Colorectal Cancer Clinical Trial Timeline — In-depth Report

Coverage: 74 landmark trials cited by NCCN Colon + NCCN Rectal V1.2026 (every PMID traceable) + 6 treatment paradigms + 5 precision pathways (MSI-H / BRAF V600E / HER2 / KRAS G12C / NTRK) + rectal TNT and organ preservation

Curated by Dual Brain Lab (csilab.net)

1. One-sentence definition

This report traces the evolution and current decision landscape of systemic therapy for colorectal cancer (CRC; comprising colon cancer and rectal cancer as two major subtypes; histology >95% adenocarcinoma) over the past 60 years (1960s 5-FU backbone → 2025 precision + immunotherapy combinations), as cited by landmark clinical trials referenced in NCCN Colon V1.2026 and NCCN Rectal V1.2026. It serves as a traceable panoramic map for frontline clinicians making “who, what, and why” decisions at the 2026 time point.

CRC is the world’s third most common (1.9 million new cases in 2022) and second most lethal (~900,000 deaths) malignancy; in China, ~510,000 new cases and ~240,000 deaths annually, with both incidence and mortality continuing to rise. Clinical staging runs in parallel across stage I-IV (colon) / locally advanced rectal (LARC) / metastatic (mCRC); biomarkers include six major stratification pathways — RAS (KRAS/NRAS, ~50%) / BRAF V600E (~8%) / MSI-H·dMMR (stage IV ~4-5%, stage II ~15-20%) / HER2 amp (~3%) / KRAS G12C (~3%) / NTRK fusion (<1%). Scope: systemic therapy (chemotherapy / targeted / immune / perioperative TNT); excludes metastasis ablation / HAI (hepatic arterial infusion) / peritoneal HIPEC / hereditary CRC (Lynch syndrome screening itself).

Iron rule: every data point of every trial is traceable to PubMed (PMID) or ClinicalTrials.gov (NCT id) — every [PMID xxxxxxxx] in the text can be opened directly to verify against the PubMed source.

2. Longitudinal: evolution timeline of six treatment paradigms

CRC systemic therapy has gone through six paradigm shifts in the past 60 years: 5-FU backbone established (1960s-2000) → FOLFOX / FOLFIRI dual-backbone era (2000-2004) → metastatic biologic-targeted breakthrough vs concurrent adjuvant triple failure (2004-2014) → MSI-H immune reversal (2015-2025) → BRAF V600E / HER2 / KRAS G12C / NTRK precision rockets (2019-2025) → rectal TNT and organ preservation (2020-2023).

Each shift used 2-4 phase III trials to push the old standard of care (SoC) into 2L. The biggest lesson across 60 years, in one sentence: metastatic-effective ≠ adjuvant-effective — bevacizumab, cetuximab, and irinotecan are backbones in metastatic disease; transplanted into adjuvant, all failed. FOLFOX is the sole exception (MOSAIC successfully moved the metastatic doublet into adjuvant). This extrapolation trap produced 4 negative phase IIIs in CRC (NSABP C-08 / AVANT / N0147 / CALGB 89803 + PETACC-3), cumulatively enrolling over 10,000 patients.

2.1 Establishing the 5-FU / FOLFOX / FOLFIRI backbone (1990s-2004): from monotherapy to doublet

Story: 5-FU (5-fluorouracil) was synthesized in 1957 and remained the sole CRC systemic drug for the next 40 years; in the 1990s leucovorin (LV) sensitization + infusional schedules pushed ORR from 10% to 20%. In 2000, two independent phase III trials were published in the same year: de Gramont 2000 (FOLFOX2) + Saltz 2000 (IFL, irinotecan + 5-FU/LV) pushed 1L metastatic mOS from 11 months to 14-17 months. In 2004, N9741 three-arm head-to-head firmly placed FOLFOX above IFL. Also in 2004, MOSAIC (adjuvant FOLFOX4 vs 5-FU/LV) moved FOLFOX from metastatic into stage III adjuvant — the only case across CRC’s 60 years where “a metastatic-successful drug moved into adjuvant with a positive phase III.”

• de Gramont FOLFOX2 [PMID 10944126] (de Gramont 2000 J Clin Oncol, N=420): advanced CRC 1L oxaliplatin + 5-FU/LV infusion (FOLFOX2) vs 5-FU/LV alone. mPFS 9.0 vs 6.2 months, mOS 16.2 vs 14.7 months, ORR 50.7% vs 22.3%. Foundation of the FOLFOX backbone — platinum crossed from ovarian / lung cancer into CRC.
• SALTZ IFL [PMID 11006366] (Saltz 2000 N Engl J Med, N=683): advanced CRC 1L irinotecan + 5-FU/LV (IFL regimen) vs 5-FU/LV. mPFS 7.0 vs 4.3 months, mOS 14.8 vs 12.6 months. First CRC 1L phase III positive for irinotecan — but toxicity was high (neutropenia, diarrhea); later replaced by the infusional FOLFIRI schedule.
• N9741 [PMID 14665611] (Goldberg 2004 J Clin Oncol, N=795): advanced CRC 1L FOLFOX4 vs IFL vs IROX three-arm head-to-head. FOLFOX4 mOS 19.5 months > IFL 15.0 months > IROX 17.4 months; ORR 45% vs 31% vs 35%. FOLFOX decisively beat IFL in 1L metastatic — “FOLFOX is the backbone, FOLFIRI is the backup” entered guidelines from here.
• GERCOR-TOURNIGAND [PMID 14657227] (Tournigand 2004 J Clin Oncol, N=220): FOLFIRI-then-FOLFOX6 vs FOLFOX6-then-FOLFIRI as 1L→2L sequences. No difference in mOS: 21.5 vs 20.6 months — classic evidence that “which doublet comes first doesn’t matter, but both should be used.”
• MOSAIC [PMID 15175436] (André 2004 N Engl J Med, N=2,246): stage II-III colon cancer after curative resection FOLFOX4 × 6 months vs 5-FU/LV × 6 months adjuvant. Stage III subgroup 5-year DFS 66.4% vs 58.9% (HR 0.80); stage II overall negative. First phase III to successfully move a metastatic-effective doublet into adjuvant — but stage II did not benefit, planting the seed for “don’t add oxaliplatin in low-risk stage II.” 10-year follow-up [PMID 26527776] (André 2015 J Clin Oncol): stage III DFS HR still 0.80, OS HR 0.85 (p=0.046) — long-term benefit maintained, but BRAFm / dMMR subgroups did not benefit.
• X-ACT [PMID 15987918] (Twelves 2005 N Engl J Med, N=1,987): stage III colon cancer after curative resection capecitabine monotherapy vs IV 5-FU/LV adjuvant. 3-year DFS 64.2% vs 60.6% (HR 0.87, p=0.053 non-inferiority met). Oral fluoropyrimidine monotherapy equivalent to IV 5-FU/LV — added a convenience option to chemotherapy regimens.
• XELOXA (NO16968) [PMID 21383294] (Haller 2011 J Clin Oncol, N=1,886): stage III colon cancer after curative resection CAPOX (capecitabine + oxaliplatin) vs 5-FU/LV adjuvant. 3-year DFS 70.9% vs 66.5% (HR 0.80). CAPOX equivalent to FOLFOX — “IV or PO doublet both usable” became SoC.
• QUASAR [PMID 18083404] (QUASAR Collaborative Group 2007 Lancet, N=3,239): stage II (91%) + low-risk stage III colorectal cancer after curative resection, adjuvant 5-FU/LV vs observation. 5-year OS 80.3% vs 77.4% (HR 0.82), absolute benefit 3.6%. Largest phase III for stage II adjuvant chemotherapy — absolute benefit small; decisions require individualization (T4 / poorly differentiated / lymphovascular invasion / obstruction / perforation are high-risk features).

Takeaway: 2000-2004 laid CRC’s backbone in four years — 1L metastatic FOLFOX > IFL, stage III adjuvant FOLFOX4 or CAPOX replaced 5-FU/LV monotherapy; infusional irinotecan (FOLFIRI) as 1L alternative and 2L; capecitabine monotherapy left an option for those intolerant of IV. But this backbone benefited only low-risk stage II, and BRAFm / dMMR subgroups did not benefit — planting the seed for the MSI-H reversal in 2015.

2.2 Metastatic biologic-targeted breakthrough vs adjuvant triple failure (2004-2014): the same drug, different fate — the extrapolation trap

Story: In 2004, Hurwitz’s AVF2107 added bevacizumab (anti-VEGF) to IFL — mOS pushed from 15.6 to 20.3 months (HR 0.66), CRC’s first metastatic targeted-drug phase III positive. That same year, cetuximab and panitumumab (anti-EGFR) entered 1L KRAS/RAS wild-type populations through five phase IIIs across 2004-2009. But in the five years 2011-2014, three parallel attempts to “move metastatic-effective targeted drugs into adjuvant” all failed: NSABP C-08 / AVANT (bev) + N0147 (cetuximab) + CALGB 89803 / PETACC-3 (irinotecan) — cumulatively enrolling >10,000 patients. This is CRC’s biggest clinical lesson across 60 years.

