Genome-wide interrogation of structural variation reveals novel African-specific prostate cancer oncogenic drivers

Abstract

Background: African ancestry is a significant risk factor for advanced prostate cancer (PCa). Mortality rates in sub-Saharan Africa are 2.5-fold greater than global averages. However, the region has largely been excluded from the benefits of whole genome interrogation studies. Additionally, while structural variation (SV) is highly prevalent, PCa genomic studies are still biased towards small variant interrogation.

Methods: Using whole genome sequencing and best practice workflows, we performed a comprehensive analysis of SVs for 180 (predominantly Gleason score ≥ 8) prostate tumours derived from 115 African, 61 European and four ancestrally admixed patients. We investigated the landscape and relationship of somatic SVs in driving ethnic disparity (African versus European), with a focus on African men from southern Africa.

Results: Duplication events showed the greatest ethnic disparity, with a 1.6- (relative frequency) to 2.5-fold (count) increase in African-derived tumours. Furthermore, we found duplication events to be associated with CDK12 inactivation and MYC copy number gain, and deletion events associated with SPOP mutation. Overall, African-derived tumours were 2-fold more likely to present with a hyper-SV subtype. In addition to hyper-duplication and deletion subtypes, we describe a new hyper-translocation subtype. While we confirm a lower TMPRSS2-ERG fusion-positive rate in tumours from African cases (10% versus 33%), novel African-specific PCa ETS family member and TMPRSS2 fusion partners were identified, including LINC01525, FBXO7, GTF3C2, NTNG1 and YPEL5. Notably, we found 74 somatic SV hotspots impacting 18 new candidate driver genes, with CADM2, LSAMP, PTPRD, PDE4D and PACRG having therapeutic implications for African patients.

Conclusions: In this first African-inclusive SV study for high-risk PCa, we demonstrate the power of SV interrogation for the identification of novel subtypes, oncogenic drivers and therapeutic targets. Identifying a novel spectrum of SVs in tumours derived from African patients provides a mechanism that may contribute, at least in part, to the observed ethnic disparity in advanced PCa presentation in men of African ancestry.

Keywords: Advanced disease; African ancestry; Chromosomal instability; Ethnic disparity; Prostate cancer; Whole genome sequencing.

Fig. 1

The spectrum of somatic structural variations (SVs) based on type classification. Top two panels are the total count and relative frequency of each SV type. Samples in the horizontal axis are placed in the order of increasing relative frequency of deletions. The “Recurrent genes” heatmap presents the 24 most recurrently mutated genes, each with at least three samples in each hyper-SV group. Coding, SNVs/indels in coding region; Noncoding, SNVs/indels in noncoding region; HRPCa, high-risk PCa; IRPCa, intermediate-risk PCa; LRPCa, low-risk PCa

Fig. 2

Genome-wide structural variation (SV) frequency across all 180 samples. Dots show the number of SV breakpoints (A) and the number of samples with SVs (B) within 1 Mbp windows, with hotspot regions (>3 SD from mean, threshold shown as dashed horizontal line) highlighted as green. Hotspots where genes are interrupted by >50% breakpoints or recurrent in >50% genomes are labelled

Fig. 3

Structural variation (SV) types and ethnic groups in SV hotspot regions. The count of SV breakpoints per SV hotspot, coloured by SV types (A) and ethnic groups (B). The count of samples per SV hotspot coloured by ethnic groups (C). SV hotspots represented by 1 Mbp non-overlapping bin groups are ordered by the percentage of samples of African ancestry

Fig. 4

TMPRSS2 and ETS family gene fusions. A shows the count of SVs involved with colour representing SV types and B shows the count of the sample having the corresponding gene fusion, with colour representing the ethnic groups. Abbreviations: SV, structural variation; TRA, translocation; INV, inversion; INS, insertion; DUP, duplication; DEL, deletion

Fig. 5

Spectrum of gene fusion junctions. A–C panels show the three forms of structural variation (SV) breakpoint clusters based on different transcripts of TMPRSS2 and ERG, shown in the top right. The number of samples with breakpoint in different exon positions of TMPRSS2-ERG fusion junction is shown in brackets