The local microenvironment drives activation of neutrophils in human brain tumors.

Neutrophils are abundant immune cells in the circulation and frequently infiltrate tumors in substantial numbers. However, their precise functions in different cancer types remain incompletely understood, including in the brain microenvironment. We therefore investigated neutrophils in tumor tissue of glioma and brain metastasis patients, with matched peripheral blood, and herein describe the first in-depth analysis of neutrophil phenotypes and functions in these tissues. Orthogonal profiling strategies in humans and mice revealed that brain tumor-associated neutrophils (TANs) differ significantly from blood neutrophils and have a prolonged lifespan and immune-suppressive and pro-angiogenic capacity. TANs exhibit a distinct inflammatory signature, driven by a combination of soluble inflammatory mediators including tumor necrosis factor alpha (TNF-ɑ) and Ceruloplasmin, which is more pronounced in TANs from brain metastasis versus glioma. Myeloid cells, including tumor-associated macrophages, emerge at the core of this network of pro-inflammatory mediators, supporting the concept of a critical myeloid niche regulating overall immune suppression in human brain tumors.

Genetic editing of colonic organoids provides a molecularly distinct and orthotopic preclinical model of serrated carcinogenesis.

OBJECTIVE: Serrated colorectal cancer (CRC) accounts for approximately 25% of cases and includes tumours that are among the most treatment resistant and with worst outcomes. This CRC subtype is associated with activating mutations in the mitogen-activated kinase pathway gene, BRAF, and epigenetic modifications termed the CpG Island Methylator Phenotype, leading to epigenetic silencing of key tumour suppressor genes. It is still not clear which (epi-)genetic changes are most important in neoplastic progression and we begin to address this knowledge gap herein. DESIGN: We use organoid culture combined with CRISPR/Cas9 genome engineering to sequentially introduce genetic alterations associated with serrated CRC and which regulate the stem cell niche, senescence and DNA mismatch repair. RESULTS: Targeted biallelic gene alterations were verified by DNA sequencing. Organoid growth in the absence of niche factors was assessed, as well as analysis of downstream molecular pathway activity. Orthotopic engraftment of complex organoid lines, but not BrafV600E alone, quickly generated adenocarcinoma in vivo with serrated features consistent with human disease. Loss of the essential DNA mismatch repair enzyme, Mlh1, led to microsatellite instability. Sphingolipid metabolism genes are differentially regulated in both our mouse models of serrated CRC and human CRC, with key members of this pathway having prognostic significance in the human setting. CONCLUSION: We generate rapid, complex models of serrated CRC to determine the contribution of specific genetic alterations to carcinogenesis. Analysis of our models alongside patient data has led to the identification of a potential susceptibility for this tumour type.

Phenotypic diversity of T cells in human primary and metastatic brain tumors revealed by multiomic interrogation.

The immune-specialized environment of the healthy brain is tightly regulated to prevent excessive neuroinflammation. However, after cancer development, a tissue-specific conflict between brain-preserving immune suppression and tumor-directed immune activation may ensue. To interrogate potential roles of T cells in this process, we profiled these cells from individuals with primary or metastatic brain cancers via integrated analyses on the single-cell and bulk population levels. Our analysis revealed similarities and differences in T cell biology between individuals, with the most pronounced differences observed in a subgroup of individuals with brain metastasis, characterized by accumulation of CXCL13-expressing CD39+ potentially tumor-reactive T (pTRT) cells. In this subgroup, high pTRT cell abundance was comparable to that in primary lung cancer, whereas all other brain tumors had low levels, similar to primary breast cancer. These findings indicate that T cell-mediated tumor reactivity can occur in certain brain metastases and may inform stratification for treatment with immunotherapy.

CXCL9:SPP1 macrophage polarity identifies a network of cellular programs that control human cancers.

Tumor microenvironments (TMEs) influence cancer progression but are complex and often differ between patients. Considering that microenvironment variations may reveal rules governing intratumoral cellular programs and disease outcome, we focused on tumor-to-tumor variation to examine 52 head and neck squamous cell carcinomas. We found that macrophage polarity-defined by CXCL9 and SPP1 (CS) expression but not by conventional M1 and M2 markers-had a noticeably strong prognostic association. CS macrophage polarity also identified a highly coordinated network of either pro- or antitumor variables, which involved each tumor-associated cell type and was spatially organized. We extended these findings to other cancer indications. Overall, these results suggest that, despite their complexity, TMEs coordinate coherent responses that control human cancers and for which CS macrophage polarity is a relevant yet simple variable.

