Consensus Molecular Subtype 4 (CMS4) Identifies RAS Wild-Type mCRC Patients Who May Benefit Preferentially from Anti‑EGFR Therapy

Highlights

– An individual patient data meta-analysis (n = 790) found CMS4 tumors had significantly longer PFS and OS than CMS1 and that anti‑EGFR (vs anti‑VEGF) antibodies substantially improved PFS and OS in CMS4 RAS wild‑type metastatic colorectal cancer (mCRC).

– The interaction between consensus molecular subtype (CMS) and antibody class (anti‑EGFR vs anti‑VEGF) was statistically significant for PFS and OS, supporting CMS4 as a predictive biomarker candidate in first‑line RAS WT mCRC.

– Findings are hypothesis‑generating; prospective validation and standardization of CMS testing will be required before routine clinical implementation.

Background

Metastatic colorectal cancer (mCRC) remains a leading cause of cancer morbidity and mortality worldwide. Molecular biomarkers—most importantly RAS (KRAS/NRAS) and BRAF mutational status and microsatellite instability (MSI)—guide systemic therapy decisions, especially selection of anti‑EGFR monoclonal antibodies (cetuximab, panitumumab) for RAS wild‑type tumors. The Consensus Molecular Subtypes (CMS1–4), derived from transcriptomic profiling, capture biologically distinct CRC subgroups with prognostic and potential predictive implications. CMS4 (the ‘mesenchymal’ subtype) is characterized by stromal activation, TGF‑β signaling, epithelial‑to‑mesenchymal transition (EMT), and generally worse prognosis. Whether CMS classification can refine first‑line biologic antibody choice (anti‑EGFR vs anti‑VEGF) in RAS wild‑type mCRC remains controversial and clinically important.

Study design

Stahler and colleagues performed a one‑step individual patient data (IPD) meta‑analysis including five first‑line mCRC randomized trials (FIRE1, FIRE3 [NCT00433927], XELAVIRI [NCT01249638], PanaMa [NCT01991873], TRIBE2 [NCT02339116]) that enrolled patients treated per protocol with fluoropyrimidine‑based chemotherapy plus irinotecan or oxaliplatin and either anti‑VEGF (bevacizumab) or anti‑EGFR antibodies, and for whom evaluable CMS calls and RAS wild‑type status were available. The analysis included 790 patients with RAS wild‑type tumors and clustered models to account for study effects. Primary endpoints were objective response rate (ORR), progression‑free survival (PFS) and overall survival (OS). Statistical methods: generalized estimating equations for ORR and stratified Cox models for PFS/OS, with formal tests for interaction between CMS and antibody type.

Key findings

Population: Among 790 patients, CMS distribution was: CMS1 n = 77 (9.7%), CMS2 n = 345 (43.7%), CMS3 n = 74 (9.4%), CMS4 n = 294 (37.2%). Between‑study heterogeneity was negligible.

Objective response rate (ORR)

Compared with CMS1, both CMS2 and CMS4 demonstrated numerically higher ORR but did not reach conventional statistical significance after adjustment: CMS2 OR 1.668 (95% CI 0.982–2.836; P = .059); CMS4 OR 1.369 (95% CI 0.874–2.146; P = .170). This suggests CMS4 tumors are at least as chemosensitive in combination with biologic antibodies as other subtypes, but ORR differences were modest.

Progression‑free survival (PFS)

Using CMS1 as reference, CMS2 and CMS4 demonstrated significantly longer PFS: CMS2 HR 0.64 (95% CI 0.48–0.85; P = .002); CMS4 HR 0.67 (95% CI 0.50–0.91; P = .009). Importantly, within CMS4 tumors, the use of anti‑EGFR (vs anti‑VEGF) antibodies was associated with improved PFS: HR 0.67 (95% CI 0.46–0.97; P = .03). This PFS benefit for anti‑EGFR was consistent in CMS4 subsets that were RAS/BRAF wild‑type (HR 0.55; 95% CI 0.37–0.83; P = .004) and microsatellite‑stable (HR 0.52; 95% CI 0.32–0.86; P = .01).

Overall survival (OS)

CMS2 and CMS4 both had longer OS than CMS1: CMS2 HR 0.59 (95% CI 0.43–0.80; P < .001); CMS4 HR 0.67 (95% CI 0.49–0.92; P = .01). Within CMS4, anti‑EGFR versus anti‑VEGF treatment conferred a pronounced OS benefit: HR 0.49 (95% CI 0.33–0.72; P < .001). The OS advantage persisted in RAS/BRAF wild‑type CMS4 tumors (HR 0.55; 95% CI 0.37–0.83; P = .004) and in microsatellite‑stable CMS4 tumors (HR 0.52; 95% CI 0.32–0.86; P = .01).

Interaction tests

Significant interaction was documented between antibody class and CMS for both PFS (P < .001) and OS (P < .001) in the overall cohort, and for OS in the RAS/BRAF wild‑type subset (P = .02). These statistical interactions support a predictive role of CMS for antibody selection rather than simply a prognostic association.

