Search

Malignant peritoneal mesothelioma (MPeM) is a rare but aggressive malignancy with limited treatment options. VEGF inhibition enhances efficacy of immune-checkpoint inhibitors by reworking the immunosuppressive tumor milieu. Efficacy and safety of combined PD-L1 (atezolizumab) and VEGF (bevacizumab) blockade (AtezoBev) was assessed in 20 patients with advanced and unresectable MPeM with progression or intolerance to prior platinum–pemetrexed chemotherapy. The primary endpoint of confirmed objective response rate per RECISTv1.1 by independent radiology review was 40% [8/20; 95% confidence interval (CI), 19.1–64.0] with median response duration of 12.8 months. Six (75%) responses lasted for >10 months. Progression-free and overall survival at one year were 61% (95% CI, 35–80) and 85% (95% CI, 60–95), respectively. Responses occurred notwithstanding low tumor mutation burden and PD-L1 expression status. Baseline epithelial–mesenchymal transition gene expression correlated with therapeutic resistance/response (r = 0.80; P = 0.0010). AtezoBev showed promising and durable efficacy in patients with advanced MPeM with an acceptable safety profile, and these results address a grave unmet need for this orphan disease.Significance:Efficacy of atezolizumab and bevacizumab vis-à-vis response rates and survival in advanced peritoneal mesothelioma previously treated with chemotherapy surpassed outcomes expected with conventional therapies. Biomarker analyses uncovered epithelial–mesenchymal transition phenotype as an important resistance mechanism and showcase the value and feasibility of performing translationally driven clinical trials in rare tumors.See related commentary by Aldea et al., p. 2674.This article is highlighted in the In This Issue feature, p. 2659

Heatmap showing fRMA level gene data from Affymetrix subsets in training and validation datasets.

Supplementary methods and results

Validation of CMS assay at the research molecular diagnostic laboratory. Each sample classified in to CMS subtype 1-4 or mixed (M) subtype.

3A. Classification accuracy of CMS1 samples on training data set V1 using 4-fold cross validation as function of number of genes.3B.able-3B: Classification accuracy of CMS1 samples on validation data set (V2) as function of number of genes.3C. Classification accuracy CMS1 samples with out of sample (V2o) prediction set as function of number of genes.3D. Classification accuracy CMS1 samples with TCGA dataset (RNA-seq subset) as function of number of genes. 3E. Classification accuracy CMS1 samples with Affymetrix set (V2a) as function of number of genes.

DEG and GSEA gene set analysis post vs pre treatment

Post treatment and pre treatment DEG in RFS long vs short

Supplementary Fig S3. Association of gene size and the differential expression of IFN-γ pathway in genes commonly mutated in MSS CRC

Classification accuracy on validation data set, the entire set (V2), out-of-sample prediction (V2o), Affymetrix subset (V2a), and TCGA RNAseq subset (TCGA) for various methods as function of number of genes.

Purpose:Despite the prognostic importance of immune infiltrate in colorectal cancer, immunotherapy has demonstrated limited clinical activity in refractory metastatic proficient mismatch-repair (pMMR) colorectal cancer. This study explores combining anti–CTLA-4 and an anti–PD-L1 therapy in the preoperative management of resectable colorectal cancer liver metastases with the intent to improve immune responses in this disease setting.Patients and Methods:Patients with resectable colorectal cancer liver-only metastases received one dose of tremelimumab and durvalumab preoperatively followed by single-agent durvalumab postoperatively. Primary objectives were to determine feasibility and safety.Results:A total of 24 patients were enrolled between November 2016 and November 2019. Twenty-three patients received treatment [21 pMMR and 2 deficient mismatch-repair (dMMR)] and subsequently 17 (74%; 95% CI: 53%–88%) underwent surgical resection. Grade 3/4 treatment-related immune toxicity and postoperative grade 3/4 toxicity were seen in 5/23 (22%; 95% CI: 10%–44%) and 2/17 (12%; 95% CI: 2%–38%) patients. The median relapse-free survival (RFS) was 9.7 (95% CI: 8.1–17.8) months, and overall survival was 24.5 (95% CI: 16.5–28.4) months. Four patients demonstrated complete pathologic response, two dMMR patients and two POLE mutation patients. Pre- and post-tumor tissue analysis by flow cytometry, immunofluorescence, and RNA sequencing revealed similar levels of T-cell infiltration, but did demonstrate evidence of CD8+ and CD4+ activation posttreatment. An increase in B-cell transcriptome signature and B-cell density was present in posttreatment samples from patients with prolonged RFS.Conclusions:This study demonstrates the safety of neoadjuvant combination tremelimumab and durvalumab prior to colorectal cancer liver resection. Evidence for T- and B-cell activation following this therapy was seen in pMMR metastatic colorectal cancer.

