Distinct patterns of somatic genome alterations in lung adenocarcinomas and squamous cell carcinomas

Authors

Joshua D. Campbell

Anton Alexandrov

Jaegil Kim

Jeremiah Wala

Alice H. Berger

Chandra Sekhar Pedamallu

Sachet A. Shukla

Guangwu Guo

Angela N. Brooks

Bradley A. Murray

Marcin Imielinski

Xin Hu

Shiyun Ling

Rehan Akbani

Mara Rosenberg

Carrie Cibulskis

Aruna Ramachandran

Eric A. Collisson

David J. Kwiatkowski

Michael S. Lawrence

John N. Weinstein

Roel G. W. Verhaak

Catherine J. Wu

Peter S. Hammerman

Andrew D. Cherniack

Gad Getz

Cancer Genome Atlas Research Network

Maxim N. Artyomov

Robert Schreiber

Ramaswamy Govindan

Matthew Meyerson

Doi

10.1038/ng.3564

PMID: 27158780 · DOI: 10.1038/ng.3564 · Journal: Nature Genetics (2016)

TL;DR

Campbell et al. compared exome sequences and SNP-array copy-number profiles from 660 lung adenocarcinoma (ADC) and 484 lung squamous cell carcinoma (SqCC) tumor/normal pairs (1,144 NSCLCs total) to identify novel drivers and to contrast molecular landscapes between the two histologies. Recurrently altered genes in lung SqCC more closely resembled those of head & neck squamous and bladder carcinomas than they did lung ADC. The authors nominated novel significantly mutated genes (PPP3CA, DOT1L, CMTR2/FTSJD1 in ADC; RASA1 in SqCC; KLF5, EP300, CREBBP in both); novel amplification peaks (MIR21 in ADC, MIR205 and YES1 in SqCC, MAPK1 in both); and additional Ras/Raf/RTK pathway candidates (SOS1, VAV1, RASA1, ARHGAP35) in oncogene-negative lung ADCs. 47% of lung ADC and 53% of lung SqCC tumors had ≥5 predicted neoepitopes, supporting broad immunotherapy applicability across both subtypes (PMID:27158780).

Cohort & data

Lung ADC: 660 tumor/normal exome pairs — 274 previously unpublished, 227 from prior TCGA (luad_tcga_pub), 159 from the Imielinski et al cohort (luad_broad) (PMID:27158780).
Lung SqCC: 484 tumor/normal exome pairs — 308 previously unpublished, 176 previously described in TCGA (lusc_tcga_pub) (PMID:27158780).
Cancer types: LUAD and LUSC (combined cohort referred to as “Pan-Lung” / NSCLC).
Cohort/study: nsclc_tcga_broad_2016.
Assays: Exome capture with Agilent SureSelect Human All Exon 50MB followed by Illumina paired-end sequencing (whole-exome-seq); SNP-array copy number on Affymetrix SNP 6.0 (affymetrix-snp6); RNA-seq for 495 ADCs and 476 SqCCs (rna-seq); mutation calling with mutect and indelocator; driver discovery with MutSig2CV (mutsig); recurrent SCNAs with gistic (GISTIC2.0); purity/ploidy with absolute; mutational signatures via non-negative matrix factorization (bayesian-nmf, mutational-signatures); fusion calls via prada (PMID:27158780).

