Clinical sequencing defines the genomic landscape of metastatic colorectal cancer

Authors

Rona Yaeger

Walid K. Chatila

Marla D. Lipsyc

Jaclyn F. Hechtman

Andrea Cercek

Francisco Sanchez-Vega

Gowtham Jayakumaran

Sumit Middha

Ahmet Zehir

Mark T.A. Donoghue

Daoqi You

Agnes Viale

Nancy Kemeny

Neil H. Segal

Zsofia K. Stadler

Anna M. Varghese

Ritika Kundra

Jianjiong Gao

Aijazuddin Syed

David M. Hyman

Efsevia Vakiani

Neal Rosen

Barry S. Taylor

Marc Ladanyi

Michael F. Berger

David B. Solit

Jinru Shia

Leonard Saltz

Nikolaus Schultz

Doi

10.1016/j.ccell.2017.12.004

PMID: 29316426 · DOI: 10.1016/j.ccell.2017.12.004 · Journal: Cancer Cell (2018)

TL;DR

Yaeger and colleagues prospectively sequenced 1,134 colorectal adenocarcinomas (1,011 tumors from 979 metastatic CRC patients plus 123 early-stage tumors) at MSKCC using the MSK-IMPACT targeted capture panel. The study refines the genomic landscape of metastatic CRC: it discovers a recurrent intronic splice-acceptor variant near APC exon 9 and large in-frame CTNNB1 exon 3 deletions that together push WNT pathway alteration frequency to 96% of CRCs. Right-sided microsatellite-stable mCRC has shorter survival, higher mutation burden, and enrichment of oncogenic KRAS, BRAF, PIK3CA, AKT1, RNF43, and SMAD4 alterations versus left-sided primaries; left-sided tumors more often lack mitogenic-pathway mutations but show higher RTK ligand expression. The full dataset is the cBioPortal crc_msk_2017 study.

Cohort & data

  • 1,134 colorectal adenocarcinomas total: 1,011 tumors (478 primaries, 533 metastases) from 979 patients with metastatic CRC, plus 123 tumors from 120 patients with early-stage CRC (PMID:29316426).
  • Cancer types: colorectal adenocarcinoma — COADREAD (with subsites COAD and READ).
  • Dataset: crc_msk_2017 — public on cBioPortal at cbioportal.org/study?id=crc_msk_2017; raw data deposited at European Variation Archive accession PRJEB23844.
  • Assay: MSK-IMPACT hybridization capture NGS — panel size expanded during study from 341 to 410 to 468 genes (IMPACT341 used in 214 cases, IMPACT410 in 911, IMPACT468 in 9). Mean coverage 747×; matched tumor/normal calling in a CLIA-certified lab.
  • MSI status assigned by MSIsensor score ≥10 (98.6% concordance with MMR IHC).
  • Cohort breakdown: 1,027 MSS (90.6%), 99 MSI-H/hypermutated (8.7%), 8 POLE-mutant (0.7%). 4% of mCRC cases were MSI-H. 70% resection / 30% needle biopsy; 52% pre-treatment.
  • Median follow-up: 23.7 months for MSS early-stage, 28.6 months for MSS mCRC.

Key findings

  • 47 significantly recurrently mutated genes in MSS CRC identified via MutSig and MuSiC. Top frequencies: APC 79%, TP53 78%, KRAS 44%, PIK3CA 18%, SMAD4 16%.
  • Novel recurrently mutated genes in MSS CRC (1–4% each): PTPRS, PIK3CG, FLT4, MAP2K4, IKZF1, JUN, TBX3, FOXP1, INHBA, CDKN1B.
  • Only TP53 alterations were significantly enriched in mCRC vs early-stage; FBXW7 was enriched in early-stage tumors, suggesting a possible protective effect.
  • WNT pathway alterations rise to 96% of CRCs when intronic APC variants and large CTNNB1 exon-3 in-frame deletions are included (vs 85% MSS / 93% MSI-H using only conventional callers).
  • Recurrent APC intronic splice-acceptor mutation: chr5:112151184 A>G (NM_000038), 8 nt upstream of exon 9, found in 35 MSS cases. Specific to MSS CRC (absent in 9,939 other MSK-IMPACT tumors). Confirmed to produce a frameshifted splice product by droplet-digital-pcr in 5/5 tested tumors, with nuclear β-catenin staining on IHC consistent with WNT activation.
  • Large CTNNB1 exon-3 in-frame deletions in 29 MSS cases — exclusive to MSS, mutually exclusive with APC and other WNT pathway alterations (p<0.001). 26 deleted the entire exon. Re-analysis of TCGA CRC found 3 additional cases (<1%). Short deletions seen at lower frequencies in hepatobiliary cancer (2%) and melanoma (1%); not in endometrial cancer.
  • Right-sided MSS mCRC has shorter overall survival than left-sided: 5-year OS 45% vs 67% (p<0.001).
  • Right-sided MSS mCRC has higher mutation burden (median 7.15 vs 5.92 mut/MB, p<0.001) and higher oncogenic-mutation burden (4.11 vs 3.44, p<0.001), but fewer copy-number alterations (FGA 0.18 vs 0.21, p=0.001).
  • Enrichment of oncogenic alterations by side (right vs left): KRAS, BRAF, PIK3CA, PTEN, AKT1, RNF43, SMAD2, SMAD4 enriched right-sided; APC and TP53 enriched left-sided.
  • 37% of left-sided MSS mCRC have no detectable mitogenic-signaling alteration; these tumors show higher expression of amphiregulin, epiregulin, neuregulin, and HGF (TCGA RNA-seq), suggesting ligand-driven RTK activation.
  • Multivariate survival model (correcting for primary site): oncogenic alterations in APC (HR=0.57, p<0.01), BRAF (HR=2.02, p<0.001), KRAS (HR=1.40, p<0.01), and NRAS (HR=2.59, p<0.01) were independent predictors; primary site was not (HR=1.29, p=0.14).
  • Five genomic mitogenic-pathway subgroups defined: (1) RTK-only, (2) RAS-MAPK, (3) PI3K, (4) RAS-MAPK + PI3K concurrent, (5) no alteration. >75% of right-sided tumors fall in subgroup 4; ~50% of left-sided tumors fall in subgroups 1 or 5. Subgroups significantly differ in survival and metastatic patterns.
  • High primary–metastasis genomic concordance in 18 patients with multiple sequenced specimens.

