Characteristics of Patients with Central Nervous System Relapse in Veterans with Diffuse Large B-Cell Lymphoma within the Veterans Health Administration

Introduction Around 2-5% of patients with diffuse large B-cell lymphoma (DLBCL) will experience central nervous system (CNS) relapse resulting in a poor prognosis. The Central Nervous System International Prognostic Index (CNS-IPI) is a validated risk model used to help identify DLBCL patients receiving R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone) chemotherapy who are at risk for CNS relapse. CNS prophylaxis is recommended for those with high-risk CNS-IPI scores; however, the role of CNS prophylaxis has been called into question given recent large datasets showing no impact on CNS relapse. The purpose of this study is to evaluate the characteristics of Veterans who experienced CNS relapse within the Veterans Healthcare Administration (VHA) and to validate the CNS-IPI risk stratification within this population. Methods Trained abstractors performed a retrospective chart review of 3287 lymphoma patients seen in the VHA nationwide between 01/01/2011 and 12/31/2017. Figure 1 describes the selection of the study cohort. We evaluated baseline patient and disease characteristics including CNS-IPI score, performance of diagnostic lumbar puncture (LP) and first-line chemotherapy regimen. Pathology reports to identify cell of origin (COO), and additional risk factors for CNS relapse present at the time of original diagnosis including HIV associated lymphoma, testicular lymphoma, and high-grade B-cell lymphomas (HGBLs) defined as double or triple-hit DLBCL were evaluated. We also assessed response to first-line treatment, type of CNS prophylaxis used, including number of doses, and time to CNS relapse. Results A total of 1621 patients met criteria for analysis. Patients were predominately male, white, had a median age of 67, and presented with advanced disease (Table 1). At the time of diagnosis, 81% of the cohort had an ECOG performance status of 0-2, 73% received a CHOP based regimen, and 52% were designated as having a high-risk CNS-IPI score. Patients were about twice as likely to have a germinal center B-cell (GCB) COO rather than activated B-cell (ABC), but the COO was unavailable for almost 30% of the cohort. About 6% of the patients were known to have HGBLs, but the “hit status” of around 65% of the patients was unknown. Diagnostic LP was performed in 10%, 14%, and 19% of patients in the low-, intermediate-, and high-risk CNS-IPI groups respectively. The median follow-up time for the subjects in the study was 44 months. The lo