The long tail of oncogenic drivers in prostate cancer

Authors

Joshua Armenia

Stephanie A.M. Wankowicz

David Liu

Jianjiong Gao

Ritika Kundra

Ed Reznik

Walid K. Chatila

Debyani Chakravarty

Barry S. Taylor

Howard I. Scher

Peter S. Nelson

Johann S. de Bono

Mark A. Rubin

Charles L. Sawyers

Arul M. Chinnaiyan

Nikolaus Schultz

Eliezer M. Van Allen

Doi

10.1038/s41588-018-0078-z

PMID: 29610475 · DOI: 10.1038/s41588-018-0078-z · Journal: Nature Genetics (2018)

TL;DR

Armenia and colleagues aggregated and uniformly re-analyzed whole-exome sequencing of 1,013 prostate tumor/normal pairs (680 primary, 333 metastatic castration-resistant) from prior cohorts to define the “long tail” of prostate cancer drivers. Mutational significance analysis (MutSig2CV plus biological filters) yielded 97 significantly mutated genes (SMGs), 70 of them not previously implicated in prostate cancer. The work defines a new class of ETS-fusion-negative tumors driven by mutations in epigenetic regulators / chromatin remodelers (20% of samples), nominates the spliceosome (SF3B1, U2AF1) and ubiquitin-pathway gene CUL3 as novel prostate-cancer pathways, identifies SPEN as a metastasis-enriched AR-pathway driver, and shows that PI3K, DNA-repair, cell-cycle, WNT/CTNNB1 and epigenetic-regulator alterations are all enriched in metastatic vs primary disease.

Cohort & data

  • N = 1,013 tumor/germline pairs aggregated from prior WES studies — 680 primary (treatment-naive) and 333 metastatic (castration-resistant) PRAD tumors. Cohort deposited in cBioPortal as prad_p1000.
  • Validation cohorts: 706 advanced prostate cancers from MSK-IMPACT (msk_impact_2017) and 204 prostate cancers from Foundation Medicine (phs001179).
  • Uniform re-alignment via PICARD; SNVs called with MuTect v1.1.6, indels with Strelka v1.0.11; copy-number from ReCapSeg + GISTIC 2.0; allele-specific copy-number, purity and ploidy from FACETS v0.5.10; clonality (cancer-cell fraction) via FACETS / ABSOLUTE-style framework.
  • Significance analysis: MutSig2CV (integrating MutSigCV, MutSigFN, MutSigCL) with additional biological filters (≥10 altered samples, allelic-fraction floors, length / expression / oncogenic-variant filters) plus OncoKB rescue of genes with ≥5 oncogenic variants.
  • Tumor mean target coverage 104.7×; normals 103.8×; ContEst < 5% (mean 0.6%).

