Whole-Genome Sequencing of Salivary Gland Adenoid Cystic Carcinoma

Authors

Rettig EM

Talbot CC Jr

Sausen M

Jones S

Bishop JA

Wood LD

Tokheim C

Niknafs N

Karchin R

Fertig EJ

Wheelan SJ

Marchionni L

Considine M

Ling S

Fakhry C

Papadopoulos N

Kinzler KW

Vogelstein B

Ha PK

Agrawal N

Doi

10.1158/1940-6207.CAPR-15-0316

PMID: 26862087 · DOI: 10.1158/1940-6207.CAPR-15-0316 · Journal: Cancer Prevention Research (2016)

TL;DR

Rettig et al. performed whole-genome sequencing of 25 fresh-frozen salivary-gland adenoid cystic carcinomas (ACCs) with matched normal tissue, plus mRNA expression analysis of 17 of those tumors. Beyond the canonical t(6;9) MYB–NFIB fusion (44%), they uncovered five additional rearrangements involving NFIB (with MAP3K5, MYBL1, RIMS1, MYO6, and RPS6KA2), pushing total NFIB translocation prevalence to 60% (15/25). NFIB mRNA was overexpressed in tumors versus normal tissue (p=0.002) regardless of fusion status, suggesting an MYB-independent oncogenic role for NFIB in ACYC. The study also reports recurrent somatic mutations in Rho-family GTPase signaling (44% of tumors, p=3.9×10⁻⁶) and axon-guidance signaling (56%, p=8.3×10⁻⁵) pathways, and corroborates earlier findings of chromatin-regulator and Notch-pathway disruption.

Cohort & data

  • n=25 fresh-frozen surgically resected ACC tumors with matched blood/normal tissue, accrued under IRB at Johns Hopkins Hospital and the Salivary Gland Tumor Biorepository (PMID:26862087).
  • Cancer type: salivary-gland ACYC (adenoid cystic carcinoma); tumor sites included parotid, submandibular, sublingual, minor salivary glands of palate/maxilla/lip/septum/ear (Table 1).
  • Neoplastic cellularity ≥60% (mean) achieved via macro-dissection from frozen-section histology.
  • Dataset: acyc_jhu_2016 (genomic and mRNA data deposition described as “in process” at time of publication).
  • Assays: whole-genome-seq on all 25 tumor/normal pairs (mean coverage 65.2× tumor / 38.1× normal, 98.2% of targets ≥10×); rna-seq on 17 tumors (Illumina HiSeq 2000, ≥50M paired 100×100 bp reads/sample).
  • Bioinformatics: variantdx custom somatic-mutation caller (matched tumor/normal, hg19), tophat-fusion for fusion-transcript discovery, rsem for transcript quantification, snpEff for functional annotation, ingenuity-pathway-analysis for pathway enrichment.