Supplementary figures and supplemental methods

Purpose:Consensus molecular subtyping (CMS) of colorectal cancer has potential to reshape the colorectal cancer landscape. We developed and validated an assay that is applicable on formalin-fixed, paraffin-embedded (FFPE) samples of colorectal cancer and implemented the assay in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory.Experimental Design:We performed an in silico experiment to build an optimal CMS classifier using a training set of 1,329 samples from 12 studies and validation set of 1,329 samples from 14 studies. We constructed an assay on the basis of NanoString CodeSets for the top 472 genes, and performed analyses on paired flash-frozen (FF)/FFPE samples from 175 colorectal cancers to adapt the classifier to FFPE samples using a subset of genes found to be concordant between FF and FFPE, tested the classifier's reproducibility and repeatability, and validated in a CLIA-certified laboratory. We assessed prognostic significance of CMS in 345 patients pooled across three clinical trials.Results:The best classifier was weighted support vector machine with high accuracy across platforms and gene lists (>0.95), and the 472-gene model outperforming existing classifiers. We constructed subsets of 99 and 200 genes with high FF/FFPE concordance, and adapted FFPE-based classifier that had strong classification accuracy (>80%) relative to “gold standard” CMS. The classifier was reproducible to sample type and RNA quality, and demonstrated poor prognosis for CMS1–3 and good prognosis for CMS2 in metastatic colorectal cancer (P < 0.001).Conclusions:We developed and validated a colorectal cancer CMS assay that is ready for use in clinical trials, to assess prognosis in standard-of-care settings and explore as predictor of therapy response.

Nanostring CMS Classifier Performance: Performance of Nanostring 100 gene CMS classifier applied to FFPE samples (6A and 6C) and FF samples (6B and 63D). The top two figures plot 4-class accuracy vs. classification confidence, with the dots marking individual samples either correctly (1.0) or incorrectly (0.0), with color indicat-ing correct CMS. The line contains a generalized additive model (GAM) fit to these data with 95% pointwise confidence bands, and demonstrates that samples classified with greater confidence were more likely to be cor-rectly classified. Figure 6C and 6D plot 4-class accuracy vs. RNA quality, defined as 0nt (FFPE) or RIN (FF). Note that there is little if any association of CMS accuracy with RNA quality, suggesting that the perfor-mance of classifier is robust to RNA quality in this study.

Classification accuracy of different modelling strategies on training data set (V1) using 4-fold cross validation.

Supplementary Table S1

Comparison of Classifiers. Comparison of random forest (RF) and single sample predictors (SSP) in Guinney et al. (2015) paper with our wSVM classifier with 472 genes, either forcing a single CMS call for each sample (all calls) or only when classified with high confidence

RFS long and short microbiome-taxa abundance

5A. Classification accuracy of CMS 3 samples on training data set V1 using 4-fold cross validation as function of number of genes. 5B.Classification accuracy of CMS 3 samples on validation data set (V2) as function of number of genes. 5C. Classification accuracy CMS3 samples with out of sample prediction set (V2o) as function of number of genes. 5D. Classification accuracy CMS 3 samples with TCGA dataset (RAN-seq subset) as function of number of genes. 5E. Classification accuracy CMS3 samples with Affymetrix dataset (V2a) as function of number of genes.

6A. Classification accuracy of CMS 4 samples on training data set V1 using 4-fold cross validation as function of number of genes. 6B. Classification accuracy of CMS 4 samples on validation data set (V2) as function of number of genes. 6C. Classification accuracy CMS 4 samples with out of sample prediction set (V2o) as function of number of genes. 6D. Classification accuracy CMS 3 samples with TCGA dataset (RAN-seq subset) as function of number of genes. 6E. Classification accuracy of CMS 3 samples with Affymetrix data set (V2a) as function of number of genes.

