Prospective Genomic Profiling of Prostate Cancer Across Disease States Reveals Germline and Somatic Alterations That May Affect Clinical Decision Making

Authors

Wassim Abida

Joshua Armenia

Anuradha Gopalan

Ryan Brennan

Michael Walsh

David Barron

Daniel Danila

Dana Rathkopf

Michael Morris

Susan Slovin

Brigit McLaughlin

Kristen Curtis

David M. Hyman

Jeremy C. Durack

Stephen B. Solomon

Maria E. Arcila

Ahmet Zehir

Aijazuddin Syed

Jianjiong Gao

Debyani Chakravarty

Hebert Alberto Vargas

Mark E. Robson

Joseph Vijai

Kenneth Offit

Mark T. A. Donoghue

Adam A. Abeshouse

Ritika Kundra

Zachary J. Heins

Alexander V. Penson

Christopher Harris

Barry S. Taylor

Marc Ladanyi

Diana Mandelker

Liying Zhang

Victor E. Reuter

Philip W. Kantoff

David B. Solit

Michael F. Berger

Charles L. Sawyers

Nikolaus Schultz

Howard I. Scher

Doi

10.1200/PO.17.00029

PMID: 28825054 · DOI: 10.1200/PO.17.00029 · Journal: JCO Precision Oncology (2017)

TL;DR

Abida et al. prospectively profiled 504 tumors from 451 patients with prostate cancer spanning the full clinical spectrum (locoregional, metastatic noncastrate, and metastatic castration-resistant prostate cancer [mCRPC]) using the MSK-IMPACT hybridization-capture targeted DNA panel, with paired germline analysis in 221 patients (PMID:28825054). They identified potentially actionable alterations in 36% of patients (per OncoKB annotation), including a 27% combined germline-or-somatic alteration rate in DNA damage repair genes BRCA2, BRCA1, ATM, and CHEK2. Comparative analyses across disease states showed that APC alterations are enriched in metastatic disease, ATM alterations in mCRPC, and that TP53 and somatic BRCA2 alterations arise as early clonal events in patients who later progress to metastasis (PMID:28825054). The authors argue for routine paired germline + somatic profiling as standard of care in advanced PRAD.

Cohort & data

  • Cohort: 451 patients with prostate adenocarcinoma, 504 tumors profiled (44 patients had >1 tumor sequenced); enrolled at Memorial Sloan Kettering (PMID:28825054).
  • Disease-state composition (last known): 348 metastatic (77%), 53 biochemically recurrent (12%), 50 locoregional (11%) (PMID:28825054).
  • Disease-state at tissue acquisition (504 tumors): 140 locoregional, 51 metastatic noncastrate prostate, 54 metastatic noncastrate metastasis sites, 95 mCRPC prostate, 164 mCRPC metastasis sites (PMID:28825054).
  • Metastatic biopsy sites: lymph node 45%, bone 22%, liver 14%, lung 5%, other soft tissue 14% (PMID:28825054).
  • Sequencing success rate: 504/746 (68%); bone and lung samples were the most challenging (42–52% success) (PMID:28825054).
  • Assay: MSK-IMPACT hybridization-capture targeted DNA panel covering >300 cancer-related genes for SNVs, indels, somatic copy number alterations, and structural rearrangements; matched normal blood for germline filtering (PMID:28825054).
  • Germline analysis: 221 patients consented to germline analysis of 76 known cancer-predisposing genes (the first 124 also reported in Pritchard et al.) (PMID:28825054).
  • Other tools: clonality estimated via cancer cell fraction with the FACETS algorithm; actionability called using OncoKB (PMID:28825054).
  • Dataset: released as prad_mskcc_2017 on cBioPortal (PMID:28825054).
  • Comparison cohorts: the SU2C-PCF mCRPC dataset (prad_su2c_2015, PMID:26000489) and the TCGA primary prostate cancer cohort (prad_tcga_pub, PMID:26544944).

