Punctuated Evolution of Prostate Cancer Genomes

Authors

Sylvan C. Baca

Davide Prandi

Michael S. Lawrence

Juan Miguel Mosquera

Alessandro Romanel

Yotam Drier

Kyung Park

Naoki Kitabayashi

Theresa Y. MacDonald

Mahmoud Ghandi

Eliezer Van Allen

Gregory V. Kryukov

Andrea Sboner

Jean-Philippe Theurillat

T. David Soong

Elizabeth Nickerson

Daniel Auclair

Ashutosh Tewari

Himisha Beltran

Robert C. Onofrio

Gunther Boysen

Candace Guiducci

Christopher E. Barbieri

Kristian Cibulskis

Andrey Sivachenko

Scott L. Carter

Gordon Saksena

Douglas Voet

Alex H. Ramos

Wendy Winckler

Michelle Cipicchio

Kristin Ardlie

Philip W. Kantoff

Michael F. Berger

Stacey B. Gabriel

Todd R. Golub

Matthew Meyerson

Eric S. Lander

Olivier Elemento

Gad Getz

Francesca Demichelis

Mark A. Rubin

Levi A. Garraway

Doi

10.1016/j.cell.2013.03.021

PMID: 23622249 · DOI: 10.1016/j.cell.2013.03.021 · Journal: Cell (2013)

TL;DR

Baca and colleagues performed whole-genome sequencing (WGS) of 57 prostate tumors (55 treatment-naïve primary adenocarcinomas plus 2 neuroendocrine prostate cancer metastases) paired with matched normal DNA to characterize somatic structural alterations. They identified a class of complex, interdependent chromosomal rearrangements they term “chromoplexy” — chains of translocations and deletions that, unlike chromothripsis, frequently span 5+ chromosomes and arise alongside ETS gene fusions. Chromoplectic chains involving ≥5 rearrangements were detected in 50/57 tumors (88%), often co-disrupting multiple tumor-suppressor genes in a single coordinated event. Clonality analysis of deletions sketched a consensus progression path: early loss of NKX3-1 and TMPRSS2-ERG fusion → intermediate loss of CDKN1B/TP53 → late loss of PTEN. The data support a “punctuated” model of tumor evolution sitting between gradual mutation and catastrophic chromothripsis PMID:23622249.

Cohort & data

  • Tumors: 57 prostate cancers — 55 treatment-naïve primary PRAD adenocarcinomas spanning Gleason score 6–9 and pathological stages pT2N0–pT4N1, plus 2 neuroendocrine (PRNE) metastases that emerged after castration-based therapy.
  • Dataset: prad_broad_2013 (cBioPortal study ID), reference genome hg19. BAM, RNA-seq, and SNP-array data deposited at dbGaP phs000447.v1.p1.
  • Assays:
    • Whole-genome sequencing on tumor and normal (mean coverage 61× and 34× respectively) — Illumina GAIIx paired-end 101 bp reads, aligned with BWA.
    • Affymetrix SNP 6.0 arrays for somatic DNA copy number profiling.
    • RNA-seq on 20 tumors with matched benign prostate tissue for 16 cases.
    • Validation by Sanger resequencing/PCR and FISH for ETS fusions.
  • Algorithms introduced/used: ChainFinder (new graph-theory algorithm for chromoplexy detection), CLONET (new clonality estimator), MuTect, Indelocator, dRanger, BreakPointer, GISTIC v2, ABSOLUTE.
  • Extended cohort of 199 prostate adenocarcinomas (this study plus PMID:22610119) was used for SCNA recurrence/GISTIC analysis.