Sample-wise and Gene-wise correlation of FF/FFPE paired Samples: Scatterplot of gene-specific Spearman correlation of FF/FFPE computed separately for CRCSC samples and non-CRCSC samples , with red and blue marking the overall top 100 and top 200 genes, respectively, in terms of FF/FFPE correlation computed across all samples. Note the high level of agreement across batches that genes with high FF/FFPE correlations for one batch tended to also have high FF/FFPE correlation in the other batch, suggesting high FF/FFPE correla-tion is a characteristic of the gene/probe set.

4A. Classification accuracy of CMS2 samples on training data set V1 using 4-fold cross validation as function of number of genes.4B. Classification accuracy of CMS2 samples on validation data set (V2) as function of number of genes.4C. Classification accuracy CMS2 samples with out of sample(V2o) prediction set as function of number of genes. 4D. Classification accuracy CMS2 samples with TCGA dataset (RNA-seq subset) as function of number of genes.4E. Classification accuracy CMS2 samples with Affymentrix dataset (V2a) as function of number of genes.

Algorithm showing strategy to de-velop the Nanostring based FFPE CMS classifier using Affymetrix gene expression data. FF: flash frozen, FFPE: Formalin fixed par-affin embedded, NFFPE: Nanos-tring FFPE, AFF: Flash frozen Affymetrix, GES: Gene expres-sion score, CRCSC: Colorectal cancer subtyping consortium

Treatment related adverse events and list of antibodies used.

Classification Confidence of Affymetrix Validation Samples vs. Classification Accuracy of 472-gene wSVM clas-sifier: Distribution of classification confidence (�i) across 713 validation samples (top), and classification accura-cy (1=correct, 0=incorrect) as a function of classification confidence (�i), with fitted loess curve and 95% confidence interval.

Classification accuracy for training data V1 (based on 4-fold cross val-idation, 2A) and the full validation data V2 (2B) for various methods as function of number

PDF file - 1127K, Deletion of β2SP in MEFs leads to up-regulation of EMT transcription factors. Indirect immunofluorescence was performed on MEFs cell with or without b2SP deletion cells using anti-Slug or anti-Snail antibodies and followed by DAPI counterstaining (blue). The merge of Slug or Snail (red) with DAPI (blue) is also shown.

PDF file - 1574K, YAP1 and SOX9 expression in EC cell lines.

PDF file - 59K

PDF file - 1015K, YAP1 Induces SOX9 promoter activity in 293T cells and primer design for SOX9 ChIP-PCR.

PDF file - 481K, Effects of β2SP on Notch receptors, ligand and γ-secretase complex.

PDF file - 127K, SOX9 is elevated in mouse liver tumors with inactivating Hippo pathway mutations.

PDF file - 740K, Pathological lymph node positivity categorized by SOX9 nuclear staining intensity.

PDF file - 6199K, Increased expression of stem cell markers in shβ2SP EA cells, while shSOX9 and Notch inhibitor GSI rescue the phenotype of shβ2SP. Indirect immunofluorescence was performed on SKGT-4 cells with shControl, shb2SP, double shβ2SP and shSOX9 and sh2SP treated with GSI 5M using anti-Oct3/4 or anti-Lgr5 antibodies and followed by DAPI counterstaining (blue). The merge of Oct3/4 or Lgr5 (red) with DAPI (blue) is also shown.

PDF file - 6469K, Activated Notch 1 receptor (ICN1) and stem cell markers were increased in xenograft mouse tumor tissues with knockdown β2SP. Expression of ICN1, ALDH-1, Oct3/4 and Lgr5 were determined by immunohistochemistry as described in Materials & Methods.