Key findings

  • Mutation burden rises with castration resistance: mean nonsynonymous mutations/megabase increased from 1.74 (locoregional) to 4.02 (mCRPC), P < .001; metastatic noncastrate tumors were similar to locoregional (2.08 mut/Mb) (PMID:28825054).
  • PI3K/AKT pathway alterations in 24% of patients, including PTEN, PIK3CA, PIK3CB, PIK3R1, AKT1, AKT3; most PIK3CA, AKT1, AKT3 point mutations were known activating hotspots (PMID:28825054).
  • MAPK pathway alterations in 5%, including hotspot mutations in BRAF, HRAS, KRAS, and MAP2K1 (PMID:28825054).
  • Wnt/β-catenin pathway alterations in 15%, including APC, CTNNB1, and RNF43 (PMID:28825054).
  • Somatic homologous recombination (HR) gene alterations in 22% of the 451 patients — BRCA2 7%, ATM 5%, CDK12 7%, FANCA 3%, PALB2 2%, BRCA1 1%, RAD50 1% (PMID:28825054).
  • Germline pathogenic alterations in 19% of 221 patients consented: BRCA2 9%, CHEK2 4%, ATM 2%, BRCA1 1%, FH 1%, plus PMS2, NBN, PALB2, BRIP1 (<1% each) (PMID:28825054).
  • Combined germline-or-somatic alterations in BRCA2, BRCA1, ATM, or CHEK2 in 27% of the 221 germline-tested patients; germline-only analysis would have captured ~half of these patients (PMID:28825054).
  • Mismatch repair (MMR) gene alterations in 3% of patients (MSH2, MLH1, PMS2, MSH6); MMR-deficient tumors carried the highest mutation counts and were enriched for MMR / microsatellite instability mutational signatures (PMID:28825054).
  • mCRPC-enriched alterations vs. locoregional: AR (2% → 52%), TP53 (27% → 48%), PTEN (12% → 29%), RB1 (2% → 18%), ATM (2% → 11%), APC (4% → 15%), FANCA (1% → 7%), CDK12 (4% → 11%) (PMID:28825054).
  • mCRPC-enriched vs. metastatic noncastrate: AR, TP53, RB1, PTEN, ATM all significantly enriched, implicating them in development of castration resistance — particularly ATM, a DDR gene (PMID:28825054).
  • Metastatic-noncastrate-enriched vs. locoregional: only APC and (less strongly) ARID5B, implicating APC in metastasis (PMID:28825054).
  • SPOP alterations were enriched in earlier disease states (locoregional 12%, metastatic noncastrate 11%, mCRPC 5%), suggesting better outcomes / increased sensitivity to androgen-deprivation therapy for SPOP-mutant tumors (pending functional validation) (PMID:28825054).
  • Truncal/clonal events in matched samples: TP53 and somatic BRCA2 alterations were present early and clonal (cancer cell fraction > 0.9) in primary tumors of patients who later developed metastasis; conversely, AR alterations and a recurrent activating PIK3CA E545K mutation were acquired late (PMID:28825054).
  • Phylogenetic analyses identified truncal alterations in epigenetic regulators EP300 and KDM5A in metastatic tumors of patient P-0000377, with subclonal KMT2D alteration in metachronous bone metastases (PMID:28825054).
  • OncoKB-annotated actionability: 36% of patients carried at least one potentially actionable alteration; this excludes non-BRCA/ATM germline events, VUS missense, and genes of less clear significance such as CDK12 and FANCA, so the true rate may be higher (PMID:28825054).

Genes & alterations

  • AR — amplification and resistance mutations (e.g., F877L conferring resistance to enzalutamide/apalutamide, and H875Y conferring resistance to flutamide) enriched in mCRPC; 4% of metastatic noncastrate tumors already carried AR alterations, consistent with subclinical transition to castration resistance (PMID:28825054).
  • TP53 — present at 27% locoregional / 30% metastatic noncastrate / 48% mCRPC; alterations are early clonal events in matched primaries of patients who later progress to metastasis, consistent with prior reports of aggressive behavior of TP53-altered prostate cancer (PMID:28825054).
  • PTEN, RB1 — frequencies rise sharply with castration resistance (PTEN 12→29%, RB1 2→18%) (PMID:28825054).
  • BRCA2 — 9% germline pathogenic in 221 tested; somatic alterations clonal/truncal in matched samples; combined germline+somatic frequency drives PARP-inhibitor / platinum sensitivity rationale (see olaparib) (PMID:28825054).
  • BRCA1, ATM, CHEK2 — additional germline DDR alterations contributing to the 27% combined HR-deficiency rate; ATM also shows somatic enrichment specifically in mCRPC (PMID:28825054).
  • CDK12, FANCA, PALB2, RAD50 — additional somatic HR pathway genes; CDK12 loss is hypothesized to confer PARP inhibitor sensitivity (per cited preclinical work) (PMID:28825054).
  • APC — enriched in both metastatic disease states relative to locoregional, implicating it as a driver of metastasis (PMID:28825054).
  • SPOP — enriched in earlier disease states (12% locoregional → 5% mCRPC); paired with FOXA1, defines a possibly androgen-deprivation-sensitive subset (functional validation pending) (PMID:28825054).
  • PIK3CA, AKT1, AKT3, PIK3R1 — predominantly known activating hotspots; one patient acquired a PIK3CA E545K hotspot ~3 years after prostatectomy, illustrating late-emergent actionable events (PMID:28825054).
  • BRAF, HRAS, KRAS, MAP2K1 — hotspot MAPK-pathway mutations in ~5% of patients (PMID:28825054).
  • CTNNB1, RNF43 — Wnt-pathway co-alterations alongside APC (PMID:28825054).
  • MSH2, MLH1, MSH6, PMS2 — MMR genes altered in 3%; produce hypermutator phenotype with MMR/MSI signatures, raising potential for immune-checkpoint blockade response (PMID:28825054).
  • BRIP1, NBN, FH — additional germline cancer-predisposition findings (each <1–1%) (PMID:28825054).
  • EP300, KDM5A, KMT2D — epigenetic regulators identified as truncal/subclonal events in matched-tumor phylogenetic analysis (PMID:28825054).
  • FOXA1, ARID5B — additional context genes in disease-state enrichment analyses (PMID:28825054).