Expert commentary and biological plausibility

The principal and somewhat unexpected finding is that CMS4, traditionally associated with adverse prognosis and stromal enrichment, appears to derive clinically meaningful benefit from anti‑EGFR therapy relative to anti‑VEGF therapy in RAS wild‑type mCRC. Several mechanistic hypotheses could account for this observation:

• EGFR signaling in tumor‑associated stromal cells or cancer‑associated fibroblasts may contribute to CMS4 biology; blockade of EGFR might disrupt tumor–stroma crosstalk that drives progression in mesenchymal tumors.
• Anti‑EGFR antibodies have immune‑modulatory effects including antibody‑dependent cellular cytotoxicity (ADCC) that could be more consequential in an inflamed or stromal microenvironment characteristic of CMS4.
• CMS4 tumors, often microsatellite‑stable with fewer actionable genomic alterations, may rely on epidermal growth factor receptor pathways for growth signals that are not reflected solely by tumor epithelial RAS status.

These remain plausible but speculative mechanisms. Preclinical and translational studies examining EGFR expression and signaling dynamics in CMS4 stroma versus tumor cells, immune cell infiltration patterns, and response biomarkers (e.g., amphiregulin/epiregulin expression) would help to elucidate causality.

Limitations and generalizability

Key limitations temper immediate clinical translation:

• Although IPD meta‑analysis improves statistical power and harmonizes endpoint definitions, the included data derive from trials with differing designs, chemotherapeutic backbones, and patient selection criteria. Residual confounding is possible despite clustering adjustment.
• CMS assignment requires high‑quality transcriptomic data and standardized pipelines; variability in specimen source (primary tumor vs metastasis), preanalytic factors, and classifier versions can affect subtype calls and reproducibility.
• Sample sizes for some subgroups (e.g., CMS1 and CMS3) were modest, limiting precision of estimates in those strata.
• Interaction testing was significant, but multiplicity concerns and the post hoc nature of biomarker analyses necessitate prospective confirmation before guideline changes.
• Sidedness (left vs right primary), which impacts anti‑EGFR benefit, was not the primary focus here and may interact with CMS effects; further stratified analyses are needed.

Clinical implications and next steps

This IPD meta‑analysis provides compelling hypothesis‑generating evidence that CMS4 may be a predictive biomarker of benefit from anti‑EGFR therapy in RAS (and BRAF) wild‑type mCRC. However, before changing practice, the following steps are recommended:

• Prospective validation: Incorporate CMS classification prospectively into randomized trials comparing anti‑EGFR versus anti‑VEGF strategies in first‑line RAS WT mCRC, with stratified randomization to test predictive value robustly.
• Assay standardization: Develop and validate clinically feasible CMS assays (e.g., targeted transcript panels compatible with formalin‑fixed paraffin‑embedded specimens) with clear analytic performance characteristics and interlaboratory reproducibility.
• Integrated biomarker models: Evaluate CMS alongside established biomarkers (primary tumor sidedness, amphiregulin/epiregulin expression, MSI, BRAF) to build multivariable predictive models that can guide individualized antibody selection.
• Translational work: Investigate mechanistic bases for CMS4 sensitivity to EGFR blockade, focusing on tumor–stroma interactions and immune modulation to identify companion biomarkers and combination strategies.

Conclusion

This individual patient data meta‑analysis suggests that CMS4 is associated with improved PFS and OS when anti‑EGFR agents are used instead of anti‑VEGF therapy in first‑line RAS wild‑type mCRC. The statistical interaction between CMS and antibody class supports a predictive signal for CMS4, but these results remain hypothesis‑generating. Prospective trials and standardized CMS testing are required to confirm CMS4 as a clinically actionable biomarker for biologic antibody selection in mCRC.

Funding and clinicaltrials.gov identifiers

The pooled trials contributing IPD included: FIRE1 (no identifier), FIRE3 (ClinicalTrials.gov NCT00433927), XELAVIRI (NCT01249638), PanaMa (NCT01991873), and TRIBE2 (NCT02339116). Funding sources for the original trials are acknowledged in the primary publications; the meta‑analysis authors reported their funding and disclosures in the cited JCO manuscript.

References

1. Guinney J, Dienstmann R, Wang X, et al. The consensus molecular subtypes of colorectal cancer. Nat Med. 2015;21(11):1350–1356. PMID: 26445588.

2. Stahler A, Modest DP, Stintzing S, et al. Individual Patient Data Meta‑Analysis of Consensus Molecular Subtypes as Biomarkers of First‑Line Treatment in RAS Wild‑Type Metastatic Colorectal Cancer. J Clin Oncol. 2025 Nov 18: JCO2500596. doi: 10.1200/JCO-25-00596. Epub ahead of print. PMID: 41252656.

Thumbnail image prompt

A multidisciplinary tumor board scene: clinicians (oncologist, pathologist, molecular biologist) gathered around a screen showing heatmap CMS classification, Kaplan‑Meier survival curves comparing anti‑EGFR vs anti‑VEGF, and a highlighted CMS4 row; subtle elements include a DNA helix and antibody molecular icons, neutral color palette, photorealistic, high detail.