Patients (2025) were categorised according to initial central nervous system (CNS) status: CNS-1 (CNS negative, n = 1866), CNS-2 (<5 leucocytes/mm3, CSF with blasts, n = 50), CNS-3 (CNS positive, n = 49), TLP+ (TLP with blasts, n = 60). CNS-directed therapy consisted in intravenous (i.v.) methotrexate (5 g/sqm) in 4–10 courses, and intrathecal methotrexate injections (10–20), according to CNS status. Cranial irradiation was omitted in all patients.
In the CNS1, TLP+, CNS2 and CNS3 group the 8-year EFS rate (SE % ) was 69.7 % (1.1 % ), 68.8 % (6.2 % ), 71.3 % (6.5 % ) and 68.3 % (6.2 % ), respectively. The 8-year incidence of isolated CNS relapse (SE % ) was 3.4 % (0.4 % ), 1.7 % (1.7 % ), 6.1 % (3.5 % ) and 9.4 % (4.5 % ), respectively, whereas the 8-year isolated or combined CNS relapse incidence was 7.6 % (0.6 % ), 3.5 % (2.4 % ), 10.2 % (4.4 % ) and 11.7 % (5.0 % ), respectively. Patients with CSF blasts had a higher rate of initial bad risk features. Multivariate analysis indicated that presence of blasts in the CSF had no prognostic value: (i) for EFS and OS; (ii) for isolated and isolated or combined CNS relapse; WBC count < 25 × 109/L and Medac E-coli asparaginase treatment were each related to a lower CNS relapse risk.
The presence of initial CNS involvement has no prognostic significance in EORTC 58881. Intensification of CNS-directed chemotherapy, without CNS radiation, is an effective treatment of initial meningeal leukaemic involvement.