Prospective comprehensive molecular characterization of lung adenocarcinomas for efficient patient matching to approved and emerging therapies

Authors

Emmet J. Jordan

Hyunjae R. Kim

Maria E. Arcila

David Barron

Debyani Chakravarty

JianJiong Gao

Matthew T. Chang

Andy Ni

Ritika Kundra

Philip Jonsson

Gowtham Jayakumaran

Sizhi Paul Gao

Hannah C. Johnsen

Aphrothiti J. Hanrahan

Ahmet Zehir

Natasha Rekhtman

Michelle S. Ginsberg

Bob T. Li

Helena A. Yu

Paul K. Paik

Alexander Drilon

Matthew D. Hellmann

Dalicia N. Reales

Ryma Benayed

Valerie W. Rusch

Mark G. Kris

Jamie E. Chaft

Jose Baselga

Barry S. Taylor

Nikolaus Schultz

Charles M. Rudin

David M. Hyman

Michael F. Berger

David B. Solit

Marc Ladanyi

Gregory J. Riely

Doi

10.1158/2159-8290.CD-16-1337

PMID: 28336552 · DOI: 10.1158/2159-8290.CD-16-1337 · Journal: Cancer Discovery (2017)

TL;DR

Jordan et al. report the first 860 patients with recurrent or metastatic LUAD prospectively profiled by the MSK-IMPACT hybrid-capture NGS assay (versions IMPACT341 and IMPACT410) at Memorial Sloan Kettering between January 2014 and March 2016 (PMID:28336552, study lung_msk_2017). Potentially actionable somatic alterations were identified in 86.9% of patients and stratified into OncoKB levels 1–4 by strength of supporting evidence. Overall 37.1% (319/860) received a molecularly matched therapy; excluding standard-of-care matches (sensitizing EGFR mutations, ALK and ROS1 fusions), only 14.4% (69/478) received matched therapy, with 52% of those deriving clinical benefit. Among 103 patients without any level 1–4 driver (“unknown mitogenic driver”, UMD), STK11 and KEAP1 were significantly enriched and nominated as candidate targetable mitogenic drivers.

Cohort & data

  • Patients: 860 patients with recurrent or metastatic lung adenocarcinoma (LUAD); 915 tumors profiled.
  • Dataset: lung_msk_2017 — clinical and genomic data deposited in cBioPortal.
  • Assay: MSK-IMPACT matched tumor:normal hybrid-capture NGS — IMPACT341 (341 genes) for 296 samples; IMPACT410 (410 genes) for 619 samples. Average coverage 615X; matched normal available for 97% of samples.
  • Demographics (Table 1): women 58.8% (506); men 41.2% (354); age 51–75 years 71.5%; never smokers 32.2% (277); former heavy smokers (>15 pack-years) 48.8% (420); 8.4% Asian.
  • Tissue origin: lung 420, lymph node 169, pleura/pleural fluid 110, liver 59, brain 47, bone 29.
  • Sample timing: median 28 days from collection to genomic analysis (range 0–3274); 89% used tissue ≤1 year old; mean turnaround 17 days from receipt to report.
  • Comparator cohort: retrospective TCGA LUAD (luad_tcga_pub, PMID:25079552) used as reference for treatment-naïve, primary-resection LUAD.
  • Tumor purity in UMD subset: estimated via FACETS allele-specific copy number analysis (range 12–92%, mean 35%).
  • Trial registration: NCT01775072 (genomic testing protocol); subsequent basket trials NCT02201212, NCT02675829, NCT02352844.

