Comprehensive molecular characterization of human colon and rectal cancer

Author

The Cancer Genome Atlas Network

Doi

10.1038/nature11252

PMID: 22810696 · DOI: 10.1038/nature11252 · Journal: Nature (2012)

TL;DR

The Cancer Genome Atlas performed genome-scale analysis of 276 colorectal carcinomas using exome sequencing, DNA copy number arrays, promoter methylation profiling, and mRNA/miRNA expression. The study identified 16% of CRCs as hypermutated (75% MSI-H with MLH1 silencing, 25% with mismatch-repair gene or POLE mutations). Excluding hypermutated tumours, colon and rectal cancers showed indistinguishable genomic alteration patterns. Twenty-four genes were significantly mutated, with novel frequent mutations in ARID1A, SOX9, and AMER1 (FAM123B). The study identified drug-targetable ERBB2 amplifications and novel IGF2 amplifications, a recurrent NAV2-TCF7L1 fusion, and demonstrated near-ubiquitous WNT pathway activation and MYC-directed transcriptional programs.

Cohort & data

  • 276 colorectal carcinoma tumour/normal pairs from the TCGA coadread_tcga_pub cohort
  • 224 pairs analysed by whole-exome sequencing; 97 by low-pass whole-genome sequencing
  • 257 tumours profiled for somatic copy-number alterations with Affymetrix SNP 6.0 arrays
  • 236 tumours profiled for promoter DNA methylation (Illumina HumanMethylation27)
  • mRNA expression (Agilent microarrays + RNA-Seq), miRNA expression
  • Cancer type: COADREAD (colorectal adenocarcinoma)

Key findings

  • 16% of CRCs are hypermutated: 77% of these are MSI-H (usually with MLH1 methylation/CIMP), 23% have somatic mismatch-repair or POLE mutations without MSI-H.
  • Among non-hypermutated tumours, the eight most frequently mutated genes are APC (81%), TP53 (60%), KRAS (43%), PIK3CA (18%), FBXW7 (11%), SMAD4 (10%), TCF7L2 (9%), and NRAS (9%).
  • Non-hypermutated colon and rectal cancers are genomically indistinguishable (copy number, methylation, mRNA, miRNA patterns).
  • WNT signalling pathway altered in 93% of all tumours (biallelic APC inactivation or activating CTNNB1 mutations in ~80%).
  • 55% of non-hypermutated tumours have alterations in KRAS, NRAS, or BRAF, with mutual exclusivity.
  • ERBB2 amplification at 17q21.1 found in 4% of tumours; potentially targetable with trastuzumab.
  • Focal amplification of 11p15.5 (7% of tumours) targets IGF2 and miR-483; IGF2 overexpression is mutually exclusive with PI3K pathway mutations (P < 0.01).
  • Recurrent NAV2-TCF7L1 fusion found in 3 cases, joining NAV2 exons 1-2 with the 3’ portion of TCF7L1, lacking the beta-catenin binding domain.
  • Co-occurrence of RAS and PI3K pathway alterations in one-third of tumours (P = 0.039).
  • PARADIGM integrative analysis revealed near-universal changes in MYC transcriptional targets across all tumours.

