Validation of candidate protein biomarkers previously identified by genetic instruments for prostate cancer risk: A prospective cohort analysis of directly measured protein levels in the ARIC study

Abstract Background Multiple novel protein biomarkers have been shown to be associated with prostate cancer risk using genetic instruments. This study aimed to externally validate the associations of 30 genetically predicted candidate proteins with prostate cancer risk using aptamer‐based levels in US Black and White men in the Atherosclerosis Risk in Communities (ARIC) study. Plasma protein levels were previously measured by SomaScan® using the blood collected in 1990–1992. Methods Among 4864 eligible participants, we ascertained 667 first primary prostate cancer cases through 2015. Hazard ratios (HRs) of prostate cancer and 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression for tertiles of each protein. We adjusted for age, race, and other risk factors. Results Of the 30 proteins and considering a nominal p trend < 0.05, two were positively associated with prostate cancer risk—RF1ML (tertile 3 vs. 1: HR = 1.23; 95% CI 1.02–1.48; p trend = 0.037) and TPST1 (1.28, 95% CI 1.06–1.55; p trend = 0.0087); two were inversely associated—ATF6A (HR = 0.80, 95% CI 0.65–0.98; p trend = 0.028) and SPINT2 (HR = 0.74, 95% CI 0.61–0.90; p trend = 0.0025). One protein, KDEL2, which was nonlinearly associated (test‐for‐linearity: p < 0.01) showed a statistically significant lower risk in the second tertile (HR = 0.79, 95% CI 0.65–0.95). Of these five, four proteins—ATF6A, KDEL2, RF1ML, and TPST1—were consistent in the direction of association with the discovery studies. Conclusion This study validated some pre‐diagnostic protein biomarkers of the risk of prostate cancer.