Key findings

  • Burden: 356,136 somatic base-pair mutations across the cohort; average of 33 non-silent exonic mutations per primary tumor. dRanger identified 5,596 high-confidence somatic rearrangements after filtering against a panel of 172–176 non-cancerous genomes; 113 rearrangements were validated by PCR/resequencing PMID:23622249.
  • Chromoplexy is prevalent: Chromoplectic chains of ≥5 rearrangements (≥10 breakpoints) were detected in 50/57 tumors (88%); 36/57 tumors (63%) carried ≥2 such chains. Overall, 39% of rearrangements participated in chains, versus 2.8% in simulated and 0.2% in scrambled control genomes (p < 10⁻⁴) PMID:23622249.
  • Independence rejected: For 50% of observed rearrangements, both breakpoints sat closer to other breakpoints than expected under an independent model (p < 10⁻⁴), arguing that chains arise via a coordinated process, not by sequential independent events. The authors also rule out a sequential-dependent model based on the existence of 121 “closed” chain cycles that would require unfeasibly elevated local rearrangement rates.
  • ETS-status stratifies chromoplexy mechanism:
    • ETS+ tumors (most commonly TMPRSS2-ERG): significantly more inter-chromosomal fusions (p < 10⁻⁴), larger maximum chromosomes per chain (p = 0.009), with breakpoints enriched in highly expressed prostate-tumor genomic regions — consistent with transcription-hub-mediated DNA injury (possibly androgen-receptor coupled).
    • ETS−, CHD1del tumors: predominantly intra-chromosomal rearrangements within chains (p = 2×10⁻⁴) and overall (p = 4×10⁻⁴); enriched in GC-poor, late-replicating, gene-poor heterochromatin, resembling a chromothripsis-like pattern. Across an extended 199-tumor cohort, CHD1 loss was associated with increased recurrent SCNAs (p = 1.5×10⁻⁸).
  • Oncogenic ERG fusion via chromoplexy: 15/26 (58%) ERG-positive cases acquired the ERG fusion in the context of a chromoplexy chain.
  • Chromoplexy is not unique to prostate cancer: ChainFinder detected chains of ≥5 rearrangements in every tumor type tested across an additional 59 genomes (melanoma, NSCLC, HNSC, breast adenocarcinoma).
  • Coordinated tumor-suppressor disruption: Of 17 prostate-related TSGs from KEGG, 26/57 tumors (46%) had ≥1 disrupted in a chain of ≥3 rearrangements; adding TMPRSS2-ERG and 10 putative prostate cancer genes raised this to 35/57. Genes recurrently hit by chromoplexy included PTEN (9 cases), NKX3-1 (8 cases), TP53 (4 cases), CDKN1B (3 cases), and RB1 (2 cases). Single chains were shown to coordinately disrupt TMPRSS2-ERG + SMAD4, or CDKN1B+ETV6+ETV3, or co-delete PIK3R1+PTEN and TP53+CHEK2.
  • Clonal progression path: Using CLONET-based clonality estimates from germline SNP allelic fractions:
    • Early/strictly clonal: NKX3-1 deletion, the 3 Mb 21q deletion producing TMPRSS2-ERG, FOXP1 deletion, and point mutations in SPOP/FOXA1.
    • Intermediate: CDKN1B and TP53 loss.
    • Late/often subclonal: PTEN deletion (significantly more subclonal than NKX3-1; p = 10⁻⁵).
    • Some chains contained strictly subclonal deletion bridges, indicating chromoplexy can recur during subclonal expansion.
  • Histologic grade tracks genomic derangement: In 199 tumors, those with predominant Gleason pattern 4 had more recurrent SCNAs than predominantly pattern 3 tumors (p = 0.0059), independent of overall SCNA load, purity, and mutation burden.

Genes & alterations

  • TMPRSS2–ERG — recurrent androgen-regulated fusion via 21q intronic deletion; arose within chromoplexy chains in 15/26 ERG+ cases; observed clonally early in progression PMID:23622249.
  • ETV1 — alternative ETS fusion partner detected by sequencing and validated by FISH PMID:23622249.
  • SPOP and FOXA1 — recurrent point mutations, clonal/early in the progression path (cross-confirmed against PMID:22610119).
  • NKX3-1 — strictly clonal deletion in 8 cases via chromoplexy; one of the earliest detectable lesions in prostate carcinogenesis PMID:23622249.
  • FOXP1 — early clonal deletion in the consensus path PMID:23622249.
  • CDKN1B — intermediate-clonality deletion; recurrently disrupted (3 cases via chromoplexy); co-deleted with ETV6/ETV3 in one 25-rearrangement chain PMID:23622249.
  • TP53 — intermediate-clonality deletion; 4 cases disrupted by chromoplexy; co-deleted with CHEK2 in one chain PMID:23622249.
  • PTEN — recurrently subclonal deletion (p = 10⁻⁵ vs NKX3-1); 9 cases disrupted by chromoplexy; can be hit by disruptive rearrangement; co-deleted with PIK3R1 in one chain PMID:23622249.
  • RB1 — disruptive rearrangement in 2 chromoplectic cases PMID:23622249.
  • CHD1 — focal deletion or disruptive rearrangement defines an ETS-negative subset with chromothripsis-like intra-chromosomal chains in late-replicating, GC-poor DNA; CHD1 loss correlates with elevated recurrent SCNAs (p = 1.5×10⁻⁸) in extended 199-tumor cohort PMID:23622249.
  • SMAD4 — disrupted by chromoplexy in one tumor (P05-3852) within the same chain that produced the TMPRSS2-ERG fusion across 6 chromosomes PMID:23622249.
  • MAGI2, GSK3B, FOXO1 — recurrent disruptive rearrangements affecting genes implicated in prostate cancer signaling PMID:23622249.
  • NRF1–BRAF — singleton sense-preserving fusion (tumor PR-4240) leaving the BRAF kinase domain intact, hypothesized to drive overexpression of an oncogenic kinase PMID:23622249.
  • CRKL–MAPK1 — singleton sense-preserving fusion (tumor P04-1084) involving the ERK-2 kinase, kinase domain intact PMID:23622249.
  • KDM6A, MED12 — referenced as known recurrently mutated genes from prior exome studies but not specifically called as novel hits here (PMID:22610119, PMID:22722839).