2.2.1 Metastatic: three breakthrough routes — bev / cet / FOLFOXIRI

• AVF2107 [PMID 15175435] (Hurwitz 2004 N Engl J Med, N=813): advanced CRC 1L IFL + bevacizumab vs IFL + placebo. mOS 20.3 vs 15.6 months (HR 0.66, p<0.001); mPFS 10.6 vs 6.2 months; ORR 44.8% vs 34.8%. CRC’s first metastatic targeted phase III positive — VEGF-pathway tumor angiogenesis became a druggable target, and simultaneously planted the extrapolation-trap seed for the 2011-2012 adjuvant bev failures.
• CRYSTAL [PMID 19339720] (Van Cutsem 2009 N Engl J Med, N=1,198): advanced CRC 1L FOLFIRI + cetuximab vs FOLFIRI. Overall population mPFS HR 0.85 (positive but marginal); KRAS wild-type subgroup mPFS 9.9 vs 8.4 months (HR 0.70); mOS 23.5 vs 20.0 months (HR 0.80). Foundational trial establishing KRAS as a stratification biomarker — “without biomarker stratification, no benefit is visible.”
• OPUS [PMID 19114683] (Bokemeyer 2009 J Clin Oncol, N=337): FOLFOX + cetuximab vs FOLFOX 1L. KRAS WT subgroup ORR 61% vs 37% (OR 2.55); mPFS HR 0.57. OPUS + CRYSTAL together pinned the 1L indication for anti-EGFR to KRAS WT.
• AMADO-KRAS-ANALYSIS [PMID 18316791] (Amado 2008 J Clin Oncol, N=427 chemo-refractory mCRC reanalysis): reanalysis of panitumumab in chemo-refractory mCRC phase III by KRAS status: KRAS WT ORR 17%, KRAS mut ORR 0%. Landmark subgroup analysis — pinned KRAS mutation as an “absolute resistance” predictor for anti-EGFR therapy; from then on, all anti-EGFR trials mandated KRAS/RAS testing.
• PRIME [PMID 20921465] (Douillard 2010 J Clin Oncol, N=1,183): FOLFOX4 + panitumumab vs FOLFOX4 1L. KRAS WT subgroup mPFS 9.6 vs 8.0 months (HR 0.80); mOS 23.9 vs 19.7 months. Panitumumab established in 1L. Extended RAS reanalysis [PMID 24024839] (Douillard 2013 N Engl J Med): patients with KRAS exon 3/4 + NRAS exon 2/3/4 mutations also did not benefit, and may have been harmed — the definition of “KRAS wild-type” was extended to “RAS wild-type.”
• CRYSTAL-RAS-EXTENDED [PMID 25605843] (Van Cutsem 2015 J Clin Oncol, N=1,198 reanalysis): extended RAS analysis of CRYSTAL. All-RAS wild-type subgroup mOS 28.4 vs 20.2 months (HR 0.69); RAS-mut subgroup no benefit. PRIME + CRYSTAL extended analyses together established “all-RAS wild-type” as the 2015 SoC gate for anti-EGFR eligibility.
• PEAK [PMID 24687833] (Schwartzberg 2014 J Clin Oncol, N=285 phase II): FOLFOX6 + panitumumab vs FOLFOX6 + bevacizumab in KRAS WT (later extended to RAS WT) 1L. RAS WT subgroup mOS 41.3 vs 28.9 months. First pan vs bev 1L head-to-head exploration — small-sample phase II, but 41-month mOS was a sensational number in 2014.
• FIRE-3 [PMID 25088940] (Heinemann 2014 Lancet Oncol, N=592): FOLFIRI + cetuximab vs FOLFIRI + bevacizumab head-to-head in KRAS WT 1L. Primary endpoint ORR 62.0% vs 58.0% (p=0.18) — no difference; but mOS 28.7 vs 25.0 months (HR 0.77, p=0.017) significantly favored cetuximab. First phase III to open the “anti-EGFR vs anti-VEGF in 1L — which to pick” question.
• CALGB-80405 [PMID 28632865] (Venook 2017 JAMA, N=1,137): FOLFIRI / FOLFOX + cetuximab vs + bevacizumab head-to-head in KRAS WT 1L (US NCI-sponsored). mOS 30.0 vs 29.0 months (HR 0.88, p=0.08) — no difference — contradicting FIRE-3.
• CALGB-80405-SIDEDNESS [PMID 34061178] (Yin 2021 J Natl Cancer Inst, CALGB 80405 reanalysis by tumor location): left-sided colon mOS cet 32.9 vs bev 29.3; right-sided colon mOS cet 13.7 vs bev 29.2 (right-sided reversed!). Foundational reanalysis establishing primary tumor sidedness as a stratification biomarker — the FIRE-3 vs CALGB-80405 contradiction was resolved: left-sided RAS WT favors cet; right-sided RAS WT favors bev.
• PARADIGM [PMID 37071094] (Watanabe 2023 JAMA, N=802, Japan / partial China): mFOLFOX6 + panitumumab vs mFOLFOX6 + bevacizumab dedicated prospective RCT in RAS WT + left-sided colon mCRC 1L. mOS 37.9 vs 34.3 months (HR 0.82) significantly favored panitumumab. Upgraded “sidedness guides anti-EGFR selection” from retrospective analysis to prospective phase III evidence — by 2023, the 1L anti-EGFR use case was formally pinned to left-sided RAS WT.
• TRIBE [PMID 25337750] (Loupakis 2014 N Engl J Med, N=508): FOLFOXIRI + bev triplet vs FOLFIRI + bev 1L. mPFS 12.1 vs 9.7 months (HR 0.75); mOS 29.8 vs 25.8 months. Triplet + bev established in 1L for fit patients — but with significantly higher toxicity. TRIBE-UPDATED [PMID 26338525] (Cremolini 2015 Lancet Oncol) OS update: BRAFm subgroup mOS 19.0 vs 10.7 months (HR 0.54) — BRAFm mCRC cannot wait for a chemo doublet long-term; the triplet wins the race.

2.2.2 Adjuvant: three drug classes, three concurrent failures

• NSABP-C-08 [PMID 20940184] (Allegra 2011 J Clin Oncol, N=2,672): stage II-III colon cancer after curative resection, mFOLFOX6 × 6 months + bevacizumab × 12 months vs mFOLFOX6 × 6 months adjuvant. 3-year DFS 77.4% vs 75.5% (HR 0.89, p=0.15) — negative. First large negative phase III for adjuvant bev — metastatic-1L-effective bev did not benefit in adjuvant.
• AVANT [PMID 23168362] (de Gramont 2012 Lancet Oncol, N=3,451): stage III or high-risk stage II colon cancer after curative resection, FOLFOX-4 or XELOX + bev × 12 months vs FOLFOX-4 or XELOX. mDFS HR 1.17 (FOLFOX+bev) / 1.07 (XELOX+bev) — negative vs control; mOS HR 1.27 (p=0.02) — directional harm in FOLFOX+bev. Second negative adjuvant bev phase III + OS numerical harm. Roche subsequently abandoned adjuvant bev development.
• N0147 [PMID 22474202] (Alberts 2012 JAMA, N=2,686): stage III colon cancer after curative resection, mFOLFOX6 × 6 months + cetuximab × 6 months vs mFOLFOX6 × 6 months adjuvant (KRAS WT subgroup n=1,863 main analysis). KRAS WT 3-year DFS cet+ 75.8% vs mFOLFOX6 78.2% (HR 1.21) — negative; in the ≥70-year subgroup, the HR trended unfavorably. Adjuvant cet phase III negative.
• CALGB-89803 [PMID 17687149] (Saltz 2007 J Clin Oncol, N=1,264): stage III colon cancer after curative resection IFL (5-FU + LV + irinotecan) vs 5-FU + LV adjuvant. 5-year DFS 61% vs 63% (p=0.88) — negative; G4 neutropenia 39% vs 24%. “First adjuvant defeat for metastatic-effective irinotecan.”
• PETACC-3 [PMID 19451425] (Van Cutsem 2009 J Clin Oncol, N=2,094 stage III main analysis): stage II-III colon cancer after curative resection FOLFIRI vs 5-FU/LV adjuvant. Stage III 5-year DFS 56.7% vs 54.3% (HR 0.90, p=0.106) — negative. Second adjuvant defeat for irinotecan — combined with CALGB 89803 / French Accord 02 / FFCD 9802 (four negative phase IIIs), closed the door on “moving irinotecan into adjuvant.”

