ABSTRACTnation of blood smears in cases in which the automated counts were abnormal.CNS relapse-free survival was calculated by cause-specific function, censoring patients who died without CNS relapse, in accordance with previously published studies.11-14 The starting date of the time-to-event analysis was defined at diagnosis. Univariate and multivariate analyses of estimated risk factors for CNS relapse in DLBCL patients were performed using the log-rank test and Cox proportional hazard regression analysis. P values <0.05 were considered statistically significant. Background characteristics of the patients with and without CNS involvement were compared using the Fisher exact test and Student t-test. All statistical analyses were performed with EZR 15 (Saitama, Medical Center, Jichi Medical University, Japan).