Discrepancy between central nervous system involvement at diagnosis and relapse burden in childhood acute...

Central nervous system (CNS) involvement is a critical prognostic factor in childhood acute lymphoblastic leukemia (ALL). Conventional cytology remains the diagnostic standard, but its limited sensitivity may lead to underdetection of CNS disease. This study evaluated the incidence and prognostic impact of CNS involvement at diagnosis and relapse in a Mexican pediatric cohort, focusing on the relationship between baseline CNS status and subsequent relapse.

We conducted a retrospective cohort study of 73 children (<15 years) diagnosed with ALL at a national tertiary referral center between 2019 and 2020, with a five-year follow-up. CNS involvement was defined by cerebrospinal fluid cytology and classified as CNS-1, CNS-2, and CNS-3 according to international criteria. Incidence at diagnosis, CNS relapse during follow-up, and cumulative CNS involvement were estimated. CNS relapse was analyzed among patients achieving complete remission (n =57), excluding early deaths, to define the at-risk population. Survival was analyzed using Kaplan-Meier estimates and the log-rank test.

At diagnosis, CNS involvement occurred in 2.9% of evaluable patients (n =2/69; 95% CI: 0.8-9.5%), while 33.3% of patients were classified as CNS-2. During follow-up, CNS relapse occurred in 17.5% (n =10/57; 95% CI: 9.4-28.9%), with a cumulative proportion of 16.4% (n =12/73; 95% CI: 9.3-26.2%). Half of CNS relapses occurred in patients initially classified as CNS-2. CNS-3 status at diagnosis was associated with significantly inferior overall survival (log-rank, p = 0.007).

Our findings identify a marked discordance between baseline CNS detection and subsequent relapse burden. The observed relapse burden in a cohort with a high prevalence of CNS-2 at presentation raises the possibility that conventional criteria may underestimate clinically relevant CNS disease in some patients. Prospective studies incorporating more sensitive diagnostic approaches are warranted to clarify the clinical implications of low-level cerebrospinal fluid blast detection.