Clinical implications

  • Routine paired germline + somatic profiling for advanced prostate cancer: germline-only analysis missed about half of patients with actionable BRCA2/BRCA1/ATM/CHEK2 alterations, supporting paired testing as standard of care irrespective of family history (PMID:28825054).
  • PARP inhibitor candidacy: the 22% somatic HR-gene alteration rate (and 27% combined germline+somatic for the four canonical DDR genes) identifies a substantial population of candidates for PARP inhibitors such as olaparib and platinum-based therapy, citing prior clinical evidence (TOPARP-A and platinum case series) (PMID:28825054).
  • Immune checkpoint blockade in MMR-deficient prostate cancer: the 3% MMR-altered subset with hypermutator phenotype may be candidates for PD-1 blockade by analogy to MSI-high colorectal and other malignancies (PMID:28825054).
  • AR-resistance biomarkers: detection of an AR F877L mutation in a patient progressing after 4 years on apalutamide (ARN-509), and an H875Y mutation associated with flutamide resistance, illustrate clinical-grade detection of acquired antiandrogen-resistance mechanisms (PMID:28825054).
  • TP53 as an early biomarker of progression risk: clonal TP53 alterations in localized tumors may predict eventual progression to metastasis, potentially informing more aggressive upfront therapy (PMID:28825054).
  • Family-counseling implications: the high germline pathogenic-variant rate (19% of consented patients) supports referral for cascade testing of relatives (PMID:28825054).
  • Trial enrollment: authors frame the assay as a feasible mechanism for routing patients into molecularly-guided trials within standard clinical workflow (PMID:28825054).

Limitations & open questions

  • Selection bias toward aggressive disease: locoregional tumors in MSK-IMPACT were enriched for cases that subsequently recurred or metastasized, biasing comparisons against the TCGA primary cohort (e.g., higher TP53 and FOXA1 frequencies than TCGA) (PMID:28825054).
  • Targeted panel scope: MSK-IMPACT covers >300 cancer-related genes but cannot capture exome-wide signatures, non-coding drivers, or comprehensive structural rearrangements outside the bait set (PMID:28825054).
  • Bone-biopsy success rate: only 42–52% of bone and lung biopsies yielded sequenceable DNA, leaving the dominant osseous-metastasis pattern of prostate cancer underrepresented; circulating tumor DNA assays are flagged as a possible solution (PMID:28825054).
  • Germline-cohort consent: only 221/451 patients consented to germline analysis, limiting power for rare-variant frequency estimates (PMID:28825054).
  • Statistical power on metastases-only sub-analyses: when restricted to metastatic tumors or lymph nodes, several disease-state enrichment trends did not reach significance owing to smaller sample size (Appendix Fig A12) (PMID:28825054).
  • Functional validation pending: disease-state enrichment of APC (metastasis) and ATM (castration resistance), as well as putative ADT-sensitivity of SPOP-mutant tumors, are correlative and require laboratory validation (PMID:28825054).
  • OncoKB actionability undercounts: the 36% actionability rate excludes non-BRCA/ATM germline alterations, missense VUS, and genes such as CDK12 and FANCA whose clinical significance is still emerging (PMID:28825054).
  • Trial availability: at the time of publication, the authors note PARP-inhibitor trials in prostate cancer were “due to open shortly,” constraining immediate translation of HR-deficiency findings into therapy (PMID:28825054).

Citations from this paper used in the wiki

  • “We successfully sequenced 504 tumors from 451 patients with prostate cancer.”
  • “Twenty-seven percent of patients harbored a germline or a somatic alteration in a DNA damage repair gene that may predict for response to poly (ADP-ribose) polymerase inhibition.”
  • “Profiling of matched tumors from individual patients revealed that somatic TP53 and BRCA2 alterations arose early in tumors from patients who eventually developed metastatic disease.”
  • “APC alterations were enriched in metastatic tumors, whereas ATM alterations were specifically enriched in castration-resistant prostate cancer.”
  • “Overall, 36% of patients were found to have a potentially actionable alteration by using the OncoKB annotation platform.”
  • “The number of nonsynonymous mutations per tumor increased significantly from tumors in the locoregional disease state to those in the mCRPC disease state (1.74 v 4.02 mean mutations/megabase; P < .001).”
  • “An F877L enzalutamide/ARN509 resistance mutation … was found in the tumor of a patient who experienced progression after 4 years of treatment on ARN509.”
  • “Four locoregional tumors were found to have mutations in AR, including an H875Y mutation that is known to confer resistance to flutamide.”
  • “Released as study prad_mskcc_2017 on cBioPortal” (Reference 12: cbioportal.org/study?id=prad_mskcc_2017).

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