Genes & alterations

  • APC — 79% of MSS tumors; recurrent intronic chr5:112151184 A>G splice-acceptor mutation expands oncogenic spectrum into introns; oncogenic alterations were a positive prognostic factor in mCRC (HR=0.57).
  • TP53 — 78%; only gene significantly enriched in mCRC vs early-stage primaries.
  • KRAS — 44% MSS; hotspot mutations are OncoKB Level 1 resistance markers for anti-EGFR antibodies; enriched in right-sided primaries; HR=1.40 for OS.
  • NRAS — Level 1 anti-EGFR resistance marker; HR=2.59 for OS.
  • BRAF — V600E in 22% MSI-H vs 5% MSS mCRC; right-sided enrichment; HR=2.02. Class 1 (V600), class 2 (RAS-independent dimers), and class 3 (RAS-dependent, EGFR-amplifying) mutants all observed; classes co-mutate differently with RAS-pathway genes; class 3 mutants predicted EGFR-antibody-sensitive. Two MSS mCRC cases harbored activating BRAF fusions (AGAP3-BRAF, CUL1-BRAF).
  • PIK3CA — 18% overall; 41% MSI-H vs 16% MSS (p<0.01); right-sided enrichment.
  • PTEN — 35% MSI-H vs 5% MSS (p<0.01).
  • CTNNB1 — 8% total (24% MSI-H, 6% MSS). Hotspot N-terminal exon 3 mutations (T41A, S45F, S45P) and APC/Axin binding-domain mutations (K335I, W383R) mostly in MSI-H. Large in-frame exon-3 deletions (26 whole-exon, 3 partial) exclusive to MSS — eliminate β-catenin priming and GSK3β phosphorylation sites, escape β-TRCP-mediated degradation, drive nuclear β-catenin.
  • RNF43 — 9%; right-sided enrichment in MSS mCRC.
  • SMAD4 — 16%; right-sided enrichment.
  • SMAD2 — right-sided enrichment.
  • AKT1 — right-sided enrichment.
  • ERBB2 — amplifications in 4% of MSS mCRC; the most common RTK alteration; one case co-occurred with EGFR S492R as a likely cetuximab-resistance mechanism.
  • EGFR — hotspot mutations in 4 tumors (two L861Q, one L858R, one R776H) potentially TKI-sensitizing; one S492R cetuximab-resistance allele observed after 1.5 years of cetuximab.
  • POLE — exonuclease-domain hotspots P286R, S459F, V411L defined the ultra-mutated subgroup (n=8, 0.7%); typically early-stage and male.
  • MSH6, MLH1, MSH2, PMS2 — assayed by IHC for MMR status; concordance with MSIsensor 98.6%.
  • BRCA1 / BRCA2 — alterations in 22% MSI-H vs 1% MSS.
  • NTRK1 / NTRK3 fusions — 8% MSI-H vs 1% MSS. Specific fusions: LMNA-NTRK1 (3 cases total: 1 MSS, 2 MSI-H), TPM3-NTRK1, ETV6-NTRK3.
  • RET — one NCOA4-RET fusion in MSS mCRC.
  • FBXW7 — enriched in early-stage vs mCRC (suggested protective effect).
  • Novel recurrently mutated: PTPRS, PIK3CG, FLT4, MAP2K4, IKZF1, JUN, TBX3, FOXP1, INHBA, CDKN1B.
  • Univariate survival hits in MSS mCRC also included DNMT3B and SRC, but only APC, BRAF, KRAS, NRAS held in multivariate analysis.

