Comprehensive molecular profiling of lung adenocarcinoma

Author

The Cancer Genome Atlas Research Network

Doi

10.1038/nature13385

PMID: 25079552 · DOI: 10.1038/nature13385 · Journal: Nature (2014)

TL;DR

The TCGA Research Network performed comprehensive multi-platform molecular profiling (whole-exome sequencing, RNA-seq, miRNA-seq, methylation, copy-number, and reverse-phase protein arrays) of 230 resected, treatment-naive lung adenocarcinomas. Combined with 182 previously published cases (n=412 for statistical analysis), MutSig2CV identified 18 significantly mutated genes, including newly characterized activating mutations in RIT1 and loss-of-function mutations in MGA (mutually exclusive with focal MYC amplification). The study nominated focal MET and ERBB2 amplifications, plus NF1 and RIT1 mutations, as drivers in oncogene-negative tumours, raising the fraction of lung adenocarcinomas with a known RTK/RAS/RAF activating event from 62% to 76%. Integration of DNA and RNA sequencing demonstrated MET exon 14 skipping (4% of cases) and U2AF1-associated splicing alterations.

Cohort & data

Key findings

  • Mean somatic mutation rate 8.87 mutations/Mb (range 0.5–48, median 5.78); non-synonymous rate 6.86/Mb.
  • 18 significantly mutated genes identified by MutSig2CV across n=412 tumours.
  • Most frequently mutated genes in the TCGA cohort (n=230): TP53 46%, KRAS 33%, EGFR 14%, STK11 17%, KEAP1 17%, NF1 11%, BRAF 10%, SETD2 9%, RBM10 8%, MGA 8%, ARID1A 7%, PIK3CA 7%, MET 7%, SMARCA4 6%, RB1 4%, CDKN2A 4%, U2AF1 3%, RIT1 2%.
  • MGA loss-of-function mutations were mutually exclusive with focal MYC amplification (Fisher’s exact P = 0.04), implicating a previously unappreciated mechanism of MYC pathway activation.
  • C>A transversion-high (TH) vs transversion-low (TL) classification (n=269 TH, n=144 TL): TH strongly associated with smoking (P < 2.2×10⁻¹⁶); KRAS mutations enriched in TH (P = 2.1×10⁻¹³); EGFR, PIK3CA, RB1 enriched in TL (P < 0.05).
  • Sex-biased mutations: EGFR enriched in females (P = 0.03); RBM10 loss-of-function enriched in males (P = 0.002); 16/21 RBM10 mutations in the TH subset occurred in males (P = 0.003).
  • Significant focal amplifications: NKX2-1, TERT, MDM2, KRAS, EGFR, MET, CCNE1, CCND1, TERC, MECOM, 8q24 near MYC, and a novel peak containing CCND3. Most significant deletion: CDKN2A.
  • Low-pass WGS (n=93) revealed an average of 36 gene–gene/gene–inter-gene rearrangements per tumour; chromothripsis in 6/93 (6%).
  • MET exon 14 skipping in 4% of cases (10/230); in 9 of these, a 5′ or 3′ splice site mutation or deletion was identified; one case had a MET Y1003* stop codon mutation predicted to disrupt splicing enhancers.
  • U2AF1 S34F mutations (n=8) were associated with 129 differentially-utilized splicing events, predominantly cassette exons and alternative 3′ splice sites — including alternative splicing of CTNNB1.
  • Fusions in TL tumours only (P = 1.85×10⁻⁴): ALK (3/230), ROS1 (4/230), RET (2/230).
  • Oncogene-negative subset (38%): WES identified enrichment of TP53, KEAP1, NF1, and RIT1 mutations (P < 0.01); GISTIC identified focal ERBB2 and MET amplifications specific to this group. With these added, 76% of LUAD have a defined RTK/RAS/RAF activating event, up from 62%.
  • Three mTOR activation patterns by RPPA: (1) basal, (2) STK11-inactivating mutation or low STK11/low AMPK, (3) high p-AKT, PIK3CA mutation, or both.
  • Three transcriptional subtypes (renamed): TRU (terminal respiratory unit; formerly bronchioid), PI (proximal-inflammatory; formerly squamoid), PP (proximal-proliferative; formerly magnoid). PP enriched for KRAS + STK11 inactivation; PI enriched for NF1+TP53 co-mutation; TRU contained most EGFR-mutant and kinase-fusion tumours and was prognostically favourable.
  • CIMP-H vs CIMP-L methylation clusters: CIMP-H tumours hypermethylated WNT pathway genes (P = 0.0015), CDKN2A, GATA family, HOXA9, HOXD13, RASSF1, SFRP1, SOX17, WIF1; MYC overexpression associated with CIMP-H (P = 0.003).
  • iCluster integrative analysis yielded 6 clusters; SETD2 mutation associated with CDKN2A methylation in cluster 4 (TRU-associated, low ploidy, low mutation rate).