Micrometastases, defined as microscopic cancer cells spatially separated from the macroscopically evident metastasis, are identified in 24% to 56% of resected colorectal liver metastases (CLMs). Somatic gene mutations have emerged as independent prognostic factors in CLM. This study aimed to determine the prognostic impact and risk factors for the presence of micrometastases, including somatic gene mutations.Prospective evaluation for micrometastases was performed at 2 centers in the US and France from 2015 to 2019. CLM specimens were cut radially from the tumor margin to surrounding grossly normal liver for a distance of 2 cm. Depending on CLM size, 3 to 8 specimens per patient were submitted for microscopic analysis. Somatic gene mutations were detected by next-generation sequencing.Among 140 patients undergoing CLM resection in the US (n = 84) and France (n = 56), 36 (26%) patients were found to have micrometastases. Five-year overall and recurrence-free survival rates with micrometastases were 39% and 0%, respectively, compared with 61% and 20% without micrometastases (both p0.05). In multivariable analyses, the presence of micrometastases was an independent risk factor for worse overall survival (hazard ratio 2.88, 95% CI 1.46 to 5.70, p = 0.002) and recurrence-free survival (hazard ratio 1.56, 95% CI 1.01 to 2.41, p = 0.046). In binary logistic regression analysis, RAS/TP53 co-mutation was found to significantly increase the risk of micrometastases (odds ratio 2.77, 95% CI 1.15 to 6.71, p = 0.024).Micrometastases are associated with significantly worse survival after CLM resection. RAS/TP53 co-mutation correlated with increased risk of micrometastases. Further studies are needed to determine strategies to eradicate micrometastases.

Purpose: Preclinical evidence supporting immunomodulatory effect of MEK inhibition that augments anti-tumor activity of PD-1 inhibitors is a compelling rationale for investigating combined MEK and PD-L1 inhibition in advanced microsatellite-stable (MSS) SBA wherein immune checkpoint inhibitors (ICIs) have little clinical benefit. We conducted a prospective, phase 2 trial evaluating efficacy and safety of COTEZO regimen in treatment refractory SBA. Procedures: Patients aged ≥ 18 years with metastatic MSS SBA, ECOG PS 0 - 2, and disease progression on prior systemic chemotherapy were enrolled and treated with cobimetinib (60 mg orally once daily for days 1 - 21) and atezolizumab (840 mg intravenously every 2 weeks) every 28-day cycle. The primary endpoint was objective response rate (ORR) per Response Evaluation Criteria in Solid Tumors (RECIST) Version 1.1. by independent radiology review. Key pre-specified secondary endpoints were safety, disease control rate (DCR), progression-free (PFS) and overall (OS) survival. Pre/on-treatment biopsies were examined for response biomarkers. Results: Between 4/2017 and 7/2020, 20 patients were enrolled. The confirmed ORR per RECISTv1.1 was 10% [2/20; 95%CI: 1.2 - 31.7] (1 complete response and 1 partial response). Seven patients had stable disease for a DCR of 45% (95%CI: 23.1 - 68.5). Median PFS and OS were 2.4 (95% CI 1.3-3.5) and 8.8 (95% CI 5.6 - 12) months, respectively. Grade 3 treatment-emergent adverse events (no grade 4/5) occurred in 7 (35%) patients; most common being elevated CPK (15%) and vomiting (10%). One (5%) patient had grade 3 vomiting, necessitating treatment discontinuation. There appeared to be a marked difference in OS between patients with increased pre-treatment tumor immune infiltration as compared to those lacking immune infiltration amongst patients with RNASeq data available (n=13). Specifically, increased immune infiltration as assessed by computational deconvolution of RNASeq data pointed to increased macrophage and monocyte infiltration as prognostic factors. This was additionally assessed via application of a global “immune score” algorithm, ESTIMATE. In order to generalize our findings, we expanded our computational deconvolution analysis to include additional small bowel adenocarcinoma patients who received standard-of-care treatment (n=7). For the combined standard-of-care and experimentally treated patients we saw that those with immune infiltration had significantly better OS (9 months vs greater than 3 years, HR 5.7, Padj = 0.017), suggesting that immune infiltration is a prognostic, rather than predictive biomarker. Conclusions: Although our study did not meet its primary endpoint of ORR, we identified tumor immune infiltration as a prognostic biomarker in SBA, which could select patients with better outcomes on immunotherapy. Citation Format: Nicholas Hornstein, John P. Shen, Andrew J. Pellatt, Suyu Liu, Mark Knafl, Anneleis F. Willett, Haifeng Zhu, Dipen M. Maru, Walter Darbonne, Ignacio I. Wistuba, Andy Futreal, James C. Yao, Scott E. Woodman, Michael J. Overman, Kanwal P. Raghav. Efficacy, safety, and biomarker analysis of combined MEK (Cobimetinib) and PD-L1 (Atezolizumab) inhibition (COTEZO) in advanced small bowel adenocarcinoma (SBA) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr CT148.