Connexins orchestrate progression of breast cancer metastasis to the brain by promoting FAK activation.

Brain metastasis is a complication of increasing incidence in patients with breast cancer at advanced disease stage. It is a severe condition characterized by a rapid decline in quality of life and poor prognosis. There is a critical clinical need to develop effective therapies to prevent and treat brain metastases. Here, we describe a unique and robust spontaneous preclinical model of breast cancer metastasis to the brain (4T1-BM2) in mice that has been instrumental in uncovering molecular mechanisms guiding metastatic dissemination and colonization of the brain. Key experimental findings were validated in the additional murine D2A1-BM2 model and in human MDA231-BrM2 model. Gene expression analyses and functional studies, coupled with clinical transcriptomic and histopathological investigations, identified connexins (Cxs) and focal adhesion kinase (FAK) as master molecules orchestrating breast cancer colonization of the brain. Cx31 promoted homotypic tumor cell adhesion, heterotypic tumor-astrocyte interaction, and FAK phosphorylation. FAK signaling prompted NF-κB activation inducing Lamc2 expression and laminin 332 (laminin 5) deposition, α6 integrin-mediated adhesion, and sustained survival and growth within brain parenchyma. In the MDA231-BrM2 model, the human homologous molecules CX43, LAMA4, and α3 integrin were involved. Systemic treatment with FAK inhibitors reduced brain metastasis progression. In conclusion, we report a spontaneous model of breast cancer metastasis to the brain and identified Cx-mediated FAK-NF-κB signaling as a mechanism promoting cell-autonomous and microenvironmentally controlled cell survival for brain colonization. Considering the limited therapeutic options for brain metastatic disease in cancer patients, we propose FAK as a therapeutic candidate to further pursue in the clinic.

Dynamic and adaptive cancer stem cell population admixture in colorectal neoplasia.

Intestinal homeostasis is underpinned by LGR5+ve crypt-base columnar stem cells (CBCs), but following injury, dedifferentiation results in the emergence of LGR5-ve regenerative stem cell populations (RSCs), characterized by fetal transcriptional profiles. Neoplasia hijacks regenerative signaling, so we assessed the distribution of CBCs and RSCs in mouse and human intestinal tumors. Using combined molecular-morphological analysis, we demonstrate variable expression of stem cell markers across a range of lesions. The degree of CBC-RSC admixture was associated with both epithelial mutation and microenvironmental signaling disruption and could be mapped across disease molecular subtypes. The CBC-RSC equilibrium was adaptive, with a dynamic response to acute selective pressure, and adaptability was associated with chemoresistance. We propose a fitness landscape model where individual tumors have equilibrated stem cell population distributions along a CBC-RSC phenotypic axis. Cellular plasticity is represented by position shift along this axis and is influenced by cell-intrinsic, extrinsic, and therapeutic selective pressures.

Single-cell and bulk transcriptome sequencing identifies two epithelial tumor cell states and refines the consensus molecular classification of colorectal cancer.

The consensus molecular subtype (CMS) classification of colorectal cancer is based on bulk transcriptomics. The underlying epithelial cell diversity remains unclear. We analyzed 373,058 single-cell transcriptomes from 63 patients, focusing on 49,155 epithelial cells. We identified a pervasive genetic and transcriptomic dichotomy of malignant cells, based on distinct gene expression, DNA copy number and gene regulatory network. We recapitulated these subtypes in bulk transcriptomes from 3,614 patients. The two intrinsic subtypes, iCMS2 and iCMS3, refine CMS. iCMS3 comprises microsatellite unstable (MSI-H) cancers and one-third of microsatellite-stable (MSS) tumors. iCMS3 MSS cancers are transcriptomically more similar to MSI-H cancers than to other MSS cancers. CMS4 cancers had either iCMS2 or iCMS3 epithelium; the latter had the worst prognosis. We defined the intrinsic epithelial axis of colorectal cancer and propose a refined 'IMF' classification with five subtypes, combining intrinsic epithelial subtype (I), microsatellite instability status (M) and fibrosis (F).

Detecting epistasis with restricted response patterns in pairs of biallelic loci.

