Viral infections and the risk of prostate cancer: a two-sample Mendelian randomization study with multi-cohort replication and an extended viral antibody panel

Observational evidence linking viral infection to prostate cancer (PCa) remains controversial. We evaluated the potential causal relationships between serological markers of viral infection and PCa risk using two-sample Mendelian randomization (MR). Exposures comprised serological markers of eight viruses (HPV16/18 E7 oncoproteins, BK and JC polyomaviruses, cytomegalovirus [CMV], Epstein-Barr virus [EBV], herpes simplex virus 1/2) and Helicobacter pylori, obtained from public genome-wide association studies (GWAS). PCa summary statistics came from PRACTICAL (79,148 cases, 61,106 controls), FinnGen R10 (15,199/131,266) and the Wang 2023 European-ancestry subset (122,188/604,640). Inverse-variance-weighted estimation was primary, with extensive sensitivity analyses (MR-Egger, weighted-median, MR-PRESSO, Steiger directionality testing, colocalization, multivariable, reverse and mediation MR) and cross-cohort meta-analysis. An extended 37-trait serological panel covering 12 viruses was additionally screened. EBV EBNA-1 antibody levels were associated with increased PCa risk in PRACTICAL (OR = 1.069, P = 0.032) and replicated in FinnGen (OR = 1.117, P = 0.017) and Wang 2023 (OR = 1.062, P = 0.036), with a cross-cohort random-effects meta-analytic OR of 1.074 (95% CI 1.034–1.115, P = 2.1 × 10-4, I2 = 0%), and in the fully independent PRACTICAL + FinnGen meta-analysis (OR = 1.084, P = 0.002). The association was independent of CMV pp150 in multivariable MR, free of reverse causality and directional pleiotropy, and Steiger-confirmed. CMV pp150 showed directionally consistent, meta-analytically significant effects (OR = 1.038, P = 0.020). The extended panel identified two novel Bonferroni-significant, cross-cohort-replicated associations: Merkel cell polyomavirus seropositivity with reduced risk (meta OR = 0.967, P = 1.6 × 10-5) and BK polyomavirus seropositivity with increased risk (meta OR = 1.267, P = 1.0 × 10-5). TCGA-PRAD transcriptomics showed down-regulation of CMV/EBV host-cell entry genes, compatible with a hit-and-run model. No robust evidence implicated HPV, JC virus, HSV, HHV-6/7 or VZV. Genetically proxied EBV latent infection is a candidate causal factor for PCa, and MCV (protective) and BK polyomavirus (risk) seropositivity are novel candidates warranting mechanistic and multi-ancestry validation. Effect sizes are small and, under Bonferroni correction for 37 traits, only the polyomavirus signals survive; all findings should therefore be regarded as hypothesis-generating pending further independent replication.