Takeaway: across 2004-2014, metastatic disease pushed 1L mOS from 15 to 30 months via bev / cet / pan / FOLFOXIRI, but adjuvant phase IIIs in the same period failed six times across three drug classes (bev / cet / iri), cumulatively >10,000 patients — the largest “metastatic → adjuvant extrapolation trap” in CRC history. MOSAIC’s successful move of FOLFOX into adjuvant is the lone exception (the mechanism may be that FOLFOX’s direct cytotoxic action still applies to micrometastases). The clinical lesson of this era condenses into one line: “effective in metastatic ≠ effective post-operatively” — adjuvant must prove itself with its own phase III data.

2.3 Later-line: refinement of maintenance / 2L / 3L+ (2007-2023)

Story: In 2007-2014, four phase IIIs — E3200 / RAISE / ML18147 / VELOUR — pushed 2L mOS from 9-10 months to 13 months. In 2013-2023, CORRECT / RECOURSE / FRESCO / FRESCO-2 / SUNLIGHT took 3L+ refractory mCRC from “no drug” to “at least 3 options.” China’s FRESCO / FRESCO-2 fruquintinib became one of the few mCRC 3L drugs originating in China to be pushed to global FDA approval.

• E3200 [PMID 17442997] (Giantonio 2007 J Clin Oncol, N=829): 2L mCRC (bev-naïve) FOLFOX4 + bevacizumab vs FOLFOX4 vs bev. FOLFOX4+bev mOS 12.9 months vs FOLFOX4 10.8 months (HR 0.75). Foundational data for bev in 2L.
• TML-ML18147 [PMID 23168366] (Bennouna 2013 Lancet Oncol, N=820): after progression on 1L bev-containing, 2L chemotherapy ± bevacizumab continuation (continued bev with a switched chemo backbone) vs chemo alone. mOS 11.2 vs 9.8 months (HR 0.81). Established the “bev beyond progression” concept — resistance is not because bev fails; continue its use.
• VELOUR [PMID 22949147] (Van Cutsem 2012 J Clin Oncol, N=1,226): 2L mCRC (oxaliplatin-pretreated) FOLFIRI + aflibercept (VEGF-trap) vs FOLFIRI. mOS 13.5 vs 12.1 months (HR 0.82). Aflibercept (Zaltrap) FDA-approved for 2L mCRC in 2012.
• RAISE [PMID 25877855] (Tabernero 2015 Lancet Oncol, N=1,072): 2L mCRC (after progression on 1L bev + oxaliplatin + fluoropyrimidine) FOLFIRI + ramucirumab (anti-VEGFR2) vs FOLFIRI + placebo. mOS 13.3 vs 11.7 months (HR 0.84). Third anti-angiogenic mechanism in 2L (bev / aflibercept / ramu) — anti-angiogenic therapy in 2L became standard.
• CORRECT [PMID 23177514] (Grothey 2013 Lancet, N=760): 3L+ refractory mCRC regorafenib 160 mg d1-21 q4w vs placebo. mOS 6.4 vs 5.0 months (HR 0.77); mPFS 1.9 vs 1.7 months. Absolute benefit small but the first effective option “after all roads exhausted.” FDA approved in September 2012.
• RECOURSE [PMID 25970050] (Mayer 2015 N Engl J Med, N=800): 3L+ refractory mCRC TAS-102 (trifluridine/tipiracil) vs placebo. mOS 7.1 vs 5.3 months (HR 0.68). Oral nucleoside antimetabolite — FDA approved September 2015.
• FRESCO [PMID 29946728] (Li 2018 JAMA, N=416, 28 Chinese centers): 3L+ refractory mCRC fruquintinib (selective VEGFR1/2/3 TKI) vs placebo. mOS 9.3 vs 6.6 months (HR 0.65); mPFS 3.7 vs 1.8 months. China-originating, NMPA approved September 2018 — the first registrational phase III for a China-developed oncology drug in CRC 3L.
• FRESCO-2 [PMID 37331369] (Dasari 2023 Lancet, N=691, international multi-center + partial China): 3L+ refractory mCRC (resistant to or unsuitable for TAS-102 / regorafenib) fruquintinib vs placebo, international validation. mOS 7.4 vs 4.8 months (HR 0.66). FDA approved November 2023 — a China-originating drug won a global label in mCRC.
• SUNLIGHT [PMID 37133585] (Prager 2023 N Engl J Med, N=492): 3L+ refractory mCRC TAS-102 + bevacizumab vs TAS-102 monotherapy. mOS 10.8 vs 7.5 months (HR 0.61). Old drug + bev breakthrough — 3L mOS crossed 10 months for the first time. FDA approved August 2023.
• IMBLAZE370 [PMID 31003911] (Eng 2019 Lancet Oncol, N=363): 3L MSS/pMMR mCRC atezolizumab + cobimetinib (MEK inhibitor) vs regorafenib vs atezolizumab monotherapy. All three arms failed — demonstrated that MSS/pMMR CRC does not respond to IO, and that MEK + IO combinations cannot “heat” cold tumors.
• REGOTORI [PMID 34622226] (Wang 2021 Cell Rep Med, Chinese phase Ib/II): MSS/pMMR refractory mCRC regorafenib + toripalimab (domestic PD-1). ORR 15.2%, DCR 36.4%; mPFS 2.1 months. Early signal + gut-microbiome analysis — left hypothesis-generating data for subsequent MSS CRC exploration.

Takeaway: in 2023, 3L+ refractory mCRC SoC = one of fruquintinib (FRESCO-2) / TAS-102 + bev (SUNLIGHT) / regorafenib (CORRECT). 2L is one of FOLFIRI + bev / aflibercept / ramucirumab plus anti-EGFR switching (if RAS WT and not previously used). MSS/pMMR CRC is cold to IO (IMBLAZE370 failed); only the MSI-H subgroup is CRC’s true immunotherapy battlefield (§2.4).

2.4 MSI-H / dMMR immune reversal (2015-2025): from refractory advanced to 1L to 100% cCR in rectal

Story: In 2015, early work by Le DT et al. used ~40 MSI-H / dMMR solid-tumor cases in a phase II to demonstrate pembrolizumab ORR 40-50% in this subgroup (that early foundational trial is not among the 74 main trials here), establishing the tumor-agnostic biology of “MMR status predicts IO response.” In 2017, CheckMate-142 nivo monotherapy achieved ORR 31% in MSI-H mCRC 2L+; nivo+ipi ORR 55%. In 2020, KEYNOTE-177 used phase III data to push pembrolizumab mPFS in MSI-H mCRC 1L vs chemo from 8.2 to 16.5 months (HR 0.60). In 2022, Cercek’s rectal dMMR trial achieved 100% clinical complete response (cCR) with dostarlimab monotherapy in dMMR locally advanced rectal cancer — a rare precedent of “immunotherapy replacing surgery” in oncology. In 2024, CheckMate-8HW pushed MSI-H mCRC 1L from pembro monotherapy to nivo+ipi dual IO (24-mo PFS 72% vs 14%, HR 0.21, one of the largest HRs in history). Over these 7 years, the MSI-H / dMMR subgroup went from “chemo-insensitive refractory” to “immune-SoC reversed.”

