Integrative genomic characterization of oral squamous cell carcinoma identifies frequent somatic drivers

Authors

Curtis R. Pickering

Jiexin Zhang

Suk Young Yoo

Linnea Bengtsson

Shhyam Moorthy

David M. Neskey

Mei Zhao

Marcus V. Ortega Alves

Kyle Chang

Jennifer Drummond

Elsa Cortez

Tong-xin Xie

Di Zhang

Woonbok Chung

Jean-Pierre J. Issa

Patrick A. Zweidler-McKay

Xifeng Wu

Adel K. El-Naggar

John N. Weinstein

Jing Wang

Donna M. Muzny

Richard A. Gibbs

David A. Wheeler

Jeffrey N. Myers

Mitchell J. Frederick

Doi

10.1158/2159-8290.CD-12-0537

PMID: 23619168 · DOI: 10.1158/2159-8290.CD-12-0537 · Journal: Cancer Discovery (2013)

TL;DR

Pickering et al. performed the first integrated multi-platform genomic analysis (gene expression, copy number, DNA methylation, exome sequencing) of 40 oral squamous cell carcinomas (OSCC) from MD Anderson, finding four major driver pathways (mitogenic signaling, Notch, cell cycle, TP53) plus two additional key genes (FAT1, CASP8). Notch pathway alterations were present in 66% of tumors and functional activation of NOTCH1 suppressed proliferation in HNSCC cell lines, supporting a tumor-suppressor role. CASP8 mutations (10%) defined a new molecular subtype with strikingly fewer copy-number alterations and frequent co-occurrence with HRAS mutations. Eighty percent of tumors harbored at least one alteration in a “targetable” gene, opening avenues for molecularly directed therapy in a disease whose survival has not changed in decades [PMID:23619168].

Cohort & data

  • N = 40 OSCC patients treated at MD Anderson Cancer Center; tumor and matched non-malignant adjacent tissue, surgically resected, fresh-frozen, >60% tumor nuclei content [PMID:23619168].
  • Cancer type: oral squamous cell carcinoma (OCSC), a subset of head and neck squamous cell carcinoma (HNSC).
  • Dataset: hnsc_mdanderson_2013. Expression and methylation data deposited in GEO under accession GSE41117 [PMID:23619168].
  • Assays / platforms:
    • Copy number: Affymetrix affymetrix-snp6 SNP arrays on 38 tumors; ASCAT for allele-specific CN; gistic v2.0.12 for significance.
    • Exome: Nimblegen capture, SOLiD or Illumina sequencing (whole-exome-seq).
    • DNA methylation: Illumina HumanMethylation450 arrays (450k-methylation-array).
    • Expression: Affymetrix Human Exon 1.0ST arrays (mRNA) and Agilent Human microRNA microarrays rel12.0.
    • Validation: sanger-sequencing of NOTCH1 and CASP8 in a panel of 44 HNSCC cell lines.
  • Cross-cohort validation of CASP8 findings used TCGA HNSCC and TCGA colorectal (COADREAD) cohorts PMID:22810696.

