Genetic predictors of response to systemic therapy in esophagogastric cancer

Authors

Yelena Y. Janjigian

Francisco Sanchez-Vega

Philip Jonsson

Walid K. Chatila

Jaclyn F. Hechtman

Geoffrey Y. Ku

Jamie C. Riches

Yaelle Tuvy

Ritika Kundra

Nancy Bouvier

Efsevia Vakiani

Jianjiong Gao

Zachary J. Heins

Benjamin E. Gross

David P. Kelsen

Liying Zhang

Vivian E. Strong

Mark Schattner

Hans Gerdes

Daniel G. Coit

Manjit Bains

Zsofia K. Stadler

Valerie W. Rusch

David R. Jones

Daniela Molena

Jinru Shia

Mark E. Robson

Marinela Capanu

Sumit Middha

Ahmet Zehir

David M. Hyman

Maurizio Scaltriti

Marc Ladanyi

Neal Rosen

David H. Ilson

Michael F. Berger

Laura Tang

Barry S. Taylor

David B. Solit

Nikolaus Schultz

Doi

10.1158/2159-8290.CD-17-0787

PMID: 29122777 · DOI: 10.1158/2159-8290.CD-17-0787 · Journal: Cancer Discovery (2018)

TL;DR

Janjigian et al. report the first 295 patients with stage IV esophagogastric adenocarcinoma prospectively profiled at MSKCC using the MSK-IMPACT targeted next-generation sequencing assay (341/410/468-gene panels), with mature clinical annotation including treatment response and survival on first-line platinum chemotherapy, trastuzumab, and immune checkpoint inhibitors. Three principal findings: (1) homologous recombination deficiency (LST score) was not predictive of response to platinum chemotherapy in this disease; (2) MSI-H tumors (only 9 of 295, 3%) were intrinsically resistant to chemotherapy but enriched for durable responses to anti-PD-1 therapy, and the sole EBV+ patient achieved a complete, durable immunotherapy response; (3) ERBB2 amplification level by NGS predicted trastuzumab benefit better than IHC/FISH, while loss of ERBB2 amplification, ERBB2 exon 16 deletion, and co-mutations in the RTK-RAS-PI3K pathway emerged as recurrent mechanisms of intrinsic and acquired trastuzumab resistance. The dataset is publicly available as egc_msk_2017 on cBioPortal.

Cohort & data

  • Sample size: 295 patients with metastatic (stage IV) esophageal, gastric, and gastroesophageal junction adenocarcinoma; 318 tumor samples (some patients with paired pre- and post-treatment biopsies). Consented February 2014–February 2017 PMID:29122777.
  • Cancer types: Esophagogastric Adenocarcinoma (EGC), spanning esophageal carcinoma (ESCA), stomach adenocarcinoma (STAD) including diffuse-type stomach adenocarcinoma (DSTAD), and gastroesophageal junction (GEJ) tumors PMID:29122777.
  • Dataset: egc_msk_2017 — publicly available at cbioportal.org/study?id=egc_msk_2017; committed to AACR Project GENIE PMID:29122777.
  • Assay: MSK-IMPACT, a CLIA-certified hybrid-capture targeted NGS assay run as 341- (IMPACT341), 410- (IMPACT410), or 468-gene (IMPACT468) panels on paired tumor/normal blood, mean coverage 744X, capable of detecting mutations, copy-number alterations, and select rearrangements PMID:29122777.
  • Auxiliary methods: MSI status from sequencing data via MSIsensor (MSI-H ≥ 10); allele-specific copy number and tumor purity via FACETS; LST scores from copy-number data (HRD surrogate) on samples with ≥20% tumor purity; oncogenic effect via OncoKB annotation; HER2 status by IHC (3+ positive) and FISH per CAP/ASCO criteria; EBV by EBER-ISH on 26 immunotherapy-treated patients PMID:29122777.
  • Treatment subsets analyzed: 187 HER2-negative patients on first-line fluoropyrimidine/platinum; 50 HER2+ patients (pre-treatment biopsy) on first-line trastuzumab/chemotherapy; 23 paired pre-/post-trastuzumab tumor pairs (3 prospective + 20 retrospective augmentation); 40 patients on immune checkpoint inhibitors (5 MSI-H, 35 non-MSI-H) PMID:29122777.

