Comparative sequencing analysis reveals high genomic concordance between matched primary and metastatic colorectal cancer lesions

Authors

A Rose Brannon

Efsevia Vakiani

Brooke E Sylvester

Sasinya N Scott

Gregory McDermott

Ronak H Shah

Krishan Kania

Agnes Viale

Dayna M Oschwald

Vladimir Vacic

Anne-Katrin Emde

Andrea Cercek

Rona Yaeger

Nancy E Kemeny

Leonard B Saltz

Jinru Shia

Michael I D’Angelica

Martin R Weiser

David B Solit

Michael F Berger

Doi

10.1186/s13059-014-0454-7

PMID: 25164765 · DOI: 10.1186/s13059-014-0454-7 · Journal: Genome Biology (2014)

TL;DR

Brannon and colleagues at Memorial Sloan Kettering performed deep targeted next-generation sequencing of 230 cancer-associated genes on 69 patient trios (matched primary tumor, metastasis, and normal tissue) from microsatellite-stable colorectal cancer patients, supplemented with whole genome sequencing of four trios. The mutation profile was 100% concordant between primary and metastatic lesions for KRAS, NRAS, and BRAF, and 93% concordant overall for recurrent driver mutations. The authors conclude that diagnostic sequencing of either primary or metastatic tissue is acceptable in current clinical practice for COADREAD patients, though heterogeneity in PI3K-pathway and RTK-RAS-pathway genes remains clinically relevant.

Cohort & data

  • N = 69 matched trios (primary CRC + metastasis + normal) from MSKCC, all microsatellite-stable, with 62/69 (90%) presenting at stage IV.
  • Primary tumor location: right colon 28 (40%), left colon 30 (44%), rectum 11 (16%).
  • Treatment: 30 (43.5%) chemonaive prior to resection; 39 (56.5%) received prior therapy. None received anti-EGFR therapy before resection.
  • Concurrent primary/metastasis resection: 52 (75%); subsequent: 17 (25%, mean interval 15.3 months).
  • Cancer type: COADREAD (colorectal adenocarcinoma).
  • Dataset: coadread_mskcc — cBioPortal study “Colorectal Adenocarcinoma Triplets”; raw data in dbGaP phs000790.v1.p1.
  • Assay: msk-impact-panel IMPACT custom capture covering all protein-coding exons of 230 actionable/cancer-related genes; mean target coverage 692X.
  • Validation assay: whole-genome-seq (WGS) of four trios at New York Genome Center / Illumina (median 87X tumor, 50X normal).
  • Adjunct methods: fish for EGFR amplification, msi-pcr-pentaplex (BAT25, BAT26, D17S250, D2S123, D5S346) for MSI status, sanger-sequencing for plasmid verification, mutect/GATK SomaticIndelDetector for calling, bwa for alignment, oncotator + COSMIC v65 for annotation, strelka for WGS variant calling.

