Cross-population metabolome-wide Mendelian randomization study of prostate cancer risk

Prostate cancer (PCa) is the most common cancer in US men, with Black men experiencing the highest incidence rates and Black and Hispanic men experiencing higher aggressive PCa rates than White men. Metabolomic dysregulation is a cancer hallmark; however, PCa metabolomic epidemiological evidence is heterogeneous and limited in diverse populations.

We conducted a metabolome-wide two-sample Mendelian randomization (MR) of PCa risk in African (AFR) and European (EUR) genetic ancestry and Hispanic (HIS) ethnicity populations. MR was performed in each population using serum metabolomic genome-wide association study (GWAS) summary statistics from ARIC (N_Metabolites=250, N_AFR=1,740, N_EUR=1,498) and HCHS/SOL (N_Metabolites=711, N_HIS=3,166) and PCa GWAS summary statistics from PRACTICAL (N_AFR=19,391/61,608, N_EUR=122,188/604,640, N_HIS=3,931/26,405 cases/controls). MR results were meta-analyzed across populations.

We identified 61 significant associations representing 50 unique metabolites in population-specific or cross-population analyses, with enrichment for lipids, including polyunsaturated fatty acids. Upon evaluating the strength of evidence, considering sensitivity analyses and the consistency of findings across populations, fourteen metabolites had strong evidence, including three drug-modifiable and six dietary-modifiable metabolites. Five were novel metabolites not previously reported in PCa MR studies, including two identified in AFR-specific associations: lysophospholipid 1-linoleoyl-GPE (18:20) and fatty acid hexadecanedioate, with colocalization suggesting a shared causal variant (rs28864441) between hexadecanedioate and PCa, and three identified in cross-population associations: amino acid 3-methoxytryosine, fatty acid 2-hydroxysterate, and peptide gamma-glutamylleucine.

Our findings provide evidence of associations between serum metabolites, particularly fatty acids, and PCa development across populations. The biological mechanisms and clinical utility of these metabolites as biomarkers of PCa risk warrant further investigation.