Risk and predictors of late second primary malignancies in long-term breast, prostate, colon, and rectal cancer survivors.

1642 Background: In older, long-term (5-year) cancer survivors, mortality risks from aging and treatment-related effects may surpass those of their index cancer. Second primary malignancies (SPMs) occurring 5-10 years post-diagnosis are understudied in older patients. This study aims to quantify SPM risk, identify predictors, and describe prevalent SPM sites in older survivors of breast, prostate, colon, and rectal cancer to guide survivorship care. Methods: This retrospective cohort study analyzed patients aged 66+ with stage I-III cancer diagnosed between 2003-2011 using the SEER-Medicare database. Eligible patients survived ≥5 years post-diagnosis and had continuous Medicare Parts A & B enrollment from 1 year pre-diagnosis to 1 year post-diagnosis. The primary outcome was late SPMs occurring 5-10 years after index cancer diagnosis. Covariates included demographics, comorbidities, index cancer characteristics, treatment, and early SPMs (diagnosed within 5 years). Least absolute shrinkage and selection operator for variable selection and 5-year restricted mean survival time regression models were used. Cumulative late SPM incidence was calculated with mortality as a competing risk. The prevalence of specific SPMs was calculated as a proportion relative to the total number of SPMs within each cohort, and categorized as hematologic, predominantly screen-detected (breast, prostate, colorectal), or other solid tumors. Results: Of the 88,227 long-term survivors included with median age of 73.3 (IQR 69.5-78), 6.2% developed early SPMs and 8.2% (7,231) developed late SPMs. The 5-year cumulative incidence of late SPMs was 8.6%, highest in prostate (9.2%) and lowest in breast (6.7%) cancer survivors. Non-screenable cancers had the highest 5-year risk (6.2%), followed by screen-detected (1.3%) and hematologic malignancies (1.1%). Lung was the most common SPM overall (18.4% of SPMs), including in survivors of breast (21%), rectal (19.2%) and colon (16.5%) cancers, while prostate was most common in rectal cancer survivors (17.0%). Diagnosis of SPM in the early ( < 5 years) survivorship cohort was associated with shorter time to a new late SPM, particularly in prostate cancer survivors (RMST Ratio 0.97, 95% CI 0.96-0.98). Treatments and high-risk disease features showed no significant associations with occurrence of late SPMs. Conclusions: Late SPMs were diagnosed in 8.6% of older, long-term cancer survivors. Lung cancer was the most common SPM overall. Some screen-detected SPMs, such as prostate were also common which is notable in a population aging out of screening guidelines. Prediction of SPMs was limited by the absence of modifiable risk factors, genetic data, and family history in SEER-Medicare data. Early SPMs were the sole predictor of late SPMs while treatment and index cancer features showed no effect, suggesting other drivers of late SPM development in older survivors.