• KEYNOTE-164 [PMID 31725351] (Le 2020 J Clin Oncol, N=124 single-arm phase II): MSI-H / dMMR mCRC ≥2L (cohort A ≥3L / cohort B ≥2L) pembrolizumab monotherapy. Overall ORR 33% (cohort A) / 34% (cohort B); mDoR not reached; mPFS 2.3-4.1 months, mOS 31.4-47.0 months. KEYNOTE-164 + the earlier Le 2015 NEJM established the core CRC data underpinning the FDA 2017 tumor-agnostic MSI-H approval.
• CHECKMATE-142 [PMID 29355075] (Overman 2018 J Clin Oncol, nivo+ipi cohort, N=119): MSI-H mCRC ≥2L (76% ≥2L) nivolumab + ipilimumab. ORR 55%, 9-month PFS 76%, 9-month OS 87%. Dual IO far outperformed monotherapy in MSI-H.
• CHECKMATE-142-NIVO-MONO [PMID 28734759] (Overman 2017 Lancet Oncol, nivo mono cohort, N=74): MSI-H mCRC 2L+ nivolumab monotherapy. ORR 31%, 12-month PFS 50%, 12-month OS 73%. First large-cohort signal for MSI-H / IO — from then on, all CRC IO studies were stratified by MMR status.
• KEYNOTE-177 [PMID 33264544] (André 2020 N Engl J Med, N=307): MSI-H / dMMR mCRC 1L pembrolizumab vs investigator’s choice chemo (FOLFOX/FOLFIRI ± bev ± cet). mPFS 16.5 vs 8.2 months (HR 0.60, 95% CI 0.45-0.80, p=0.0002); ORR 43.8% vs 33.1%; G3-5 AE 22% vs 66%. First positive IO 1L phase III in CRC — MSI-H became the hard gate for 1L immunotherapy selection.
• KEYNOTE-177-5YR [PMID 39631622] (André 2025 Ann Oncol): 5-year follow-up. mPFS 16.5 vs 8.2 months maintained; mOS HR 0.73 (95% CI 0.54-0.99) long-term benefit reached significance; OS remained significant despite an initial 60% crossover rate — long-tail OS benefit for MSI-H 1L pembro was confirmed.
• CHECKMATE-8HW [PMID 39602630] (André 2024 N Engl J Med, N=839): MSI-H / dMMR mCRC 1L (main analysis) or 2L+ nivolumab + ipilimumab vs nivo mono vs chemo. 1L nivo+ipi vs chemo 24-mo PFS 72% vs 14% (HR 0.21, 95% CI 0.13-0.35) — one of the largest HRs ever in a CRC phase III. nivo+ipi vs nivo mono HR 0.62 significant. FDA approved nivo+ipi for MSI-H 1L in 2025 — MSI-H 1L moved from pembro monotherapy toward dual IO (in fit patients).
• CERCEK-DMMR-RECTAL-CANCER-DOSTARLIMAB [PMID 35660797] (Cercek 2022 N Engl J Med; expanded cohort 2025): dMMR stage II/III locally advanced rectal adenocarcinoma neoadjuvant dostarlimab (PD-1) monotherapy × 6 months with organ-preservation intent (if cCR achieved, avoid CRT + surgery). Initial cohort N=12 all reached cCR (100%); expanded to 42 patients, all with sustained cCR. A rare precedent in oncology of “IO monotherapy replacing surgery + radiotherapy” — a must-discuss clinical branchpoint for dMMR rectal cancer in 2026.
• NICHE-2 [PMID 38838311] (Chalabi 2024 N Engl J Med, N=115, Netherlands): dMMR non-metastatic locally advanced stage II-III colon cancer, neoadjuvant nivolumab + ipilimumab × 2 doses followed by surgery. pCR 67%, MPR 95%; 3-yr DFS 100%; no recurrences. Neoadjuvant IO in non-metastatic dMMR colon cancer — an astonishing early signal of “neoadjuvant IO pushes DFS to 100% in dMMR colon”; awaiting phase III-scale validation.
• ATEZOTRIBE [PMID 35636444] (Antoniotti 2022 Lancet Oncol, N=218 phase II): mCRC 1L unselected for MMR status FOLFOXIRI + bev ± atezolizumab. Overall mPFS 13.1 vs 11.5 months (HR 0.69) positive; dMMR subgroup HR 0.27 / pMMR subgroup HR 0.78. A phase II hope signal for pMMR CRC + IO — but phase III-level MSS CRC + IO data is still lacking (IMBLAZE370 / REGOTORI both small / negative).

Takeaway: in 2026, the MSI-H / dMMR decision = advanced 1L SoC is dual IO (nivo+ipi, CheckMate-8HW) or mono IO (pembro, KEYNOTE-177); non-metastatic locally advanced dMMR colon — consider neoadjuvant IO (NICHE-2 early signal); dMMR rectal — prioritize dostarlimab monotherapy for organ preservation (Cercek expanded cohort still 100% cCR); MSS/pMMR CRC — IO mono / IO+ICI all negative. CRC is the cancer type where “MMR determines everything.”

2.5 Biomarker-matched precision rockets (2016-2025): four dedicated pathways — BRAF / HER2 / KRAS G12C / NTRK

Story: In 2019, BEACON CRC pushed encorafenib + cetuximab (2-drug) in BRAF V600E mutant mCRC 2L+ from “chemo-given-up” to SoC (mOS 9.3 vs 5.9 months). In 2025, BREAKWATER advanced encorafenib + cetuximab + mFOLFOX6 (3-drug) into 1L — BRAFm finally had “a usable 1L regimen.” In 2016, HERACLES opened HER2+ CRC (trastu + lapatinib); in 2021, DESTINY-CRC01 used T-DXd (ADC) to win HER2-amp CRC 2L+ with ORR 45.3%; in 2023, MOUNTAINEER achieved ORR 38% with tucatinib + trastu (two drugs, avoiding ADC ILD risk). In 2022-2023, KRYSTAL-1 + CodeBreaK 300 extended KRAS G12C (CRC ~3%) from NSCLC-exclusive to CRC. In 2018-2020, NAVIGATE + STARTRK (larotrectinib / entrectinib) opened the NTRK-fusion tumor-agnostic pathway; the CRC subgroup is <1% but ORR >50%. The four pathways combined cover ~13-15% of CRC patients, forming the core map of metastatic CRC precision therapy.

2.5.1 BRAF V600E pathway (mCRC ~8%, poor prognosis)

• SWOG-S1406 [PMID 33356422] (Kopetz 2021 J Clin Oncol, N=106 phase II RCT): BRAF V600E mut mCRC 2L irinotecan + cetuximab ± vemurafenib (triplet) vs irinotecan + cetuximab (doublet). mPFS 4.4 vs 2.0 months (HR 0.50); ORR 17% vs 4%. Mechanistic validation for the triplet — BRAF inhibition + EGFR inhibition synergy opened the CRC BRAFm therapeutic window.
• BEACON-CRC [PMID 31566309] (Kopetz 2019 N Engl J Med, N=665 phase III): BRAF V600E mut mCRC 2L (1-2 prior lines) encorafenib + cetuximab (2-drug) vs encorafenib + binimetinib + cetuximab (3-drug) vs chemo (FOLFIRI + cet or irinotecan + cet). 2-drug vs chemo: mOS 9.3 vs 5.9 months (HR 0.61); ORR 20% vs 2%; 3-drug vs chemo HR 0.52. FDA approved enco+cet for 2L BRAFm mCRC in April 2020. Triplet-arm toxicity was significantly higher; the doublet became standard — 2L BRAFm SoC established.
• BREAKWATER [PMID 40444708] (Elez 2025 N Engl J Med, N=637 phase III): BRAF V600E mut + RAS WT mCRC 1L encorafenib + cetuximab + mFOLFOX6 (3-drug) vs investigator’s choice chemo (FOLFOX ± bev or FOLFOXIRI ± bev). mPFS 12.8 vs 7.1 months (HR 0.53); ORR 60.9% vs 40.0%; mOS interim HR 0.49 (p=0.0015). First positive phase III in BRAFm 1L — pushed 1L mOS from 12-15 months to 30+ months (interim); FDA granted accelerated approval in 2024, full NEJM publication 2025. BRAF flipped from “metastatic’s worst-prognosis subgroup” to a turnaround story.

2.5.2 HER2 pathway (mCRC ~3%, enriched in RAS WT)

• HERACLES [PMID 27108243] (Sartore-Bianchi 2016 Lancet Oncol, N=27 phase II): HER2+ + KRAS exon 2 wild-type mCRC 3L+ trastuzumab + lapatinib doublet. ORR 30%, mPFS 21 weeks. First prospective phase II signal for CRC HER2 + established CRC-specific HER2+ criteria (differing from breast-cancer Dako scoring — HERACLES required IHC 3+ or IHC 2+/FISH+ with >50% positive cells).
• MOUNTAINEER [PMID 37142372] (Strickler 2023 Lancet Oncol, N=117 phase II): HER2+ + RAS WT mCRC 3L+ tucatinib (HER2-selective TKI) + trastuzumab. ORR 38.1%, mDoR 12.4 months, mPFS 8.2 months. The two-drug combination avoids ADC ILD risk + small-molecule TKI is convenient — FDA granted accelerated approval for HER2+ mCRC 3L+ in January 2023.
• DESTINY-CRC01 [PMID 33961795] (Siena 2021 Lancet Oncol, N=78 phase II): HER2-expressing mCRC 3L+ trastuzumab deruxtecan (T-DXd, ADC) across three cohorts by IHC 3+ / IHC 2+ FISH+ / IHC 2+ FISH-. Cohort A (IHC 3+, n=53) ORR 45.3%, mPFS 6.9 months, mOS 15.5 months. Landmark ADC data in HER2+ CRC; ILD risk must be monitored.
• DESTINY-CRC02 [PMID 39116902] (Raghav 2024 Lancet Oncol, N=122 phase II): HER2+ mCRC 2L+ T-DXd dose-finding (5.4 mg/kg vs 6.4 mg/kg). 5.4 mg/kg ORR 37.8%, 6.4 mg/kg ORR 27.5%. Established 5.4 mg/kg as the CRC dose + demonstrated activity in RAS-mut population — first benefit signal for RAS mut in the HER2 pathway.