Key findings

Median somatic mutation rate was 8.7/Mb in lung ADC and 9.7/Mb in lung SqCC; MutSig2CV identified 38 significantly mutated genes in ADC and 20 in SqCC (q < 0.1) (PMID:27158780).
Only six genes — TP53, RB1, ARID1A, CDKN2A, PIK3CA, NF1 — were significantly mutated in both histologies; TP53, CDKN2A, and PIK3CA were significantly more frequently mutated in SqCC than ADC (p < 0.01, Fisher’s exact) (PMID:27158780).
Lung SqCC most closely resembled HNSC and BLCA (>25% overlap of significantly mutated genes), while lung ADC most resembled GBM and CRC; the two NSCLC subtypes overlapped only ~12% with each other (p = 0.105) (PMID:27158780).
Six mutational signatures identified by NMF, mapping to COSMIC SI4 (smoking, C>A transversions), SI7 (UV), SI13/SI2 (APOBEC), SI15/SI6 (MMR), and SI5 (clock-like). Smoking-signature SI4 separated never- vs ever-smokers far better in ADC (AUC = 0.87) than in SqCC (AUC = 0.62) — implying inaccurate smoking annotation in some SqCC cases (PMID:27158780).
87% of lung ADCs from never-smokers were transversion-low (TV-L; ≤0.696 SI4 mutations/Mb; p = 8.5 × 10⁻³⁷), but only 45% of TV-L lung ADCs came from never-smokers (PMID:27158780).
Three lung SqCCs (~1%) showed UV-pattern mutations (SI7) and one (TCGA-18-3409) had a documented prior basal-cell carcinoma — likely cutaneous metastases misclassified as lung primaries (PMID:27158780).
Seven tumors (4 ADC, 3 SqCC) had MMR-like signature (SI15/SI6) with elevated SSNVs and indels (p < 0.001) and reduced MLH1 expression (p = 0.011) (PMID:27158780).
Pan-Lung joint analysis revealed 14 additional significantly mutated genes including KLF5 (zinc-finger hotspot), the EP300/CREBBP HAT-domain hotspot region, FBXW7, and B2M (FDR q = 0.006) (PMID:27158780).
Novel focal amplifications: CCND3 and MIR21/TUBD1 in ADC; YES1 and MIR205 in SqCC; MAPK1 in the Pan-Lung analysis (PMID:27158780).
Novel focal deletions: SMARCA4 and ARID2 in ADC; ZMYND11, CREBBP, ROBO1, USP22, KDM6A in SqCC; B2M and TRAF3 recurrently lost in both (PMID:27158780).
Sex-specific enrichment: EGFR mutations in females and SMARCA4 mutations in males in ADC; RB1 mutations in females and PASK mutations exclusively in males in SqCC (FDR q < 0.1) (PMID:27158780).
Oncogene-negative lung ADC: 242/660 ADCs lacked a known activating RTK/Ras/Raf alteration. Among them, SOS1 (recurrent autoinhibitory-domain p.N233Y, n=4), VAV1 (p.S67Y near CH/Ac/PH interface), RASA1, and ARHGAP35 were significantly enriched (q < 0.1), as were amplifications around FGFR1/WHSC1L1, PDGFRA/KIT/KDR, and MAPK1 (q < 0.25) (PMID:27158780).
After incorporating these new candidates, 499/660 (76%) lung ADCs harbored an alteration in a known or putative RTK/Ras/Raf driver; in the 227-tumor expert-pathology-reviewed subset with RNA-seq, the rate rose to 193/227 (85%) (PMID:27158780).
Co-occurrence: STK11 mutations significantly co-occurred with activating KRAS (p = 1.1 × 10⁻⁶), and 28 oncogene-negative ADCs additionally carried STK11 mutations — suggesting an unrecognized KRAS-related event complementary to STK11 loss (PMID:27158780).
High EGFR amplification overlapped with activating EGFR mutations (p = 1.9 × 10⁻⁸); MET amplifications co-occurred with NF1 mutations (p = 0.019) (PMID:27158780).
Neoantigens: Recurrent predicted neoepitopes included PIK3CA p.E542K, NFE2L2 p.E79Q, BRAF p.G466V, EGFR p.G719A, TP53 p.V157F/p.G154V/p.R175G/p.P278A, and a previously unappreciated recurrent MB21D2 (C3orf59) p.Q311E. 47% of lung ADC and 53% of lung SqCC tumors had ≥5 predicted neoepitopes (PMID:27158780).
Neoepitope and nonsynonymous mutation counts were significantly lower in never-smoker ADCs vs ever-smoker ADCs (p < 0.001, Wilcoxon rank-sum), but did not differ between ever-smoker ADCs and SqCCs (PMID:27158780).