Key findings

  • High clinical actionability. 53% of patients had at least one OncoKB-defined potentially actionable alteration. Most frequent oncogenic/likely oncogenic alterations in the chromosomal-instability (CIN) subset: ERBB2 (25%), KRAS (16%), EGFR (8%), ERBB3 (7%), PIK3CA (7%), FGFR2 (5%), MET (5%) PMID:29122777.
  • Molecular subtype distribution differs from TCGA. CIN subtype dominated (63%); genomically stable (GS, 34%) tumors were enriched for diffuse histology (32% vs 9%, P=3e-5) and CDH1 mutations (20% vs 7%, P=0.01). Only 9/295 (3%) tumors were MSI-H — significantly lower than TCGA’s 16% (P=8e-10), attributed to the metastatic enrichment of the MSK cohort PMID:29122777.
  • Most-mutated genes: TP53 (73%), ARID1A (15%), CDKN2A (12%) PMID:29122777.
  • No primary-vs-metastatic differences. Alteration frequencies did not differ significantly between primary and metastatic samples (Supplementary Figure 1B) PMID:29122777.
  • HRD/LST is not a chemotherapy biomarker in EGC. LST score did not predict PFS on first-line platinum (HR=0.99, P=0.947, log-rank) and was not elevated in patients with prolonged response (>24 months, P=0.6); the two longest outlier responders (68 and 104 months) harbored no somatic alterations in known HR genes including BRCA1/BRCA2 PMID:29122777.
  • MSI-H tumors are chemotherapy-refractory but immunotherapy-sensitive. Median PFS on first-line chemotherapy: 4.8 months (MSI-H) vs 6.9 months (non-MSI-H), HR=0.4, P=0.027. Of 5 MSI-H patients who received immunotherapy, multiple achieved durable responses PMID:29122777.
  • TMB predicts immunotherapy outcome. Patients in the top quartile of tumor mutational burden (>9.7 mut/Mb) had median OS 16.8 vs 6.62 months and 2-year OS 44% vs 14% (HR=0.40, log-rank P=0.058) PMID:29122777.
  • EBV+ exceptional response. The single EBV+ patient (1/26 EBER-ISH tested) achieved a complete and durable response (>30–32 months and ongoing) to anti-PD-1 therapy PMID:29122777.
  • B2M loss is a recurrent immunotherapy-resistance mechanism. Acquired loss-of-function mutation in exon 1 of B2M emerged in a post-progression sample of an MSI-H/PD-L1+ patient who initially achieved CR on anti-PD-1, with confirmation by IHC. Across the cohort, 4/9 (44%) MSI-H tumors carried likely deleterious B2M alterations; some patients still responded despite baseline B2M mutations PMID:29122777.
  • NGS-based ERBB2 amplification is concordant with IHC/FISH. PPV 90%, NPV 96.9%, overall concordance 93.7%. 92% (46/50) of clinical HER2+ patients receiving first-line trastuzumab were ERBB2-amplified by NGS PMID:29122777.
  • NGS-discordant HER2+ patients have inferior trastuzumab outcomes. Median PFS 5.8 months (4 NGS-negative) vs 14.0 months (NGS-positive); P=1e-6, log-rank — discordance attributable to tumor heterogeneity or equivocal IHC/FISH PMID:29122777.
  • Higher ERBB2 amplification level predicts longer trastuzumab benefit. Patients in the top quartile of ERBB2 amplification by NGS had median PFS of 24.3 months on first-line trastuzumab PMID:29122777.
  • Co-alterations in RTK-RAS-PI3K pathway shorten trastuzumab PFS. Median PFS in HER2+ patients with such co-alterations was 8.4 months vs longer in pathway-wildtype patients; in multivariate analysis, both ERBB2 level and PI3K-pathway co-alteration independently contributed to PFS differences PMID:29122777.
  • Cell-cycle alterations did not affect trastuzumab benefit in the MSK cohort (12.2 vs 14.0 months median time on treatment; P=0.11), in contrast to a prior Asian-population report PMID:29122777.
  • Loss of ERBB2 amplification is a common acquired-resistance mechanism. Of 44 post-trastuzumab samples from clinical HER2+ patients, 7 (16%) were ERBB2-negative by NGS; HER2 protein loss was confirmed by IHC in all 7. Two patients showed loss of ERBB2 amplification and one had a focal ERBB2 exon 16 deletion in the post-progression sample only PMID:29122777.
  • Acquired RTK-RAS-PI3K activating mutations. Activating KRAS co-mutations were found in 1/50 (2%) pre-trastuzumab vs 3/23 (13%) post-trastuzumab tumors; activating PIK3CA mutations in 1/50 (2%) vs 2/23 (8.6%) PMID:29122777.
  • Comparison with TCGA non-MSI-H EGC. TP53 was enriched in the MSK cohort (73% vs 62%, q=0.11), while KMT2C, GRIN2A, PTPRD, and CTNNB1 were less frequently mutated in MSK at 15% FDR PMID:29122777.