Genes & alterations

  • APC: Inactivating mutations in 81% non-hypermutated, 51% hypermutated; biallelic inactivation common.
  • TP53: Mutations in 60% non-hypermutated, 20% hypermutated (P < 0.0001).
  • KRAS: Oncogenic codon 12/13 mutations in 43% non-hypermutated; mutually exclusive with NRAS and BRAF.
  • NRAS: Codon 61 mutations in 9% non-hypermutated.
  • BRAF: V600E mutations dominant in hypermutated tumours.
  • PIK3CA: Mutations in 18% non-hypermutated; mutually exclusive with PIK3R1 and PTEN loss.
  • PIK3R1: Mutations in 2% non-hypermutated.
  • PTEN: Deletions/mutations in 4% non-hypermutated.
  • SMAD4: Mutations and deletions; chromosome 18 lost in 66% of tumours.
  • SMAD2: Mutated in non-hypermutated tumours.
  • CTNNB1: Activating mutations in tumours lacking APC mutations.
  • SOX9: Novel — all 9 mutated alleles in non-hypermutated CRCs were frameshift or nonsense; not previously associated with human cancer.
  • ARID1A: Frameshift/nonsense mutations; chromatin remodelling gene.
  • AMER1 (FAM123B/WTX): X-linked negative WNT regulator; virtually all mutations were loss-of-function.
  • FBXW7: Inactivating mutations in 11% non-hypermutated; never co-occurred with distant metastasis (P = 0.0019).
  • TCF7L2: Deleted or mutated in 12% non-hypermutated; focal deletion at 10p25.2.
  • ERBB2: Focal amplification at 17q21.1 in 4%; recurrent V842I mutation.
  • ERBB3: Recurrent V104M mutation in 2 non-hypermutated cases.
  • IGF2: Focal amplification at 11p15.5 in 7%; overexpression mutually exclusive with PI3K pathway events.
  • IRS2: Overexpression on chromosome 13; mutually exclusive with IGF2 overexpression and PI3K mutations.
  • MLH1: Epigenetic silencing via methylation in 19 of 23 MSI-H hypermutated tumours.
  • TGFBR2: Recurrently mutated in hypermutated cancers (60%).
  • ACVR2A: Most frequently mutated gene in hypermutated tumours (63%).
  • MYC: Amplification at 8q24; transcriptional targets universally altered per PARADIGM.
  • NAV2-TCF7L1: Recurrent fusion in 3 cases; TCF7L1 loses beta-catenin binding domain.
  • POLE: Somatic mutations in ultramutated subset lacking MSI-H.
  • MSH3, MSH6, MSH2, PMS2: Mismatch-repair gene mutations in hypermutated tumours.

Clinical implications

  • ERBB2 amplifications (4%) represent a targetable alteration; breast and gastric cancers with ERBB2 amplification respond to trastuzumab.
  • IGF2-IGF1R-IRS2 axis signals to PI3K, suggesting therapeutic targeting could block PI3K activity in a subset of patients.
  • Simultaneous inhibition of RAS and PI3K pathways may be required given their co-occurrence in one-third of tumours.
  • WNT signalling inhibitors and small-molecule beta-catenin inhibitors are potential therapeutic avenues (93% WNT activation).
  • Proteins in the RTK-RAS and PI3K pathways (IGF2, IGF1R, ERBB2, ERBB3, MEK, AKT, MTOR) are potential targets for inhibition.
  • Mutation rate (hypermutated vs non-hypermutated) may be a better prognostic indicator than MSI status alone, given survival differences.
  • Non-hypermutated colon and rectal cancers are genomically equivalent, supporting unified treatment strategies.

Limitations & open questions

  • Survival data were not available for most tumours (prospective collection); aggressiveness was inferred from tumour stage, lymph node status, and vascular invasion.
  • Low-pass whole-genome sequencing (~3-4x coverage) limits sensitivity for detecting structural variants.
  • FAM123B (AMER1) is X-linked, complicating interpretation of loss-of-function in males vs females.
  • The functional significance of the NAV2-TCF7L1 fusion and its therapeutic implications remain unclear.
  • Why hypermutated tumours predominantly arise from the right colon is unexplained.
  • The role of FZD10 overexpression (~17% of samples, up to 100x normal) in WNT-driven CRC warrants further investigation.

Citations from this paper used in the wiki

  • “16% of colorectal carcinomas were found to be hypermutated: three-quarters of these had the expected high microsatellite instability, usually with hypermethylation and MLH1 silencing, and one-quarter had somatic mismatch-repair gene and polymerase e (POLE) mutations.”
  • “Excluding the hypermutated cancers, colon and rectum cancers were found to have considerably similar patterns of genomic alteration.”
  • “Twenty-four genes were significantly mutated, and in addition to the expected APC, TP53, SMAD4, PIK3CA and KRAS mutations, we found frequent mutations in ARID1A, SOX9 and FAM123B.”
  • “An amplicon at 17q21.1, found in 4% of the tumours, contains seven genes, including the tyrosine kinase ERBB2.”
  • “We found a pattern of near exclusivity (corrected P < 0.01) of IGF2 overexpression with genomic events known to activate the PI3K pathway.”
  • “We found that the WNT signalling pathway was altered in 93% of all tumours.”
  • “We found that 55% of non-hypermutated tumours have alterations in KRAS, NRAS or BRAF, with a significant pattern of mutual exclusivity.”
  • “Co-occurrence of alterations involving the RAS and PI3K pathways in one-third of tumours (P = 0.039, Fisher’s exact test).”

This page was processed by paper-compiler on 2026-05-06.