Colorectal cancer (CRC) is the second leading cause of cancer deaths worldwide. One major factor contributing to the high mortality rate of CRC is the presence of microscopic minimal residual disease (MRD) that remains radiographically undetectable and persist regardless of therapeutic interventions. In fact, many CRC patients who have received therapy with curative intent will experience recurrence years later due to MRD, which can progress to recurrent disease that results in clinical morbidity. While immune modulation is crucial for mediating progression and metastasis of primary CRC tumors, how micrometastases suppress or evade antitumor immunity and/or maintain a state of dormancy before growing into macrometastases remains poorly understood. Filling this knowledge gap requires establishing and utilizing models capable of accurately recapitulating the immune phenotype of human microsatellite stable (MSS) metastatic CRC. To this end, we generated genetically engineered, syngeneic mouse colonic organoids using CRE recombinase technology to recapitulate MSS metastatic CRC models with APC and TP53 mutations. After establishing hematologic experimental metastases, histopathological examination confirmed that those present in the liver bore a close resemblance to human colorectal adenocarcinoma liver metastases. We then leveraged our model to test the hypothesis that the activation of suppressive immune cell populations is key for enabling the establishment and progression of MRD. We performed high-plex immunofluorescent co-localization analysis to delineate the immune compartments of micro- and macrometastases. Our results showed that, overall, leukocytes (CD45+ cells) decreased as metastatic tumor foci in the liver progressed. Specifically, cytotoxic CD8+ T cell densities were markedly reduced as tumor metastases progressed to macroscopic disease. Furthermore, populations of CD4+FOXP3+ (T-reg) cells increased as tumors progressed, with an increase in the ratio of CD4+FOXP3+ to CD8+ cells during tumor progression. Additionally, M2 macrophage density (IBA-1+, CD163+) increased as the liver metastases progressed in size. Overall, our findings support our hypothesis, and initiate the first step for identifying suppressive immune infiltrates that may be exploited for therapeutic targeting. Importantly, our somatic mutation-based modeling strategy enables a highly precise recapitulation of the mutational drivers and heterogeneity that occurs in CRC patients as well as provides a valuable resource for research aimed at elucidating the immunobiology of MRD. Citation Format: Alaa M. Mohamed, Melinda Soeung, Jumanah Alshenaifi, Natalie Fowlkes, Ganiraju Manyam, Jennifer Davis, Amanda Anderson, Will Norton, Angela K. Deem, Sisi Gao, Isha Khanduri, Christopher Bistow, Federica Carbone, Stefania Napolitano, Justin Huang, Dipen M. Maru, David G. Menter, Giulio F. Draetta, Giannicola Genovese, Scott Kopetz. Recapitulating metastatic colorectal cancer in somatic mutation models for investigating the tumor immune microenvironment in minimal residual disease [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 77.

PURPOSE BRAF V600E mutations are rarely associated with objective responses to the BRAF inhibitor vemurafenib in patients with metastatic colorectal cancer (CRC). Blockade of BRAF V600E by vemurafenib causes feedback upregulation of EGFR, whose signaling activities can be impeded by cetuximab. METHODS One hundred six patients with BRAF V600E-mutated metastatic CRC previously treated with one or two regimens were randomly assigned to irinotecan and cetuximab with or without vemurafenib (960 mg PO twice daily). RESULTS Progression-free survival, the primary end point, was improved with the addition of vemurafenib (hazard ratio, 0.50, P = .001). The response rate was 17% versus 4% ( P = .05), with a disease control rate of 65% versus 21% ( P < .001). A decline in circulating tumor DNA BRAF V600E variant allele frequency was seen in 87% versus 0% of patients ( P < .001), with a low incidence of acquired RAS alterations at the time of progression. RNA profiling suggested that treatment benefit did not depend on previously established BRAF subgroups or the consensus molecular subtype. CONCLUSION Simultaneous inhibition of EGFR and BRAF combined with irinotecan is effective in BRAF V600E-mutated CRC.