Genes & alterations

  • ERBB2 — Focal amplification in 25% of CIN tumors. Level of amplification by NGS predicts trastuzumab PFS (top quartile: median PFS 24.3 mo). Loss of ERBB2 amplification (in 7/44 post-trastuzumab samples; 16%) and acquired ERBB2 exon 16 deletion (constitutively hyperphosphorylated delta-16 HER2 variant) are recurrent acquired-resistance mechanisms PMID:29122777.
  • TP53 — Most frequently mutated gene (73%); enriched relative to TCGA EGC non-MSI-H (62%, q=0.11) PMID:29122777.
  • ARID1A — Mutated in 15% of tumors PMID:29122777.
  • CDKN2A — Mutated in 12% of tumors PMID:29122777.
  • KRAS — Oncogenic alterations in 16% of CIN tumors. Activating co-mutations associated with intrinsic and acquired trastuzumab resistance (2% pre vs 13% post-trastuzumab tumors) PMID:29122777.
  • PIK3CA — Oncogenic alterations in 7% of CIN tumors. Activating co-mutations enriched post-trastuzumab progression (2% pre vs 8.6% post); newly acquired E545K mutation observed in one post-progression sample PMID:29122777.
  • EGFR — Oncogenic alterations in 8% of CIN tumors; among potentially targetable kinases PMID:29122777.
  • ERBB3 — Oncogenic alterations in 7% of CIN tumors PMID:29122777.
  • FGFR2 — Oncogenic alterations in 5% of CIN tumors PMID:29122777.
  • MET — Oncogenic alterations in 5% of CIN tumors; named among potentially targetable kinases PMID:29122777.
  • CDH1 — Mutated in 20% of GS tumors vs 7% of CIN tumors (P=0.01); enriched in diffuse-histology GS subset PMID:29122777.
  • B2M — Likely deleterious alterations in 4/9 (44%) MSI-H tumors. Acquired exon-1 loss-of-function mutation associated with acquired resistance to anti-PD-1 monotherapy in one MSI-H patient (β2-microglobulin loss confirmed by IHC); however, baseline B2M mutations did not preclude response in other patients PMID:29122777.
  • BRCA1 / BRCA2 — No association with response to platinum-based chemotherapy in this cohort. Authors flag as Level-2 alterations potentially predictive of PARP-inhibitor benefit PMID:29122777.
  • KMT2C, GRIN2A, PTPRD, CTNNB1 — All less frequently mutated in MSK metastatic cohort vs TCGA non-MSI-H (q≤0.10) PMID:29122777.
  • CDK4, FGFR1 — Among potentially targetable kinase targets identified in the cohort; often concurrent with other actionable kinases, suggesting combination strategies are needed PMID:29122777.