Well-established examples of genetic epistasis between a pair of loci typically show characteristic patterns of phenotypic distributions in joint genotype tables. However, inferring epistasis given such data is difficult due to the lack of power in commonly used approaches, which decompose the epistatic patterns into main plus interaction effects followed by testing the interaction term. Testing additive-only or all terms may have more power, but they are sensitive to nonepistatic patterns. Alternatively, the epistatic patterns of interest can be enumerated and the best matching one is found by searching through the possibilities. Although this approach requires multiple testing correction over possible patterns, each pattern can be fitted with a regression model with just one degree of freedom and thus the overall power can still be high, if the number of possible patterns is limited. Here we compare the power of the linear decomposition and pattern search methods, by applying them to simulated data generated under several patterns of joint genotype effects with simple biological interpretations. Interaction-only tests are the least powerful; while pattern search approach is the most powerful if the range of possibilities is restricted, but still includes the true pattern.

Impact of normalization on miRNA microarray expression profiling.

Profiling miRNA levels in cells with miRNA microarrays is becoming a widely used technique. Although normalization methods for mRNA gene expression arrays are well established, miRNA array normalization has so far not been investigated in detail. In this study we investigate the impact of normalization on data generated with the Agilent miRNA array platform. We have developed a method to select nonchanging miRNAs (invariants) and use them to compute linear regression normalization coefficients or variance stabilizing normalization (VSN) parameters. We compared the invariants normalization to normalization by scaling, quantile, and VSN with default parameters as well as to no normalization using samples with strong differential expression of miRNAs (heart-brain comparison) and samples where only a few miRNAs are affected (by p53 overexpression in squamous carcinoma cells versus control). All normalization methods performed better than no normalization. Normalization procedures based on the set of invariants and quantile were the most robust over all experimental conditions tested. Our method of invariant selection and normalization is not limited to Agilent miRNA arrays and can be applied to other data sets including those from one color miRNA microarray platforms, focused gene expression arrays, and gene expression analysis using quantitative PCR.

BCL-XL is crucial for progression through the adenoma-to-carcinoma sequence of colorectal cancer.

Evasion of apoptosis is a hallmark of cancer, which is frequently mediated by upregulation of the antiapoptotic BCL-2 family proteins. In colorectal cancer (CRC), previous work has highlighted differential antiapoptotic protein dependencies determined by the stage of the disease. While intestinal stem cells (ISCs) require BCL-2 for adenoma outgrowth and survival during transformation, ISC-specific MCL1 deletion results in disturbed intestinal homeostasis, eventually contributing to tumorigenesis. Colon cancer stem cells (CSCs), however, no longer require BCL-2 and depend mainly on BCL-XL for their survival. We therefore hypothesized that a shift in antiapoptotic protein reliance occurs in ISCs as the disease progresses from normal to adenoma to carcinoma. By targeting antiapoptotic proteins with specific BH3 mimetics in organoid models of CRC progression, we found that BCL-2 is essential only during ISC transformation while MCL1 inhibition did not affect adenoma outgrowth. BCL-XL, on the other hand, was crucial for stem cell survival throughout the adenoma-to-carcinoma sequence. Furthermore, we identified that the limited window of BCL-2 reliance is a result of its downregulation by miR-17-5p, a microRNA that is upregulated upon APC-mutation driven transformation. Here we show that BCL-XL inhibition effectively impairs adenoma outgrowth in vivo and enhances the efficacy of chemotherapy. In line with this dependency, expression of BCL-XL, but not BCL-2 or MCL1, directly correlated to the outcome of chemotherapy-treated CRC patients. Our results provide insights to enable the rational use of BH3 mimetics in CRC management, particularly underlining the therapeutic potential of BCL-XL targeting mimetics in both early and late-stage disease.

A single-array preprocessing method for estimating full-resolution raw copy numbers from all Affymetrix genotyping arrays including GenomeWideSNP 5 & 6.

MOTIVATION: High-resolution copy-number (CN) analysis has in recent years gained much attention, not only for the purpose of identifying CN aberrations associated with a certain phenotype, but also for identifying CN polymorphisms. In order for such studies to be successful and cost effective, the statistical methods have to be optimized. We propose a single-array preprocessing method for estimating full-resolution total CNs. It is applicable to all Affymetrix genotyping arrays, including the recent ones that also contain non-polymorphic probes. A reference signal is only needed at the last step when calculating relative CNs. RESULTS: As with our method for earlier generations of arrays, this one controls for allelic crosstalk, probe affinities and PCR fragment-length effects. Additionally, it also corrects for probe sequence effects and co-hybridization of fragments digested by multiple enzymes that takes place on the latest chips. We compare our method with Affymetrix's CN5 method and the dChip method by assessing how well they differentiate between various CN states at the full resolution and various amounts of smoothing. Although CRMA v2 is a single-array method, we observe that it performs as well as or better than alternative methods that use data from all arrays for their preprocessing. This shows that it is possible to do online analysis in large-scale projects where additional arrays are introduced over time.