2.5.3 KRAS G12C pathway (mCRC ~3%)

• KRYSTAL-1 [PMID 36546659] (Yaeger 2023 N Engl J Med, N=43+28 phase II): KRAS G12C mut mCRC 3L+ adagrasib monotherapy vs adagrasib + cetuximab combination. Adagrasib monotherapy ORR 19%, mPFS 5.6 months; adagrasib + cetuximab ORR 46%, mPFS 6.9 months. Dual targeting (KRAS + EGFR feedback blockade) significantly outperformed KRAS G12C monotherapy — confirmed CRC-specific adaptive resistance (EGFR feedback activation).
• CODEBREAK-300 [PMID 37870968] (Fakih 2023 N Engl J Med, N=160 phase III): KRAS G12C mut mCRC 3L+ (≥2 prior lines) sotorasib + panitumumab vs sotorasib + panitumumab low-dose vs investigator’s choice trifluridine/tipiracil or regorafenib. Sotorasib 960 mg + pan arm mPFS 5.6 vs 2.2 months (HR 0.49), ORR 26% vs 0%. FDA approved sotorasib + panitumumab for KRAS G12C+ mCRC 3L+ in January 2024 — the second positive phase III in CRC precision therapy (after BEACON).

2.5.4 NTRK fusion (mCRC <1%, tumor-agnostic)

• LAROTRECTINIB-NAVIGATE [PMID 29466156] (Drilon 2018 N Engl J Med, N=55 tumor-agnostic, CRC n=5): TRK fusion+ solid tumors larotrectinib. Overall ORR 75% (95% CI 61-85); CRC subgroup data sparse but responsive. FDA’s first tumor-agnostic biomarker approval in November 2018 — regulatory milestone.
• ENTRECTINIB-STARTRK [PMID 31838007] (Doebele 2020 Lancet Oncol, N=54 tumor-agnostic, CRC subgroup reported separately): NTRK fusion+ solid tumors entrectinib (pan-TRK/ROS1/ALK, good CNS penetration). Overall ORR 57.4%; CRC subgroup small N but consistent ORR. FDA tumor-agnostic approval in August 2019.

Takeaway: in 2026, CRC precision therapy = “every newly diagnosed advanced CRC must undergo comprehensive molecular profiling” — among RAS / BRAF V600E / MMR·MSI / HER2 IHC + ISH / KRAS G12C / NTRK fusion, any positive biomarker → a prospective phase III / phase II evidence-level SoC is available. Missing detection = missing a high-yield response subgroup with ORR 30-60%.

2.6 Rectal TNT and organ preservation (1990s-2023): from “surgery is mandatory” to “surgery is optional”

Story: over the past 30 years, rectal cancer moved from the basic question of “preoperative vs postoperative concurrent chemoradiotherapy (neoadjuvant vs adjuvant CRT)” → the optimization questions of “short-course vs long-course radiation,” “add oxaliplatin or not,” “radiation or chemo first” → the structural paradigm shift of 2020-2023 TNT (total neoadjuvant therapy) + organ preservation. Core tension: rectal-cancer radiotherapy side effects (sexual function / proctitis / bowel function) are heavy, and surgery itself carries permanent ostomy risk → don’t resect the organ if you can avoid it.

• GERMAN-CAO-ARO-AIO-94 [PMID 15496622] (Sauer 2004 N Engl J Med, N=823): T3-4 or N+ rectal cancer, preoperative CRT (50.4 Gy + 5-FU) vs postoperative CRT. 5-yr local recurrence 6% vs 13% (p=0.006); acute G3-4 toxicity preop 27% vs postop 40%. Preoperative CRT replaced postoperative CRT as LARC (locally advanced rectal cancer) SoC — all rectal trials over the next 20 years iterated on top of preoperative CRT.
• EORTC-22921 [PMID 24440473] (Bosset 2014 Lancet Oncol, N=1,011): T3-4 rectal 2×2 factorial design (preop CRT vs preop RT + postop 5-FU vs observation) long-term follow-up. Postop 5-FU adjuvant did not improve 10-yr OS (HR 0.97); preop CRT significantly improved local recurrence but did not affect OS. Cast doubt on “post-preop-CRT adjuvant chemotherapy.”
• POLISH-I [PMID 16983741] (Bujko 2006 Br J Surg, N=312): T3-T4 rectal short-course radiation (5×5 Gy + immediate surgery) vs long-course CRT (50.4 Gy + concurrent 5-FU). No difference in 4-yr local recurrence or OS. Short-course radiation as equivalent alternative to long-course CRT — shorter duration (1 week vs 5-6 weeks) + better patient compliance.
• ACCORD-12 [PMID 20194850] (Gérard 2010 J Clin Oncol, N=598): T3-4 rectal preop capecitabine + 45 Gy vs capecitabine + oxaliplatin + 50 Gy (dose escalation + oxaliplatin). ypCR 13.9% vs 19.2% (p=0.09, not significant); toxicity clearly higher in oxaliplatin arm. Adding oxaliplatin to preop CRT did not benefit — consistent with STAR-01 / NSABP R-04.
• CAO-ARO-AIO-04 [PMID 26189067] (Rödel 2015 Lancet Oncol, N=1,236): LARC preop 5-FU + oxaliplatin + CRT followed by FOLFOX adjuvant vs standard 5-FU + CRT + 5-FU adjuvant. 3-yr DFS 75.9% vs 71.2% (HR 0.79, p=0.03). The only phase III to demonstrate “throughout-oxaliplatin inclusion (preop CRT + postop adjuvant) improves DFS” — at a toxicity cost.
• FOWARC [PMID 31557064] (Deng 2019 J Clin Oncol, N=495, China, Sun Yat-sen University, led by Deng Y): LARC cT3-4 or N+ neoadjuvant mFOLFOX6 + RT vs 5-FU + RT vs mFOLFOX6 alone (no RT). No difference in 3-yr DFS / OS across three arms. Challenged the dogma that “all LARC needs CRT” — mFOLFOX6-alone arm did not receive RT but had similar outcomes. A Chinese forerunner for PROSPECT 2023’s RT-omission decision.
• RAPIDO [PMID 33301740] (Bahadoer 2021 Lancet Oncol, N=920): high-risk LARC (cT4a/b, extramural vascular invasion, cN2, MRF+, lateral lymph nodes) short-course 5×5 Gy radiation → 18 weeks CAPOX/FOLFOX total neoadjuvant (TNT) → TME surgery vs standard long-course CRT + TME + postop adjuvant. 3-yr disease-related treatment failure 23.7% vs 30.4% (HR 0.75); pCR 28% vs 14%. First phase III to show TNT short-course branch superior to standard CRT — TNT short-course + CAPOX pathway established.
• PRODIGE-23 [PMID 33862000] (Conroy 2021 Lancet Oncol, N=461): LARC cT3-4 N0-2 FOLFIRINOX × 6 cycles induction → CRT → surgery → postop adjuvant (TNT induction branch) vs standard CRT → surgery → postop adjuvant. 3-yr DFS 75.7% vs 68.5% (HR 0.69); 3-yr mOS HR 0.65; ypCR 28% vs 12%. TNT induction branch also positive — together with RAPIDO’s short-course branch, pinned two TNT pathways into 2021 SoC.
• STELLAR [PMID 35263150] (Jin 2022 J Clin Oncol, N=599, China, CAMS, led by Jin J): Chinese LARC cT3-4 or N+ short-course radiation + CAPOX TNT vs long-course CRT + postop adjuvant. 3-yr DFS 64.5% vs 62.3% (HR 0.883, non-inferiority met); 3-yr OS 86.5% vs 75.1% (HR 0.67) — a China-led TNT short-course phase III, with DFS non-inferior + OS even superior to the long-course CRT control. Asian-population-specific TNT data.
• PROSPECT [PMID 37272534] (Schrag 2023 N Engl J Med, N=1,194): cT2 N+ or cT3 N0-N+ low-risk LARC suitable for sphincter-preserving surgery FOLFOX × 6 cycles → selective CRT reserved only for those not achieving ≥20% tumor shrinkage vs standard long-course CRT + surgery + postop adjuvant. 5-yr DFS 80.8% vs 78.6% (HR 0.92, non-inferiority met); 5-yr OS 89.5% vs 90.2%. RT-omission decision — low-risk LARC can skip RT, using only FOLFOX + surgery. From Sauer 2004’s “all LARC needs CRT” to 2023’s “low-risk skip CRT” — a full circle in 20 years.
• OPRA [PMID 35483010] (Garcia-Aguilar 2022 J Clin Oncol, N=324): stage II-III rectal TNT (induction chemo + CRT vs CRT + consolidation chemo, with post-TNT assessment) organ preservation: patients achieving cCR enter watch-and-wait (W&W). 3-yr organ preservation 40% vs 58% (consolidation superior to induction); 3-yr DFS 75% vs 78% (comparable). Phase II-level evidence for organ preservation rate 40-58% + DFS uncompromised — the “cCR→W&W” pathway after TNT was established.
• IWWD [PMID 29976470] (van der Valk 2018 Lancet, N=880, international registry across 47 centers in 15 countries): observational registry of rectal cancers achieving cCR after neoadjuvant therapy and entering watch-and-wait. 3-yr distant metastasis-free rate 91.9%; 3-yr regrowth rate 25.2% (97% salvageable by surgery) — after W&W regrowth, salvage surgery remains possible. Largest real-world W&W dataset.
• IDEA [PMID 29590544] (Grothey 2018 N Engl J Med, N=12,834, 6-trial pooled): stage III colon cancer adjuvant FOLFOX or CAPOX 3 months vs 6 months. Main analysis non-inferiority not met (HR 1.07); subgroups: T1-3 N1 non-inferior at 3 months with significantly reduced toxicity (G3 neuropathy); T4 or N2 still favored 6 months. 3-vs-6-month by risk group — global SoC: low-risk 3-month CAPOX, high-risk 6-month FOLFOX. Important de-escalation evidence for oxaliplatin-induced neuropathy.
• DYNAMIC [PMID 35657320] (Tie 2022 N Engl J Med, N=455): stage II colon cancer ctDNA-guided adjuvant: ctDNA-positive (week 4 or 7 post-op) → chemotherapy; ctDNA-negative → observation, vs standard clinicopathologic decision. ctDNA-positive chemotherapy rate 15% vs standard chemotherapy rate 28% (~50% reduction in chemotherapy use); 2-yr RFS 93.5% vs 92.4% (non-inferiority met). Phase III-level evidence for ctDNA-guided de-escalation — significantly reduced overtreatment in low-risk stage II.