Genes & alterations

TP53, CDKN2A, PIK3CA — significantly more frequently mutated in lung SqCC than ADC (p < 0.01) (PMID:27158780).
STK11, RBM10, KEAP1, RAF1, RIT1, MET — significantly mutated exclusively in lung ADC vs other TCGA tumor types (q < 0.1) (PMID:27158780).
NFE2L2, KDM6A, RASA1, NOTCH1, HRAS — significantly mutated in lung SqCC but not other cancer types (excluding HNSC, BLCA) (PMID:27158780).
AKT1 — recurrent p.E17K mutation in lung ADC; CDK4 — recurrent p.R24L; DNMT3A — modest q-value in ADC (PMID:27158780).
PPP3CA (calcineurin catalytic subunit) — novel SMG in lung ADC; missense mutations cluster in the autoinhibitory C-terminal domain (suggesting gain of function) and co-occur with activating KRAS (p = 0.033) (PMID:27158780).
DOT1L (H3K79 methyltransferase) — novel SMG in 3% of lung ADCs, enriched for truncating mutations (PMID:27158780).
KMT2C (MLL3) and SETD2 — significantly mutated methyltransferases in lung ADC (PMID:27158780).
CMTR2 / FTSJD1 (cap methyltransferase) — novel SMG in lung ADC enriched for frameshifts; also recurrent in SF3B1 and SNRPD3 (PMID:27158780).
RASA1 (p120GAP) and CUL3 (KEAP1 partner) — novel SMGs in lung SqCC enriched for frameshift mutations (p < 0.001 for RASA1 frameshift enrichment) (PMID:27158780).
EGFR — kinase-domain duplication in lung ADC sample TCGA-49-4512; complex indels in EGFR or MET in 11 tumors (PMID:27158780).
MAP2K1 — recurrent in-frame insertion in lung ADC (PMID:27158780).
MET fusions: novel MET–CAPZA2 fusion (with neighboring gene) and previously reported KIF5B–MET (PMID:27158780).
NTRK2 fusions: TRIM24–NTRK2 in ADC and a novel NTRK2–TP63 fusion in lung SqCC (PMID:27158780).
MET and ERBB2 high-level amplifications — enriched in tumors lacking other RTK/Ras/Raf activating events (p < 0.01) (PMID:27158780).
Novel amplification target genes: KAT6A, ZNF217, MYCL (ADC); IGF1R, KDM5A, PTP4A1/PHF3, MYCL (SqCC) (PMID:27158780).
B2M — focally deleted in both ADC and SqCC, enriched for loss-of-function mutations (p < 0.01), Pan-Lung FDR q = 0.006 — implicated in MHC-I antigen presentation loss (PMID:27158780).
MB21D2 (C3orf59) — recurrent p.Q311E with predicted neoepitope properties; not previously implicated in lung cancer (PMID:27158780).

Clinical implications

Targeted therapy is largely subtype-specific. The subtype-distinct landscapes of recurrently mutated drivers and SCNAs imply that approved RTK-directed therapies (e.g., EGFR-TKI for EGFR mutant ADC, ALK and ROS1 inhibitors for ALK/ROS1 fusion ADC) are expected to apply almost exclusively to lung ADC, while lung SqCC has few directly actionable kinase drivers (PMID:27158780).
Expanded druggable pool in oncogene-negative lung ADC. Adding SOS1, VAV1, RASA1, ARHGAP35, and amplifications at FGFR1/WHSC1L1, PDGFRA/KIT/KDR, and MAPK1 raises the proportion of lung ADCs with a candidate Ras/Raf/RTK pathway driver to 76% (overall) and 85% (in expert-reviewed subset) — narrowing the unexplained fraction (PMID:27158780).
Predicted EGFR kinase-domain duplication (KDD) in TCGA-49-4512 — separately reported as afatinib-responsive in lung cancer (citation 31 in the paper) (PMID:27158780).
Immunotherapy applies broadly across NSCLC histologies. With ≥5 predicted neoepitopes in 47% of ADCs and 53% of SqCCs, the authors argue that checkpoint immunotherapy is expected to benefit both subtypes (in contrast to histology-restricted targeted therapies) (PMID:27158780).
Smoking history is the dominant correlate of neoantigen load in ADC. Never-smoker ADCs have significantly fewer neoepitopes than ever-smoker ADCs (p < 0.001) — relevant for biomarker stratification of immunotherapy candidates (PMID:27158780).
Recurrent neoepitope hotspots — PIK3CA p.E542K, NFE2L2 p.E79Q, BRAF p.G466V, EGFR p.G719A, multiple TP53 hotspots, and MB21D2 p.Q311E — are candidate shared neoantigens for off-the-shelf vaccine design (PMID:27158780).
No mutation-status association with patient survival or tumor stage survived multiple-hypothesis correction (with or without controlling for stage) (PMID:27158780).

