ASPO Abstracts
Marginal Structural Model to Determine the Causal Effect of Aspirin on Reducing the Risk of Prostate Cancer
Category: Inflammation & Cancer, Inflammation & Cancer
Conference Year: 2018
Abstract Body:
Purpose: In the United States, prostate cancer is the most commonly diagnosed malignancy among men, accounting for about 1 in 5 new cancer diagnoses. Chronic inflammation plays a major role in carcinogenesis and it is hypothesized that aspirin, a nonsteroidal anti-inflammatory drug could play a preventive role in prostate carcinogenesis through its inhibitory effect on the COX enzyme. Previously, baseline aspirin use at least once per day was associated with a modest reduced risk of prostate cancer in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. We updated the analyses to include information on aspirin use at follow-up and to adjust for time-varying covariates. Method: Incident cases of prostate cancer were obtained from the screened arm of the PLCO. Enrollment was from 1993 until 2001 and subjects were followed up until 2009. Aspirin use during the previous 12 months was ascertained through self-report at baseline and in 2006, 13 years into the trial, to capture updated information. Subjects with missing baseline information on aspirin use (n=973), race (n=22), education (n=45), marital status (n=14), and no follow up time (n=164) were excluded from the analysis. The final sample included 36303 men aged 49 to 77 years [mean (SD) = 62.70 (5.32)] at baseline. A marginal structural model (MSM) was used to estimate the causal effect of aspirin use on the risk of prostate cancer by creating stabilized inverse-probability-weights to adjust for time dependent covariates; smoking, diabetes, arthritis, heart disease, hypertension, and stroke. Other covariates included in the model were age, race, education, marital status, family history of prostate cancer, BMI, and study site. Censored weights also were created to adjust for bias from loss to follow-up. Results: A total of 4246 incident cases of prostate cancer were reported. MSM analysis for the causal effect of aspirin use at baseline on prostate cancer incidence produced an OR of 0.98 (95% CI: 0.91 — 1.06) and the total effect of aspirin use over the study period produced an OR of 1.00 (95% CI: 0.91 — 1.10). Conclusion: These results suggest that the cumulative use of aspirin over the follow-up period does not have a significant causal effect on the risk of prostate cancer.
Keywords: nonsteroidal anti- inflammatory drugs, cyclooxygenase, prostate cancer, marginal structural model