Molecular risk assessment of BIG 1-98 participants by expression profiling using RNA from archival tissue.

BACKGROUND: The purpose of the work reported here is to test reliable molecular profiles using routinely processed formalin-fixed paraffin-embedded (FFPE) tissues from participants of the clinical trial BIG 1-98 with a median follow-up of 60 months. METHODS: RNA from fresh frozen (FF) and FFPE tumor samples of 82 patients were used for quality control, and independent FFPE tissues of 342 postmenopausal participants of BIG 1-98 with ER-positive cancer were analyzed by measuring prospectively selected genes and computing scores representing the functions of the estrogen receptor (eight genes, ER_8), the progesterone receptor (five genes, PGR_5), Her2 (two genes, HER2_2), and proliferation (ten genes, PRO_10) by quantitative reverse transcription PCR (qRT-PCR) on TaqMan Low Density Arrays. Molecular scores were computed for each category and ER_8, PGR_5, HER2_2, and PRO_10 scores were combined into a RISK_25 score. RESULTS: Pearson correlation coefficients between FF- and FFPE-derived scores were at least 0.94 and high concordance was observed between molecular scores and immunohistochemical data. The HER2_2, PGR_5, PRO_10 and RISK_25 scores were significant predictors of disease free-survival (DFS) in univariate Cox proportional hazard regression. PRO_10 and RISK_25 scores predicted DFS in patients with histological grade II breast cancer and in lymph node positive disease. The PRO_10 and PGR_5 scores were independent predictors of DFS in multivariate Cox regression models incorporating clinical risk indicators; PRO_10 outperformed Ki-67 labeling index in multivariate Cox proportional hazard analyses. CONCLUSIONS: Scores representing the endocrine responsiveness and proliferation status of breast cancers were developed from gene expression analyses based on RNA derived from FFPE tissues. The validation of the molecular scores with tumor samples of participants of the BIG 1-98 trial demonstrates that such scores can serve as independent prognostic factors to estimate disease free survival (DFS) in postmenopausal patients with estrogen receptor positive breast cancer.

Neutrophils suppress tumor-infiltrating T cells in colon cancer via matrix metalloproteinase-mediated activation of TGFß.

High T-cell infiltration in colorectal cancer (CRC) correlates with a favorable disease outcome and immunotherapy response. This, however, is only observed in a small subset of CRC patients. A better understanding of the factors influencing tumor T-cell responses in CRC could inspire novel therapeutic approaches to achieve broader immunotherapy responsiveness. Here, we investigated T cell-suppressive properties of different myeloid cell types in an inducible colon tumor mouse model. The most potent inhibitors of T-cell activity were tumor-infiltrating neutrophils. Gene expression analysis and combined in vitro and in vivo tests indicated that T-cell suppression is mediated by neutrophil-secreted metalloproteinase activation of latent TGFß. CRC patient neutrophils similarly suppressed T cells via TGFß in vitro, and public gene expression datasets suggested that T-cell activity is lowest in CRCs with combined neutrophil infiltration and TGFß activation. Thus, the interaction of neutrophils with a TGFß-rich tumor microenvironment may represent a conserved immunosuppressive mechanism in CRC.

Zeb2 drives invasive and microbiota-dependent colon carcinoma.

Colorectal cancer (CRC) is highly prevalent in Western society, and increasing evidence indicates strong contributions of environmental factors and the intestinal microbiota to CRC initiation, progression and even metastasis. We have identified a synergistic inflammatory tumor-promoting mechanism through which the resident intestinal microbiota boosts invasive CRC development in an epithelial-to-mesenchymal transition-prone tissue environment. Intestinal epithelial cell (IEC)-specific transgenic expression of the epithelial-to-mesenchymal transition regulator Zeb2 in mice (Zeb2IEC-Tg/+) leads to increased intestinal permeability, myeloid cell-driven inflammation and spontaneous invasive CRC development. Zeb2IEC-Tg/+ mice develop a dysplastic colonic epithelium, which progresses to severely inflamed neoplastic lesions while the small intestinal epithelium remains normal. Zeb2IEC-Tg/+ mice are characterized by intestinal dysbiosis, and microbiota depletion with broad-spectrum antibiotics or germ-free rederivation completely prevents cancer development. Zeb2IEC-Tg/+ mice represent the first mouse model of spontaneous microbiota-dependent invasive CRC and will help us to better understand host-microbiome interactions driving CRC development in humans.

