Polygenic Risk Score Modifies Prostate Cancer Risk of Pathogenic Variants in Men of African Ancestry

Abstract Prostate cancer risk is influenced by rare and common germline variants. We examined the aggregate association of rare germline pathogenic/likely pathogenic/deleterious (P/LP/D) variants in ATM, BRCA2, PALB2, and NBN with a polygenic risk score (PRS) on prostate cancer risk among 1,796 prostate cancer cases (222 metastatic) and 1,424 controls of African ancestry. Relative to P/LP/D non-carriers at average genetic risk (33%–66% of PRS), men with low (0%–33%) and high (66%–100%) PRS had Odds Ratios (ORs) for overall prostate cancer of 2.08 [95% confidence interval (CI) = 0.58–7.49] and 18.06 (95% CI = 4.24–76.84) among P/LP/D carriers and 0.57 (95% CI = 0.46–0.71) and 3.02 (95% CI = 2.53–3.60) among non-carriers, respectively. The OR for metastatic prostate cancer was 2.73 (95% CI = 0.24–30.54) and 28.99 (95% CI = 4.39–191.43) among P/LP/D carriers and 0.54 (95% CI = 0.31–0.95) and 3.22 (95% CI = 2.20–4.73) among non-carriers, for men with low and high PRS, respectively. Lifetime absolute risks of overall prostate cancer increased with PRS (low to high) from 9.8% to 51.5% in P/LP/D carriers and 5.5% to 23.9% in non-carriers. Lifetime absolute risks of metastatic prostate cancer increased with PRS from 1.9% to 18.1% in P/LP/D carriers and 0.3% to 2.2% in non-carriers These findings suggest that assessment of prostate cancer risk for rare variant carriers should include PRS status. Significance: These findings highlight the importance of considering rare and common variants to comprehensively assess prostate cancer risk in men of African ancestry.


Introduction
Pathogenic variants in DNA repair genes are associated with risk of advanced prostate cancer and determine eligibility for targeted therapies (1-5).We recently reported in European ancestry men that prostate cancer risk associated with carrying pathogenic variants in established prostate cancer risk genes is more accurately determined when incorporating information regarding a man's genetic risk profile based on common risk variants, as measured by a polygenic risk score (PRS; ref. 6).Pathogenic variants and a prostate cancer PRS have been 1 Center for Genetic Epidemiology, Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, California. 2 Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, California. 3 Department of Quantitative Health Sciences, Mayo Clinic, Rochester, Minnesota. 4Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, Minnesota. 5Department of Oncology, Mayo Clinic, Rochester, Minnesota. 6Uro Care, Kampala, Uganda. 7Mulago Hospital, Kampala, Uganda. 8Makerere University College of Health Sciences, Kampala, Uganda. 9SurgPath, Kampala, Uganda. 10Kagando Hospital, Kasese, Uganda. 11Public Health Sciences, Fred Hutchinson Cancer Center, Seattle, Washington.
separately shown to impact prostate cancer risk in African ancestry men (5, 7-10); however, the combined association of exceedingly rare pathogenic variants and common variants found through genome-wide association study has not been explored in men of African ancestry.Understanding the degree to which PRS modifies risk for African ancestry rare variant carriers may inform risk assessment and have clinical value.We investigated the combined association of rare variants in BRCA, ATM, NBN, and PALB (5) and common variants, evaluated with a multi-ancestry prostate cancer PRS (7), on prostate cancer risk in a study of 3,220 men of African ancestry.

Targeted Gene Sequencing and Genome-wide Genotyping
Gene sequencing and quality control procedures have been described previously (5).Briefly, next-generation sequencing was performed using an Illumina HiSeq 4000 with 150 bp paired end reads.Genomic DNA libraries were prepared using a custom QIAseq amplicon-based targeted panel of all coding regions and essential splice sites.Reads were aligned to the human reference genome (GRCh37) using BWA-MEM v0.7.10.The Genome Atlas Toolkit (GATK v3.4-46) was utilized for sequence realignment and recalibration, joint sample calling, and evaluation of depth of coverage.

AACRJournals.org
Cancer Res Commun; 3(12) December 2023 Genome-wide genotyping was performed using the Illumina Human 1M and Illumina OncoArray in the MEC, LAC, and LAAPC studies and the Illumina OncoArray in UGPCS.All studies were imputed to phase III of the 1000 Genomes Project reference panel (18).Genotyping and postimputation quality control details are summarized elsewhere (7).

PRS
The prostate cancer PRS was calculated on the basis of a previously developed multi-ancestry PRS of 269 risk variants (7).Of the 269 variants, 254 were polymorphic in the current dataset (minor allele frequency >1% in controls).The PRS was calculated as a weighted sum of the number of risk alleles carried by each participant, using the previously reported variant-specific multi-ancestry weights (7), and analyzed as high-(66.7%-100%),intermediate-(33.3%-66.7%),and low-(0%-33.3%)risk groups (PRS tertiles) based on the PRS distribution among controls.

