Prognostic factors in patients with chronic myeloid leukemia according to competing-risks analysis: a study based on the SEER database

Abstract

Abstract Background: The classic Cox proportional-hazards model is generally used to derive risk factors affecting patients with chronic myeloid leukemia (CML). However, when competing risk factors are present, the results of Cox analysis need to be revisited.Objective: This study aimed to develop a competing-risks model to assess the factors that influence predictions of patients with CML in an attempt to produce results that are more accurate than those from Cox analyses.Methods: The CML data in the SEER (Surveillance, Epidemiology, and End Results) database that met our requirements during 1975–2019 were analyzed. Univariate analyses in this study were performed using cumulative incidence functions and Gray’s tests, and the multivariate analysis was performed using three models: Fine-Gray, cause-specific, and Cox proportional-hazards models.Results: Of the 8331 included cases, 4827 (57.94%) died (2459 [29.52%] from CML and 2368 [28.42%] from other causes) and 3504 (42.60) survived. Gray’s test indicated that the outcome was significantly affected by year of diagnosis, age, total number of in situ/malignant tumors, type of reporting source, marital status, and primary indicator of malignancy. The results of the multivariate competing-risks analyses suggested that age, year of diagnosis, total number of in situ/malignant tumors, type of reporting source, marital status, and primary indicator of malignancy were independent risk factors for the prognosis of patients with CML (P<0.05). Conclusions: This study has developed a competing-risks analysis model for assessing the risk factors for patients with CML. Our findings may allow for more-accurate formulation of clinical decisions, saving healthcare resources in the current individualized treatment environment for the benefit of the patients.