Genes & alterations

  • TP53 — mutated in 46% of TCGA cases; cluster 1–3 enrichment; co-mutated with NF1 in the PI subtype.
  • KRAS — activating mutations in 33% (n=74 cancer-associated); mutually exclusive with EGFR; enriched in TH/smoker cohort and the PP transcriptional subtype.
  • EGFR — mutations in 14% (n=26 cancer-associated activating); enriched in females, TL, and TRU subtype; in-frame insertions specifically enriched in TL.
  • BRAF — mutations in 10% (n=16 cancer-associated activating).
  • PIK3CA — mutations in 7%; enriched in TL; drives mTOR pathway activation via p-AKT in subset.
  • MET — DNA mutations in 7%; exon 14 skipping (alternative splicing) in 4% (10/230); focal amplification enriched in oncogene-negative tumours.
  • RIT1 — activating mutations clustered around residue Q79 (homologous to RAS Q61) in 2% of cases; exclusively in the oncogene-negative subset; nominated as a novel RTK/RAS/RAF driver.
  • STK11 — inactivating mutations in 17%; mTOR activation via loss of LKB1/AMPK signalling; defining co-feature of PP subtype.
  • KEAP1 — mutations in 17%; enriched in oncogene-negative tumours (P < 0.01).
  • NF1 — loss-of-function mutations in 11%; enriched in oncogene-negative subset; co-mutated with TP53 in PI subtype; nominated as a driver event.
  • RB1 — mutated in 4%; frameshift indels enriched in TL adenocarcinomas vs SCLC (P < 0.05).
  • CDKN2A — mutated in 4%; most significant deletion peak in copy-number data; hypermethylated in CIMP-H tumours; methylation associated with low ploidy, low mutation rate, and SETD2 mutation.
  • SETD2 — mutated in 9% (chromatin modifier); associated with CDKN2A methylation in iCluster 4.
  • ARID1A — mutated in 7% (chromatin modifier).
  • SMARCA4 — mutated in 6% (chromatin modifier).
  • RBM10 — RNA-binding protein on X chromosome; LOF mutations in 8%; strongly enriched in males (P = 0.002).
  • U2AF1 — S34F mutations in 3% (8 samples); drive 129 alternative splicing events including CTNNB1.
  • MGA — newly described loss-of-function (frameshift/nonsense) mutations in 8%; mutually exclusive with focal MYC amplification (P = 0.04); encodes a Max-interacting protein on the MYC pathway.
  • MYC — 8q24 amplification peak; overexpression associated with CIMP-H phenotype (P = 0.003).
  • ERBB2 — activating in-frame insertion and point mutations (n=5); focal amplification specific to oncogene-negative tumours.
  • ALK — fusions in 3/230; transversion-low only.
  • ROS1 — fusions in 4/230; transversion-low only.
  • RET — fusions in 2/230; transversion-low only.
  • NKX2-1 — significant focal amplification.
  • TERT, MDM2, CCNE1, CCND1, CCND3, TERC, MECOM — recurrent focal amplifications.
  • MAP2K1 (n=2), NRAS (n=1), HRAS (n=1) — additional RTK/RAS/RAF pathway mutations.
  • CTNNB1 — alternative splicing strongly associated with U2AF1 S34F mutations.

Clinical implications

  • Therapeutic targets in oncogene-negative LUAD: The paper argues for “increased implementation of MET and ERBB2/HER2 inhibitors” in tumours otherwise lacking an activated oncogene, given the enrichment of focal MET and ERBB2 amplifications in this 38% subset.
  • RIT1 mutations transform NIH3T3 cells and activate MAPK and PI(3)K signalling, supporting RIT1 as a candidate driver and potential therapeutic target.
  • MET exon 14 skipping is highlighted as a clinically actionable splicing alteration; ~9/10 cases have an identifiable cis-acting splice-site mutation suitable for genomic detection.
  • TRU transcriptional subtype membership was prognostically favourable, replicating prior reports.
  • The discrepancy between genomic alterations and protein-level pathway activity (RPPA) suggests additional, undetected mechanisms of MAPK/mTOR activation — implying that pathway-state biomarkers may complement DNA-based stratification.
  • MGA–MYC mutual exclusivity underscores MYC pathway dysregulation as a recurrent feature, motivating MYC-pathway-directed therapeutic strategies.

Limitations & open questions

  • Many tumours show MAPK or mTOR pathway activation by RPPA without identifiable underlying genomic alterations — mechanisms unexplained.
  • The mechanism by which the MET Y1003* mutation produces exon 14 skipping remains unknown.
  • The oncogenic contribution of U2AF1 S34F-driven splicing changes (e.g. CTNNB1) is not established functionally.
  • Median follow-up (19 months) limits durable conclusions on prognostic subtype performance.
  • Driver assignment depends on the published list of known driver events (Pao & Hutchinson 2012); some additional uncharacterized KRAS, EGFR, and BRAF variants observed were not classified as drivers and would inflate frequencies if reclassified.
  • Cluster 6 of the iCluster analysis is largely defined by lower cellularity and lacks clear molecular features — biological identity unclear.
  • Statistical significance for sex-biased EGFR enrichment (P = 0.03) is modest; replication needed.

Citations from this paper used in the wiki

  • “High rates of somatic mutation were seen (mean 8.9 mutations per megabase). Eighteen genes were statistically significantly mutated” (Abstract).
  • “Loss-of-function (frameshift and nonsense) mutations in MGA were mutually exclusive with focal MYC amplification (Fisher’s exact test P = 0.04)” (page 3).
  • “Ten tumours had somatic MET DNA alterations with MET exon 14 skipping in RNA. In nine of these samples, a 5′ or 3′ splice site mutation or deletion was identified” (page 4).
  • “We nominate amplifications in MET and ERBB2 as well as mutations of NF1 and RIT1 as driver events specifically in otherwise oncogene-negative lung adenocarcinomas. This analysis increases the fraction of lung adenocarcinoma cases with somatic evidence of RTK/RAS/RAF activation from 62% to 76%” (Conclusions, page 7).
  • “EGFR mutations were enriched in tumours from the female cohort (P = 0.03) whereas loss-of-function mutations within RBM10 … were enriched in tumours from men (P = 0.002)” (page 3).

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