Lineage-dependent gene expression programs influence the immune landscape of colorectal cancer.

Immunotherapy for metastatic colorectal cancer is effective only for mismatch repair-deficient tumors with high microsatellite instability that demonstrate immune infiltration, suggesting that tumor cells can determine their immune microenvironment. To understand this cross-talk, we analyzed the transcriptome of 91,103 unsorted single cells from 23 Korean and 6 Belgian patients. Cancer cells displayed transcriptional features reminiscent of normal differentiation programs, and genetic alterations that apparently fostered immunosuppressive microenvironments directed by regulatory T cells, myofibroblasts and myeloid cells. Intercellular network reconstruction supported the association between cancer cell signatures and specific stromal or immune cell populations. Our collective view of the cellular landscape and intercellular interactions in colorectal cancer provide mechanistic information for the design of efficient immuno-oncology treatment strategies.

Selecting control genes for RT-QPCR using public microarray data.

BACKGROUND: Gene expression analysis has emerged as a major biological research area, with real-time quantitative reverse transcription PCR (RT-QPCR) being one of the most accurate and widely used techniques for expression profiling of selected genes. In order to obtain results that are comparable across assays, a stable normalization strategy is required. In general, the normalization of PCR measurements between different samples uses one to several control genes (e.g. housekeeping genes), from which a baseline reference level is constructed. Thus, the choice of the control genes is of utmost importance, yet there is not a generally accepted standard technique for screening a large number of candidates and identifying the best ones. RESULTS: We propose a novel approach for scoring and ranking candidate genes for their suitability as control genes. Our approach relies on publicly available microarray data and allows the combination of multiple data sets originating from different platforms and/or representing different pathologies. The use of microarray data allows the screening of tens of thousands of genes, producing very comprehensive lists of candidates. We also provide two lists of candidate control genes: one which is breast cancer-specific and one with more general applicability. Two genes from the breast cancer list which had not been previously used as control genes are identified and validated by RT-QPCR. Open source R functions are available at http://www.isrec.isb-sib.ch/~vpopovic/research/ CONCLUSION: We proposed a new method for identifying candidate control genes for RT-QPCR which was able to rank thousands of genes according to some predefined suitability criteria and we applied it to the case of breast cancer. We also empirically showed that translating the results from microarray to PCR platform was achievable.

Relationship between estrogen receptor alpha location and gene induction reveals the importance of downstream sites and cofactors.

BACKGROUND: To understand cancer-related modifications to transcriptional programs requires detailed knowledge about the activation of signal-transduction pathways and gene expression programs. To investigate the mechanisms of target gene regulation by human estrogen receptor alpha (hERalpha), we combine extensive location and expression datasets with genomic sequence analysis. In particular, we study the influence of patterns of DNA occupancy by hERalpha on expression phenotypes. RESULTS: We find that strong ChIP-chip sites co-localize with strong hERalpha consensus sites and detect nucleotide bias near hERalpha sites. The localization of ChIP-chip sites relative to annotated genes shows that weak sites are enriched near transcription start sites, while stronger sites show no positional bias. Assessing the relationship between binding configurations and expression phenotypes, we find binding sites downstream of the transcription start site (TSS) to be equally good or better predictors of hERalpha-mediated expression as upstream sites. The study of FOX and SP1 cofactor sites near hERalpha ChIP sites shows that induced genes frequently have FOX or SP1 sites. Finally we integrate these multiple datasets to define a high confidence set of primary hERalpha target genes. CONCLUSION: Our results support the model of long-range interactions of hERalpha with the promoter-bound cofactor SP1 residing at the promoter of hERalpha target genes. FOX motifs co-occur with hERalpha motifs along responsive genes. Importantly we show that the spatial arrangement of sites near the start sites and within the full transcript is important in determining response to estrogen signaling.

Biological processes associated with breast cancer clinical outcome depend on the molecular subtypes.

