Postdiagnosis Statin Use and Mortality in Danish Patients With Prostate Cancer

Purpose Increasing evidence indicates that statin use may reduce mortality from prostate cancer. In this work, we examined whether postdiagnosis statin use was associated with reduced cancer-specific mortality or all-cause mortality among patients with prostate cancer in Denmark. Material and Methods From nationwide Danish registries, we identified all patients with incident prostate adenocarcinoma from 1998 to 2011 and retrieved data on tumor and patient characteristics, drug use, and primary treatment. We defined postdiagnosis use (two or more prescriptions) of statins as a time-varying covariate with 1-year lag. Cox proportional hazards regression models used to compute hazard ratios (HRs) for prostate cancer-specific mortality and all-cause mortality through 2013 associated with postdiagnosis statin use. In secondary and sensitivity analyses, we assessed statin use within exposure periods of 1 year or 5 years after prostate cancer diagnosis and evaluated the influence of prediagnosis statin use. Results Among 31,790 patients, 7,365 died of prostate cancer and 11,811 died from other causes during a median follow-up of 2.8 years (interquartile range, 1.3 to 5.1 years) from 1 year after diagnosis. Postdiagnosis statin use was associated with adjusted HRs of 0.83 (95% CI, 0.77 to 0.89) for prostate cancer mortality and 0.81 (95% CI, 0.76 to 0.85) for all-cause mortality. Similar results were observed in 1-year and 5-year sensitivity analyses. No substantial effect measure modification was found with estimated dose or type of statin, clinical stage, Gleason score, or with prediagnosis statin use; however, patients who were diagnosed early in the study period or who underwent radical prostatectomy or endocrine therapy exhibited slightly lower HRs for prostate cancer mortality with postdiagnosis statin use than did those in the overall analyses. Conclusion Postdiagnosis statin use was associated with reduced mortality from prostate cancer; however, it remains to be established whether this association is causal.