Copy number alteration burden predicts prostate cancer relapse

Authors

Haley Hieronymus

Nikolaus Schultz

Anuradha Gopal

Brett S. Carver

Matthew T. Chang

Yonghong Xiao

Adriana Heguy

Kety Huberman

Melanie Bernstein

Melissa Assel

Rajmohan Murali

Andrew Vickers

Peter T. Scardino

Chris Sander

Victor Reuter

Barry S. Taylor

Charles L. Sawyers

Doi

10.1073/pnas.1411446111

PMID: 25024180 · DOI: 10.1073/pnas.1411446111 · Journal: Proc Natl Acad Sci USA (2014)

TL;DR

Hieronymus and colleagues at Memorial Sloan–Kettering Cancer Center test whether the fraction of the autosomal genome bearing copy number alterations — termed CNA burden — predicts prostate cancer relapse. They profile 104 new prostatectomy cases (the “contemporary cohort”, prad_mskcc_2014) by array-CGH and update long-term outcome for their earlier 168-case cohort (prad_mskcc, Taylor et al. 2010). CNA burden is significantly associated with biochemical recurrence (BCR) in both cohorts and with metastasis in the initial cohort, independent of pretreatment PSA, Gleason score, and the Stephenson postoperative nomogram. The association holds in the intermediate-risk Gleason 7 subpopulation, adds prognostic information beyond an RNA cell-cycle progression (CCP) signature, and can be measured in FFPE diagnostic needle biopsies using low-pass whole-genome sequencing.

Cohort & data

  • Cancer type: Primary prostate cancer (PRAD).
  • Initial cohort (prad_mskcc, Taylor et al. 2010): 181 prostatectomy cases (168 with high-resolution CNA), median follow-up 8 y; BCR n=46 (43 per 1,000 person-years), metastasis n=19 (15 per 1,000 person-years).
  • Contemporary cohort (prad_mskcc_2014): 104 prostatectomy cases + matched normals from MSKCC, median follow-up 6 y; BCR n=24 (53 per 1,000 person-years), metastasis n=3 (5 per 1,000 person-years).
  • Assay (contemporary cohort): Agilent 1M-feature array-CGH (array-cgh-agilent-1m) on snap-frozen samples with >70% tumor content; Promega pooled-reference DNA.
  • Biopsy feasibility sub-study: Four FFPE needle biopsies and adjacent benign controls profiled by Agilent 244K array-CGH (agilent-244k) and by low-pass paired-end 100-bp whole-genome sequencing (whole-genome-seq) at 1–3× coverage from 100–250 ng input DNA (KAPA LTP / Illumina HiSeq 2000); reads aligned with bwa to hg19.
  • Data deposition: GEO accession GSE54691; also distributed via the MSKCC Prostate Cancer Genomics Data Portal.

Key findings

  • CNA burden in primary prostate cancer ranges from <0.1% to >50% of the autosomal genome, with mean 5% (initial) and 4% (contemporary), substantially lower than the ~32% mean burden seen in metastatic prostate cancer and lower than bladder and renal cell carcinomas.
  • As a continuous variable, CNA burden is associated with BCR in both cohorts: HR 1.09 (95% CI 1.06–1.13, P<0.001) for the initial cohort and HR 1.05 (95% CI 1.01–1.09, P=0.008) for the contemporary cohort per 1% change in CNA burden (PMID:25024180, Table 1).
  • Stratified at the intermediate-cluster threshold (5.41% CNA burden), HRs for BCR are 1.99 (P=0.021) and 3.85 (P=0.001) in the initial and contemporary cohorts respectively; stratified at the low-cluster threshold (1.34%), HRs are 2.03 (P=0.048) and 2.42 (P=0.037).
  • Patients with CNA burden in the top vs. bottom quartile of their cohort had HRs for BCR of 2.65 (initial, P=0.001) and 4.43 (contemporary, P<0.001).
  • CNA burden is associated with metastasis in the initial cohort as a continuous variable (HR 1.10, 95% CI 1.04–1.15, P=0.0004); contemporary cohort had too few metastatic events (n=3) for analysis.
  • In multivariate Cox models, CNA burden remained associated with BCR after adjustment for pretreatment PSA (HR 1.09 and 1.04 per 1% CNA, P<0.001 and P=0.025), biopsy Gleason score (HR 1.08 and 1.05, P<0.0001 and P=0.026), and pathology Gleason score (HR 1.04 and 1.05, P=0.043 and P=0.026). After adjustment for the postoperative Stephenson 5-y nomogram, CNA burden remained associated with BCR in the contemporary cohort (HR 1.05, P=0.011) but not the initial cohort; concordance index of the nomogram (0.812 / 0.817) was not improved.
  • In Gleason 7 intermediate-risk patients, CNA burden ranges 0.05–25% (initial) and 0.003–50% (contemporary); univariate HRs for BCR are 1.08 (P=0.011) and 1.06 (P=0.017). The association persists after adjusting for pretreatment PSA (P=0.006 and 0.019) and for the Stephenson nomogram in the contemporary cohort (P=0.014).
  • Of the principal focal CNAs in prostate cancer, PTEN loss was positively associated with BCR in the initial cohort (HR 3.04, 95% CI 1.55–6.00, P=0.001) whereas TP53 loss and MYC gain were not. CNA burden remained significant after adjusting for PTEN loss (HR 1.09, 95% CI 1.05–1.13, P=0.001).
  • CNA burden contributes independent prognostic information beyond a non-proprietary version of the RNA cell-cycle progression (CCP) signature (CCP HR 1.124, P=0.05; CNA burden HR 1.097, P<0.001 in n=113 cases with both data types) and outperforms CCP in Gleason 7 cancers (Table S3).
  • Low-pass whole-genome sequencing (1–3× coverage) of FFPE needle biopsies from four patients yields CNA profiles concordant with high-resolution aCGH, including focal 3p13 deletions spanning FOXP1, RYBP, and SHQ1, and arm-level gains/losses on 7, 8p, 8q, and 13q.