Key findings

  • 97 SMGs identified, including 70 not previously implicated in prostate cancer and 9 not previously reported as recurrently altered in any cancer type PMID:29610475. Most novel SMGs are mutated in <5% of tumors and required >900 samples to detect.
  • Mutational and copy-number burden are higher in metastatic disease. Mean non-synonymous mutation rate 1.36 mut/Mb (primary) vs 2.93 mut/Mb (metastatic); after adjustment for tumor purity and sequencing depth, metastatic tumors carry an estimated 1.43 mut/Mb excess (p<0.001). Copy-number burden also higher in metastatic (p<0.001). In primary tumors, increased age and higher Gleason score associate with higher mutation burden (p=0.02, p<0.001) and copy-number burden (p<0.001, p<0.001).
  • Epigenetic/chromatin-remodeler mutations define a class of ETS-fusion-negative tumors. 20% of samples carry alterations (often truncating) in epigenetic modifiers or chromatin-remodeling genes; SWI/SNF members ARID1A (1.6%), ARID4A (1%), ARID2 (1.3%) and SMARCA1 (1.1%) are the most frequent. In primary tumors these alterations associate with higher Gleason score (10% Gleason 3+4 vs 22% Gleason 8–10, p=0.001 Fisher’s exact). Within the ETS-status-known subset (n=765), epigenetic-regulator mutations are enriched in ETS-fusion-negative tumors (p=1×10^-4) and in tumors lacking any previously known driver (ETS fusion, IDH1/SPOP/CUL3, FOXA1) (p=0.007).
  • CUL3 is a new SPOP-like prostate-cancer driver. CUL3 is mutated in 1.3% of cases, primarily at the recurrent p.Met299Arg hotspot (5 tumors), exclusively missense, and mutually exclusive with SPOP mutations (cohort underpowered to reach statistical significance). CUL3-mutant tumors show SPOP-like copy-number profiles (5q, 6q, 13 losses). The MSK-IMPACT validation cohort confirmed 9 additional somatic CUL3 mutations (1.3%), including three p.Met299Arg events. Including USP28 (1.4%) and USP7 (1.2%), 12% of tumors harbor ubiquitin-pathway alterations.
  • Spliceosome pathway altered in 4% of tumors. Hotspot mutations in SF3B1 (1.1%, clustered in C-terminal HEAT repeats) and U2AF1 (0.5%) — the first nomination of splicing-factor mutations as a recurrent prostate-cancer pathway.
  • SPEN is a new metastasis-enriched AR-pathway driver. SPEN is mutated in 2.4% of the cohort, mostly truncating, significantly enriched in metastatic samples (q=0.008, Fisher’s exact), and clonal — consistent with a driver role in advanced disease.
  • Novel PI3K subunit driver. PI3K pathway altered in 25% (PTEN homozygous loss / truncation 16%); PIK3R2 is mutated in 1% and is paralogous to PIK3R1. The PIK3R2 p.Asp557Tyr mutation is paralogous to the known oncogenic PIK3R1 p.Asp560Tyr and was reproduced in the validation cohort.
  • WNT/CTNNB1 pathway altered in 10%. CTNNB1 mutations cluster in the N-terminal domain plus a novel p.Lys335Ile hotspot near the AXIN-binding interface; APC carries loss-of-function alterations.
  • DNA-repair SMGs include MRE11A and PALB2. DNA-repair pathway altered in 16% overall. Novel prostate-cancer-specific SMGs include MRE11 and PALB2; CDK12 mutations are predominantly truncating (p<0.001 binomial), and 15/31 CDK12-mutant primary tumors (and 27/56 in validation) carry two CDK12 mutations, suggestive of frequent biallelic inactivation. CDK12 missense variants cluster in the kinase domain (p<0.001 binomial).
  • RAS/RAF/MAPK pathway altered in 5%, including SMGs in KRAS and BRAF at established hotspots that had not previously reached significance in prostate cancer.
  • Primary-vs-metastatic comparison defines a high-risk genomic signature. Genes enriched in metastatic disease: TP53, AR, PTEN, RB1, FOXA1, APC, BRCA2, plus epigenetic regulators KMT2C and KMT2D. SPOP is enriched in primary tumors; after mutation-rate correction, IDH1 and ZMYM3 are also enriched in primary (p=0.01, weighted permutation test). At pathway level, PI3K, DNA-repair, cell-cycle, WNT/CTNNB1 and epigenetic regulators are all enriched in metastatic vs primary (p<0.0001 Fisher’s exact).

Genes & alterations

  • CUL3 — recurrent missense hotspot p.Met299Arg in 1.3% of primary cohort and 1.3% of MSK-IMPACT validation; mutually exclusive with SPOP mutations and shows SPOP-like copy-number profile (losses on 5q, 6q, 13). Encodes part of the BTB-CUL3-RBX1 E3 ubiquitin ligase complex with SPOP.
  • SPEN — 2.4% mutation rate, predominantly truncating, clonal, significantly enriched in metastatic samples (q=0.008). Encodes a hormone-inducible transcriptional repressor; nominated as a driver in advanced/metastatic prostate cancer.
  • SF3B1 (1.1%) and U2AF1 (0.5%) — first significant nomination of splicing-factor mutations as a recurrent prostate-cancer alteration; SF3B1 mutations cluster in C-terminal HEAT repeats.
  • PIK3R2 (1%) — novel PI3K-pathway SMG; p.Asp557Tyr is paralogous to oncogenic PIK3R1 p.Asp560Tyr and reproduced in validation cohort.
  • CTNNB1 — N-terminal hotspot mutations plus novel p.Lys335Ile hotspot near the AXIN-binding domain.
  • CDK12 — truncating-biased mutation pattern (p<0.001) with frequent biallelic inactivation (15/31 in discovery, 27/56 in validation); missense variants cluster in the kinase domain.
  • MRE11 and PALB2 — novel prostate-specific DNA-repair SMGs.
  • KMT2C, KMT2D — epigenetic-regulator SMGs significantly enriched in metastatic tumors.
  • ARID1A, ARID2, ARID4A, SMARCA1 — SWI/SNF members altered in ~5% of cohort, enriched in ETS-fusion-negative tumors.
  • USP28, USP7 — additional ubiquitin-pathway SMGs.
  • KRAS, BRAF — established hotspots, newly significant in prostate cancer.
  • Established drivers (TP53, AR, PTEN, RB1, FOXA1, APC, BRCA2, SPOP, IDH1, ZMYM3, ERG/TMPRSS2 fusions) confirmed; metastasis-vs-primary enrichment quantified.

