Determinants of Isolated CNS Relapse in Adults with Philadelphia-Negative Acute Lymphoblastic Leukemia. from Graall-2005 to -2014 Trials

Introduction: Treatment of adults with Philadelphia chromosome-negative acute lymphoblastic leukemia (Ph-negative ALL) has advanced through more intensive chemotherapy regimens. Despite these improvements, relapse occurs in approximately 30-40% of patients. The central nervous system (CNS) is the most common extra-medullary site of relapse. Historically, isolated CNS (iCNS) relapses have been considered as high-risk events among adults with relapsed ALL. This study aims at identifying factors associated with iCNS relapses in two consecutive GRAALL trials and evaluating the prognosis of these relapses. Patients and methods. The pediatric-inspired GRAALL-2005 and GRAALL-2014 trials included 1,530 patients with Ph-negative ALL aged 18-59 years between May 2006 and December 2020. In the 2014 trial, prophylactic CNS irradiation (18 Gy) previously used in the 2005 study was replaced by an increased number of triple intrathecal therapy (ITT), from 6 to 13 in patients not eligible for allogeneic hematopoietic stem cell transplant (HSCT). In both trials, patients with initial CNS infiltration received 24 Gy CNS irradiation before maintenance therapy or a CNS boost before total body irradiation used for HSCT conditioning regimen. High-dose methotrexate was also increased from 3,000 mg to 5,000 mg/m² for patients younger than 45 years. Additionally, criteria for HSCT shifted from baseline and early peripheral and bone marrow response-based in the 2005 to minimal residual disease (MRD)-based only in the 2014 trial. The cumulative incidence of iCNS relapse (CI-iCNS) was assessed from the time of complete remission (CR) to iCNS relapse, considering death in CR and other relapses as competing events. Results. Among the 1,530 patients, median age was 35.6 years, 38.9% were female, 33.7% had a T-ALL phenotype, and 9.3% presented with initial CNS disease. Complete remission was achieved in 1,416 patients (92.6%). The median follow-up was 5.2 years in the 2005 trial and 3.2 years in the 2014 trial. A total of 445 relapses occurred, with 355 (79.8%) involving the bone marrow (31 [7.0%] BM and CNS, 17 [3.8%] BM and extra-medullary site other than CNS), 57 (12.8%) were iCNS, 31 (7.0%) were extramedullary other than CNS relapses, and 2 (0.4%) were not characterized. The cumulative incidence of relapse (CIR) and CI-iCNS at 4 years were 33.3% (95% CI, 30.8-36.0) and 4.3% (95% CI, 3.3-5.5), respectively. Time to iCNS relapse was similar to other relapses (11.9 vs 11.7 months, p=0.86). In univariate analysis, participation in the most recent 2014 trial, male status, T-lineage, higher WBC, but not age, were associated with an increased risk of iCNS relapse. In multivariate analysis, participation in the 2014 trial (SHR 2.3; 95% CI, 1.3-4.1), T-lineage (SHR 2.7; 95% CI, 1.5-4.8), and higher logWBC (SHR 1.4; 95% CI, 1.1-1.6) were associated with an increased CI-iCNS. Univariate analysis in BCP-ALL identified higher logWBC, TCF3::PBX1 gene fusion, and participation in the 2014 trial, but not KMT2A-rearrangement, as being associated with iCNS relapse. In T-ALL, higher logWBC and non-ETP ALL were the two factors associated with iCNS relapse. Overall, none of the patients with high hyperdiploidy, IgH::DUX4 fusion, or ETP-ALL experienced iCNS relapse. End-of-induction positive MRD was a strong predictor of an increased risk of relapse (SHR 1.8; 95% CI, 1.4-2.2) but was associated with a decreased risk of iCNS relapse (SHR 0.37; 95% CI, 0.17-0.77), reflecting the competition between BM and iCNS relapse events. While 30.7% of patients underwent HSCT in first CR, only 5 patients experienced iCNS relapse after HSCT (SHR 0.20; 95% CI, 0.08-0.52). Median post-relapse follow-up was 2.7 years. Patients with iCNS relapse had similar post-relapse survival compared to other relapses (at 2 years: 32.0% [95% CI, 19.2-45.6] versus 28.5% [95% CI, 23.8-33.5]; p=0.28). Conclusion. This study identifies disease profiles associated with an increased risk of isolated CNS relapse and suggests that both prophylactic CNS irradiation and TBI-based HSCT provide protective benefits. Our findings may inform the refinement of future frontline strategies, including immunotherapy, which has been linked to higher rates of CNS-involving progressions.