Key findings

  • Genomic instability is a dominant feature of OSCC. 32% (12/38) tumors diploid, 32% triploid, 37% tetraploid+; mean 16.3 focal and 7.3 arm-level CNAs per sample; 74% (28/38) had ≥20 CNAs [PMID:23619168].
  • Recurrent arm-level CNAs (>50% of tumors): gains of 8q (63%, 24/38) and 3q (58%, 22/38); losses of 3p (76%, 29/38), 8p (53%, 20/38) and 18q (58%, 22/38). Loss of 8p correlated with reduced disease-free survival (p=0.047), increased extracapsular spread (p=0.021), and trended toward reduced overall survival (p=0.067) [PMID:23619168].
  • High-level focal amplifications in >10% of samples: CCND1 (22%, 8/38) and EGFR (16%, 6/38). TP63, MYC, and regions on 2q11.2, 5p, 9p23, 14q24.3, 20p12.1 were amplified in ≥5% of samples [PMID:23619168].
  • Relative copy number (normalized to tumor ploidy) correlated more strongly with gene expression than absolute copy number — 1721 genes vs. 224 genes at FDR<0.01 — suggesting polyploid tumors compensate gene expression for ploidy [PMID:23619168].
  • mRNA/miRNA expression clusters correlate with differentiation state (p<0.002). Cluster 1 contained 7/8 poorly differentiated tumors; cluster 2 contained 6/6 well-differentiated tumors. Cluster 1 tumors had fewer mutations (median 53 vs. 81, p=0.0073) and higher ploidy (3.33 vs. 2.62, p=0.033); all non-tobacco users fell in cluster 1 (p<0.007) [PMID:23619168].
  • A CIMP-like methylation subgroup exists: methylation cluster 1 had higher methylation frequency and fewer CNAs, similar to CRC CIMP, but lacked MLH1 methylation and the high-mutation phenotype of CRC CIMP. Tobacco use was associated with methylation clusters (p<0.009); 8/9 non-tobacco users were in cluster 2 [PMID:23619168].
  • Notch pathway altered in 66% (23/35) of tumors when including NOTCH1 mutation, TP63 alterations, and gains in JAG1/JAG2/NUMB and losses in MAML1. NOTCH1 was mutated in 9% of the primary cohort; in 44 HNSCC cell lines, 16 candidate NOTCH1 alterations were identified including 3 homozygous frameshift and 2 homozygous nonsense mutations. Truncating mutants showed complete loss of NOTCH1 protein on Western blot [PMID:23619168].
  • Functional NOTCH1 signaling inhibits OSCC proliferation. Retroviral expression of activated NOTCH1 (ICN1) reduced GFP-positive cell fraction to <40% of original within 16 days in 5 cell lines harboring NOTCH1 missense or truncating mutations. ICN1 caused G1 arrest (67% G1 at day 3), induced p21 (CDKN1A), and induced senescence-associated β-galactosidase. Full-length wild-type NOTCH1 (NFL) also inhibited cells; effect enhanced by JAG1 ligand. HN31 and UM-SCC-47 xenografts in the orthotopic tongue model showed dramatic tumor-size reduction with ICN1 or NFL versus vector control [PMID:23619168].
  • FAT1 inactivation in 46% (16/35) when combining mutation (30%, 12/40 — the highest rate yet reported for HNSCC; two-thirds inactivating nonsense/frameshift/splice-site) and focal deletion [PMID:23619168].
  • Mitogenic signaling altered in 63% (22/35), including activating mutations in HRAS, PIK3CA, BRAF and copy gains in EGFR, PIK3CA, AKT1, RPS6KB1, MYC [PMID:23619168].
  • Cell cycle pathway altered in 94% (33/35) of tumors via CCND1 amplification and/or CDKN2A loss [PMID:23619168].
  • TP53 mutations in 60% (24/40); mostly missense, but 13% (12/40) were splice-site mutations — markedly enriched versus the 2% baseline reported across tumor types in IARC. TP53 mutations associated with more CNAs (p=0.0051) [PMID:23619168].
  • CASP8 mutations in 10% (4/40) define a new subtype: significantly fewer CNAs in the discovery cohort (p=0.0136) and validated in TCGA HNSCC (p<0.0001) and TCGA COAD (p<0.0001) [PMID:23619168] PMID:22810696.
  • CASP8 mutations co-occur with HRAS mutations (p=0.0016): 3/4 CASP8-mutant tumors had HRAS mutation; 3/4 HRAS-mutant tumors had CASP8 mutation. Validated in TCGA HNSCC where 7/8 HRAS-mutant tumors had CASP8 mutation. In TCGA CRC, CASP8 correlated with BRAF (5/21 BRAF-mutant CRC had CASP8 mutation; p=0.0009) but not RAS [PMID:23619168] PMID:22810696.
  • CASP8 loss-of-function promotes aggressive OSCC. CASP8-mutant HNSCC cell lines were more tumorigenic and produced larger, more lethal orthotopic tongue tumors. CASP8 shRNA knockdown in HN4 cells reproduced the aggressive phenotype [PMID:23619168].
  • 80% (28/35) of tumors harbored at least one “targetable” alteration, 54% (19/35) had ≥2. Notable targets: SRC-family kinases (SRC/LYN/YES1, 29%, 10/35), PI3K signaling (PIK3CA/AKT1/2/3, 34%, 12/35), TNK2 (17%, 6/35), PTK2/FAK (14%, 5/35), Aurora kinases (AURKA/C, 14%, 5/35) [PMID:23619168].

