Colorectal Cancer Trial Design Patterns

Definition

Colorectal cancer (CRC) encompasses colon and rectal cancers. Phase 3 CRC trial design is stratified by: treatment setting (metastatic vs. adjuvant), molecular biomarkers (KRAS/NRAS/BRAF mutation status, MSI-H/dMMR vs. MSS/pMMR, HER2 amplification), and treatment line (1L, 2L+, 3L+). The majority of CRC Phase 3 trials use OS as primary endpoint, reflecting the historically short post-progression survival (< 12 months) and the availability of multiple active subsequent lines that dilute PFS signals.

CTG Phase 3 CRC dataset (300 trials): OS and PFS primaries each approximately 50–55/300 (17–18%); DFS primary 38/300 (13%). Median enrollment 418 patients. Double-blind 40/300 (13%).

Regulatory Position

OS (preferred for metastatic CRC): Historically the standard primary endpoint. FDA Cancer Endpoints 2018: OS is the gold standard for metastatic CRC where post-progression survival is typically 8–18 months with effective subsequent lines. Multiple active lines (FOLFOX, FOLFIRI, bevacizumab, cetuximab/panitumumab in RAS WT, regorafenib, TAS-102, immunotherapy in MSI-H) mean OS is measurable and clinically relevant.

DFS (adjuvant CRC): DFS is the accepted primary endpoint for adjuvant Stage III CRC. FDA accepted DFS for adjuvant colon cancer based on ODAC 2004 consensus. FOLFOX adjuvant approval (MOSAIC) was based on DFS. 3-year DFS rate is the FDA-accepted surrogate for 5-year DFS in Stage III colon cancer.

PFS (emerging in IO-eligible subsets): KEYNOTE-177 (pembrolizumab 1L MSI-H/dMMR) used PFS as co-primary with OS. MSI-H/dMMR represents ~5% of metastatic CRC; IO is standard 1L in this biomarker-defined population.

Status: FDA Cancer Endpoints 2018 = Final

When to Use Each Endpoint

Setting	Primary Endpoint	Key Secondary	Notes
1L metastatic (KRAS/NRAS WT, EGFR)	OS	PFS, ORR	FIRE-3, CALGB 80405
1L metastatic (VEGF, doublet+bev)	OS	PFS, ORR	NO16966 (bev + FOLFOX), TRIBE (FOLFOXIRI+bev)
1L metastatic (MSI-H/dMMR)	PFS + OS co-primary	ORR	KEYNOTE-177 (pembro, PFS HR 0.60; OS HR 0.74)
2L metastatic (post-1L)	OS	PFS, ORR	RAISE (ramucirumab+FOLFIRI, OS HR 0.84)
3L+ (refractory)	OS	PFS, ORR	RECOURSE (TAS-102, OS HR 0.68); CORRECT (regorafenib, OS HR 0.77)
Adjuvant Stage III	DFS (3-year rate or time-to-event)	OS	MOSAIC (FOLFOX, DFS HR 0.77); IDEA collaboration
Adjuvant Stage II (high-risk)	DFS	OS	Subset analyses; FOLFOX benefit uncertain in Stage II
Localized rectal cancer (neoadjuvant + adjuvant)	DFS or OS	pCR (for neoadjuvant)	TME + chemoradiation trials

Design Considerations

OS-Primary Design Requirements

For metastatic CRC OS-primary trials:

Event-driven design: Typically 400–600 OS events for 80% power at HR 0.80
Median OS assumptions: Control arm 20–28 months (modern 1L); experimental arm target HR 0.75–0.85
Accrual: 400–600 patients; 12–24 month accrual; 24–36 month follow-up to event maturity
Interim OS analysis: One interim at ~50% events (O'Brien-Fleming boundary α ≈ 0.005) is typical; strong efficacy signals allow early stopping

RAS/BRAF Stratification (Mandatory)

All modern CRC trials must stratify by:

KRAS/NRAS mutation status (wild-type vs. mutant) — EGFR antibodies (cetuximab, panitumumab) active only in RAS WT
BRAF V600E status — BRAF-mutated CRC (~10%) has poor prognosis; IO combinations and BRAF+MEK inhibitors active in BRAF+ (BEACON-CRC)
MSI-H/dMMR vs. MSS/pMMR — IO immunotherapy active only in MSI-H (~5% metastatic); pembrolizumab standard 1L in MSI-H
Primary tumor sidedness (right vs. left colon) — right-sided tumors have worse prognosis and differential EGFR benefit

Biomarker analysis hierarchy: Enrich for MSI-H or RAS WT populations when those subgroups are the primary indication; all-comers as secondary analysis with pre-specified gatekeeping.