Clinical implications

  • Reflex ERBB2 and MSI testing should be standard. The authors recommend that, on the basis of FDA-approval of trastuzumab and pembrolizumab, reflex ERBB2 and MSI testing should now be standard practice in EGC PMID:29122777.
  • NGS may be a more robust HER2 biomarker than IHC/FISH. ERBB2 amplification by NGS captures both the binary status (NPV 96.9%, PPV 90% vs IHC/FISH) and the dose-dependent response gradient (top-quartile patients had median PFS 24.3 months on first-line trastuzumab) better than the categorical IHC/FISH readout, where 30% of clinical HER2+ patients lacked NGS-confirmed amplification or harbored RTK-RAS-PI3K co-mutations and progressed rapidly PMID:29122777.
  • Co-mutational profile should inform trastuzumab decisions. Patients with ERBB2-amplified, RTK/RAS/PI3K-wildtype tumors derive the greatest trastuzumab benefit; PI3K-pathway co-alteration is an independent negative predictor in multivariate analysis. Combination strategies (HER2 + PI3K/RAS pathway agents) may be required for HER2+ co-mutated patients PMID:29122777.
  • Acquired-resistance mechanisms suggest re-biopsy at progression. Loss of ERBB2 amplification (16% of post-trastuzumab samples), ERBB2 exon 16 deletion (delta-16 HER2 variant), and acquired KRAS/PIK3CA mutations all argue for tumor re-sequencing at progression to guide next-line therapy PMID:29122777.
  • Move immunotherapy earlier in MSI-H EGC. MSI-H patients fail rapidly on cytotoxic therapy (median PFS 4.8 months) but include the longest immunotherapy responders; the authors advocate considering anti-PD-1 (pembrolizumab, nivolumab, durvalumab) or combinations with anti-CTLA4 (ipilimumab, tremelimumab) early in the disease course PMID:29122777.
  • Routine EBV testing may identify exceptional responders. The single EBV+ patient achieved durable CR on immunotherapy (>30 months); the authors recommend routine EBV testing to prospectively identify likely beneficiaries PMID:29122777.
  • HRD/LST should not gate platinum decisions in EGC. Unlike ovarian or breast cancer, HRD surrogate markers did not predict platinum response — even outlier long responders lacked HR-gene alterations. Authors flag BRCA1/BRCA2 as Level-2 PARP-inhibitor candidates but not as platinum biomarkers PMID:29122777.
  • High TMB (>9.7 mut/Mb) supports immunotherapy. Top-quartile TMB was associated with median OS 16.8 vs 6.62 months on immunotherapy (HR=0.40, P=0.058) PMID:29122777.

Limitations & open questions

  • Targeted-capture blind spots. The MSK-IMPACT panel cannot detect alterations in genes outside its 341/410/468-gene design, epigenetic mechanisms (e.g., BRCA1/BRCA2 promoter methylation), or viral EBV DNA. Future panel iterations will include EBV-capture probes PMID:29122777.
  • Spatial heterogeneity unaddressed. Single-site biopsies cannot fully assess clonal complexity in multi-site metastatic disease. Authors recommend cell-free DNA approaches for future studies PMID:29122777.
  • Selection bias toward a fitter cohort. Favorable OS in the MSK cohort vs published literature may reflect 90% ECOG 0–1 status, multidisciplinary specialty care, and access to novel therapies — not solely sequencing-guided treatment PMID:29122777.
  • Small MSI-H and immunotherapy subsets. Only 9 MSI-H patients (5 receiving immunotherapy) and 40 total immunotherapy patients constrain power for MSI/TMB/EBV biomarker validation PMID:29122777.
  • EBV n=1. The dramatic EBV+ response is a single observation; EBV biomarker generalizability requires prospective replication PMID:29122777.
  • Heterogeneous post-trastuzumab paired cohort. Only 3 of 23 paired pre-/post-trastuzumab tumors came from the prospective series; 20 were retrospectively augmented, and biopsy site varied with progression site (11 of 23 same anatomical site) PMID:29122777.
  • Co-occurring kinase amplifications imply combination strategies. Among targetable kinase targets (ERBB2, EGFR, MET, CDK4, FGFR1), many were concurrent in individual patients — clinical actionability of any single marker is uncertain without combination trials PMID:29122777.
  • B2M paradox unresolved. Some patients with baseline B2M mutations responded to checkpoint blockade despite B2M loss being a recognized resistance mechanism; the determinants of why some β2-microglobulin-deficient tumors still respond are not characterized.

