCexpression (Fig. 3; Supplementary Table S6), and are thus mostlikely poorly immunogenic, which may explain the low tumor T-cell infiltration in these subgroups. The use of bispecific antibodiestargeting a tumor-associated antigen (42), such as carcinoembryo-nic antigen (43), could enhance these tumors' immunogenicity.

Because the transcriptomic classification of colorectal cancer isstrongly associated with different immune and stromal contex-tures, the present work paves the way of novel classifications ofcolorectal cancer tumors, based on the relationships between thephenotype of the cancer cell and the corresponding immune andstromal profile of its microenvironment, potentially identifyingthe most appropriate treatments, including antiangiogenic drugsand immunotherapies.

Cell linesThe cell lines correspond to transcriptomes of cell lines down-

loaded fromGEO (dataset GSE36133), andwere not cultivated inthe laboratory.

Disclosure of Potential Conflicts of InterestP. Laurent-Puig is a consultant/advisory board member for Amagen, Astra-

Zeneca, Boehringer-Ingelheim, INTEGRAGEN,Merck-Serono, and Sanofi. W.H.Fridman is a consultant/advisory board member for Curetech, Efranat, PierreFabre Medicament, Sandoz, Sanofi, and Servier. No potential conflicts ofinterest were disclosed by the other authors.

Authors' ContributionsConception and design: E. Becht, A. de Reyni�es, N.A. Giraldo, P. Laurent-Puig,W.H. FridmanDevelopment of methodology: E. Becht, A. de Reyni�es, N.A. Giraldo, L. LacroixAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): E. Becht, N.A. Giraldo, B. Buttard, J. Selves,C. Saut�es-Fridman, P. Laurent-PuigAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): E. Becht, A. de ReynièsWriting, review, and/or revision of the manuscript: E. Becht, A. de Reyni�es,N.A. Giraldo, C. Pilati, J. Selves, C. Saut�es-Fridman, W.H. FridmanAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): L. Lacroix, W.H. FridmanStudy supervision: A. de Reyni�es, W.H. Fridman

AcknowledgmentsThe authors acknowledge members of the Centre d'Imagerie Cellulaire et de

Cytom�etrie "CICC" plateform of the Cordeliers Research Center and the"PlateformeBiopuces et S�equencage" of the IGBMC for their respective technicalexpertise. The efforts of Gene ExpressionOmnibus, arrayExpress, the expressionproject for Oncology and the International Genomics Consortium, and all theteams that shared their GEP results are greatly acknowledged. They also thankIvo Natario for his help with data collection and Lubka Roumenina, GabrielaBindea, Jerome Galon, Bernhard Mlecnik, Estelle Devevre, Audrey Lupo, andMarie-Caroline Dieu-Nosjean for their fruitful discussions.

Grant SupportThis work was supported by the "Institut National de la Sant�e et de la

Recherche M�edicale," the University Paris-Descartes, the University Pierre etMarie Curie, the Institut National du Cancer (2011-1-PLBIO-06-INSERM 6-1),CARPEM (CAncer Research for PErsonalizedMedicine), Labex Immuno-Oncol-ogy (LAXE62_9UMS872 FRIDMAN), the Fondation ARC pour la recherche surle cancer, the Canc�eropole Ile-de-France, Institut National du Cancer (2011-1-PLBIO-06-INSERM 6-1,PLBIO09-088-IDF-KROEMER), the Universidad de losAndes School of Medicine (N.A. Giraldo), and Colciencias (N.A. Giraldo). E.Becht is supported by B3MI doctorate fellowship, and N.A. Giraldo by PPATHdoctorate fellowship.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received December 2, 2015; revised March 5, 2016; accepted March 8, 2016;published OnlineFirst March 18, 2016.

References1. Sobin L, Gospodarowicz M, Wittekind C. Tnm classification of malignant

tumors, 7th ed. Oxford: Wiley-Blackwell; 2009.2. Li�evre A, Bachet J, Le Corre D, Boige V, Landi B, Emile J, et al. Krasmutation

status is predictive of response to cetuximab therapy in colorectal cancer.Cancer Res 2006;66:3992–5.

3. Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, et al.Distinct types of diffuse large b-cell lymphoma identified by gene expres-sion profiling. Nature 2000;403:503–11.

4. Perou CM, Sørlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, et al.Molecular portraits of human breast tumours. Nature 2000;406:747–52.






5. Sørlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, et al. Geneexpression patterns of breast carcinomas distinguish tumor subclasses withclinical implications. Proc Natl Acad Sci USA 2001;98:10869–74.

6. Beuselinck B, Job S, Becht E, Karadimou A, Verkarre V, Couchy G, et al.Molecular subtypesof clearcell renal cell carcinomaarepredictiveof sunitinibresponse in the metastatic setting. Clin Cancer Res 2015;21:1329–39.

