Integrated genomic analyses reveal frequent aberrations in acral melanoma

Authors

Winnie S. Liang

William Hendricks

Jeffrey Kiefer

Jessica Schmidt

Shobana Sekar

John Carpten

David W. Craig

Jonathan Adkins

Lori Cuyugan

Zarko Manojlovic

Rebecca F. Halperin

Adrienne Helland

Sara Nasser

Christophe Legendre

Laurence H. Hurley

Karthigayini Sivaprakasam

Douglas B. Johnson

Holly Crandall

Klaus J. Busam

Victoria Zismann

Valerie Deluca

Jeeyun Lee

Aleksandar Sekulic

Charlotte E. Ariyan

Jeffrey Sosman

Jeffrey Trent

Doi

10.1101/gr.213348.116

PMID: 28373299 · DOI: 10.1101/gr.213348.116 · Journal: Genome Research (2017)

TL;DR

Liang et al. performed integrated whole-exome, long-insert whole-genome, and RNA-seq analysis of 38 acral lentiginous melanoma (ALM) tumors from 34 patients — a rare, sun-shielded melanoma subtype with worse prognosis than cutaneous melanoma. Unlike cutaneous melanoma, ALM was dominated by structural variation rather than UV-derived single-nucleotide variants: only 38% of patients carried canonical BRAF/NRAS/NF1 driver mutations. The authors identified PAK1 copy gains in 15% of patients (exclusively in BRAF/NRAS wild-type tumors) and aberrations in TERT (translocations, copy gains, missense and promoter mutations, or germline events) in 41% of patients. In vitro, Telomerase Inhibitor IX produced ≥75% loss of viability in two primary ALM cell lines (one with a TERT CNV gain, one with a homozygous TERT promoter mutation) but not in normal melanocytes — providing preliminary evidence for TERT as a therapeutic target in ALM.

Cohort & data

  • Cohort: 38 ALM tumors from 34 patients (acral melanoma, ACRM) enrolled at Vanderbilt University and Memorial Sloan-Kettering Cancer Center; 3 patients contributed multiple tumors (patient 25: two metastases; patient 29: primary + two metastases; patient 34: primary + metastasis). Median tumor cellularity = 50%.
  • Dataset: mel_tsam_liang_2017; raw sequencing deposited in dbGaP under accession phs001036.v1.p1.
  • Assays: paired tumor/constitutional whole-exome sequencing (33 patients; tumor-only for 1), long-insert (~900 bp) whole-genome sequencing for SV/CNV detection (31 patients), and tumor RNA-seq (33 patients). Sanger sequencing was used to interrogate the TERT promoter (poorly covered by exome baits) in 28 patients.
  • Bioinformatics: alignment with BWA (DNA) and STAR (RNA); somatic SNV calling required 2-of-3 callers — Seurat (Q>30), MuTect, and Strelka; RNA fusions detected with TopHat-Fusion (quality>100); differential expression with Cufflinks/Cuffdiff and DESeq2.
  • Comparator: the cutaneous melanoma (SKCM) genomic landscape per The Cancer Genome Atlas Network (2015) and other prior CM cohorts.