Adjuvant CRC Design (Stage III)

DFS endpoint: Time from surgery to recurrence (local, regional, distant) or death from any cause
3-year DFS rate as surrogate: FDA-accepted based on ACCENT collaboration meta-analysis — 3-year DFS rate predicts 5-year OS with high concordance in Stage III colon cancer
Assessment schedule: Every 3–6 months for 3 years, then annually to 5 years (CT chest/abdomen/pelvis + CEA)
IE: subsequent surgery or chemotherapy — treatment policy; adjuvant therapy changes do not censor DFS

KEYNOTE-177 MSI-H Design

Co-primary OS + PFS in the MSI-H/dMMR subpopulation (5% of mCRC):

PFS: hierarchical testing — PFS in MSI-H first, then OS in MSI-H
Biomarker-selected population: all enrolled patients tested centrally for MSI status; only MSI-H treated
PFS HR 0.60, OS HR 0.74 — both met pre-specified boundaries

Intercurrent Events

Subsequent Lines (Metastatic CRC — 5–6+ Lines)

OS (primary): Treatment policy universally — all deaths counted regardless of subsequent lines
Hypothetical OS would require censoring at every subsequent therapy change → >80% of patients censored → uninformative
SAP language: "Overall survival will be analyzed using the treatment policy strategy. All deaths from any cause will be counted as events regardless of subsequent anti-cancer therapies received."

Bevacizumab/Aflibercept Beyond Progression (BEP)

Standard of care in 2L mCRC includes continuing VEGF inhibition beyond progression (bevacizumab beyond progression). This is an IE in 2L trials:

If BEP is permitted in control arm and not experimental arm: potential IE creating imbalanced subsequent therapy
Strategy: Treatment policy (BEP counted as part of natural treatment course); pre-specify whether BEP is allowed and stratify by BEP intention

Treatment Discontinuation Without Progression

CRC patients often stop oxaliplatin due to neurotoxicity (adjuvant FOLFOX) while remaining on 5-FU/leucovorin. In adjuvant trials:

Treatment policy: Continue DFS assessments; the endpoint is DFS not oxaliplatin-specific DFS
Oxaliplatin dose modification does not censor

Regulatory Precedent

Trial	Drug	Setting	Primary Endpoint	HR / Result	Approval
MOSAIC	FOLFOX4 vs. LV5FU2	Adjuvant Stage II/III	DFS	HR 0.77 (Stage III)	Regular (2004)
NO16966	Bevacizumab + FOLFOX/XELOX	1L metastatic	PFS	HR 0.83	Supplemental (2004)
CORRECT	Regorafenib	3L+ refractory	OS	HR 0.77	Regular (2012)
RECOURSE	TAS-102 (trifluridine/tipiracil)	3L+ refractory	OS	HR 0.68	Regular (2015)
RAISE	Ramucirumab + FOLFIRI	2L	OS	HR 0.84	Regular (2015)
KEYNOTE-177	Pembrolizumab	1L MSI-H/dMMR	PFS + OS co-primary	PFS HR 0.60; OS HR 0.74	Regular (2020)
BEACON-CRC	Encorafenib + binimetinib + cetuximab	2L BRAF V600E	OS	HR 0.52 (triplet vs control)	Regular (2020)
SUNLIGHT	TAS-102 + bevacizumab	3L+	OS	HR 0.61	Regular (2023)

Limitations and Pitfalls

OS dilution by multiple active subsequent lines: FOLFOX, FOLFIRI, bevacizumab, cetuximab/panitumumab, regorafenib, TAS-102 all available — 5+ active subsequent lines dilute OS signal. This is why OS HRs in 1L CRC are modest (0.75–0.85) despite meaningful PFS benefit.

3-year DFS as OS surrogate (Stage III only): The ACCENT surrogate is validated for Stage III colon cancer treated with fluoropyrimidine ± oxaliplatin. Not validated for newer agents (bevacizumab adjuvant QUASAR 2 failed; cetuximab PETACC-8 negative). New agents require new surrogacy validation.

MSI-H subgroup size: Only ~5% of metastatic CRC is MSI-H. Phase 3 trials enriched for MSI-H are small (n=307 in KEYNOTE-177). Trials including both MSI-H and MSS patients require large sample sizes to show signals in the small MSI-H subgroup.

PFS in MSS metastatic CRC as primary: PFS is less accepted than OS for MSS metastatic CRC because post-progression survival is long (10–18 months) and subsequent therapies are active. FDA has historically required OS for MSS metastatic CRC.

Backlinks

Source: FDA Cancer Endpoints 2018 (Final); ACCENT Group meta-analysis Compiled from FDA guidance + CTG Phase 3 CRC dataset (300 trials) + known Phase 3 trial results