Mutational landscape of aggressive cutaneous squamous cell carcinoma

Authors

Curtis R. Pickering

Jane H. Zhou

J. Jack Lee

Jennifer A. Drummond

S. Andrew Peng

Rami E. Saade

Kenneth Y. Tsai

Jonathan L. Curry

Michael T. Tetzlaff

Stephen Y. Lai

Jun Yu

Donna M. Muzny

Harshavardhan Doddapaneni

Eve Shinbrot

Kyle R. Covington

Jianhua Zhang

Sahil Seth

Carlos Caulin

Gary L. Clayman

Adel K. El-Naggar

Richard A. Gibbs

Randal S. Weber

Jeffrey N. Myers

David A. Wheeler

Mitchell J. Frederick

Doi

10.1158/1078-0432.CCR-14-1768

PMID: 25303977 · DOI: 10.1158/1078-0432.CCR-14-1768 · Journal: Clinical Cancer Research (2014)

TL;DR

Pickering et al. performed whole-exome sequencing of 39 aggressive cutaneous squamous cell carcinomas (CSCC) of the head and neck region, paired with matched normal lymphocytes, to define the driver landscape of this clinically aggressive, UV-driven disease. Despite an extreme UV-induced background mutation rate (median 61.2 mutations/Mb — one of the highest reported), [MutSig|mutsig] CV plus two newly developed inactivation-bias algorithms identified 23 candidate driver genes including the well-known tumor suppressors TP53, CDKN2A, NOTCH1, NOTCH2, AJUBA, HRAS, CASP8, FAT1, and KMT2C (MLL3), and the novel candidate tumor suppressors NOTCH2, PARD3, and RASA1. KMT2C mutations were associated with bone invasion (p=0.008) and shorter recurrence-free survival (HR 5.16; p=0.003).

Cohort & data

  • 39 patients with aggressive CSCC of the head and neck (snap-frozen tumor + matched normal blood), MD Anderson Cancer Center.
  • Aggression defined by regional/distant metastasis or features previously linked to mortality (Clayman et al. 2005): 38.5% recurrent, 25.6% persistent, 35.9% previously untreated; 7 metastatic-site samples; 71.8% invaded beyond subcutaneous space; 48.7% with perineural invasion (PNI); 43.6% poorly differentiated; 100% of evaluable cases with Clark level ≥4.
  • Assay: whole-exome-seq on Illumina HiSeq 2000 using the HGSC VCRome 2.1 42 Mb capture, mean 115× coverage.
  • Variant calling via the HGSC Mercury pipeline; copy number estimated with [ABSOLUTE|absolute] adjusting for purity and ploidy.
  • Driver discovery: [[MutSig|mutsig]] CV v1.4, IntOGen v2.3.0, plus two custom chi-square and multinomial inactivation-bias algorithms.
  • cBioPortal dataset: cscc_hgsc_bcm_2014.