Takeaway: 2026 rectal cancer = risk-adapted, organ-preserving, TNT-centered. High-risk LARC (cT4 / MRF+ / cN2+) → TNT (RAPIDO short-course or PRODIGE-23 induction) + surgery; low-risk LARC (cT2 N+ / cT3 N0) → PROSPECT — RT omission possible; dMMR rectal cancer → prioritize Cercek dostarlimab monotherapy for organ preservation (§2.4); post-TNT cCR → consider W&W (OPRA + IWWD real-world). Chinese data plays an important role in FOWARC + STELLAR (Asian-population validation + RT-omission forerunner).

3. Cross-sectional: 2026 decision landscape (six dimensions)

Projecting the longitudinal evolution onto 2026’s concrete clinical decision trees — the following are six key branchpoints and the evidence behind each.

3.1 Newly diagnosed mCRC: order comprehensive molecular profiling immediately

NCCN Colon V1.2026 + NCCN Rectal V1.2026 both explicitly recommend comprehensive molecular testing (tissue or ctDNA or both) for all newly diagnosed advanced CRC, covering: RAS (full-exon KRAS + NRAS) + BRAF V600E + MMR/MSI (IHC + PCR or NGS) + HER2 IHC + ISH + KRAS G12C (naturally covered during RAS sequencing) + NTRK fusion (optional, <1%). Molecular results directly affect:

• 1L chemotherapy backbone selection: BRAFm → BREAKWATER 3-drug (enco + cet + mFOLFOX6); MSI-H → dual IO (CheckMate-8HW) or mono IO (KEYNOTE-177); RAS WT left-sided → sidedness-guided anti-EGFR (PARADIGM); RAS WT right-sided → bev; RAS mut → FOLFOX/FOLFIRI + bev.
• 2L+ targeted accessibility: HER2 amp → MOUNTAINEER / DESTINY-CRC02; KRAS G12C → CodeBreaK 300 (sotorasib + pan); NTRK → larotrectinib / entrectinib; MSI-H 2L (if IO not used in 1L) → pembro / nivo+ipi.
• Clinical trial enrollment: BREAKWATER follow-up updates / NICHE-2 follow-up phase III / new pMMR IO combination trials.

Missing RAS → wrong anti-EGFR; missing BRAF → miss BREAKWATER 1L; missing MMR → miss IO; missing HER2 → miss ADC / small-molecule TKI; missing KRAS G12C → miss the sotorasib/adagrasib + pan pathway.

3.2 mCRC 1L: five parallel biomarker-stratified pathways

2026 mainstream: 1L is no longer “FOLFOX + bev” one-size-fits-all — five biomarker-stratified pathways.

Contraindications:

• RAS mut — do not use cetuximab / panitumumab (PRIME [PMID 24024839] + CRYSTAL [PMID 25605843] extended-RAS analyses confirmed harm)
• MSS/pMMR — do not use IO monotherapy / MEK + IO combinations (IMBLAZE370 [PMID 31003911] three-arm failure)
• BRAFm — pure chemo + bev 1L not recommended (mOS typically <15 months, given BREAKWATER’s better option)

NCCN Colon V1.2026: MSI-H 1L = nivo+ipi / pembro / ipilimumab+nivolumab each Category 1; BRAFm + RAS WT 1L = enco+cet+mFOLFOX6 Category 1; RAS WT left-sided 1L = mFOLFOX6/FOLFIRI + pan/cet Category 1; RAS WT right-sided / RAS mut 1L = mFOLFOX6/FOLFIRI + bev Category 1.

3.3 mCRC 2L+: precision-targeted routing + anti-angiogenic re-breakthrough after resistance

2026 mainstream: 2L routes by what was used in 1L + biomarker status —

Controversies:

• 3L+ three-drug sequence: fruquintinib vs TAS-102+bev vs regorafenib — no direct H2H; cross-trial mOS comparisons (7.4 / 10.8 / 6.4 months) versus controls are biased; clinical selection by toxicity profile + prior exposure.
• MSS CRC 2L+ IO combinations: IMBLAZE370 negative + REGOTORI early signal; no positive phase III; not recommended outside clinical trials.

3.4 Colon adjuvant: MOSAIC foundation + IDEA de-escalation + DYNAMIC ctDNA pathway

2026 mainstream:

Contraindications 2026 (three classes + irinotecan):

• Adjuvant bev contraindicated: NSABP C-08 [PMID 20940184] + AVANT [PMID 23168362] (OS HR 1.27 potential harm)
• Adjuvant cetuximab contraindicated: N0147 [PMID 22474202] (3-yr DFS HR 1.21)
• Adjuvant panitumumab contraindicated: no positive phase III, and cet-class mechanism unfavorable
• Adjuvant irinotecan contraindicated: CALGB 89803 [PMID 17687149] + PETACC-3 [PMID 19451425] (four negative phase IIIs)

Decision insight: any adjuvant “drug addition” idea must have positive phase III evidence — in 60 years, only FOLFOX successfully moved into adjuvant.

3.5 Rectal: three-way branch — TNT / organ preservation / low-risk RT omission

2026 mainstream:

Historical gradient:

• German CAO/ARO/AIO-94 [PMID 15496622] (2004) established preoperative CRT
• POLISH-I [PMID 16983741] (2006) short-course and long-course equivalent
• EORTC-22921 [PMID 24440473] (2014) questioned post-operative adjuvant chemotherapy value
• ACCORD-12 [PMID 20194850] (2010) + CAO-ARO-AIO-04 [PMID 26189067] (2015) mixed results for adding oxaliplatin to CRT
• FOWARC [PMID 31557064] (2019) China challenged “all LARC needs CRT”
• RAPIDO / PRODIGE-23 (2021) TNT positive
• STELLAR (2022) Asian short-course TNT
• PROSPECT (2023) RT omission

3.6 The dedicated role of Chinese data in CRC

Unlike HCC’s “China-led global IO combinations” or NSCLC’s “parallel independent Chinese PD-1 phase IIIs,” Chinese data in CRC plays the role of “3L+ breakthrough + Asian-population TNT validation + RT-omission forerunner”:

• FRESCO [PMID 29946728] + FRESCO-2 [PMID 37331369]: fruquintinib pushed from Chinese 3L (FRESCO 2018) to global 3L+ (FRESCO-2 2023 FDA approval) — a significant case of a China-originating oncology drug obtaining first FDA approval in mCRC.
• STELLAR [PMID 35263150] (Jin Jing, CAMS): China-led short-course TNT phase III, OS even superior to long-course CRT — Asian-population-dedicated LARC TNT data.
• FOWARC [PMID 31557064] (Deng Yanhong, Sun Yat-sen University): challenged the classical notion that “all LARC needs CRT” — 2019 logical foundation for PROSPECT 2023’s RT omission.
• REGOTORI [PMID 34622226] (Wang Feng): regorafenib + toripalimab exploring IO+TKI direction in MSS CRC + gut-microbiome correlative — left hypotheses for subsequent pMMR CRC IO combinations.
• PARADIGM (Japan + partial China [PMID 37071094]): though Japan-led, Asian-population data — prospective validation of sidedness.