PURPOSE: Recently, several prognostic gene expression signatures have been identified; however, their performance has never been evaluated according to the previously described molecular subtypes based on the estrogen receptor (ER) and human epidermal growth factor receptor 2 (HER2), and their biological meaning has remained unclear. Here we aimed to perform a comprehensive meta-analysis integrating both clinicopathologic and gene expression data, focusing on the main molecular subtypes. EXPERIMENTAL DESIGN: We developed gene expression modules related to key biological processes in breast cancer such as tumor invasion, immune response, angiogenesis, apoptosis, proliferation, and ER and HER2 signaling, and then analyzed these modules together with clinical variables and several prognostic signatures on publicly available microarray studies (>2,100 patients). RESULTS: Multivariate analysis showed that in the ER+/HER2- subgroup, only the proliferation module and the histologic grade were significantly associated with clinical outcome. In the ER-/HER2- subgroup, only the immune response module was associated with prognosis, whereas in the HER2+ tumors, the tumor invasion and immune response modules displayed significant association with survival. Proliferation was identified as the most important component of several prognostic signatures, and their performance was limited to the ER+/HER2- subgroup. CONCLUSIONS: Although proliferation is the strongest parameter predicting clinical outcome in the ER+/HER2- subtype and the common denominator of most prognostic gene signatures, immune response and tumor invasion seem to be the main molecular processes associated with prognosis in the ER-/HER2- and HER2+ subgroups, respectively. These findings may help to define new clinicogenomic models and to identify new therapeutic strategies in the specific molecular subgroups.

Identifying synergistic regulation involving c-Myc and sp1 in human tissues.

Combinatorial gene regulation largely contributes to phenotypic versatility in higher eukaryotes. Genome-wide chromatin immuno-precipitation (ChIP) combined with expression profiling can dissect regulatory circuits around transcriptional regulators. Here, we integrate tiling array measurements of DNA-binding sites for c-Myc, sp1, TFIID and modified histones with a tissue expression atlas to establish the functional correspondence between physical binding, promoter activity and transcriptional regulation. For this we develop SLM, a methodology to map c-Myc and sp1-binding sites and then classify sites as sp1-only, c-Myc-only or dual. Dual sites show several distinct features compared to the single regulator sites: specifically, they exhibit overall higher degree of conservation between human and rodents, stronger correlation with TFIID-bound promoters, and preference for permissive chromatin state. By applying regression models to an expression atlas we identified a functionally distinct signature for strong dual c-Myc/sp1 sites. Namely, the correlation with c-Myc expression in promoters harboring dual-sites is increased for stronger sp1 sites by strong sp1 binding and the effect is largest in proliferating tissues. Our approach shows how integrated functional analyses can uncover tissue-specific and combinatorial regulatory dependencies in mammals.

Impact of Consensus Molecular Subtype on Survival in Patients With Metastatic Colorectal Cancer: Results From CALGB/SWOG 80405 (Alliance).

PURPOSE: To determine the predictive and prognostic value of the consensus molecular subtypes (CMSs) of colorectal cancer (CRC) that represent a merging of gene expression-based features largely in primary tumors from six independent classification systems and provide a framework for capturing the intrinsic heterogeneity of CRC in patients enrolled in CALGB/SWOG 80405. PATIENTS AND METHODS: CALGB/SWOG 80405 is a phase III trial that compared the addition of bevacizumab or cetuximab to infusional fluorouracil, leucovorin, and oxaliplatin or fluorouracil, leucovorin, and irinotecan as first-line treatment of advanced CRC. We characterized the CMS classification using a novel NanoString gene expression panel on primary CRCs from 581 patients enrolled in this study to assess the prognostic and predictive value of CMSs in these patients. RESULTS: The CMSs are highly prognostic for overall survival (OS; P < .001) and progression-free survival (PFS; P < .001). Furthermore, CMSs were predictive for both OS (P for interaction < .001) and PFS (P for interaction = .0032). In the CMS1 cohort, patients treated with bevacizumab had a significantly longer OS than those treated with cetuximab (P < .001). In the CMS2 cohort, patients treated with cetuximab had a significantly longer OS than patients treated with bevacizumab (P = .0046). CONCLUSION: These findings highlight the possible clinical utility of CMSs and suggests that refinement of the CMS classification may provide a path toward identifying patients with metastatic CRC who are most likely to benefit from specific targeted therapy as part of the initial treatment.