Statistical Analysis
The

Absolute Risk Estimation
Absolute risks of prostate cancer were estimated (7) for a given age for each PRS category, carrier status, and the combined PRS category and carrier status variable using the ORs from each respective analysis (as described above), combined with mortality and incidence rates for Black men in the United States (23, 24), while accounting for competing causes of death.Because our study population was enriched for aggressive cases, for overall prostate cancer absolute risk estimates, we calculated ORs that weighted aggressive and nonaggressive prostate cancer ORs based on the expected proportions of aggressive and nonaggressive prostate cancer cases in a general Black population (37% and 63%, respectively, based on data from the African Ancestry Prostate Cancer Consortium; Supplementary Tables S2 and S3).For metastatic prostate cancer absolute risk, we used metastatic prostate cancer ORs (Supplementary Tables S4-S6).Overall and metastatic prostate cancer incidence rates were obtained from the SEER Program (23, 24)  Results are also provided in Supplementary Table S6.
PRS (Fig. 1B; Supplementary Table S6).No multiplicative interaction between PRS and P/LP/D carrier status was identified on the basis of the likelihood ratio test and Wald statistic, which suggests an additive effect between the PRS and carrier status (Supplementary Table S9).In analyses stratified by country, results were similar in African American and Ugandan men, although CIs were wide given the reduced sample sizes (Supplementary Tables S10-S15).When repeating analyses for pathogenic variants in four additional genes (BRCA, RAD, MLH, and MSH) with weaker evidence of association with prostate cancer risk (5), the risk modifying effect of PRS was similar, albeit the associations were diminished (Supplementary Tables S16 and S17).

Discussion
These findings highlight the importance of considering rare and common variants to comprehensively assess prostate cancer risk in men of African ancestry.We found that the association of P/LP/D variants in BRCA, ATM, PALB, and NBN with prostate cancer risk in men of African ancestry varied by a man's PRS profile, with enhanced risk stratification for metastatic and aggressive prostate cancer, suggesting an additive effect of the PRS and carrier status.These findings are highly consistent with what has been reported in studies among men of European ancestry (6) that have examined P/LP/D variants in several highrisk genes (e.g., HOXB, BRCA, ATM, and CHEK).In the current study, the set of genes (ATM, BRCA, PALB, and NBN) were those previously shown to be strongly associated with prostate cancer risk in this African ancestry sample.A similar risk-modifying effect of PRS on carrier status was also observed for genes that were found to have weaker independent associations with overall prostate cancer and aggressive disease (BRCA, RAD, MLH, and MSH).
While previous larger studies in men of European ancestry permitted finer stratification of PRS (deciles), the smaller sample size of the current study permitted testing only broad PRS strata (tertiles) when examining aggregate effects by carrier status.Despite these differences, the trends between studies when examining this question were very similar.The higher PRS observed in noncarriers than carriers of P/LP/D variants also suggests independent and separate processes involved in prostate cancer development.
We estimate that absolute risk of prostate cancer for men who carry P/LP/D variants is heavily influenced by PRS status.At age 60, absolute risk ranges from 2% for men with a low PRS to 16% for men with a high PRS for men who carry P/LP/D variants; by age 70 these estimates range from 6% to 40%, respectively (Supplementary Table S18

FIGURE 1
FIGURE 1 ORs for overall prostate cancer (A) and metastatic prostate cancer (B) in African ancestry men by PRS and carrier status of rare P/LP/D variants aggregated across BRCA2, ATM, NBN, and PALB2.The x-axis indicates the PRS tertile.The y-axis represents OR values for the indicated PRS tertile compared with non-carriers within the intermediate PRS category as the reference group and adjusted by age, study, and the first 10 principal components.Dots represent ORs and error bar lines represent 95% CIs.Black represents results in non-carriers, and green represents results in carriers.Results are also provided in Supplementary TableS6.

TABLE 1
Descriptive characteristics of prostate cancer cases and controls , P/LP/D variants were defined as variants predicted to result in protein truncation or to significantly alter the protein sequence (frameshift insertion/deletion, gain of stop codon, or loss of essential splice site donor/acceptor) and missense variants (nonsynonymous codon change and exon start/end codon change) that were reported as pathogenic or likely pathogenic in ClinVar (19) by one or more clinical laboratories (Ambry, SCRP, InVitae, GeneDX, Emory, and In-SiGHT).Missense variants not identified as pathogenic or likely pathogenic in ClinVar were annotated with dbNSFP version 3.3a (20) using five in silico algorithms (Polyphen2-HumDiv, PolyPhen2-HumVar, LRT, Mutation Taster, and SIFT) to predict potential functional effects(21)and determine P/LP/D status.Variants in introns, 5 -UTR (untranslated region), 3 -UTR, in-frame insertions/deletions, synonymous variants, and those located 2 bp outside of the consensus splice site according to CAVA(22)were excluded.BRCA, ATM, NBN, and PALB P/LP/D variants were found in 1% of nonaggressive cases, 3.3% of aggressive cases, 4.2% of metastatic cases, and 0.6% of controls.
Men were classified on the basis of the combination of carrier status and PRS in the following six categories: low PRS non-carriers, low PRS carriers, in- combined effect of the PRS and carrier status on prostate cancer risk was evaluated by aggregating across P/LP/D variants in BRCA, ATM, NBN, and PALB.
. Population frequencies for combined PRS P/LP/D groups were represented by weighted average frequencies calculated from the frequency of cases and controls in each PRS P/LP/D group within the study weighted by the frequency of prostate cancer in African American men (16.7%; ref.25)for combined PRS P/LP/D groups.Absolute risk confidence intervals (CI) were generated using estimates averaged from 1,000 iterations of the Monte Carlo sampling for a specific age for PRS categories and carrier status.
).While BRCA carriers are recommended to initiate prostate cancer screening at age 40 or 10 years prior to the youngest prostate cancer diagnosis in a family (with consideration for BRCA, ATM, HOXB, and mismatch repair gene carriers; ref. 4), our findings in African ancestry men and others in European ancestry men(6, 26-29)suggest that PRS should also be incorporated into prostate cancer risk assessment to inform age-based screening guidelines.Future studies with larger sample sizes of diverse African ancestry men will allow for finer PRS subgroups to be examined together with rare variant carrier status for more accurate estimation of genetic risk for overall prostate cancer, as well as metastatic and aggressive prostate cancer across the African diaspora.