Figure. Association Between Radiotherapy Type and Second Primary Cancers Among Male Prostate Cancer Survivors in the Linked SEER-Medicare Cohort.

Models were adjusted for time-fixed variables of age at diagnosis, race, grade at prostate diagnosis, Charlson comorbidity index, and prostatectomy using the categories presented in the Table and time-dependent covariates of chemotherapy (for analyses of hematologic cancers this only includes chemotherapy received after the first year), brachytherapy, and hormone therapy. Treatment variables were ascertained from Medicare claims. Age, calendar year, and tumor characteristics were ascertained from SEER. Follow-up began at 2 years and 5 years (corresponding to the expected minimum latency periods for radiation-related cancers) following prostate cancer diagnosis, for hematologic cancers and solid cancers, respectively. Men were followed until the earliest of age 85 years, second cancer, Medicare claim for proton therapy, death or December 31, 2015 (last available data). Case counts of less than 11 were suppressed to protect patient confidentiality. Abbreviations: AML/MDS, acute myeloid leukemia/myelodysplastic syndrome; 3DCRT, 3-dimensional conformal radiation therapy; CNS, central nervous system; GI, gastrointestinal; HR, hazard ratio; IMRT, intensity-modulated radiation therapy; SEER, Surveillance, Epidemiology, and End Results.