Genes & alterations

  • PTEN — focal copy number loss; independently associated with BCR (HR 3.04, P=0.001) but CNA burden remains prognostic after adjusting for it.
  • MYC — focal copy number gain; not significantly associated with BCR in this analysis.
  • TP53 — focal copy number loss (16p); not significantly associated with BCR in this analysis.
  • RB1 — focal copy number loss on 13q; noted as a recurrently altered focal CNA in the cohorts.
  • FOXP1, RYBP, SHQ1 — co-deleted in a 3p13 focal deletion identified by both aCGH and low-pass WGS in the FFPE biopsy proof-of-principle.
  • Recurrent broad CNAs: genomic losses on 6p, 8p, 13q, 16p; gains on 7 and 8q.

Clinical implications

  • CNA burden, a single global measure of genomic instability, is prognostic for biochemical recurrence and metastasis after radical prostatectomy and adds information beyond established clinical variables (PSA, Gleason score, Stephenson nomogram) and an RNA-based cell-cycle progression signature.
  • The biomarker is informative in the intermediate-risk Gleason 7 subpopulation, where existing clinical variables struggle to distinguish indolent from aggressive disease.
  • Because CNA burden can be measured from 100–250 ng of FFPE needle-biopsy DNA by low-pass WGS, it is technically feasible to deploy in the pretreatment / diagnostic setting and could in principle inform active-surveillance versus immediate-treatment decisions.

Limitations & open questions

  • Metastasis analysis is restricted to the initial cohort; the contemporary cohort had only 3 metastatic events, so metastasis-endpoint findings are not independently validated.
  • CNA burden did not improve the Harrell’s concordance index of the postoperative Stephenson nomogram (0.812 / 0.817) in either cohort, raising the question of incremental clinical utility on top of fully-postoperative pathological variables.
  • The proportional hazards assumption is acknowledged as not strictly met for long-term BCR/metastasis follow-up.
  • The FFPE-biopsy WGS demonstration is a proof-of-principle on only four cases; large biopsy cohorts (especially active-surveillance cohorts) are needed to test the prognostic value of CNA burden in untreated populations.
  • The mechanism linking elevated CNA burden to relapse is not addressed; the authors suggest exploring structural rearrangement breakpoints and underlying genomic-instability mechanisms.
  • The CNA-burden thresholds (5.41% and 1.34%) are derived from the initial cohort’s hierarchical clustering and have not been prospectively validated.

Citations from this paper used in the wiki

  • “We find that CNA burden across the genome, defined as the percentage of the tumor genome affected by CNA, was associated with biochemical recurrence and metastasis after surgery in these two cohorts, independent of the prostate-specific antigen biomarker or Gleason grade.” (Abstract)
  • “In primary prostate cancer genomes, CNA burden ranged from less than 0.1% to greater than 50%, with an average of 5% and 4% for the initial and contemporary cohorts, respectively. This finding differs significantly from the higher CNA burden of metastatic prostate cancers (∼32%), as well as of other urogenital neoplasms…” (Results)
  • “CNA burden, when examined as a continuous variable, was significantly associated with BCR in both the initial and contemporary cohorts [hazard ratios (HR) = 1.09 and 1.05, respectively, for each percent change in CNA burden; P values ≤ 0.0001 and 0.008].” (Results, Table 1)
  • “CNA burden was significantly associated with metastasis in the initial cohort as a continuous variable in a univariate Cox proportional hazards regression model (HR = 1.10; P value = 0.0004).” (Results, Table 1)
  • “PTEN copy number loss was positively associated with disease relapse in our cohort, whereas TP53 loss and MYC gain were not. CNA burden remained significantly associated with BCR [P value = 0.001, HR = 1.09, 95% confidence interval (CI) 1.05–1.13]… when adjusted for PTEN copy number loss.” (Results)
  • “Both the CCP signature and CNA burden were independently associated with BCR, with HRs of 1.124 and 1.097, respectively (P value = 0.05 and <0.001, initial cohort cases with both expression and CNA data, n = 113).” (Results)
  • “The data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database… accession no. GSE54691.” (Data deposition footnote)

This page was processed by crosslinker on 2026-05-11.