Key findings

  • Actionability rate: Potentially actionable OncoKB level 1–4 alterations in 747/860 patients (86.9%) (PMID:28336552, Figure 1A).
  • Matched therapy uptake by evidence level: 92% (level 1), 52% (level 2A), 17% (level 2B), 25% (level 3), 2% (level 4) — strongly correlated with level of evidence (p<0.0001, Cochran-Armitage trend, Figure 2A).
  • Sensitizing EGFR mutations: 95.3% received matched therapy with clinical benefit in 84.8%; ALK fusions 90.9% / 93.3%; ROS1 fusions 59.1% / 84.6%.
  • Level 2A drivers: MET exon 14 alterations matched in 65.4% (76.5% benefit), BRAF V600E 55.6% (75% benefit), RET fusions 53.3% (72.7% benefit), wild-type MET amplification 16.7% (50% benefit).
  • Cohort enrichment vs TCGA: activating EGFR alterations 27% vs 11% (p<0.001); EGFR T790M 5.5% (47/860) vs 0.4% (1/230) in TCGA (p<0.001), all post-EGFR-TKI. NF1 truncating/deletion lower at 2% vs 8.3% (p<0.001); BRAF mutations 3.6% vs 7% (p=0.042) — reflecting referral bias and a more clinically aggressive, treatment-experienced cohort.
  • Co-occurring actionable alterations: 239 patients (27.8%) had ≥2 level 1–4 alterations; 46 had ≥3; 52 had co-occurring level 1–3 mutations (25 with concurrent EGFR+PIK3CA); no patient was treated on a trial that simultaneously targeted two actionable alterations.
  • EGFR/KRAS mutual exclusivity: Only 2/214 EGFR-mutant tumors had concurrent KRAS mutations (p<0.0001); allele frequencies in both cases suggested subclonal KRAS-mutant populations.
  • PI3K pathway as highest-level driver: 32/860 (3.7%) — 17 PIK3CA, 6 inactivating PTEN, 3 each TSC1/TSC2, 2 AKT1 E17K, 1 MTOR S2215Y; 9/17 PIK3CA-driver tumors co-harbored a KRAS mutation.
  • Rare actionable drivers: 2 ARAF S214Y/S214P (sorafenib-sensitive), 2 RAF1 S257L, 1 HRAS, 10 NRAS (9 Q61, 1 G13), 6 MAP2K1 (E203K/K57N/Q56P/G128V/E102_I103del), 1 CD74-NRG1 fusion, 1 FGFR3-TACC3 fusion, 1 KIT E490Q, 1 ERBB3 D297H.
  • EGFR allele-specific response: clinical benefit 87.3% overall in sensitizing EGFR, but significantly lower for L861Q (43%; p=0.039 vs L858R; p=0.01 vs exon 19 del) and exon 18 deletions (40%; p=0.02 vs L858R; p=0.005 vs exon 19 del). Both EGFR-kinase domain duplication (EGFR-KDD) patients (treated with erlotinib and afatinib) derived clinical benefit.
  • EGFR exon 20 insertions: 17 patients identified; one treated with erlotinib had no response, consistent with prior data.
  • UMD cohort (n=103, 12.0% of patients): alterations in TP53, STK11, KEAP1, KMT2D, and PDGFRA significantly enriched (p<0.05) vs level 1–4 cohort. In never/former-light smokers within UMD: chromatin modifiers (KMT2C, SETD2, CREBBP) and homologous recombination genes (MRE11, BRCA2) enriched. Heavy smokers within UMD: STK11 and TP53 elevated. Unlike TCGA, no RIT1 mutations in UMD.
  • KEAP1 in UMD: altered in 35/103 (34%) UMD samples (10 truncations, 25 missense); 3D structural analysis localized missense variants to the Kelch domain that interacts with NFE2L2/Nrf2.
  • Level 3/4 patients and immunotherapy: Only 7% (26/373) of level 3/4 patients enrolled on a matched-therapy trial, but 19% (70/373) enrolled on a therapeutic trial; 62.9% (44/70) on immunotherapy trials. Inverse trend toward higher mutational load in level 3/4 vs level 1/2 patients (Supplementary Figure S7), driven by smoking history.