Genes & alterations

  • NOTCH1 — mutated in 9% of OSCC tumors and in 16/44 HNSCC cell lines (frameshift, nonsense, missense); functions as tumor suppressor in OSCC. Activated NOTCH1 (ICN1) induced G1 arrest, p21 induction, senescence, and inhibited xenograft growth [PMID:23619168].
  • CASP8 — mutated in 10% (4/40) of OSCC; nonsense/splice/frameshift consistent with loss-of-function. Defines a new molecular subtype with fewer CNAs and co-occurring HRAS mutations; CASP8 loss promotes aggressive xenograft phenotype [PMID:23619168].
  • FAT1 — mutated in 30% (12/40) of OSCC (two-thirds inactivating); 46% combined inactivation rate when adding focal deletion. Candidate tumor suppressor, third most frequently inactivated gene; linked elsewhere to aberrant Wnt activation [PMID:23619168].
  • TP53 — mutated in 60% (24/40); 13% splice-site mutations (vs. 2% baseline in IARC). Associated with higher CNA burden (p=0.0051) [PMID:23619168].
  • TP63 — altered in 34% (12/35) via ΔN-p63 overexpression; mutated in 9% (3/35). ΔN-p63 has oncogenic functions and directly inhibits NOTCH1 [PMID:23619168].
  • HRAS — activating mutation; strongly co-occurs with CASP8 mutations in HNSCC. Part of mitogenic signaling pathway [PMID:23619168].
  • PIK3CA — activating mutation and/or copy gain (on 3q); part of mitogenic pathway altered in 63% of tumors [PMID:23619168].
  • BRAF — activating mutation; part of mitogenic signaling. In CRC, BRAF (not RAS) correlates with CASP8 mutation [PMID:23619168].
  • EGFR — focal high-level amplification in 16% (6/38) of OSCC; mitogenic pathway driver. Drug target for cetuximab in HNSCC [PMID:23619168].
  • AKT1 — copy gain; part of PI3K/AKT mitogenic signaling [PMID:23619168].
  • CCND1 — focal high-level amplification in 22% (8/38) of OSCC; strongly correlated with gene expression (p=5.66e-9) [PMID:23619168].
  • CDKN2A — focal deletion; combined with CCND1 amplification yields cell-cycle alteration in 94% (33/35) of tumors [PMID:23619168].
  • CDKN1A — p21 protein induced by activated NOTCH1, mediating G1 arrest in OSCC cells [PMID:23619168].
  • MYC — on 8q; focal amplification in ≥5% of OSCC; mitogenic pathway gain [PMID:23619168].
  • FHIT — focal deletion on 3p; candidate tumor suppressor in HNSCC [PMID:23619168].
  • CSMD1 — focal deletion on 8p; candidate tumor suppressor [PMID:23619168].
  • NFIB — candidate oncogenic driver on 9p; high-level amplification in 3/38 tumors despite background 9p loss in others [PMID:23619168].
  • JAG1, JAG2 — Notch ligand gains; may inhibit Notch via cis-inhibition or activate it, but consistent with tumor-suppressive Notch role inferred for OSCC [PMID:23619168].
  • NUMB — copy gain; Notch pathway modulator [PMID:23619168].
  • MAML1 — copy loss; Notch pathway modulator [PMID:23619168].
  • FADD, BIRC2, BIRC3, IKBKB — within or near the 11q13 CCND1 amplicon with strong CN-expression correlation; possibly contribute alongside CCND1 [PMID:23619168].
  • LTO1 (ORAOV1) — within 11q13 amplicon; previously shown to promote tumor growth in OSCC [PMID:23619168].
  • RPS6KB1 — copy gain in mitogenic pathway [PMID:23619168].
  • TNK2 — copy gain with high expression in 17% (6/35); candidate target [PMID:23619168].
  • PTK2 (FAK) — copy gain with high expression in 14% (5/35); candidate target [PMID:23619168].
  • SRC, LYN, YES1 — SRC-family kinases collectively altered in 29% (10/35) [PMID:23619168].
  • AURKA, AURKC — Aurora kinases altered in 14% (5/35); drugs in clinical trials [PMID:23619168].
  • TERC — on 3q; candidate cancer driver in arm gain [PMID:23619168].
  • PRKCI — on 3q; candidate cancer driver [PMID:23619168].