Key findings

  • Low SNV burden, high SV burden. Median 42 somatic coding mutations per tumor (range 0–869); 90% of somatic coding point mutations were confirmed in matched RNA-seq. Across 31 LIWG-sequenced patients, 2,490 somatic breakpoints were identified across 74% of patients (median 31, range 0–683). Patient 14 showed chromothripsis with 503 breakpoints on Chr 12.
  • UV signature largely absent. C>T transitions made up 55.0% of point mutations, but only 39.4% occurred at dipyrimidines — below the 60% threshold defining a dominant UV signature. Two ALMs (patients 7 and 15; both BRAF-mutant) individually displayed a UV signature; signature analysis correlated dominant ALM signature S1 with the Alexandrov UV signature S7.
  • Canonical MAPK drivers under-represented vs CM. BRAF was mutated in 6 patients (V600E in 4 patients [12%], plus V600K/R462K and G466E); NRAS Q61K was mutated in 3 patients (9%) plus a fourth A59G event; the events were mutually exclusive. Combined, BRAF/NRAS driving events occurred in only 10/34 patients (29%). A single KIT L576P mutation was identified.
  • NF1 loss in 12% of patients. Homozygous loss in 9% (3 patients); a fourth (patient 19) had LOH plus a nonsense mutation (E2578*) on the second allele.
  • Low-frequency mutations in other key genes. Single events in EGFR (R334C, V726M), KRAS (V14L with CNV gain in patient 5), TP53 (R248W), and ERBB3 (S1119C). Two novel nonsynonymous events in PREX2 (S1167T, A355T).
  • Recurrent CNVs. 1,115 somatic focal CNVs (median 12, range 0–211); 48 statistically significant consensus CNVs (40 deletions, 8 amplifications). Key events: TERT and CLPTM1L gains on Chr 5p, CDKN2A loss on Chr 9, and CDK4 gain on Chr 12. TERT gains were validated by qPCR in 6/8 ALMs (75%). PTEN and NF1 deletions appeared only in primary tumors; MDM2 gain appeared only in metastases.
  • PAK1 copy gains in 5/34 patients (15%). Validated by real-time PCR in 4/5 (80%); elevated PAK1 expression in 2/5. All PAK1-gain patients were BRAF/NRAS wild-type — one NF1-subtype, four triple-wild-type — supporting PAK1 as an alternative MAPK-pathway dysregulation route in TWT ALM.
  • Most-rearranged genes. ADCY2 (32 breakpoints across 7 samples = 21% of patients) and CLPTM1L (22 breakpoints across 6 samples = 15%). ADCY2 partners include CLPTM1L, HECTD4, TERT, and UBE2QL1; CLPTM1L partners include ADCY2, PDZD2, RAI14, and TRIO. All occurred in BRAF wild-type tumors.
  • Pathway-level perturbation. MAPK/PI3K (proliferation/survival) altered in 66% of patients; TERT (telomere maintenance) in 37%; CDK4/CDKN2A (cell cycle) in 51%; MDM2/TP53 (apoptosis/senescence) in 17%.
  • RNA fusions. 106 fusions across 74% of patients (median 2); 13% had supporting DNA breakpoints. Notable fusions include MDM2:GNS, MDM2:CCT2, PTEN:RPL11, PAK2:LOC646214, and MAP3K8:DEK.
  • Neo-antigen burden. Across 22 patients receiving immune checkpoint blockade (10 anti-CTLA-4 only; 10 anti-CTLA-4 + anti-PD-1), neo-antigen burden correlated strongly with mutation burden (Pearson r = 0.89). Three complete responders (two to anti-PD-1, one to anti-CTLA-4) had low mutation (<75) and neo-antigen (<60) burdens — opposite of the trend reported in cutaneous melanoma.

Genes & alterations

  • TERT — 41% of patients (14/34) with somatic or germline aberrations. Includes consensus CNV gains (TERT-CLPTM1L locus on Chr 5p), promoter mutations in 4/28 patients sequenced by Sanger (Chr5:1,295,113:G>A; patients 15, 24, 25a, 25b), one exonic nonsynonymous mutation (F919L; patient 12), one intronic SNV in the RNA-interacting/oligomerization domain (patient 24), TERT breakpoints in 4 patients (12, 17, 29a/29c, 33), and rare germline polymorphisms (A1062T in patients 1, 4; T1110M in patient 8) predicted damaging by FATHMM and mutually exclusive of somatic TERT events. All patients with somatic TERT events expressed TERT. Patient 12 expressed a TERT:ADCY2 RNA fusion. Patient 29 (both metastases) carried a TERT:PDCD1LG2 interchromosomal rearrangement plus a possible TERT-FER inversion.
  • PAK1 — focal copy gains in 5/34 patients (15%); validated by qPCR in 4/5; all in BRAF/NRAS wild-type tumors (1 NF1-subtype, 4 triple-wild-type).
  • BRAF — 6 patients (18%): V600E (4), V600K + R462K (1), G466E (1). Mutually exclusive of NRAS.
  • NRAS — 4 patients: Q61K (3, 9%) and A59G (1).
  • **NF1 — homozygous loss in 9% (3 patients) plus LOH+nonsense (E2578*) in patient 19.**
  • KIT — single L576P mutation (1 patient).
  • CDKN2A — recurrent deletion (Chr 9), retained significance in both primary and metastatic subsets.
  • CDK4 — recurrent gain (Chr 12), significance retained only in metastases.
  • CLPTM1L — frequent breakpoints (15% of patients) and consensus copy gain on Chr 5p, adjacent to TERT.
  • ADCY2 — most-rearranged gene (21% of patients).
  • MDM2 — copy gain in metastases-only; MDM2:GNS and MDM2:CCT2 RNA fusions.
  • PTEN — focal deletion in primary tumors only; PTEN:RPL11 fusion (single tumor, no other PTEN alterations).
  • PREX2 — two novel nonsynonymous events (S1167T, A355T); not previously reported in melanoma although S1167N was reported in CM metastasis.
  • EGFR, KRAS, TP53, ERBB3 — single-patient missense events.
  • CCND1, GAB2 — discussed as previously reported ALM amplifications (cited literature).
  • MITF — notably absent from the top 421 differentially expressed genes used for clustering, in contrast to its key role in CM.