7. FridmanWH, Pag�es F, Saut�es-Fridman C, Galon J. The immune contexturein human tumours: Impact on clinical outcome. Nat Rev Cancer2012;12:298–306.

8. Andre F, Pusztai L. Molecular classification of breast cancer: Implicationsfor selection of adjuvant chemotherapy. Nat Clin Pract Oncol 2006;3:621–32.

9. Taube JM,KleinA, Brahmer JR, XuH,PanX,Kim JH, et al. Associationofpd-1, pd-1 ligands, and other features of the tumor immune microenviron-ment with response to anti-pd-1 therapy. Clin Cancer Res . 2014;20:5064–74.

10. Galon J, Costes A, Sanchez-Cabo F, Kirilovsky A, Mlecnik B, Lagorce-Pag�esC, et al. Type, density, and location of immune cells within humancolorectal tumors predict clinical outcome. Science 2006;313:1960–4.

11. Pag�es F, Berger A, Camus M, Sanchez-Cabo F, Costes A, Molidor R, et al.Effector memory T cells, early metastasis, and survival in colorectal cancer.N Engl J Med 2005;353:2654–66.

12. Mlecnik B, Bindea G, Angell HK, SassoMS,Obenauf AC, Fredriksen T, et al.Functional network pipeline reveals genetic determinants associated within situ lymphocyte proliferation and survival of cancer patients. Sci TranslMed 2014;6:228ra37.

13. Mlecnik B, Tosolini M, Charoentong P, Kirilovsky A, Bindea G, Berger A,et al. Biomolecular network reconstruction identifies T-cell homing factorsassociated with survival in colorectal cancer. Gastroenterology 2010;138:1429–40.

14. Bindea G, Mlecnik B, Tosolini M, Kirilovsky A, Waldner M, Obenauf AC,et al. Spatiotemporal dynamics of intratumoral immune cells reveal theimmune landscape in human cancer. Immunity 2013;39:782–95.

15. Dieu-Nosjean M, Goc J, Giraldo NA, Saut�es-Fridman C & Fridman WH.Tertiary lymphoid structures in cancer and beyond. Trends Immunol2014;35:571–80.

16. Llosa NJ, Cruise M, Tam A, Wicks EC, Hechenbleikner EM, Taube JM, et al.The vigorous immune microenvironment of microsatellite instable coloncancer is balanced by multiple counter-inhibitory checkpoints. CancerDiscov 2015;5:43–51.

17. Le DT, Uram JN, Wang H, Bartlett BR, Kemberling H, Eyring AD, et al. Pd-1blockade in tumors with mismatch-repair deficiency. N Engl J Med2015;372:2509–20.

18. Guinney J, Dienstmann R, Wang X, de Reyni�es A, Schlicker A, Soneson C,et al. The consensus molecular subtypes of colorectal cancer. Nat Med2015;21:1350–1356.

19. Chen DS, Mellman I. Oncologymeets immunology: The cancer-immunitycycle. Immunity 2013;39:1–10.

20. Edgar R, Domrachev M, Lash AE. Gene expression omnibus: NCBI geneexpression and hybridization array data repository. Nucleic Acids Res2002;30:207–10.

21. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S,et al. The cancer cell line encyclopedia enables predictive modelling ofanticancer drug sensitivity. Nature 2012;483:603–7.

21a. McCall MN, Bolstad BM, Irizarry RA. Frozen robust multiarray analysis(fRMA). Biostatistics 2010;11(2):242–253.

22. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarrayexpressiondatausing empirical Bayesmethods. Biostatistics 2007;8:118–27.

23. de Chaisemartin L, Goc J, DamotteD, Validire P,Magdeleinat P, AlifanoM,et al. Characterization of chemokines and adhesion molecules associatedwith T cell presence in tertiary lymphoid structures in human lung cancer.Cancer Res 2011;71:6391–9.

24. Gu-Trantien C, Loi S, Garaud S, Equeter C, Libin M, deWind A, et al. Cd4þ

follicular helper T cell infiltration predicts breast cancer survival. J ClinInvest 2013;123:2873–92.

25. Soldati R, Berger E, Zenclussen AC, Jorch G, Lode HN, Salatino M, et al.Neuroblastoma triggers an immunoevasive program involving galectin-1-dependent modulation of T cell and dendritic cell compartments. Int JCancer 2012;131:1131–41.

26. Isella C, Terrasi A, Bellomo SE, Petti C, Galatola G, Muratore A, et al.Stromal contribution to the colorectal cancer transcriptome. Nat Genet2015;47:312–9.