Clinical implications

  • The metastasis-enrichment signature (TP53, AR, PTEN, RB1, FOXA1, APC, BRCA2, KMT2C, KMT2D and aggregated PI3K / DNA-repair / cell-cycle / WNT / epigenetic alterations) defines a candidate genomic marker set the authors propose for prospective risk stratification in localized prostate cancer.
  • CUL3 hotspot p.Met299Arg defines a small but recurrent SPOP-like tumor subset and may merit annotation in clinical prostate-cancer panels.
  • Splicing-factor mutations (SF3B1, U2AF1) introduce a new class of potentially actionable alterations in prostate cancer, given splicing-modulator drug development in other diseases.
  • PIK3R2 p.Asp557Tyr is paralogous to a known oncogenic PIK3R1 mutation, suggesting it should be evaluated as a candidate PI3K-pathway biomarker.
  • Frequent biallelic CDK12 inactivation in metastatic prostate cancer is consistent with CDK12 loss as a therapeutically relevant DNA-repair phenotype.
  • All mutation calls and clinical annotation are publicly available in cBioPortal as study prad_p1000 (http://www.cbioportal.org/study?id=prad_p1000).

Limitations & open questions

  • Cohort assembled from heterogeneous prior WES projects — uniform re-alignment and joint QC mitigate but cannot fully eliminate batch effects (bait-set differences, sequencing-depth variation, FFPE artifacts addressed by orientation-bias filtering).
  • Many novel SMGs are mutated in <3% of cases; cohort still underpowered to establish statistical significance for rare phenomena (e.g., CUL3-vs-SPOP mutual exclusivity is biologically clear but not statistically resolved).
  • Functional consequences of the long-tail SMGs are largely unknown; the authors call out the need for functional follow-up to discriminate prostate-cancer-relevant drivers from passengers among the statistically nominated genes.
  • Even larger uniformly analyzed cohorts will be needed to refine SMG calls and to address stochastic differences in variant detection across source projects.
  • Saturation analyses suggest expanded primary-prostate cohorts may be near saturation for gene discovery; metastatic-cohort expansion remains the higher-yield direction for future driver discovery.

Citations from this paper used in the wiki

  • “Here we aggregate and uniformly analyze exome sequencing data from 1013 prostate cancers” (Abstract) — defines prad_p1000.
  • “We identify a total of 97 SMGs, including 70 not previously implicated in prostate cancer, such as the ubiquitin ligase CUL3 and the transcription factor SPEN” (Abstract) — headline result.
  • “20% of prostate cancer samples with mutations, frequently truncating, in epigenetic modifiers or chromatin remodeling genes” (p.3) — epigenetic-regulator class.
  • “alterations in epigenetic regulators and chromatin remodelers were significantly more common in tumors that lack an ETS fusion (p=1e-04, Fisher’s exact test)” (p.3) — ETS-negative subclass definition.
  • “CUL3 mutations were primarily in a hotspot, p.Met299Arg, and were mutually exclusive with SPOP mutations” (p.4) — CUL3 finding.
  • “we identified nine additional somatic CUL3 mutations in an independent cohort of advanced prostate cancers (1.3% in the MSK-IMPACT data)” (p.4) — MSK-IMPACT validation.
  • “the splicing pathway was altered in 4% of prostate tumors … hotspot mutations in SF3B1 (1.1%) and U2AF1 (0.5%)” (p.4) — spliceosome pathway.
  • “SPEN mutations were significantly enriched in metastatic samples (q=0.008, Fisher’s exact test) and clonal” (p.4) — SPEN driver claim.
  • “PIK3R2 (1%), which, like PIK3R1, encodes a PI3K regulatory subunit … p.Asp557Tyr, is paralogous to the known oncogenic p.Asp560Tyr mutation in PIK3R1” (p.4) — novel PI3K driver.
  • “Genes with enrichment in metastatic samples include TP53, AR, PTEN, RB1, FOXA1, APC, and BRCA2 … KMT2C and KMT2D are also significantly enriched in metastatic tumors” (p.5) — metastasis signature.
  • “After correction for differences in mutational load, IDH1 and ZMYM3 mutations were also enriched in primary tumors (p=0.01, mutation rate-adjusted permutation test)” (p.5) — primary-enrichment.
  • “PI3K, DNA repair, Cell cycle, WNT/CTNNB1, and epigenetic regulators were significantly more frequently altered in metastatic compared to primary tumors (p<0.0001 Fisher’s exact test)” (p.5) — pathway-level metastasis enrichment.

This page was processed by crosslinker on 2026-05-15.