Association of Cumulative Proton Pump Inhibitor Use with Prostate Cancer Risk and Outcomes: A Population-Based Cohort...

Proton pump inhibitor (PPI) use has been associated with increased prostate cancer risk in case-control and cohort studies of men with a negative prostate biopsy. We estimated the association of cumulative PPI use with prostate cancer diagnosis and outcomes in a biopsy-naïve cohort of men using provincewide-linked administrative data from Ontario, Canada, including men ≥66 years with no prior prostate cancer diagnosis, biopsy, or treatment and no PPI/histamine-2-blocker prescriptions within the year preceding study inclusion (January 2003-December 2018). The primary outcome was time to prostate cancer diagnosis. Associations were evaluated using univariable/multivariable logistic regression models with complementary log-log modeling. Median follow-up was 9.2 years (n = 559,425), and 30.2% had any PPI use during follow-up. Annual PPI use increased from 2.4% (2003) to 13.8% (2019). Cumulative PPI use was associated with lower prostate cancer diagnosis rates [hazard ratio (HR) for highest user quintile vs. nonusers: 0.75, 95% confidence interval (CI), 0.71-0.79]. Following adjustment for the frequency of general practitioner visits and prostate-specific antigen tests in the preceding 2 years, PPI use was no longer associated with prostate cancer diagnosis (HR, 0.99; 95% CI, 0.93-1.05). No association was observed between PPI use and rates of clinically significant or high-grade prostate cancer (HRs, 0.99 and 1.04, respectively). Cumulative PPI use was associated with up to 17% lower rates of a first androgen deprivation therapy prescription or bilateral orchiectomy. Cumulative PPI use was not associated with prostate cancer diagnosis rates, including clinically significant and high-grade prostate cancer. These findings do not support an independent association between PPI use and prostate cancer risk.

Among biopsy-naïve older men, cumulative PPI use was not associated with prostate cancer diagnosis, including clinically significant and high-grade disease, after adjusting for healthcare resource utilization. These findings challenge prior observational signals suggesting increased risk and indicate that detection bias likely explains earlier associations.