4. Research Gaps: ten unresolved clinical questions

This report identifies the following gaps, all definable specific questions (not boilerplate “more research needed”):

1. IO breakthrough in MSS/pMMR mCRC: ~85-95% of CRC; IMBLAZE370 [PMID 31003911] three-arm failure; REGOTORI [PMID 34622226] early signal with small N. ATEZOTRIBE [PMID 35636444] overall-population PFS positive but pMMR subgroup HR 0.78 marginal — how to “heat” cold tumors (MEK / VEGF / LAG-3 / TIGIT / STING agonist) is the largest unmet need in CRC IO.
2. BRAFm 1L missing H2H — BREAKWATER 3-drug vs FOLFOXIRI + bev: BREAKWATER [PMID 40444708] vs investigator’s choice chemo positive but the control did not fully utilize FOLFOXIRI + bev + anti-EGFR; the cross-trial comparison between TRIBE-UPDATED [PMID 26338525] BRAFm subgroup mOS 19 months and BREAKWATER interim 30 months is biased.
3. Adjuvant IO in MSI-H colon: NICHE-2 [PMID 38838311] neoadjuvant pCR 67%; ATOMIC phase III 2027+ readout. 2026 SoC remains FOLFOX × 6 months — when to switch to IO, decision tree undefined.
4. Generalization of ctDNA-guided adjuvant de-escalation: DYNAMIC [PMID 35657320] is stage II phase III; stage III ctDNA stratification (CIRCULATE-US / BESPOKE trials ongoing) 2026-2028 readout.
5. Left-vs-right sidedness in the RAS-mut subgroup: PARADIGM [PMID 37071094] was done only in RAS WT left-sided; the interaction of RAS mut + sidedness (whether right-sided RAS mut benefits more from FOLFOXIRI) has not been phase III validated.
6. 1L positioning of HER2+ CRC: MOUNTAINEER / DESTINY-CRC01/02 all in 3L+; whether HER2+ 1L should start with cetuximab + HER2 targeted therapy (not FOLFOX + bev) — no phase III.
7. Non-G12C KRAS mutations (G12D / G12V / G13D) beyond KRAS G12C: currently ~40% of CRC has KRAS mut but only G12C (~3%) is druggable; new pan-KRAS / G12D inhibitors (MRTX1133 / RMC-6236) are in early CRC phase II, phase III pending.
8. Long-term safety of W&W after TNT: OPRA [PMID 35483010] 3-yr organ preservation + IWWD [PMID 29976470] 3-yr regrowth 25% — 5-10 year data insufficient; early biomarkers (ctDNA + MRI-RECIST) for regrowth prediction not yet standardized.
9. Replication of dMMR rectal dostarlimab at more centers: Cercek [PMID 35660797] 42-patient expansion still 100% cCR is extremely rare; global 20+ center replication ongoing; long-term local recurrence / distant metastasis rates and cumulative IO toxicity are key observation points.
10. Modern cost-benefit of stage II adjuvant chemotherapy: QUASAR [PMID 18083404] showed stage II 5-FU/LV absolute OS +3.6%; dMMR subgroup did not benefit; whether ctDNA / circulating immune signatures can refine “who benefits / who doesn’t” in stage II — currently still measured with one stick of clinical high-risk features.

5. 2024-2026 latest updates

5.1 FDA / NMPA new approvals (10 CRC-relevant excerpts)

(This section shows 10 key approvals; early approvals such as larotrectinib 2018-11 NTRK tumor-agnostic / entrectinib 2019-08 / pembrolizumab MSI-H tumor-agnostic 2017-05 are also CRC-applicable.)

5.2 Key conference readouts (2024-2026, weighted-down)

The following entries are candidate pool only prior to formal peer review; PMID-traceable ones have been promoted to the main library.

• BREAKWATER (ASCO GI 2024 + 2025 oral + NEJM 2025 [PMID 40444708]): peer-reviewed published — main library.
• CheckMate-8HW (ASCO GI 2024 oral + NEJM 2024 [PMID 39602630]): peer-reviewed published — main library.
• KEYNOTE-177 5-yr (ASCO 2024 + Ann Oncol 2025 [PMID 39631622]): peer-reviewed — main library.
• Cercek dostarlimab 42-patient expansion (ASCO 2025 + NEJM follow-up): PMID 35660797 primary; expanded data ASCO 2025 cited with down-weighting (no independent peer-reviewed new PMID).
• NICHE-3 (dMMR colon adjuvant IO phase III): ongoing, no readout.
• ATOMIC (MSI-H stage III adjuvant atezolizumab vs chemo): 2027+ readout expected.
• CheckMate-9X8 ([PMID 38485190], MSS 1L mFOLFOX6 + bev ± nivo phase II): peer-reviewed — main library; primary endpoint not met.

5.3 Ongoing phase III (selected 2025-2028 readouts)

• ATOMIC (NCT02912559, atezolizumab + mFOLFOX6 vs mFOLFOX6 adjuvant, dMMR stage III colon cancer) — 2027 readout
• NICHE-3 / NICHE-4 (dMMR colon neoadjuvant IO series) — 2026-2028
• CIRCULATE-US (NCT05174169, stage III colon cancer ctDNA-guided adjuvant de-escalation / escalation phase III) — 2027-2028
• BESPOKE CRC (ctDNA + stage II/III CRC observational extension to interventional) — 2026-2028
• POLO-like CRC KRAS G12D / pan-KRAS (MRTX1133 / RMC-6236 phase II → III) — early readouts 2026-2027
• BREAKWATER long-term OS + subgroups (BRAF extended updates) — 2026 H2

6. Convergence insights and judgments

6.1 Longitudinal × cross-sectional: the 2026 CRC landscape is shaped by four “resonances”

Overlaying the longitudinal paradigm evolution onto the cross-sectional current-decision landscape, the 2026 CRC landscape is a superposition of four resonances:

1. “5-FU → FOLFOX / FOLFIRI → bev/cet 1L → 2L/3L backfilling” — the refined chemotherapy + targeted pipeline, plus the 60-year biggest lesson “metastatic → adjuvant extrapolation trap”: AVF2107 + CRYSTAL + PRIME + FIRE-3 + CALGB-80405 + PARADIGM — six phase IIIs pushed mCRC 1L mOS from 15 months (2000) to 35 months (2023 PARADIGM left-sided RAS WT + pan arm); in the same period, NSABP C-08 / AVANT / N0147 / CALGB 89803 / PETACC-3 — five adjuvant phase IIIs cumulatively >10,000 patients negative — the only successful adjuvant extrapolation was FOLFOX in MOSAIC. This is the biggest difference between CRC and NSCLC (adjuvant osi / alec both positive) and BTC (adjuvant BILCAP / ASCOT positive) — CRC is the textbook cancer of the “adjuvant extrapolation trap.”

2. “MSI-H / dMMR from chemo-insensitive to immune-sanctuary” — a complete 7-year reversal: 2015 Le NEJM foundation → 2017 CheckMate-142 → 2020 KEYNOTE-177 1L reversal → 2022 Cercek rectal 100% cCR → 2024 CheckMate-8HW HR 0.21 (among the largest ever) → 2024 NICHE-2 neoadjuvant pCR 67% DFS 100%. The MSI-H / dMMR subgroup flipped over 7 years from worst prognosis (chemo-insensitive) to best prognosis (IO sanctuary) — simultaneously driving Lynch syndrome screening + clinical-routine MMR IHC for every newly diagnosed CRC. This pathway has a cadence similar to BTC’s 15-year FGFR2 / IDH1 precision buildup and NSCLC’s 15-year EGFR TKI iteration — but CRC MSI-H’s endpoint is more disruptive (non-metastatic dMMR rectal 100% cCR replacing surgery).

3. “BRAF V600E / HER2 / KRAS G12C / NTRK — four precision pathways jointly covering 13-15%” — the precision rockets: BEACON-CRC 2019 → BREAKWATER 2025 pushed BRAFm from “worst prognosis” to “a 1L SoC exists”; MOUNTAINEER + DESTINY-CRC01/02 pushed HER2+ from HERACLES 2016 validation to a 3-drug SoC; CodeBreaK 300 + KRYSTAL-1 extended KRAS G12C from NSCLC to CRC; NAVIGATE + STARTRK opened NTRK-fusion tumor-agnostic therapy. CRC’s precision density in 2015-2025 caught up with NSCLC’s early-10-year cadence — but the core difference is that every CRC biomarker was first established in metastatic (BRAFm / HER2 / G12C / NTRK all unvalidated in adjuvant).

4. The “subtractive” paradigm of rectal TNT + organ preservation: 2004 German CAO/ARO/AIO-94 established preoperative CRT → 2021 RAPIDO + PRODIGE-23 TNT established → 2023 PROSPECT RT omission → 2022 Cercek dMMR 100% cCR → 2022 OPRA + 2018 IWWD W&W established. Rectal cancer is one of the few CRC subgroups where “treatment shifts from addition to subtraction” — Sauer 2004 “CRT mandatory” → Schrag 2023 “low-risk can skip RT”; Cercek 2022 “dMMR can skip surgery + skip RT.” In direct contrast to colon adjuvant’s “add bev/cet/iri” subtractive failure — rectal subtraction succeeded.

These four resonances together explain a clinical phenomenon: the treatment-intent decision for a newly diagnosed CRC patient in 2026 has 8 more decision layers than in 2000 (mCRC vs locally advanced → colon vs rectal → RAS panel → BRAF V600E → MMR/MSI → HER2 → KRAS G12C → NTRK → sidedness → rectal TNT vs traditional nCRT). The chemo backbone is still 5-FU → FOLFOX/FOLFIRI, but “what to add on top of the backbone” is fully determined by molecular subtype + location.

