Cytofluorimetric study of cerebrospinal fluid at staging of diffuse large B cell lymphoma

Central nervous system (CNS) involvement is a devastating manifestation of diffuse large B-cell lymphoma (DLBCL), historically associated with poor outcomes. However, the criteria for accurately identifying patients at highest risk of leptomeningeal involvement at diagnosis or of future CNS relapse remain unclear. Currently, the Central Nervous System-International Prognostic Index (CNS-IPI) score is widely used at diagnosis to identify patients at increased risk of CNS relapse. This study aimed to evaluate the prognostic significance of cerebrospinal fluid (CSF) analysis at diagnosis in relation to subsequent CNS recurrence. As part of the staging procedures for DLBCL, we performed diagnostic CSF analyses-including immunophenotyping, morphological evaluation, and chemical-physical examination. Between 2010 and 2023, we conducted 87 lumbar punctures for CSF collection. The median patient age was 54 years (range 48-61). CSF immunophenotyping and morphology were positive in 2 of 87 patients (2%). CNS-IPI was evaluable in 86 patients: 20 (23%) were low-risk, 47 (55%) intermediate-risk, and 19 (22%) high-risk. During follow-up, 17 patients experienced disease progression or relapse: 15 had systemic relapse, and 2 had CNS relapse. The 2-year cumulative incidence of CNS relapse was 2.2%, with a median time to progression of 10 months (IQR 6-15). CNS prophylaxis with intrathecal methotrexate or high-dose intravenous methotrexate was administered to 40 patients (46%). At a median follow-up of 40 months, 72 patients were alive and 15 had died. Although the number of CNS events was low, high-risk CNS-IPI was significantly associated with CNS relapse (p = 0.04), whereas CSF immunophenotyping and morphology at diagnosis were not. Notably, both patients who developed CNS relapse had received prophylaxis-one with intrathecal methotrexate and the other with high-dose intravenous methotrexate following R-CHOP. CSF analysis at diagnosis using immunophenotyping, morphology, and chemical-physical evaluation was not predictive of CNS recurrence. Moreover, CSF positivity at diagnosis is exceedingly rare, even among patients with high CNS-IPI. Our findings reinforce that CNS-IPI remains the most reliable predictor of CNS relapse in DLBCL.