Genes & alterations

  • EGFR — 214 (24.9%) patients with sensitizing alleles; exon 19 del/ins (113) and L858R (70) most common; rarer activating: L861Q (7), E709_T710delinsD (6), G719A (4), exon 18–25 kinase domain duplication (2). 17 exon 20 insertions and 1 H773R were resistance-associated. T790M in 5.5% (47/860), all post-TKI. Allele-specific response data support afatinib/osimertinib over first-generation TKIs for L861Q.
  • KRAS — 218 patients with KRAS mutations; only 0.9% (2/218) received matched therapy. Codon 12 dominant; codon 61 in 12/235 (5%) of KRAS-mutated tumors. Mutually exclusive with EGFR (p<0.0001).
  • ALK and ROS1 — fusions; level 1; high matched-therapy uptake and benefit. Two ROS1-fusion patients died before crizotinib FDA-approval (March 2016) for that indication.
  • RET — fusions in 1.7–3.8% range; level 2A; 53.3% matched therapy with 72.7% benefit; cabozantinib cited as matched agent.
  • MET — exon 14 alterations 65.4% matched (76.5% benefit); wild-type MET amplification 16.7% matched (50% benefit).
  • BRAF — V600E (level 2A) 55.6% matched / 75% benefit. Non-V600E alleles (K601E, D594G, T599 dup, SND1-BRAF fusion): 4/13 received matched therapy (single-agent MEK/ERK inhibitor or BRAF+MEK combo for the SND1-BRAF fusion); none derived clinical benefit.
  • ERBB2 — amplification (level 2B) 5/12 (42%) matched; 1 with clinical benefit. ERBB2 mutation (level 3): 50% matched, 40% benefit.
  • ERBB3 — D297H hotspot in 1 patient.
  • NF1 — truncating/deletion in 16 patients (level 4); enriched in TCGA cohort (8.3% vs 2%, p<0.001).
  • NRAS — 10 patients (9 Q61, 1 G13).
  • HRAS — 1 patient.
  • MAP2K1 (MEK1) — 6 patients with activating alleles (E203K, K57N, Q56P, G128V, E102_I103del); none received matched therapy as highest-level alteration.
  • ARAF — 2 hotspot S214Y/S214P (sorafenib-sensitive in prior preclinical work).
  • RAF1 — 2 S257L hotspot.
  • PIK3CA — 17 patients with PIK3CA as highest-level driver; 25 EGFR+PIK3CA co-mutations across the cohort.
  • PTEN — 6 inactivating.
  • TSC1, TSC2 — 3 patients each with truncating alterations (level 2B); none received matched therapy.
  • AKT1 — 2 E17K mutations; 1 of 2 derived 12-month clinical benefit on matched therapy.
  • MTOR — 1 S2215Y as highest driver; 1 UMD patient with novel L2383F missense had 1-month no-benefit course on everolimus.
  • BRCA1, BRCA2 — 3 BRCA1 (0.3%) and 8 BRCA2 (0.9%) likely-inactivating truncating mutations; level 2B based on olaparib approval in BRCA-mutant ovarian carcinoma; no patient received matched PARP-inhibitor therapy.
  • KIT — 1 exon 9 E490Q (previously described in thymic carcinoma).
  • CD74–NRG1 fusion — 1 patient.
  • FGFR3–TACC3 fusion — 1 patient.
  • SND1–BRAF fusion — 1 patient (treated with BRAF+MEK combo, no benefit).
  • TP53, STK11, KEAP1 — significantly enriched in UMD cohort (p<0.05). KEAP1 missense variants cluster in the Kelch domain interacting with NFE2L2/Nrf2.
  • NFE2L2 (NRF2) — mutations in 2.9% (3/103) of UMD samples.
  • KMT2D, KMT2C, SETD2, CREBBP — chromatin modifiers enriched in UMD (especially never/former-light smokers).
  • MRE11 — homologous recombination gene enriched in never/former-light smoking UMD subset (paper writes “MRE11A”; current HUGO symbol is MRE11).
  • PDGFRA — significantly enriched in UMD vs level 1–4.
  • ERBB4 — alterations identified in UMD subset, but none were hotspot or HER-kinase-inhibitor-sensitivity-conferring variants.
  • ERRFI1 (MIG6) — A143D missense in 1 UMD patient with erlotinib benefit despite no EGFR mutation; ERRFI1 mutations identified in 6 (0.7%) total, 4 with co-occurring higher-level alterations (2 EGFR exon 19 del; 2 KRAS G12).
  • KDM5C — frameshift in 1 UMD patient with 6-month stable disease on off-label azacitidine.
  • CDKN2A — deletions identified; no patient with CDKN2A deletion as highest actionable alteration received matched therapy.
  • RIT1 — notably absent from this UMD cohort despite enrichment in the TCGA UMD set.

Clinical implications

  • Standard-of-care biomarkers (level 1) deliver: 93% of patients with sensitizing EGFR mutations or ALK/ROS1 fusions received corresponding matched therapy and 85.8% derived clinical benefit, supporting universal upfront genotyping for these lesions in advanced LUAD.
  • Broaden screening to level 2A drivers: Although level 2A matched-therapy uptake was only 52%, clinical benefit when treated was 76% — an argument for routine, early testing for MET exon 14, BRAF V600E, and RET fusions rather than waiting for sequential single-gene assays.
  • EGFR allele matters: L861Q and exon 18 deletions had significantly lower clinical-benefit rates with first-generation TKIs (erlotinib), supporting clinical evaluation of afatinib, osimertinib, and dacomitinib for these alleles. EGFR exon 20 insertions remained TKI-resistant.
  • Off-label barriers limit level 2B uptake: Despite FDA-approval of olaparib in BRCA-mutant ovarian and HER2-directed agents and mTOR inhibitors in other indications, no patient with BRCA1/BRCA2 or TSC1/TSC2 truncating mutations received matched therapy in this cohort. Authors opened basket trials NCT02201212 and NCT02675829 to address this gap.
  • Immunotherapy is the de facto matched option for level 3/4 patients: With higher mutational load (driven by smoking) and lacking compelling matched therapy, 62.9% of level 3/4 patients enrolled on therapeutic trials received immunotherapy.
  • UMD nominates STK11 and KEAP1 as targetable drivers: Both significantly enriched in UMD vs level 1–4 cohort, consistent with TCGA findings (PMID:25079552). Authors flag KEAP1 Kelch-domain-clustering missense variants and propose testing Nrf2 inhibitors (luteolin, brusatol) as chemosensitizers in future trials.
  • No combination matched therapy was administered despite 27.8% of patients having ≥2 actionable alterations — a recognized gap requiring novel trial designs.

