Causes of death and effect of non-cancer-specific death on rates of overall survival in adult classic Hodgkin lymphoma: a populated-based competing risk analysis

Background The improved prognosis of classic Hodgkin lymphoma (cHL) has been accompanied by elevated risks of non-cancer-specific death (non-CSD). The aim of this study was to verify the occurrence of non-CSD and its effect on rates of overall survival among adult patients with cHL. Methods To ensure sufficient follow-up time, we analyzed retrospective data from patients aged >= 20 years with cHL that was diagnosed between 1983 and 2005 in the Surveillance, Epidemiology, and End Results (SEER) database. Logistic regression was applied to analyze the non-CSD occurrence in relation to all factors. Using Fine-Gray's method, we calculated the cumulative incidences of CSD and non-CSD. Stacked cumulative incidence plots and ratio of non-CSD to all causes of death were applied to evaluate the effect of non-CSD on rates of overall survival. Finally, we analyzed long-term mortality through Cox proportional hazard regression analysis and competing risk regression analysis to emphasize a more appropriate model of survival for patients with cHL. Results Among the 18,518 patients included, there were 3768 cases of CSD (20.3%) and 3217 of non-CSD (17.4%). Older age, earlier period, male sex, unmarried status, mixed cellularity (MC) and lymphocyte-depletion (LD) histological subtype, and patients received radiotherapy (RT) only were associated with more non-CSD according to binary logistic analysis. The cumulative incidence of non-CSD exceeded CSD after approximately 280 months follow-up. The most common causes of non-CSDs were cardiovascular disease, subsequent primary neoplasms, infectious diseases, accidents, and suicide. In a Cox proportional hazards model, patients who were black, unmarried, at an advanced stage or underwent chemotherapy (CT) alone were at greater risk of mortality than were white patients, who were married, at an early stage, and underwent combined modality; these populations were also found to be at greater risk for CSD in a competing risk model, but the risk of non-CSD did not differ significantly according to race and marital status, patients with early-stage disease and who underwent RT only were found to be at higher risk of non-CSD instead. Conclusions Lymphoma was the cause of death in most patients who died, but non-CSD was not unusual. Patients with cHL should be monitored closely for signs of cardiovascular disease and malignant tumors. Rates of overall survival of patients were diminished by non-CSD, and a competing risk model was more suitable for establishing the prognosis than was the Cox proportional hazards model.