Key findings

  • 396 somatic mutations in 372 genes across 25 tumors, median 14 mutations/tumor (range 2–36) (PMID:26862087).
  • 253 chromosomal rearrangements total (median 7/tumor, range 0–42), with breakpoint hotspots on chromosome 6q (28% of breakpoints) and 9p (14%).
  • MYB-NFIB fusion detected in 11/25 tumors (44%, 95% CI 27–63%) by whole-genome-seq.
  • Five novel/expanded NFIB fusion partners identified: MAP3K5–NFIB (2 tumors), MYBL1–NFIB (2 tumors), RPS6KA2–NFIB (1), MYO6–NFIB (1), RIMS1–NFIB (1). Three tumors carried both MYB-NFIB and a non-MYB NFIB fusion. Total NFIB translocation prevalence: 15/25 (60%, 95% CI 41–77%).
  • NFIB intrachromosomal rearrangements in 3 additional tumors (2 deletions, 1 inversion), including HN 324 which lacked any NFIB translocation.
  • mRNA confirmation: 9/9 tumors with MYB-NFIB DNA rearrangement and available RNA expressed the fusion transcript; one extra tumor (HN 335 PT) had MYB-NFIB transcript without detected DNA rearrangement; only 2/6 non-MYB NFIB rearrangements (RIMS1-NFIB, MYBL1-NFIB) yielded detectable fusion transcripts.
  • NFIB overexpressed in tumors vs. normal tissue (p=0.002), with no difference between NFIB-fusion-positive and -negative tumors (p=0.91).
  • MYB overexpressed in tumors vs. normal (p<0.001), and higher in MYB-NFIB-fusion-positive than -negative tumors (p<0.001). The two MYBL1-NFIB tumors (HN 333 PT, HN 320 PT) uniquely lacked MYB expression and trended toward higher MYBL1 expression (p=0.07), consistent with mutually exclusive MYB/MYBL1 alterations.
  • Most-mutated gene: NOTCH1 (4 mutations in 3 tumors; nonsense, missense, frameshift, in-frame deletion). NOTCH2 also carried a non-synonymous substitution (Table 2).
  • Chromatin-regulator mutations in KMT2D (MLL2; 3 mutations/2 tumors), KMT2C (MLL3), EP300, SMARCA2, SMARCC1, and KDM6A; truncating mutations were enriched (6 of 11 non-silent) at p=3.8×10⁻⁶ vs. randomization-based null. Individually significant: MLL2 (p=0.008) and EP300 (p=0.01).
  • Rho-family GTPase signaling disrupted in 11/25 tumors (44%, p=3.9×10⁻⁶ by IPA), including RHOA (2 tumors with distinct mutations), CDH2, CDH13, ARHGEF3, ARHGEF5, ARHGEF18, ACTB and others (Figure 4).
  • Axon-guidance signaling disrupted in 14/25 tumors (56%, p=8.3×10⁻⁵), including PLXNB1 (2), PRKD1 (2), SLIT1, PLXNA1, SRGAP2, SRGAP3, ADAM2, ADAMTS5, ADAMTS13, SEMA3G (one each). An identical SEMA3G 474S>P substitution was also observed in the prior Ho et al. cohort (PMID:23685749).
  • 41 copy-number variations in 12/25 tumors (range 0–9). Most common: amplifications at 7p14.1 (5 tumors) and 14q11.2 (4 tumors) — both T-cell-receptor loci, possibly artifactual; deletions at 10q26.3 spanning DUX4/DUX4L homeobox genes in 3 tumors.
  • Pathway-level disruption (mutations + rearrangements) corroborated prior reports for protein-kinase-A signaling (21 tumors), FGF signaling (15), Wnt/β-catenin (13), PI3K/AKT (13), and Notch (9).

Genes & alterations

  • NFIB — translocations in 15/25 (60%) tumors with five distinct partners (MYB, MAP3K5, MYBL1, RIMS1, MYO6, RPS6KA2); plus intrachromosomal rearrangements in 3 tumors. Overexpressed at the mRNA level vs. normal (p=0.002) independent of fusion status (p=0.91), implicating a fusion-independent oncogenic role.
  • MYB — t(6;9) MYB–NFIB fusion in 11/25 tumors. Overexpressed in tumors vs. normal (p<0.001) and significantly higher in fusion-positive tumors (p<0.001).
  • MYBL1 — recurrent MYBL1–NFIB fusion in 2 tumors (HN 333 PT, HN 320 PT) which uniquely lacked MYB expression, supporting a mutually exclusive MYB/MYBL1 alteration model in ACYC.
  • MAP3K5 (ASK1) — recurrent fusion partner with NFIB in 2 tumors; encodes a serine/threonine kinase regulating apoptosis.
  • RIMS1, MYO6, RPS6KA2 — single-tumor NFIB fusion partners on chromosome 6q.
  • NOTCH1 — most-frequently mutated gene; 4 mutations in 3 tumors (nonsense, missense, frameshift, in-frame deletion) — 12% of cohort.
  • NOTCH2 — non-synonymous substitution in one tumor.
  • KMT2D (MLL2) — 3 mutations in 2 tumors; significant enrichment of truncating mutations (p=0.008).
  • KMT2C (MLL3) — 2 mutations in 2 tumors.
  • EP300 — 2 mutations in 2 tumors; significant enrichment of truncating mutations (p=0.01).
  • SMARCA2, SMARCC1, KDM6A — recurrent chromatin-regulator mutations.
  • RHOA — two distinct somatic mutations in 2 tumors; first reported in ACYC. (Recurrent RHOA hotspot mutations had previously been reported in angioimmunoblastic T-cell lymphoma; see AITL.)
  • ARHGEF3, ARHGEF5, ARHGEF18, CDH2, CDH13, ACTB — additional Rho-pathway mutations.
  • PLXNB1, PRKD1 — recurrent (2 tumors each) axon-guidance mutations.
  • PLXNA1, SLIT1, SRGAP2, SRGAP3, SEMA3G, ADAM2, ADAMTS5, ADAMTS13 — single-tumor axon-guidance mutations.
  • DUX4 — homozygous deletions at 10q26.3 in 3 tumors (interpreted with caution as DUX4/DUX4L are tandemly repeated).