Key findings

  • 434 distinct non-synonymous somatic mutations and indels were detected across the 69 trios; 344 (79%) were shared between patient-matched primary and metastatic tumors.
  • KRAS, NRAS, and BRAF mutations were 100% concordant between primary and metastasis — no discordant events in any of the 69 cases.
  • Mutation frequencies matched coadread_tcga non-hypermutated cohort with minor differences: APC and TP53 were enriched (likely due to higher sequencing depth and more advanced disease in this cohort), NRAS was less frequent.
  • Among 247/434 (57%) mutations falling in genes called significantly mutated by [[TCGA]] for non-hypermutated CRC, 229/247 (93%) were concordant between primary and metastasis.
  • 18 private mutations in recurrent driver genes were found in APC (n=7), PIK3CA (n=5), SMAD4 (n=3), and TP53 (n=3); 5/7 APC private mutations were secondary events on top of a shared clonal APC mutation.
  • Intra-tumor heterogeneity accounts for some discordance: resequencing 97 additional spatially separate regions from 22 patients resolved 17/46 originally discordant mutations, including 12/22 (55%) “metastasis-only” calls subsequently found subclonally in the primary.
  • Convergent/parallel evolution observed in two patients with distinct mutations in the same gene: one with independent TP53 R248Q and Y163* mutations, another with separate hotspot PIK3CA E542K and E545K mutations.
  • Metastasis-private RTK-RAS events in RAS/RAF wild-type tumors:
    • Patient 19: MAP2K1 Q56P (MEK1) detected only in metastasis. Transfection into HEK 293H cells showed Q56P hyperactivates ERK signaling comparable to known activating mutation K57N. An additional A106T mutation did not affect signaling.
    • Patient 3: chromosome 7p amplification present only in the metastasis, encompassing CARD11, ETV1, IKZF1, and EGFR; fish confirmed high-level focal EGFR amplification in the metastasis (primary showed only 7p polysomy).
  • PI3K-pathway heterogeneity: private mutations of unknown significance found in PIK3CD, PIK3CG, PIK3C2G, PIK3R1, and PTEN; four were detectable subclonally in other regions.
  • Clinical correlates: no difference in mutational concordance by primary tumor location or resection interval; no disease-specific survival difference between concordant and discordant cases.
  • Treatment effect on heterogeneity: among concurrently resected cases, chemonaive patients harbored discordant mutations more often than pre-treated patients (22/28, 79% vs 11/24, 46%; chi-square P=0.01). Pre-treated patients also had a slightly smaller mean primary tumor (4.2 vs 5.2 cm; T-test P=0.02), suggesting treatment may debulk or homogenize tumors.
  • WGS validates IMPACT: all SNVs/indels above 5% allele frequency detected by IMPACT were confirmed by WGS, with no additional mutations in the 230 panel genes. WGS of discordant cases (patients 3 and 19) confirmed broad genome-wide discordance (38% and 25% shared mutations genome-wide, rising to 46% and 32% for non-synonymous exonic).
  • Case-specific WGS findings: patient 14 carried a CDK8 amplification (dysregulates WNT/β-catenin) and patient 54 carried two RASA1 nonsense mutations (C372, R679) shared between primary and metastasis — RASA1 is altered in ~2% of TCGA CRC, mutually exclusive with KRAS/NRAS. This finding prompted addition of RASA1 capture probes to subsequent msk-impact-panel versions.
  • Patient 3 also harbored a metastasis-specific PTK7 Q304* nonsense mutation; PTK7 inhibits canonical Wnt signaling via frizzled receptor binding but no PTK7 nonsense mutations are reported in TCGA CRC.

Genes & alterations

  • KRAS, NRAS, BRAF — driver mutations 100% concordant between primary and metastasis; trunk events established early in CRC carcinogenesis.
  • APC — higher prevalence than TCGA; private mutations in 7 patients, mostly secondary to a shared clonal hit; in patient 10, the primary-private APC nonsense was rendered phenotypically equivalent in the metastasis by chromosomal deletion of the APC locus.
  • TP53 — higher prevalence than TCGA (attributed to metastatic-stage cohort); parallel evolution case with independent R248Q and Y163* mutations.
  • PIK3CA — convergent evolution with independent E542K and E545K hotspots; three additional private events including primary-specific E545K (×2) and metastasis-specific N107 in-frame deletion.
  • MAP2K1 — metastasis-private Q56P confirmed by transfection to hyperactivate ERK signaling at levels comparable to known-activating K57N; A106T verified inactive by Western blot.
  • EGFR — metastasis-private high-level amplification (chr 7p) confirmed by fish; co-amplified with CARD11, ETV1, IKZF1.
  • SMAD4, SMAD3 — metastasis-specific events in patient 3 including a missense mutation in the SMAD3 transcriptional-regulation domain.
  • PTK7 — novel metastasis-specific Q304* nonsense mutation in patient 3; functional relevance speculative (Wnt inhibition via frizzled).
  • RASA1 — two nonsense mutations (C372, R679) in patient 14, shared between primary and metastasis; mutually exclusive with RAS mutations; not on original msk-impact-panel design, added in subsequent panel version.
  • CDK8 — focal amplification in patient 14 in absence of β-catenin/RAS-pathway mutations; known WNT/β-catenin regulator.
  • PIK3CD, PIK3CG, PIK3C2G, PIK3R1, PTEN — private PI3K-pathway events of unknown significance; four also detected in spatially separate regions, indicating subclonality.
  • AMER1 (FAM123B), FBXW7, CTNNB1 — recurrently mutated as expected for non-hypermutated CRC; concordant per Figure 1B.