Clinical implications

  • TERT inhibition is selectively cytotoxic to ALM cells in vitro. Treatment with 2.5 µM Telomerase Inhibitor IX for 72 h produced ≥75% loss of viability in two primary ALM lines (SMC-09 with TERT CNV gain; SU2C-001-002 with homozygous TERT promoter mutation Chr5:1,295,113:G>A) versus only ~12% loss in normal melanocyte controls (NHM-002). TERT mRNA dropped ≥25% in ALM lines and was undetectable in NHM-002. The authors propose Telomerase Inhibitor IX–type compounds as a putative therapeutic strategy for ALM, particularly given limited targeted treatment options.
  • PAK1 copy gains identify a BRAF/NRAS wild-type subgroup. PAK1 has been proposed as a target in BRAF wild-type melanoma; this study supports that hypothesis in the ALM TWT context, suggesting alternate MAPK-pathway dysregulation routes.
  • Immune checkpoint blockade — atypical responder profile. Among 22 patients treated with ipilimumab and/or pembrolizumab, the three complete responders had low mutation (<75) and neo-antigen (<60) burdens. The expected mutation/neo-antigen-burden correlation with response (well-established in CM) was not observed — the authors flag low statistical power but raise the possibility that ALM responders use distinct immunogenic mechanisms.
  • TERT-PDCD1LG2 (PDL2) rearrangement (patient 29). A novel structural rearrangement linking the TERT reverse-transcriptase domain (exon 7) to intron 5 of PDCD1LG2 — clinically intriguing given PDL2’s role in PD-1 ligand biology, although no expressed RNA fusion was detected.

Limitations & open questions

  • Functional impact of TERT structural events is unclear. Patient 12’s complex TERT rearrangement was associated with elevated TERT expression (FPKM 32.5), but most other TERT SVs lacked supporting RNA fusions; the authors call out that further functional studies are needed to verify how these aberrations modulate TERT activity.
  • Sample size limits power for survival/biomarker correlations. No significant correlation was detected between mutation burden and BRAF/NRAS status, between expression clusters and clinical features (age, gender, race, primary site, in-transit status, immunotherapy response), or between mutation/neo-antigen burden and checkpoint response.
  • Three previously reported ALM-recurrent genes were not mutated here. No somatic SNVs/indels were observed in DYNC1I1, ARID1A, or APC (cited Furney et al. 2014).
  • TERT promoter coverage was poor in exome data. Required orthogonal Sanger sequencing; only 28 of 34 patients had sufficient DNA, so promoter-mutation prevalence may be undercounted.
  • Pretreatment confounds mutation burden. Patient 7 (highest SNV burden, 867 coding SNVs) was sequenced after ipilimumab → pembrolizumab; pretreated patients 4, 5, 28 also trended toward higher mutation counts — complicating comparisons of “raw” mutation burden to other ALM cohorts.
  • Etiologic relationship between ALM and sun-shielded CM remains unresolved. The authors note that characterizing ALM may also illuminate non-UV oncogenic mechanisms in other cancers, but the current cohort is too small to anchor cross-subtype claims.

Citations from this paper used in the wiki

  • “Only 38% of patients demonstrated driver BRAF/NRAS/NF1 mutations. In contrast with CM, we observed PAK1 copy gains in 15% of patients, and somatic TERT translocations, copy gains, and missense and promoter mutations, or germline events, in 41% of patients.” — Abstract.
  • “After 72 h of drug treatment, we observed at least a 75% decrease in cell viability with 2.5 µM of Telomerase Inhibitor IX, and not in normal melanocytes.” — Results, TERT inhibition in ALM.
  • “PAK1 gains were separately validated by real-time PCR for available samples (four of five ALMs; 80%). … These patients were exclusive of characteristic CM BRAF and RAS subtypes, with one patient demonstrating the NF1 subtype and the remaining four demonstrating the triple-wild-type (TWT) subtype.” — Results, Structural alterations in ALM.
  • “In addition to TERT gains, additional TERT events were also detected with a total of 14 (41%) patients demonstrating either somatic or germline aberrations in this gene.” — Results, TERT alterations in ALM.
  • “Two patients with complete response to anti-PD1, and one with complete response to anti-CTLA4 treatments had lower mutation (less than 75) and neo-antigen (less than 60) burdens.” — Results, Neo-antigen burden.
  • “The sequencing data from this study have been submitted to the NCBI Database of Genotypes and Phenotypes (dbGaP; …) under accession number phs001036.v1.p1.” — Methods, Data access.

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