Clinical implications

  • NFIB as a therapeutic target. Authors argue that the high prevalence of NFIB translocations (60% of cases) and fusion-independent NFIB overexpression suggest “targeting NFIB alone may also be of high yield,” independent of MYB. (Statement of hypothesis only; no functional/therapeutic data.)
  • Rho-pathway and axon-guidance pathway inhibition are flagged as candidate therapeutic strategies for ACYC, leveraging existing precision-anticancer Rho-targeting research and Slit/Robo pathway exploration.
  • Notch pathway disruption (12% of cases) is highlighted as a potentially targetable signaling axis, given existing Notch-targeting therapeutics in development.
  • Perineural invasion (PNI) link. The disruption of axon-guidance signaling in 56% of tumors is hypothesis-generating for ACYC’s well-known propensity for PNI; the same pathway is disrupted in pancreatic adenocarcinoma, another PNI-prone malignancy.
  • No predictive biomarker, treatment outcome, or survival analysis is reported in this study (n=25, fresh-frozen discovery cohort).

Limitations & open questions

  • Small cohort (n=25). Statistical power for rare events and pathway enrichment is limited; novel single-tumor NFIB fusion partners (RPS6KA2, MYO6, RIMS1) are not corroborated by other ACC sequencing studies.
  • mRNA confirmation incomplete. Only 2/6 non-MYB NFIB DNA rearrangements yielded detectable fusion transcripts, leaving the functional significance of those novel rearrangements unclear.
  • Functional role of NFIB overexpression unresolved. NFIB mRNA is elevated in tumors regardless of fusion status (p=0.91 between groups), but the mechanism of dysregulation in fusion-negative tumors is not characterized.
  • T-cell-receptor amplifications interpreted with caution as potential artifacts of TCR gene rearrangement in tumor-infiltrating lymphocytes.
  • No survival/outcome data; cohort is descriptive genomic-discovery only.
  • DUX4 deletion calls at 10q26.3 are flagged by the genomic context (tandem-repeat region) as needing orthogonal validation.
  • Therapeutic implications are speculative. The paper proposes NFIB, Rho, axon-guidance, and Notch pathways as therapeutic targets but provides no in vitro or in vivo functional data; downstream functional characterization is called out as required.
  • Cross-paper synthesis: authors corroborate chromatin-regulator and Notch findings from Ho et al. (PMID:23685749), Stephens et al. (PMID:23778141), and Ross et al. (PMID:24418857); the novel finding here is the pan-NFIB rearrangement landscape and the Rho-pathway involvement.

Citations from this paper used in the wiki

  • “We identified this translocation [t(6;9) MYB-NFIB] in 11 of 25 tumors (44%, 95%CI=27–63%)” — Results, p.7.
  • “Overall, NFIB translocations occurred in 15 of 25 tumors (60%, 95%CI=41–77%)” — Results, p.7.
  • “NFIB was overexpressed in tumors relative to normal tissues (p=0.002)” — Results, p.7.
  • “NFIB expression was not significantly different in tumors with NFIB fusion genes compared to those without NFIB fusion genes (p=0.91)” — Results, p.7.
  • “11 (44%, 95%CI=27–63%) tumors containing somatic mutations in one or more members of this [Rho GTPase] pathway (p=3.9x10−6)” — Results (Pathway analyses), p.8.
  • “The axon guidance signaling pathway was also highly disrupted (14 tumors [56%, 95%CI=37–73%], p=8.3x10−5)” — Results (Pathway analyses), p.8.
  • “NOTCH1 was the most frequently altered gene, with four mutations in three tumors” — Results (Sequence mutations), p.6.
  • “p=3.8 x 10−6, randomization-based test” for chromatin-regulator truncating-mutation enrichment, with MLL2 (p=0.008) and EP300 (p=0.01) individually significant — Results, p.6.
  • “Our data implicate that targeting NFIB alone may also be of high yield” — Discussion, p.10.

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