Citations from this paper used in the wiki

  • “There was no association between HRD defects and response to platinum-based chemotherapy. Patients with MSI-H tumors were intrinsically resistant to chemotherapy but more likely to achieve durable responses to immunotherapy.” (Abstract)
  • “The single EBV+ patient achieved a durable, complete response to immunotherapy. The level of ERBB2 amplification as determined by sequencing was predictive of trastuzumab benefit.” (Abstract)
  • “Selection for a tumor subclone lacking ERBB2 amplification, deletion of ERBB2 exon 16, and co-mutations in the receptor tyrosine kinase, RAS, PI3K pathways were associated with intrinsic and/or acquired trastuzumab resistance.” (Abstract)
  • “we report the results of the first 295 patients profiled along with accompanying prospectively captured detailed clinical annotation and treatment response data.” (p. 2)
  • “We achieved a mean sequencing coverage of 744X and identified an average of 5 non-synonymous mutations per tumor sample (range 1–63)” (p. 3)
  • “only nine samples in the MSK-cohort were MSI-H (3%), which is significantly lower than the fraction in the TCGA cohort (16%, P=8e-10, Fisher’s exact test)” (p. 3)
  • “TP53 was the most frequently mutated gene (73%), followed by ARID1A (15%) and CDKN2A (12%). In total, 53% of patients had at least one potentially actionable alteration as defined by OncoKB” (p. 3)
  • “frequent oncogenic or likely oncogenic alterations in ERBB2 (25%), KRAS (16%), EGFR (8%), ERBB3 (7%), PIK3CA (7%), FGFR2 (5%), and MET (5%)” (p. 3)
  • “LST score was not predictive of progression free survival (HR=0.99, p=0.947, log-rank test)” (p. 4)
  • “Patients with MSI-H tumors suffered rapid disease progression on standard cytotoxic therapy, with a significantly shorter PFS on first-line chemotherapy when compared with non-MSI-H tumors (median PFS 4.8 vs 6.9 months for non-MSI-H patients, HR=0.4; P=0.027” (p. 4)
  • “patients in the top quartile of tumor mutational burden (>9.7 mut/Mb) having the best outcomes (median OS 16.8 compared to 6.62 months for patients with lower mutational burdens; 2 year OS: 44% vs 14%; HR=0.40, log-rank test P=0.058)” (p. 4)
  • “the post-treatment sample harbored a loss-of-function mutation in exon 1 of the B2M gene, which encodes for β2-microglobulin, loss of which was confirmed by immunohistochemistry.” (p. 5)
  • “In the MSK-cohort, 44% (4 of 9) of MSI-H tumors had likely deleterious alterations in B2M.” (p. 5)
  • “we observed a concordance rate of 93.7% between IHC/FISH and NGS, with a positive predictive value (PPV) of 90% and a negative predictive value (NPV) of 96.9%.” (p. 5)
  • “the four patients with discordant cases exhibited significantly shorter PFS on first-line trastuzumab/chemotherapy compared to patients with ERBB2 amplified tumors by NGS (median PFS 5.8 vs 14.0 months; P=1e-6” (p. 5)
  • “patients in the top quartile of ERBB2 amplification levels having a significantly longer progression-free survival on trastuzumab (median PFS 24.3” (p. 5)
  • “Patients with co-alterations in receptor tyrosine kinase (RTK)-RAS-PI3K/AKT pathway genes had significantly shorter PFS (median PFS 8.4)” (p. 5)
  • “in the 44 post-trastuzumab samples from patients that were HER2+ by clinical IHC/FISH testing prior to treatment with trastuzumab, 7 (16%) were ERBB2-negative by targeted sequencing.” (p. 6)
  • “Oncogenic co-mutations in KRAS were also identified in one tumor (1/50, 2%) collected prior to trastuzumab treatment and in three tumor samples (3/23, 13%) collected following disease progression.” (p. 6)
  • “All genomic and clinical data are publically available through the cBioPortal for Cancer Genomics (http://www.cbioportal.org/study?id=egc_msk_2017)” (p. 7)

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