Limitations & open questions

  • Cohort selection bias: MSK referral pattern enriches for EGFR mutants (27% vs TCGA 11%) and EGFR T790M (5.5% vs 0.4%); 32.2% never smokers and 8.4% Asian over-represent specific demographics relative to TCGA.
  • Treatment-experienced cohort: 37.7% of samples were post at least one prior systemic therapy, potentially biasing detection of resistance alleles and limiting comparison to treatment-naïve cohorts.
  • No combinatorial matched therapy: No patient received simultaneous matched therapy for two actionable alterations even though 239 (27.8%) had multiple actionable lesions — limits inference on combination strategies.
  • Clinical-benefit definition is imaging+symptom-based: Stable disease on two consecutive imaging scans ≥30 days apart with symptom improvement counts as benefit; no formal RECIST PFS/OS reported per arm.
  • Variants of uncertain significance: All UMD BRCA2 mutations were somatic missense VUS; the MTOR L2383F UMD variant was novel and lacked functional validation.
  • Panel evolution within study: Two MSK-IMPACT versions used in parallel (IMPACT341 for 296 samples; IMPACT410 for 619 samples) — alterations restricted to v1 panel were not testable in v2-only contexts and vice versa.
  • No fusion/expression analysis beyond panel scope: Whole exome/genome/transcriptome sequencing might have identified additional actionable fusions or alterations in the 13.1% UMD cases.
  • Rapid clinical deterioration prevented matched therapy: 11.3% (8/71) of level 2A patients deteriorated before matched therapy could be initiated, suggesting a benefit to upfront (rather than progression-driven) testing.
  • No prospective validation of STK11 as predictive biomarker: Authors note no clinical trials test STK11 as a predictive biomarker of mTOR-inhibitor response in lung cancer; NCT02352844 is a phase 2 in patients with TSC1/2, NF1/2, or STK11 mutations across solid tumors but specifically lung-restricted data are pending.

Citations from this paper used in the wiki

  • “We report our experience with the first 860 patients with recurrent or metastatic lung adenocarcinoma analyzed by MSK-IMPACT.” (Introduction)
  • “Potentially actionable somatic alterations, as defined by the OncoKB classification, were identified in 747 patients (86.9%).” (Results, Known mitogenic drivers)
  • “Overall, 37.1% (319/860) of patients received a matched therapy guided by their tumor molecular profile with the likelihood of receiving a matched therapy correlating strongly with the level of evidence that the mutation identified predicts for drug response (p<0.0001).” (Results, Use of matched therapy)
  • “Excluding alterations associated with standard of care therapy, 14.4% (69/478) received matched therapy with a clinical benefit of 52%.” (Abstract)
  • “5.5% (47/860) had a EGFR T790M mutation, all detected post EGFR-TKI therapy, as compared to 0.4% (1/230) in the TCGA dataset (p<0.001).” (Results, Known mitogenic drivers)
  • “The rate of clinical benefit was statistically significantly lower in patients with L861Q mutations (43%; p=0.039 versus L858R; p=0.01 versus exon 19 deletions) or exon 18 deletions (40%; p=0.02 versus L858R; p=0.005 versus exon 19 deletions).” (Results, Clinical benefit with EGFR inhibitors)
  • “In comparison to tumors harboring a level 1 to 4 alteration, alterations in TP53, STK11, KEAP1, KMT2D, and PDGFRA were all significantly more common (p<0.05) in the UMD cohort.” (Results, UMD set)
  • “KEAP1 (kelch-like ECH-associated-protein-1)… was altered in 34% (35/103) of the UMD cases (10 truncations and 25 missense variants) with Nrf2 mutations noted in 2.9% (3/103).” (Results, UMD set)
  • “Genomic analysis was performed using assay version-1 (341 genes) for 296 samples and version-2 (410 genes) for 619 samples.” (Methods, Genomic sequencing)
  • “All clinical and genomic data are available in electronic form through the cBioPortal for Cancer Genomics.” (Methods, Analysis)

This page was processed by crosslinker on 2026-05-14.