Clinical implications

  • Targetable landscape: 80% (28/35) of OSCC tumors harbored at least one alteration in a gene that is an established cancer therapeutic target, and 54% had ≥2 — suggesting combination strategies may be feasible. 45 of 76 targetable genes have an approved cancer agent against them [PMID:23619168].
  • EGFR/cetuximab context: While EGFR is amplified in 16% of OSCC, abundance of downstream alterations (HRAS, PIK3CA, BRAF, AKT1) may explain limited EGFR-inhibitor benefit in HNSCC; authors argue PI3K-pathway inhibition should be pursued [PMID:23619168].
  • TP53 as prognostic/predictive biomarker — splice-site enrichment (13%) and association with higher CNA burden warrant attention; consistent with prior reports of TP53 mutation impact on HNSCC survival [PMID:23619168].
  • 8p loss as candidate prognostic biomarker — associated with reduced disease-free survival (p=0.047) and extracapsular spread (p=0.021) in this cohort [PMID:23619168].
  • CASP8 mutation defines a CN-quiet subtype that may have distinct clinical behavior; if CASP8 loss prevents genomic instability, caspase inhibition might sensitize wild-type tumors to chemotherapeutics that induce CN changes [PMID:23619168].
  • Standard therapy unchanged: surgery, radiation, and cisplatin remain the backbone of OSCC therapy at the time of publication; no new biomarkers were validated for prospective treatment selection [PMID:23619168].

Limitations & open questions

  • Small discovery cohort (N=40). Authors caution that only the most robust findings were reported and expect validation in larger cohorts [PMID:23619168].
  • Candidate driver status is not functional validation. Most pathway-level claims rest on integrated genomic signatures, not perturbation experiments — beyond the NOTCH1 and CASP8 functional follow-ups [PMID:23619168].
  • Mechanism of CASP8-associated CN quiescence is unknown and contradicts a mouse B-cell lymphoma CASP8-loss model that showed increased chromosomal instability — suggesting tissue/species-specific factors [PMID:23619168].
  • FAT1’s relevance to Wnt signaling in OSCC is unresolved: Wnt-pathway alterations were not frequently identified despite FAT1 inactivation [PMID:23619168] [PMID:23354438].
  • Whether the CIMP-like methylation cluster is a true CIMP is unsettled — it lacks MLH1 methylation and hypermutation typical of CRC CIMP [PMID:23619168].
  • No biomarker for cetuximab response in HNSCC was identified; remains an open question [PMID:23619168].
  • HPV status was not a focus of this analysis — relevant given HPV’s role in oropharyngeal subtypes of HNSCC.
  • Sample size for survival correlations (e.g., 8p loss) is modest; warrants replication.

Citations from this paper used in the wiki

  • “Integrated analysis revealed more somatic events than previously reported, identifying four major driver pathways (mitogenic signaling, Notch, cell cycle, TP53) and two additional key genes (FAT1, CASP8).” (Abstract)
  • “The Notch pathway was defective in 66% of patients, and in follow-up studies of mechanism, functional NOTCH1 signaling inhibited proliferation of OSCC cell lines.” (Abstract)
  • “Frequent mutation of CASP8 defines a new molecular subtype of OSCC with few copy number changes.” (Abstract)
  • “80% of patients harbored at least one genomic alteration in a targetable gene” (Abstract; p. 6 Targetable events: “80% (28/35) of the tumors exhibited one targetable genomic alteration, and 54% (19/35) exhibited two or more.”)
  • “FAT1 was mutated in 30% (12/40) of patients, and two thirds of those were inactivating nonsense, frameshift, or splice site. That is the highest frequency of FAT1 mutation yet observed in HNSCC.” (p. 5)
  • “Three of the four tumors with CASP8 mutations had an HRAS mutation, and three of the four HRAS mutant tumors had a CASP8 mutation. That association was also validated in the TCGA HNSCC cohort … seven of eight HRAS mutant tumors also had a CASP8 mutation.” (p. 6)
  • “Persistent activation of NOTCH1 led to a significant arrest in the G1 phase (i.e, 67% at 3 days post-infection)… accompanied by induction of p21 (CDKN1A).” (p. 4–5)
  • “Expression and methylation data have been deposited in the Gene Expression Omnibus database (accession no. GSE41117).” (Acknowledgments)

This page was processed by crosslinker on 2026-05-09.