Clinical implications

  • For routine clinical molecular profiling of metastatic CRC, sequencing either the primary tumor or the metastasis is appropriate — KRAS/NRAS/BRAF status, which guides cetuximab and panitumumab eligibility, is identical in both sites.
  • Exceptions where testing the metastasis may add value: patients with multiple primary tumors/polyps, and patients whose interval drug treatment may have selected for clinically actionable subclones (e.g. anti-EGFR therapy is known to select for KRAS mutant subclones in initially KRAS/NRAS/BRAF wild-type tumors).
  • In RAS/RAF wild-type CRC, the metastasis can harbor private actionable RTK-RAS events (MAP2K1 activating mutations, EGFR amplification) that would be missed by primary-only sequencing — relevant when anti-EGFR therapy is being considered.
  • PI3K-pathway heterogeneity is clinically relevant given active development of PI3K inhibitors in CRC and evidence (Domingo et al. JCO 2013) that PIK3CA mutations may predict benefit from aspirin therapy.
  • Targeted sequencing is a viable cost- and time-effective alternative to WGS for clinical CRC molecular profiling: WGS in this cohort did not surface additional actionable findings in the 230-gene panel space, though it did reveal one rare event (RASA1 nonsense) that motivated panel expansion.

Limitations & open questions

  • Cohort enriched for stage IV disease (90%) — findings may not generalize to earlier-stage primaries.
  • Only microsatellite-stable tumors included; concordance dynamics in MSI-high CRC may differ.
  • Treatment-vs-heterogeneity association (chemonaive cases more discordant; smaller post-treatment primaries) is correlational; the authors note larger studies are needed to determine whether systemic treatment causes preferential mutational selection vs general debulking.
  • IMPACT panel (230 genes) cannot detect discordance in genes outside the panel by design — addressed for four cases by WGS but not the full cohort.
  • PTK7 Q304* functional relevance is unproven; no other CRC nonsense mutations reported in TCGA.
  • Sample size (69 trios, 4 WGS) is modest; subgroup analyses (location, resection timing) underpowered.

Citations from this paper used in the wiki

  • “Mutation profiles were 100% concordant for KRAS, NRAS, and BRAF, and were highly concordant for recurrent alterations in colorectal cancer.” (Abstract, p. 1)
  • “Of the 434 total mutations, 344 (79%) were shared between patient-matched tumors.” (Results, p. 2)
  • “Among the mutations in the genes reported by the TCGA to be significantly mutated in non-hypermutated tumors (Figure 1B; 247/434, 57%), there was very high (93%, 229/247) concordance between primary tumors and matched metastases.” (Results, p. 2)
  • “In two cases, MAP2K1 (MEK1) mutations (A106T and Q56P) were detected only in the metastatic sample… Transfection of a GFP-tagged MAP2K1 Q56P plasmid into human embryonic kidney 293H cells resulted in hyperactivation of ERK signaling to a similar degree as other known activating mutations of MAP2K1.” (Results, p. 4)
  • “In another patient, we found a gain of chromosome 7p, which included the CARD11, ETV1, IKZF1, and EGFR genes in the metastasis but not in the primary tumor… FISH analysis confirms regions of high level amplification of EGFR in the metastatic tumor while the primary tumor only shows 7p polysomy.” (Results, p. 4–5)
  • “Among patients whose tumors were concurrently resected, those that did not receive prior treatment were more likely to harbor discordant mutations (22/28, 79%) compared to those that received prior therapy (11/24, 46%; chi-square P=0.01).” (Results, p. 5)
  • “RASA1, a negative regulator of RAS, was altered in 2% of CRC TCGA samples, with mutations in this gene mutually exclusive of KRAS and NRAS mutations. This result prompted the addition of capture probes for the RASA1 gene in our current version of IMPACT.” (Results, p. 7)
  • “Data are publically available through dbGaP (accession phs000790.v1.p1) and the Memorial Sloan Kettering Cancer Center cBioPortal for Cancer Genomics… study ‘Colorectal Adenocarcinoma Triplets’.” (Data availability, p. 9)

This page was processed by crosslinker on 2026-05-11.