Key findings

  • Extreme UV-driven mutation burden. Median 61.2 mutations/Mb across the cohort, >4× the melanoma rate and higher than other squamous tumor types; 75% of substitutions were C>T transitions (87% at a C following a pyrimidine) and 5.6% were dinucleotide polymorphisms — both UV signatures.
  • Four “nasal” outlier tumors lacked the UV signature (average 39% C>T, average 294 mutations), more closely resembling HPV-negative head and neck squamous cell carcinoma (HNSC, ~40% C>T). Authors hypothesize these arose from nasal mucosa rather than skin.
  • 23 candidate driver genes identified. TP53, CDKN2A, NOTCH2, NOTCH1, and AJUBA were significant by all four methods (MutSig, IntOGen, chi-square, multinomial); SNX25, EIF2D, and PARD3 were significant by the three alternative methods but not by MutSig; FAT1, KMT2C, and RASA1 were significant by the two inactivation-bias methods. MutSig alone identified 11 genes at q<0.1 (two of which — RBM46 and DCLK1 — had low allelic fractions suggesting they may not be true drivers).
  • Mutation frequencies in cSCC (this cohort, n=39): TP53 94.9%, CDKN2A 43.6%, NOTCH1 59.0%, NOTCH2 51.3%, AJUBA 23.1%, HRAS 20.5%, CASP8 23.1%, FAT1 43.6%, KMT2C 38.5%, KMT2D 69.2%, PARD3 30.8%, RASA1 12.8%, RIPK4 28% of UV-positive tumors.
  • NOTCH mutations are inactivating in cSCC. In both NOTCH1 and NOTCH2, missense mutations cluster in the N-terminal EGF-like ligand-binding repeats and truncating mutations are scattered throughout — opposite of activating T-ALL mutations which concentrate in the heterodimerization and PEST domains.
  • cSCC mutational spectrum is most similar to HNSC. Eight top genes (TP53, CDKN2A, NOTCH1, HRAS, CASP8, AJUBA, RASA1, FAT1, KMT2D) are shared with HNSC; only four are significant in LUSC and two in SKCM.
  • Mutations absent or atypical in cSCC compared with related squamous tumors. No NFE2L2 mutations (vs 7% in HNSC and 15% in LUSC); PIK3CA mutated in only 4 patients (10%) with no E545/H1047 hotspots and 2 inactivating mutations — not statistically significant.
  • No melanoma-type BRAF V600 or NRAS hotspot mutations were observed, but one RAC1 P29S and five mutations clustered around D89 in STK19 (3 D89N, 1 E88K, 1 P90S) — both reported in melanoma.
  • POLE mutations absent. None of the hypermutated tumors carried functionally relevant POLE mutations; UV exposure rather than polymerase failure drives the burden.
  • Copy number landscape parallels HNSC. Gains >25% of samples on chromosomes 7, 8q, 9q, 14, and 20; losses on 3p, 4, 5q, 8p, 9p, 11, 17p, 18, 19, 21; focal amplification of the CCND1 region on 11q13.
  • HRAS mutations are positively correlated with AJUBA (kappa 0.423, p=0.008) and inversely correlated with TP53 (kappa −0.107, p=0.004) — the same inverse TP53/HRAS relationship observed in HNSC.
  • RIPK4 emerges as a novel candidate driver. Excluding the four non-UV nasal tumors, RIPK4 became the 10th most significant gene by MutSig (q=0.053). Mutations were mutated in 28% of UV-positive tumors and clustered in exon 2 (kinase domain) and exon 8 (ankyrin repeats); 35% were truncating, consistent with tumor-suppressor inactivation.

Genes & alterations

  • TP53 — mutated in 94.9% of patients (37/39); top driver by all four methods (MutSig q-value below detection limit).
  • CDKN2A — 43.6%; significant by all four methods.
  • NOTCH1 — 59.0%; inactivating pattern (missense in EGF-like repeats, truncating throughout); significant by all four methods.
  • NOTCH2 — 51.3%; inactivating pattern paralleling NOTCH1; significant by all four methods. Novel candidate tumor suppressor. NOTCH2 mutation positively associated with perineural invasion (p=0.04; PNI in 70% of mutant vs 33% of wild-type) and with scalp/periorbital primary site (p=0.04).
  • AJUBA — 23.1%; significant by all four methods. AJUBA mutation correlated with greater depth of invasion (16.0 ± 6.4 mm vs 8.4 ± 5.6 mm, p=0.02) and with HRAS co-mutation.
  • HRAS — 20.5%; the most obvious oncogene identified, but currently undruggable. Inversely correlated with TP53 mutation.
  • CASP8 — 23.1%; differentiation-related.
  • FAT1 — 43.6%; differentiation-related tumor suppressor.
  • KMT2C — 38.5%; strongest clinical association. Bone invasion in 53% of mutant vs 10% of wild-type (p=0.008); median recurrence-free survival 21.6 vs 167.5 months (p=0.003); HR 5.16 (95% CI 1.55–17.18) for recurrence or death.
  • KMT2D — 69.2%; another histone-methylation regulator, significant by inactivation-bias methods.
  • PARD3 — 30.8%; novel candidate tumor suppressor, 33% of mutations predicted to truncate/eliminate the protein.
  • RASA1 — 12.8%; novel candidate tumor suppressor (RAS GTPase activating protein family); 66% of mutations truncating.
  • RIPK4 — 28% in UV-signature tumors; mutations clustered in kinase and ankyrin repeat domains; 35% truncating. Novel candidate driver; known to control keratinocyte differentiation.
  • STK19 — five mutations clustered around D89 (3 D89N, 1 E88K, 1 P90S); same hotspot region as in melanoma; potentially activating an as-yet uncharacterized kinase. (STK19 not in canonical HUGO list and therefore not included as a wiki gene entity.)
  • RAC1 — one P29S hotspot mutation (same as melanoma).
  • PIK3CA — 10% (4 patients, 5 mutations), no canonical E545/H1047 hotspot, 2 inactivating; not significant.
  • CCND1 — focal 11q amplification.
  • NFE2L2, BRAF V600, NRAS hotspot, POLE — absent in this cohort.
  • PTCH1 — 17%, only 2 inactivating; not a driver in cSCC (contrast with 75% inactivating in basal cell carcinoma).
  • SNX25, EIF2D, RBM46, DCLK1 — additional MutSig/inactivation-bias hits; RBM46 and DCLK1 had low allelic fractions suggesting they may be passengers.
  • EGFR — discussed as an existing but largely unsuccessful therapeutic target in cSCC (referenced for gefitinib and cetuximab trials); not significantly mutated in this cohort.