6.2 Clinical decision takeaways (for junior-mid oncologists)

1. Newly diagnosed advanced CRC must undergo comprehensive molecular profiling: full-exon RAS + BRAF V600E + MMR/MSI + HER2 IHC/ISH + KRAS G12C + NTRK fusion. Missing any one = missing a high-yield response subgroup with ORR 30-60%. The outpatient decision window is 2-4 weeks; NGS report must be in-hand.
2. Do not move metastatic-effective drugs into adjuvant: bev / cet / pan / irinotecan are backbones in metastatic disease; in adjuvant phase IIIs, four drug classes failed six times, cumulatively >10,000 patients. Next time a patient asks “should we add this metastatic-effective drug X post-op?” — first ask: is there a corresponding adjuvant phase III? In 60 years only FOLFOX successfully moved into adjuvant.
3. MSI-H / dMMR is CRC’s only immunotherapy winner: 1L preferred nivo+ipi (CheckMate-8HW, fit patients) or pembro monotherapy (KEYNOTE-177); MSS/pMMR do not use IO (IMBLAZE370 three-arm failure). Every newly diagnosed CRC must undergo MMR IHC + MSI PCR / NGS, no exceptions.
4. dMMR locally advanced rectal cancer — prioritize dostarlimab monotherapy for organ preservation: Cercek 42-patient expansion still 100% cCR — one of the most stunning datasets in oncology from 2022-2025. In the clinic, for dMMR rectal cancer, first ask: “can we pursue the dostarlimab pathway?”
5. BRAFm mCRC 1L now has a usable regimen: BREAKWATER 3-drug (enco + cet + mFOLFOX6) — flipped from “worst-prognosis subgroup 2000-2019” to “mOS 30+ months (interim)”; 2L BEACON 2-drug (enco + cet) is already SoC. BRAFm no longer waits for death.
6. Route RAS WT by sidedness: left-sided RAS WT → anti-EGFR (cetuximab or panitumumab + FOLFOX/FOLFIRI); right-sided RAS WT → bev (avoid anti-EGFR) — PARADIGM + CALGB-80405 sidedness reanalysis. PARADIGM 2023 upgraded retrospective evidence to prospective phase III.
7. Stratify adjuvant FOLFOX per IDEA: low-risk stage III (T1-3 N1) CAPOX × 3 months (non-inferior + halved neuropathy); high-risk stage III (T4 or N2) FOLFOX × 6 months (still favored). Stage II dMMR — do not use chemotherapy (dMMR does not benefit from adjuvant 5-FU monotherapy).
8. Rectal cancer — three branches by risk + biomarker: high-risk LARC TNT (RAPIDO / PRODIGE-23); low-risk LARC PROSPECT RT-omission; dMMR rectal Cercek pathway. MRI + CRM + biopsy MMR IHC — the trio decides which branch to take.
9. 3L+ refractory mCRC has 3 options + 1 China-origin: fruquintinib (FRESCO-2) / TAS-102 + bev (SUNLIGHT) / regorafenib (CORRECT) — choose by toxicity profile + prior exposure. China-developed fruquintinib won a global FDA label in 2023 — the first China CRC drug to go global.
10. 2026 must-know 15+ CRC drug classes: 5-FU/LV + capecitabine + oxaliplatin (FOLFOX/CAPOX) + irinotecan (FOLFIRI/FOLFOXIRI) as backbone; bevacizumab + aflibercept + ramucirumab anti-angiogenic; cetuximab + panitumumab anti-EGFR; encorafenib + binimetinib (BRAF/MEK); pembrolizumab + nivolumab + ipilimumab + dostarlimab (IO); tucatinib / trastuzumab / T-DXd (HER2); sotorasib + adagrasib (KRAS G12C); larotrectinib + entrectinib (NTRK); fruquintinib + regorafenib + TAS-102 (3L+) — 60 years ago only 5-FU as a yardstick; in 2026, 15+ drugs across 5 biomarker pathways + TNT / organ preservation running in parallel, a complex decision map.

7. Information sources

The metadata for 74 trials in this report are independently verified via PubMed and ClinicalTrials.gov. Every [PMID xxxxxxxx] in the text can be verified directly in PubMed.

• Published trials: 74 entries covering 2000-2025 (all PMIDs verifiable)
• NCCN guideline citations: 74/74 hit NCCN Colon V1.2026 or NCCN Rectal V1.2026 references (most hit both)
• 2020-2025 FDA / NMPA new approvals: 10+ key approvals
• Research gaps: 10 items
• China-led proportion: ~8% (FRESCO / STELLAR / FOWARC / REGOTORI 4 + PARADIGM Japan-China collaboration)

7.1 Reference list of the report body (sorted by PMID ascending)

The table below lists all 74 trials sorted by PMID ascending. Citation density in the text is high; every PMID in this table can be clicked to the PubMed URL for verification.

7.2 Verification conventions

• Each PMID can be verified at https://pubmed.ncbi.nlm.nih.gov/{PMID}/
• Each NCT id can be accessed at https://clinicaltrials.gov/study/{NCT_id}/
• Conference abstracts (ASCO / ASCO GI / ESMO / ESMO GI) are retrievable via the official conference systems; all conference citations in this report are “down-weighted” — not peer-reviewed, with final data to be confirmed by journal publication
• If a PMID in this report points to a trial name / year / conclusion inconsistent with PubMed, corrections are welcome

Clinical trial timeline is here

Chinese: /trials/colorectal/ English: /en/trials/colorectal/

Every trial has its own detail page, including:

• Complete intervention / comparator regimens
• Primary endpoint values + 95% CI
• Key findings + clinical significance
• Clickable links to PMID / NCT source

74 trials · 6 chapters · 2000 to 2025 · NCCN Colon + NCCN Rectal V1.2026 dual-guideline-synced.

Closing

CRC has completed one of oncology’s most textbook evolutions — “from single-drug backbone to multi-dimensional precision” — over the past 60 years: from 1960s 5-FU monotherapy, 1990s leucovorin sensitization, 2000’s FOLFOX / IFL dual backbones, 2004-2014’s bev / cet / pan established in metastatic but six failures in adjuvant, 2015-2025’s MSI-H immune reversal + four precision rockets (BRAFm / HER2 / KRAS G12C / NTRK), to 2021-2023’s rectal TNT + organ-preservation subtractive paradigm.

CRC’s 60 years condensed its biggest clinical lesson into one line: “metastatic-effective ≠ adjuvant-effective” — paid for by five negative phase IIIs (NSABP C-08 / AVANT / N0147 / CALGB 89803 / PETACC-3), cumulatively >10,000 patients. The only successful adjuvant transplantation was FOLFOX (MOSAIC); bev / cet / pan / irinotecan all failed. Next time a patient asks “should we use this metastatic-effective drug post-op?” — first ask: is there a corresponding adjuvant phase III?

Another equally structural lesson is “MMR determines CRC immunotherapy’s fate”: MSS/pMMR accounts for 85-95% of CRC but IO is all negative (IMBLAZE370 three-arm failure); MSI-H/dMMR accounts for 4-5% (stage IV) to 15-20% (stage II) and flipped from chemo-insensitive to an immune sanctuary — 1L dual IO (CheckMate-8HW HR 0.21) + dMMR rectal dostarlimab monotherapy 100% cCR. All newly diagnosed CRC must undergo MMR/MSI testing, no exceptions.

At the precision-therapy level, the four pathways BRAF V600E / HER2 / KRAS G12C / NTRK jointly cover 13-15% — BEACON-CRC 2019 → BREAKWATER 2025’s six-year BRAF turnaround; MOUNTAINEER + DESTINY-CRC01/02’s HER2 3-drug regimen; CodeBreaK 300 + KRYSTAL-1’s KRAS G12C extension from NSCLC to CRC; NAVIGATE + STARTRK’s NTRK tumor-agnostic regulatory milestone. In newly diagnosed advanced CRC, missing any biomarker = missing a high-yield response subgroup with ORR 30-60%.

At the rectal-cancer level, it is a subtractive paradigm — from 2004’s “mandatory preop CRT” to 2021’s TNT, to 2022’s dMMR monotherapy 100% cCR replacing surgery + radiotherapy, to 2023’s PROSPECT low-risk RT omission. Rectal cancer is one of the few CRC subgroups where “treatment shifts from addition to subtraction”.

The value of this report is not in “exhausting all trials” (PubMed can do that), but in compressing 60 years of evolution + current decisions + unresolved gaps into the cognitive bandwidth of a single reading. Next time you face a newly diagnosed CRC patient, every branch of the decision tree has this map to consult, trace, and question.

Clinician × AI = Research Superpower + Clinical Decision Amplifier

—— Dual Brain Lab · 2026-04-21