Limitations & open questions

Whole-exome only. The study could not detect mutations in non-coding regions or regulatory elements; the authors flag this as a target for future whole-genome lung cancer studies (PMID:27158780).
Single sample per patient. Intra-tumoral heterogeneity could not be assessed (in contrast to multi-region sequencing studies) (PMID:27158780).
Underpowered for rare RTK/Ras/Raf events. With 15–25% of lung ADCs still lacking a detectable activating alteration, additional rare recurrent drivers in known and novel pathway members likely remain undiscovered (PMID:27158780).
Incomplete RNA-seq coverage — fusion calling and MET exon-14 skipping may be underestimated because matched RNA-seq was not available for every tumor (PMID:27158780).
Smoking annotation inaccuracy — SI4-based classification predicted ever-smoker status well in ADC (AUC 0.87) but poorly in SqCC (AUC 0.62), suggesting some clinically annotated SqCC “never-smokers” are misclassified (PMID:27158780).
Possible cutaneous metastases — three lung SqCCs with UV signature SI7 may represent skin SCC metastatic to the lung, not true lung primaries (PMID:27158780).

Citations from this paper used in the wiki

“we examined exome sequences and copy number profiles of 660 lung ADC and 484 lung SqCC tumor/normal pairs” (Abstract).
“Novel significantly mutated genes included PPP3CA, DOT1L, and FTSJD1 in lung ADC, RASA1 in lung SqCC, and KLF5, EP300, and CREBBP in both tumor types.” (Abstract).
“Novel amplification peaks encompassed MIR21 in lung ADC, MIR205 in lung SqCC, and MAPK1 in both.” (Abstract).
“Lung ADCs lacking receptor tyrosine kinase/Ras/Raf alterations revealed mutations in SOS1, VAV1, RASA1, and ARHGAP35.” (Abstract).
“47% of the lung ADC and 53% of the lung SqCC tumors had at least 5 predicted neoepitopes.” (Abstract).
“Only 6 genes, TP53, RB1, ARID1A, CDKN2A, PIK3CA, and NF1, were significantly mutated in both tumor types.” (Results — Comparison of somatically altered genes).
“Recurrently mutated and amplified genes in lung SqCC most closely resembled the alterations in head and neck squamous cell carcinoma (HNSC) and bladder cancer (BLCA).” (Results — Comparison of somatically altered genes).
“87% of lung ADCs from never smokers were categorized as transversion-low (TV-L; ≤0.696 of SI4 per Mb; p = 8.5 × 10⁻³⁷).” (Results — Mutational signatures).
“499 (76%) lung ADCs displayed an alteration in known or putative Ras/Raf/RTK driver genes” and “193 out of 227 (85%) lung ADCs that previously underwent secondary expert pathological review and had RNA-seq data available for fusion analysis contained a predicted activating alteration in the RTK/Ras/Raf pathway.” (Results — Identifying Ras/Raf/RTK drivers).
“STK11 mutations significantly overlapped with activating KRAS mutations (p = 1.1 × 10⁻⁶)” (Results — Identifying Ras/Raf/RTK drivers).
“Both nonsynonymous mutation and neoepitope counts … were significantly lower in lung ADCs from never smokers compared to lung ADCs from ever smokers (p < 0.001; Wilcoxon rank-sum test).” (Results — Assessment of neoantigen load).

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