Disease characteristics and outcomes of acute lymphoblastic leukemia (ALL) in adults with central nervous system (CNS) involvement

Abstract Background: Despite major advances in the treatment of ALL and incorporation of CNS prophylaxis, the risk factors for and outcomes of patients (pts) presenting or relapsing with CNS disease are incompletely understood. Aims: We aimed to characterize pts and outcomes of pts with CNS involvement at the time of ALL diagnosis and/or relapse. Methods: We performed a retrospective chart review of adult ALL pts treated at Weill Cornell Medical Center between 2012-2023. CNS involvement was defined using cytology, flow cytometry and/or imaging. Patients provided informed consent for data collection and analysis. Results: Among 191 pts, 50.3% were female, mean age at diagnosis was 50.7 yrs (18-87 yrs), 75.4% had B-cell disease and 42.9% were Philadelphia chromosome positive (Ph+). Mean BMI was 27.7 (18.1-53.9). Median follow up was 46.8 months (mo). Twenty-six pts (13.6%) presented with CNS involvement at diagnosis; those pts had significantly higher baseline WBC (mean 112.9 vs 32.3 p=0.008) and were more likely to have T cell disease (46% vs 21% p=0.007), abnormal cytogenetics (85% vs 56% p=0.016) and a trend towards higher LDH (1327 U/L vs 826 U/L p=0.068). Other extramedullary disease and Ph-status were not associated with CNS disease. Molecular analysis showed that pts with ETV6 mutations had a higher likelihood of CNS involvement at diagnosis (p= 0.03). Myeloid gene mutations (any of TP53, TET2, DNMT3A, ASXL1, RUNX1) were associated with a trend toward higher rates of CNS involvement at diagnosis (p=0.076). Pts receiving intrathecal (IT) chemotherapy cleared the CSF of blasts after an average of 2.56 treatments (range 1-20). 77% of the total cohort achieved CR, with CR rates comparable between pts with and without baseline CNS involvement (73 % vs 83% p=0.25). Post induction bone marrow MRD neg status was not statistically different between pts with or without baseline CNS disease (31% vs 49% p=0.16). Overall Survival (OS) for pts with and without baseline CNS involvement was comparable, with 5 years OS 51% vs 59% (p=0.32). Subgroup analysis was limited by small numbers, but showed comparable OS between pts with CNS involvement by flow only to those without CNS involvement (p=0.23). However, OS for pts with CNS disease documented by cytology+flow was shorter vs those with no CNS involvement (p=0.03). Among 76 pts who relapsed (39.8%), 24 (12.6%) had CNS involvement and 6 (3.1%) had isolated CNS disease. Time to relapse was equal for those with and without CNS involvement at relapse (580 vs. 507 days p=0.19), but the time to relapse was shorter in those with CNS disease at diagnosis 547 vs 1092 days (p=0.038). Pts with CNS relapse had similar baseline WBC, B vs T-cell phenotype, LDH, and Ph status compared to those without CNS relapse. Though Ph+ disease was not associated with CNS relapse, pts with BCR-ABL p210 had a trend toward more CNS relapses than those with p190 (p= 0.062). Pts with p210 also had more CNS relapses compared to Ph-negative pts (p=0.027). CNS relapse rate was not significantly impacted by the use of Dexamethasone vs Prednisone during induction (p=0.87), or by the use of pediatric inspired protocols vs other protocols (p=0.54). There was no difference in the number of days between starting induction chemotherapy and the first LP in pts with and without CNS relapse (4.2 vs 4 days, p=0.53). Baseline CNS involvement was not a risk factor for either CNS or systemic relapse (p=0.45) and only 2 pts with initial CNS involvement (one by flow only and one by cytology+flow) experienced a CNS relapse. Among pts who received ITs and systemic chemotherapy, 21% never cleared their CSF. Pts who cleared CSF needed an average of 5.72 IT treatments (range 1- 25) to accomplish clearance. Median OS of relapsed pts was 11 mo with no difference between pts with CNS vs without CNS involvement at relapse (9.7mo vs 14 mo p=0.8)Conclusion: In our large cohort of ALL patients, elevated WBC, T-cell phenotype, abnormal cytogenetics and mutations in ETV6 were risk factors for CNS involvement at diagnosis. Baseline CNS disease had a trend toward inferior OS, a finding largely driven by those patients with CNS disease detected by cytology rather than flow alone. CNS involvement at diagnosis was not a specific risk factor for either CNS relapse or systemic relapse. Furthermore, pts with CNS relapse had similar OS to those with systemic relapse only. Larger cohorts and prospective data are needed to confirm these observations.