27. Calon A, Lonardo E, Berenguer-Llergo A, Espinet E, Hernando-MomblonaX, Iglesias M, et al. Stromal gene expression defines poor-prognosis sub-types in colorectal cancer. Nat Genet 2015;47:320–9.

28. Cairns P.Renal cell carcinoma. Cancer Biomark 2010;9:461–73.29. Giraldo NA, Becht E, Pag�es F, Skliris G, Verkarre V, Vano Y, et al. Orches-

tration and prognostic significance of immune checkpoints in the micro-environment of primary and metastatic renal cell cancer. Clin Cancer Res2015;21:3031–40.

30. Astorgues-Xerri L, Riveiro ME, Tijeras-Raballand A, Serova M, Neuzillet C,Albert S, et al. Unraveling galectin-1 as a novel therapeutic target for cancer.Cancer Treat Rev 2014;40:307–19.

31. Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation.Nature 2008;454:436–44.

32. Dieu-Nosjean M, Goc J, Giraldo NA, Saut�es-Fridman C, Fridman WH.Tertiary lymphoid structures in cancer and beyond. Trends Immunol2014;35:571–80.

33. Galon J,Mlecnik B, BindeaG, AngellHK, Berger A, Lagorce C, et al. Towardsthe introduction of the 'immunoscore' in the classification of malignanttumours. J Pathol 2014;232:199–209.

34. Camus M, Tosolini M, Mlecnik B, Pag�es F, Kirilovsky A, Berger A, et al.Coordination of intratumoral immune reaction and human colorectalcancer recurrence. Cancer Res 2009;69:2685–93.

35. Germain C, Gnjatic S, Tamzalit F, Knockaert S, Remark R, Goc J, et al.Presence of b cells in tertiary lymphoid structures is associated with aprotective immunity in patients with lung cancer. Am J Respir Crit CareMed 2014;189:832–44.

36. Becht E, Goc J, Germain C, Giraldo NA, Dieu-Nosjean M, Saut�es-FridmanC, et al. Shaping of an effective immune microenvironment to and bycancer cells. Cancer Immunol Immunother 2014;63:991–7.

37. 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 th1cytotoxic immune contexture and license the positive prognostic value ofinfiltrating cd8þ t cells. Cancer Res 2014;74:705–15.

38. Taube JM, Anders RA, Young GD, Xu H, Sharma R, McMiller TL, et al.Colocalization of inflammatory response with b7-h1 expression in humanmelanocytic lesions supports an adaptive resistance mechanism ofimmune escape. Sci Transl Med 2012;4:127ra37.

39. Latchman Y, Wood CR, Chernova T, Chaudhary D, Borde M, Chernova I,et al. Pd-l2 is a second ligand for pd-1 and inhibits t cell activation. NatImmunol 2001;2:261–8.

40. Francisco LM, Sage PT, Sharpe AH. The PD-1 pathway in tolerance andautoimmunity. Immunol Rev 2010;236:219–42.

41. Marisa L, de Reyni�es A, Duval A, Selves J, Gaub MP, Vescovo L, et al. Geneexpression classification of colon cancer into molecular subtypes: charac-terization, validation, and prognostic value. PLoSMed 2013;10:e1001453.

42. Bargou R, Leo E, Zugmaier G, Klinger M, Goebeler M, Knop S, et al. Tumorregression in cancer patients by very lowdoses of a t cell-engaging antibody.Science 2008;321:974–7.

43. Oberst MD, Fuhrmann S, Mulgrew K, Amann M, Cheng L, Lutterbuese P,et al. Cea/cd3 bispecific antibody medi-565/amg 211 activation of t cellsand subsequent killing of human tumors is independent of mutationscommonly found in colorectal adenocarcinomas. MAbs 2014;6:1571–84.


Becht et al.




2016;22:4057-4066. Published OnlineFirst March 18, 2016.Clin Cancer Res Etienne Becht, Aurélien de Reyniès, Nicolas A. Giraldo, et al. ImmunotherapyAssociated with Molecular Subtypes and Relevant for Precision Immune and Stromal Classification of Colorectal Cancer Is

Updated version

10.1158/1078-0432.CCR-15-2879doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2016/03/18/1078-0432.CCR-15-2879.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/22/16/4057.full#ref-list-1

This article cites 43 articles, 13 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/22/16/4057.full#related-urls

This article has been cited by 29 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/22/16/4057To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-15-2879

http://clincancerres.aacrjournals.org/content/suppl/2016/03/18/1078-0432.CCR-15-2879.DC1

http://clincancerres.aacrjournals.org/content/22/16/4057.full#ref-list-1

http://clincancerres.aacrjournals.org/content/22/16/4057.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/22/16/4057


immune and stromal classiﬁcation of colorectal ... › content › clincanres › 22 › 16 ›...

Documents