Modeling the Effect of Treatments on Prostate Cancer-Specific Mortality and the Relevant Geographical Variation and Racial Disparities.

Background/Objectives: African American (Black) prostate cancer (PCa) patients have a higher risk of dying from the disease and are less likely to undergo radical treatment than European Americans (White). The disparities in PCa-specific mortality (PCSM) and mortality rate (PCSMR) vary geographically. This study investigated the impact of treatments on PCSM, PCSMR and the relevant disparities. Methods: Using the Cox PH model and other statistical methods, we analyzed two datasets extracted from the SEER and PLCO databases. The SEER dataset contains 637,093 White patients and 110,839 Black patients. The PLCO dataset included 7463 Whites and 495 Blacks, and supplemented the SEER data with information on PCa family history (pros_fh). Results: Analysis of SEER data showed that the relative mortality risk (RR) of patients undergoing surgery alone was significantly lower than that of patients receiving radiotherapy alone or a combination of surgery and radiotherapy. Black patients' RR estimated by the model including treatment was substantially smaller than that estimated by the reduced model excluding treatment. The differences between Black and White in the three-nine-year PCSMR of patients with high-grade or non-localized cancer were significantly correlated with the differences in surgery alone rate (r < -0.65, p < 0.001). Regression-based mediation analysis indicated that treatment disparity had a significant direct effect on mortality disparity and did not mediate the effect of age disparity. Analysis of PLCO data showed that pros_fh had no significant effect on survival but confirmed the survival advantage of surgery over radiotherapy. Conclusions: The results of this study support the hypothesis that, for PCa patients in the United States, geographical variation in treatment disparity partially explains variation in mortality disparity.