Dexamethasone-Based Treatment and the Risk of CNS-Relapse in Acute Lymphoblastic Leukemia: Results from DCOG Protocol ALL-9.

Several groups recently reported on the significance of small numbers of blasts (CNS2; WBC ≤5×10⁶/l with morphological blasts) in the cerebrospinal fluid (CSF), as well as on traumatic lumbar puncture (TLP+; >10 eyrthrocytes ×10⁶/l, with blasts). In general, CNS2 does not have poor prognostic significance, whereas TLP+ does. We retrospectively analysed the prognostic influence of CSF involvement in DCOG ALL-9, which was highly dexamathasone (DEX) driven. The induction consisted of 3 drugs with DEX 6 mg/m², followed by 3 MTX courses 2 g/m² (4 × 3 gr/m² for high-risk (HR) patients), and maintenance with DEX/vincristine and 6-MP/MTX blocks. 13 intrathecal triples (ITH) were given as CNS-prophylaxis. HR-patients (29%) received anthracyclines in induction and 2 extra intensification courses. CNS3 patients (WBC >5×10⁶ with blasts) received additional ITH, and in case of persistent blasts after 4 ITH also irradiation. CSF at diagnosis was sent on culture medium to the DCOG for analysis. Classification was based on morphology, or TdT for CNS1 (WBC ≤5×10⁶/l and no blasts) and CNS2. 859 patients (1–18 years) were enrolled; of 139 no CSF was received; hence we report on 720 patients. Median age was 5 years; WBC 10.4×10⁹/l; 61% boys; 83% B-cell precursor ALL, 10% T-ALL, 7% unknown; 1.6% BCR-ABL, 0.1% MLL-AF4, 21% TEL-AML, 22% hyperdiploid. Based on morphology, 43% was CNS1, 41% CNS2, 1.9% CNS3, 11.3% TLP+ and 2.6% TLP−(ery’s>10×10⁶/l without blasts). Frequencies with TdT were similar, but of the 201 patients with morphological CNS1, 30 (15%) were TdT positive; and of 259 patients with morphological CNS2, 121 (47%) were TdT positive (p<0.001). Median follow-up was 4 years. 5-year pEFS (morph.) was: CNS1 84%, CNS2 80%, CNS3 71%, TLP+ 70% and TLP− 100% (p=0.008), with a trend for pOS (p=0.07). With longer follow-up the difference between CNS1 and -2 became larger (8-yr. pEFS 82 vs. 71%; p=0.08). The 5-yr. cumulative incidence of relapses (CIR) were 12%, 16%, 22%, 21% and 0%, respectively (p=0.067). However, if only CNS-relapses (isolated or combined) were considered, CIR rates were 2%, 4%, 15%, 7% and 0% (p=0.025), whereas CIR rates did not differ for other relapses (p=0.48). At multivariate analysis (including sex, age, RG, IPT, cytogenetics, CNS3 and TLP+), TLP+ was an independent risk-factor with a RR of 1.65 (1.01–2.69; p=0.035) for pEFS, and 1.75 (1.01–3.05, p=0.046) for pOS. When WBC was added to the model, TLP+ lost its significance, but the RR remained similar (RR1.48; p=0.11). TLP+ patients had higher WBC than other patients (WBC >50×10⁹/l: 34 vs 21%; p=0.014). pEFS for CNS1 and -2 utilizing TdT-staining did not differ (82 vs. 80%, p=0.87), nor did the CIR. Combined morphology and TdT-data did not have prognostic significance (p=0.53). In conclusion: a) TLP+ is an independent risk factor, even on a highly DEX based regimen, and is related to high WBC; b) the incidence of CNS2 in this study was relatively high; c) there was a trend for poor outcome for CNS2 with longer follow-up; d) there is a discrepancy between morphological and TdT-based CSF classification for CNS1 and -2, but morphology seems superior for prognostification. Poor outcome for TLP+ patients has now been documented in several studies using different treatment regimens. Future studies should show whether intensified CNS-directed treatment is able to reduce the 7% CIR involving the CNS in TLP+ patients.