Clinical implications

  • Diagnostic/prognostic correlate: Recurrent SCNA burden — much of which is generated by chromoplexy — is significantly higher in high-grade (Gleason pattern 4 dominant) tumors than in lower-grade tumors (p = 0.0059), suggesting chromoplexy-driven structural damage may underlie clinical aggressiveness PMID:23622249.
  • Driver-gene prioritization: The authors argue that genes recurrently disrupted by chromoplexy are more likely to be true drivers in a given tumor, because survivable chromoplexy events should be enriched for compensating oncogenic lesions — a heuristic with potential implications for clinical WGS interpretation PMID:23622249.
  • Therapeutic hypothesis (not tested here): The link between ETS+ chromoplexy and transcription/androgen-receptor activity is consistent with the prior observation that ERG-overexpressing prostate cancer cells accumulate DNA damage and are sensitive to PARP inhibition (Brenner et al. 2011, cited but not in this corpus); this paper does not perform any PARP-inhibitor experiments.
  • No drug treatments were administered or evaluated as part of this study.

Limitations & open questions

  • Mechanism is inferred, not demonstrated. Chromoplexy is defined by a statistical/computational model (ChainFinder); the authors explicitly note that the mechanistic basis must be addressed experimentally (e.g., via FISH or 3C before/after androgen exposure in prostate epithelial cells).
  • A sequential-dependent (non-coordinated) generation of chains cannot be formally excluded, only made implausibly costly by the closed-chain enumeration argument.
  • Cohort skewed to primary, treatment-naïve adenocarcinomas; only 2 NEPC metastases were included, so chromoplexy dynamics in advanced/castration-resistant disease remain undercharacterized.
  • Clonality inference rests on tumor purity and germline SNP coverage and may misclassify lesions in low-purity samples (the WGS-vs-ABSOLUTE concordance was R² = 0.99 but flagged two discrepant samples).
  • Pan-cancer claim is preliminary — 59 additional genomes across melanoma/NSCLC/HNSC/breast indicated chromoplexy is not prostate-specific, but generalization will require larger per-tissue cohorts.
  • Functional consequences of singleton fusions (NRF1-BRAF, CRKL-MAPK1) and recurrent rearrangements of MAGI2, GSK3B, FOXO1 were not experimentally validated.
  • Open question: does targeting the transcription-coupled DNA-damage processes implicated in ETS+ chromoplexy (e.g., AR-driven processes, PARP) have therapeutic value?

Citations from this paper used in the wiki

  • “We sequenced the genomes of 55 primary prostate adenocarcinomas and two neuroendocrine prostate cancer (NEPC) metastases that developed following castration-based therapy, along with paired normal tissue.” (Results, p.3)
  • “Chromoplectic chains of five or more rearrangements (ten or more breakpoints) were detected in 50 out of 57 tumors (88%; Table S5B and Figure S3C), while 36 tumors (63%) contained two or more such chains.” (Results, p.4–5)
  • “39% of rearrangements participated in chains, while ChainFinder detected chains in only 2.8% and 0.2% of rearrangements from simulated or scrambled genomes, respectively” (Results, p.5)
  • “Chromoplexy in tumors harboring oncogenic ETS fusions (ETS+) produced significantly more inter-chromosomal rearrangements than ETS− tumors (p < 10−4) and involved a greater maximum number of chromosomes in a single event (p = 0.009)” (Results, p.5)
  • “An extended cohort of 199 prostate adenocarcinomas revealed that CHD1 loss was associated with an increased number of recurrent SCNAs (p = 1.5×10−8)” (Results, p.5)
  • “Several cancer genes were recurrently deleted or rearranged by chromoplexy, including PTEN (9 cases), NKX3-1 (8 cases), CDKN1B (3 cases), TP53 (4 cases), and RB1 (2 cases)” (Results, p.5–6)
  • “deletions of PTEN were often subclonal (p = 10−5 for comparison with NKX3-1 deletion clonality), as were CDKN1B deletions” (Results, p.6)
  • “Tumors with predominantly Gleason score (GS) 4 histology were significantly enriched for recurrent SCNAs compared to GS 3 tumors (p = 0.0059)” (Results, p.7)
  • “A consensus path of tumor evolution begins with events such as loss of NKX3-1 or fusion of TMPRSS2 and ERG. The path proceeds with the loss of CDKN1B, TP53, PTEN, and other progression-associated lesions.” (Discussion, p.8)

This page was processed by crosslinker on 2026-05-09.