Purpose HER2 amplification has been implicated in resistance to therapy with anti–epidermal growth factor receptor antibodies (anti-EGFRabs) in metastatic colorectal cancer (mCRC). The purpose of the study was to validate the predictive impact of HER2 amplification in mCRC. Patients and Methods We analyzed patients with RAS/BRAF wild-type mCRC across two distinct cohorts. In cohort 1 (n = 98), HER2 amplification was tested in tumor tissue using dual in situ hybridization ( HER2 amplification: HER2/CEP17 ratio, 2.0 or greater). Cohort 2 (n = 70) included 16 patients with HER2 amplification and 54 HER2 nonamplified controls identified by next-generation sequencing ( HER2 amplification: four or more copies) who had received prior anti-EGFRabs. The primary end point was progression-free survival (PFS) on treatment with anti-EGFRab therapy, which was estimated and compared using the Kaplan-Meier method and log-rank test. Results Median PFS in cohort 1 on anti-EGFRab–based therapy was significantly shorter in patients with HER2 amplification compared with HER2 nonamplified patients (2.8 v 8.1 months, respectively; hazard ratio [HR], 7.05; 95% CI, 3.4 to 14.9; P < .001). These findings were validated in cohort 2 (median PFS for HER2 amplified v nonamplified: 2.8 v 9.3 months, respectively; HR, 10.66; 95% CI, 4.5 to 25.1; P < .001). The median PFS on therapy without anti-EGFRabs was similar among HER2-amplified and nonamplified patients in both cohort 1 (9.7 v 11.1 months, respectively; HR, 1.01; 95% CI, 0.4 to 2.4; P = .97) and cohort 2 (9.6 v 11.3 months, respectively; HR, 1.21; 95% CI, 0.5 to 3.1; P = .66). In multivariable analyses, HER2 amplification emerged as a single independent predictor of poor PFS on anti-EGFRab therapy in both cohort 1 (HR, 6.48; 95% CI, 3.1 to 13.6; P < .001) and cohort 2 (HR, 10.1; 95% CI, 4.3 to 23.9; P < .001). Conclusion HER2 amplification in RAS/RAF wild-type mCRC seems to be a predictive biomarker for lack of efficacy of anti-EGFRab therapy. Screening patients with RAS/BRAF wild-type mCRC for HER2 amplification should be considered before anti-EGFRab treatment to guide therapy and to identify patients for early referral to clinical trials.

Factors correlated with biopsy tissue adequacy and the prevalence of within-biopsy variability were evaluated. Totally, 1149 research biopsies were performed on 686 patients from which 5090 cores were assessed. Biopsy cores were reviewed for malignant percentage (estimated percentage of cells in the core that were malignant) and malignant area (estimated area occupied by malignant cells). Linear mixed models and generalized linear mixed models were used for the analysis. A total of 641 (55.8%) biopsies contained a core with p

Simple Summary Colorectal cancer (CRC) has been divided into four consensus molecular subtypes (CMS) using the unsupervised clustering of bulk transcriptomic data. The CMS, which align with known biological differences in CRC, have helped address inter-tumor heterogeneity and have been shown to be predictive of survival. However, bulk tumors are comprised of a mix of tumor cells, non-tumor stroma, and immune cells, and the relative contributions of each are not accounted for in the current CMS classification system. Here, we build an algorithm to assign CMS classification to individual cells, which we apply to single cell RNAseq data to explore the impact of intra-tumor heterogeneity on CMS. We find that the number of stromal and immune cells present has a strong influence on the bulk CMS type and that clusters of tumor epithelial cells derived from single cell RNAseq do not align with CMS. Abstract The implications of intratumor heterogeneity on the four consensus molecular subtypes (CMS) of colorectal cancer (CRC) are not well known. Here, we use single-cell RNA sequencing (scRNASeq) to build an algorithm to assign CMS classification to individual cells, which we use to explore the distributions of CMSs in tumor and non-tumor cells. A dataset of colorectal tumors with bulk RNAseq (n = 3232) was used to identify CMS specific-marker gene sets. These gene sets were then applied to a discovery dataset of scRNASeq profiles (n = 10) to develop an algorithm for single-cell CMS (scCMS) assignment, which recapitulated the intrinsic biology of all four CMSs. The single-cell CMS assignment algorithm was used to explore the scRNASeq profiles of two prospective CRC tumors with mixed CMS via bulk sequencing. We find that every CRC tumor contains individual cells of each scCMS, as well as many individual cells that have enrichment for features of more than one scCMS (called mixed cells). scCMS4 and scCMS1 cells dominate stroma and immune cell clusters, respectively, but account for less than 3% epithelial cells. These data imply that CMS1 and CMS4 are driven by the transcriptomic contribution of immune and stromal cells, respectively, not tumor cells.