Clinical implications

  • Pembrolizumab eligibility (Level 1): all 99 MSI-H/hypermutated cases are now actionable per OncoKB after FDA approval of pembrolizumab for MSI-H solid tumors.
  • Anti-EGFR antibody resistance (Level 1): hotspot KRAS and NRAS mutations are Level 1 resistance markers for cetuximab and panitumumab. One acquired-resistance case carried EGFR S492R after 1.5 years of cetuximab plus concurrent ERBB2 amplification.
  • Actionability rates by subtype (oncokb-annotation): 86% of MSI-H/hypermutated vs 37% of MSS mCRC carried potentially actionable alterations (p<0.001). 46% of right-sided vs 30% of left-sided MSS mCRC harbored targetable oncogenic alterations.
  • Class-specific BRAF therapy implications: class 1 and class 2 BRAF mutants signal independently of EGFR and are predicted insensitive to EGFR inhibition, while class 3 (kinase-dead) mutants amplify EGFR signaling and are predicted EGFR-antibody-sensitive.
  • Prognostic stratification: APC alterations are favorable; BRAF, KRAS, NRAS alterations are unfavorable in MSS mCRC. The five genomic mitogenic-pathway subgroups (RTK-only, RAS-MAPK, PI3K, concurrent, none) stratify outcomes and metastatic tropism, and largely explain the right-vs-left survival gap (primary site itself was not significant after adjusting for genomics).
  • Sidedness biology: ligand-driven RTK signaling in left-sided tumors implies selective inhibitors of other RTKs (e.g. ERBB2-directed therapy) may have higher efficacy in left-sided MSS mCRC; right-sided tumors more often have intrinsic mitogenic-pathway mutations.
  • Rare actionable fusions: NTRK fusions in 8% of MSI-H mCRC are TRK-inhibitor candidates; activating BRAF fusions in 2 MSS cases.

Limitations & open questions

  • Targeted capture (341–468 genes) misses non-coding events outside designed regions; the discovered intronic APC variant was identified only because IMPACT covers select introns. Other splice-altering events may remain hidden.
  • Selection bias: cohort enriched for KRAS/NRAS/BRAF genotyping requests (anti-EGFR decision-making), so MSI-H frequency in mCRC (4%) may slightly under-represent the true population, and MSI-H early-stage cases are over-represented.
  • Survival analysis from “date of metastatic diagnosis” — left-truncation and informative censoring not deeply addressed.
  • APC alterations were treated as a single oncogenic group; the authors note prior data (Schell et al., 2016) suggest the number of APC hits affects survival and could refine prognostic modeling.
  • Ligand-expression analysis was performed on TCGA RNA-seq, not on the MSK cohort itself (RNA was not collected on IMPACT samples).
  • Only 18 patients had multiple sequenced specimens for primary–metastasis concordance assessment; a small N for that conclusion.
  • Mechanistic explanation for why long CTNNB1 exon-3 deletions are CRC-specific while short deletions occur in hepatobiliary and melanoma is left open.

Citations from this paper used in the wiki

  • “We performed prospective targeted sequencing of 1134 CRCs… increasing oncogenic WNT pathway alterations to 96% of CRCs” (Summary).
  • “1027 MSS tumors (90.6%), 99 MSI-H/hypermutated tumors (8.7%), and 8 POLE mutant tumors (0.7%)” (Subtype classification).
  • “Forty-seven recurrently mutated genes were identified, the most frequently mutated being APC (79%), TP53 (78%), KRAS (44%), PIK3CA (18%), and SMAD4 (16%)” (Significantly recurrently mutated genes in MSS CRC).
  • “TP53 alterations were the only genomic event significantly enriched in mCRC… FBXW7 alterations were enriched in early stage tumors” (Significantly recurrently mutated genes in MSS CRC).
  • “35 MSS cases harboring a recurrent intronic single-nucleotide variant eight nucleotides upstream of APC exon 9 (chr5:112151184 A>G)” (WNT Pathway alterations).
  • “CTNNB1 in-frame deletions were largely mutually exclusive with APC alterations and other WNT pathway alterations (p<0.001)” (WNT Pathway alterations).
  • “5-year median OS for patients with right-sided tumors was 45% compared to 67% for patients with left-sided tumors (p<0.001)” (Genomic analysis by primary tumor site).
  • “median mutation burden in patients < 50 years at diagnosis was 5.1 versus 4.2 for right- versus left-sided primary site (p<0.001)” (Genomic analysis by primary tumor site).
  • “oncogenic alterations in APC (HR=0.57, p<0.01), BRAF (HR=2.02, p<0.001), KRAS (HR=1.40, p<0.01), and NRAS (HR 2.59, p<0.01) were predictive of survival, whereas primary site was not (HR=1.29, p=0.14)” (Genomic analysis by primary tumor site).
  • “potentially actionable alterations were enriched in MSI-H/hypermutated tumors compared with MSS tumors (86% versus 37%, p<0.001)” (Clinical actionability of genomic alterations).
  • “One tumor harbored EGFR S492R, a known resistance alteration to cetuximab… this tumor also harbored ERBB2 amplification, another potential mechanism of cetuximab resistance” (Clinical actionability of genomic alterations).
  • “Our full dataset is publicly available through the cBioPortal for Cancer Genomics (http://www.cbioportal.org/study?id=crc_msk_2017)” (Data and Software Availability).

This page was processed by crosslinker on 2026-05-15.