Clinical implications

  • KMT2C as a candidate prognostic biomarker for aggressive cSCC. Mutation predicts shorter recurrence-free survival (HR 5.16) and bone invasion. Authors propose KMT2C mutations may also flag an epigenetically targetable subset.
  • No directly targetable oncogenic drivers identified. The landscape is dominated by tumor suppressors. HRAS is the most obvious oncogene but remains difficult to target. STK19 hotspot mutations are speculatively activating but no current inhibitor exists.
  • UV signature as a diagnostic aid. Four tumors from the nose lacked the UV signature and resembled HPV-negative HNSC; the authors suggest the UV signature could help adjudicate ambiguous cutaneous-vs-mucosal origin in head-and-neck squamous tumors.
  • Biological similarity to HNSC suggests therapeutic strategies effective in HNSC may translate to aggressive cSCC.
  • NOTCH2 mutation may flag perineural-invasion-prone tumors (70% mutant PNI vs 33% wild-type), though this was an exploratory uncorrected analysis.
  • AJUBA mutation correlates with deeper invasion, another exploratory association.

Limitations & open questions

  • Single-institution cohort of 39 patients; exploratory clinicopathological correlations were uncorrected for multiple testing.
  • No non-aggressive cSCC control cohort: it remains unknown whether the identified drivers are enriched in aggressive vs indolent cSCC.
  • Functional roles of NOTCH2, PARD3, RASA1, and RIPK4 in cSCC require validation; mouse NOTCH2 knockouts are not predisposed to tumors, complicating interpretation.
  • STK19 hotspot mutations are intriguing but the kinase function and druggability are unknown.
  • Four “nasal” tumors lacked a UV signature and may represent misclassified mucosal HNSC; the cell of origin in such cases is genuinely ambiguous.
  • The HRAS oncogene, the most obvious therapeutic target, remains pharmacologically intractable at the time of writing (predating clinical KRAS G12C inhibitor approvals).

Citations from this paper used in the wiki

  • “Whole exome sequencing was performed on 39 cases of aggressive cSCC to identify driver genes and novel therapeutic targets.” (Abstract, p. 1)
  • “23 candidate drivers were identified including the well-known cancer-associated genes TP53, CDKN2A, NOTCH1, AJUBA, HRAS, CASP8, FAT1, and KMT2C (MLL3). Three novel candidate tumor suppressors with putative links to cancer or differentiation, NOTCH2, PARD3 and RASA1, were also identified as possible drivers in cSCC.” (Abstract, p. 1)
  • “KMT2C mutations were associated with poor outcome and increased bone invasion.” (Abstract, p. 1)
  • “A median of 61.2 mutations/Mb were detected in this cohort… more than 4 times as high as the rate in melanoma.” (Results, p. 5)
  • “75% of events were C>T transitions… 87% of those were at a C following a pyrimidine base. Additionally, 5.6% of events were dinucleotide polymorphisms (DNPs).” (Results, p. 5)
  • “Only 10% of patients with wild type KMT2C had bone invasion, compared to 53% of patients with KMT2C mutation (p=0.008)… The hazard ratio for recurrence or death for patients with KMT2C mutation was 5.16 (1.55 to 17.18, 95% CI).” (Clinical Significance, p. 8)
  • “Approximately 70% of patients with NOTCH2 mutations had PNI present compared to just 33% of patients with no NOTCH2 mutation (p=0.04).” (Clinical Significance, p. 8)
  • “HRAS was highly correlated with AJUBA (0.423, p=0.008) and inversely associated with TP53 (−0.107, p=0.004).” (Results, p. 7)
  • “RIPK4 was mutated in 28% of the tumors with a UV signature, with all mutations clustering in either exon 2 or exon 8 which encode the kinase and ankyrin repeat domains.” (Results, p. 7)
  • “TCGA data were collected from cBioPortal.” (Table 1 footnote)

This page was processed by crosslinker on 2026-05-12.