Risk Factors for Central Nervous System (CNS) Relapse after Autologous Stem Cell Transplant (ASCT) in Diffuse Large B-Cell Lymphoma (DLBCL)

Background:Although the addition of rituximab to CHOP regimen improved prognosis in DLBCL patients with more than 80% of a long-term survival rate, CNS relapse did not decrease and 5 to 10% of patients experienced CNS relapse after rituximab-containing chemotherapy. Risk factors for CNS relapse after standard chemotherapy have been aggressively investigated, and a risk model, CNS-international prognostic index (IPI), has been widely used. However, risk factors for CNS relapse after high-dose chemotherapy following ASCT, which is recognized as an important treatment option for high-risk DLBCL patients, have not been elucidated. So, we performed this retrospective analysis to address this unsolved issue. Patients and methods:This study analyzed 87 adult patients who underwent ASCT against chemo-sensitive DLBCL including intravascular large B-cell lymphoma (IVLBCL) between 1997 and 2015 at the four institutions in Gunma, Japan. There was no restriction on the type of conditioning regimens. CNS-directed regimens were defined as chemotherapy or conditioning regimens containing high dose cytarabin, high dose methotrexate, busulfan, ranimustine, or total body irradiation. Only the first relapse after ASCT was assessed in this study. Fisher's exact test was used to compare binary variables. Cumulative incidences (CIs) of CNS relapse were compared using the stratified Gray test, considering relapse without CNS lesions and death without the event as a competing risk. The Fine-Gray proportional hazard model was used for multivariate analysis of risk factors for CNS relapse. The potential risk factors evaluated in this analysis were age at transplant, gender, clinical stage, IPI (high-intermediate/high or not), and CNS-IPI (high or not) at diagnosis, CD5 positivity, CNS involvement prior to ASCT, CNS-directed chemotherapy prior to ASCT, and CNS-directed conditioning regimen. Values of p < 0.05 were considered significant. Results:Of the 87 patients assessed in this study, 48 were male and 39 were female, and the median age was 57 years (range: 23 to 66 years). CD5 was expressed in 19% of the patients, and 10% were diagnosed as IVLBCL. CNS-IPI at diagnosis was high in 53%, and rituximab and CNS-directed chemotherapy was administered prior to ASCT in 83% and 16%, respectively. CNS involvement was observed prior to ASCT and at the time of ASCT in 9% and 0%, respectively. Disease status at the time of transplant was first complete remission (CR) in 47%, advanced CR in 23%, and partial remission in 30%. CNS-directed conditioning regimens were used in 38%. With a median observation time of 21.9 months, seven patients experienced CNS relapse as the first relapse after ASCT. The 3-year CI of CNS relapse was 7.3% with 5.7 months of median duration from ASCT (range: 2.7 to 69.0 months). In univariate analysis, only CD5 positivity was identified as a significant risk factor for CNS relapse (3-year CIs in patients with and without CD5 expression: 27.0% vs. 2.2%, respectively; p < 0.01). In multivariate analysis, CD5 positivity, CNS-IPI at diagnosis, CNS-directed chemotherapy prior to ASCT, and CNS-directed conditioning regimen were evaluated, and only CD5 positivity was identified as an independent risk factor for CNS relapse (relative risk=20.1; p < 0.01). Of the seven patients with CNS relapse after ASCT, four expressed CD5 and five received CNS-directed chemotherapy prior to ASCT and/or conditioning regimens. All seven patients died from DLBCL within two years after CNS relapse. Conclusion:These results suggested that CNS relapse occurred in DLBCL patients even after ASCT with similar incidence to that after chemotherapy. Although prophylactic strategies for CNS relapse should be investigated especially in patients with CD5-positive patients, use of CNS-directed chemotherapy prior to ASCT and/or conditioning regimens did not affect the CIs of CNS relapse. A future study with a larger cohort is warranted to develop a risk model for CNS relapse after ASCT in DLBCL patients.