Paper Details

ABSTRACT

Introduction: Central nervous system (CNS) involvement in patients diagnosed with acute myeloid leukemia (AML) relapse has a rare occurrence and a very poor prognosis. Several clinical and biological features at diagnosis have been associated with a higher risk of CNS relapse: age ≤ 45 years, hyperleukocytosis, FAB M5 subtypes, inversion of chromosome 16 and KMT2Ar. While FLT3-ITD mutations have been associated with CNS involvement, the potential role of other molecular markers remains poorly explored. Moreover, there is no data on the potential risk of decreasing cytarabine consolidation intensity (from high-dose cytarabine [HDAC: 3g/m²/12h D1-3] to intermediate-dose [IDAC: 1.5g/m²/12h D1-3] or standard dose cytarabine [SDAC: 50 mg/m²/12h, D1-5, SC]).

Methods: All adult AML patients (except APL) treated with intensive chemotherapy registered in DATAML between 2000 and 2024 were included in this study. Initial characteristics, treatments and outcome were compared in CNS relapses (CNS+Rel) and a control group (Cont) composed of relapses without CNS involvement (CNS-Rel) and AML without relapse (NoRel). Intrathecal prophylaxis was performed in hyperleukocytic AML (especially inv16) or FAB M5 subtypes according to centers’ guidelines.

Results: 3720 cases were analysed, including 66 CNS+Rel and 3654 Cont (2044 NoRel and 1610 CNS-Rel). The incidence of CNS relapse was 1.77%.

AML characteristics and treatment: in comparison to Cont, CNS+Rel pts were younger (median age, 57 vs. 61 years), with 27% being <45 (vs. 17%). Sixty-two vs 54% were male and 88 vs. 79% had de novo AML. Half of them had an extra-medullary involvement (EMI) vs. 28.5%: hepatomegaly (15 vs. 7%), splenomegaly (12 vs. 8%), lymphadenopathies (24 vs. 11%), skin (8 vs. 5%), gingival hypertrophy (24 vs. 9%). Leukostasis was observed in 15 vs. 5%. Median WBC was 25 (IQR, 6-75) vs. 8 G/L (IQR, 3-39), with 30 vs. 21% > 50 G/L. Median LDH was 786 (IQR, 469-1722) vs 512 UI/L (IQR, 297-936) and median ferritin was 958 (IQR, 518-1992) vs 653 mg/L (IQR, 333-1279). FAB M5 presentation was observed in 30% vs. 15%. Inv(16) was present in 9 vs. 5%; KMT2Ar in 9 vs. 3% and trisomy 8 in 20 vs. 9%. CD56 was ≥20% in 37% vs. 20%. Some mutations were more frequent: FLT3-ITD (29 vs 21%), FLT3-TKD (19% vs. 8%), EZH2 (10% vs. 5%), JAK2 (8 vs. 3%), KIT (11 vs. 5%), KMT2A-PTD (8 vs. 5%), UBTF (8 vs. 3%) whereas other were less frequent: IDH2 (2% vs. 12%), ASXL1 (4 vs. 14%), DDX41 (0 vs. 5%), RUNX1 (8 vs 15%), SRSF2 (8 vs. 12%), TP53 (2 vs. 9%).

In CNS+Rel, first-line treatment was ‘7+3’, CPX-351 or other in 38/52/9/1.5 vs. 35/59/4/2%. Midostaurin was administered since 2018 in 15 vs. 6%. Consolidations consisted in IDAC, HDAC, CPX-351 or SDAC in 22/37/1/35 vs. 24/31/2/43%. AlloHSCT in CR1 was performed in 36 vs. 30%.

CNS+Rel characteristics, treatment and outcome: CNS+Rel occurred at 1^st, 2^nd, 3^rd or 4^th relapse in 74/20/5/2%. CNS+Rel was associated to BM involvement in 55% and/or with other EMI (skin: 8%, breast: 2%, brain or eye: 20%, testis: 4%). Median WBC was 7.9 G/L (IQR, 5.6-12.9). At CNS+Rel, there were NPM1 mutations in 31%, FLT3-ITD 12%, FLT3-TKD 13%, IDH1 12%, IDH2 6%. Ninety-two percent received systemic salvage; 86% intrathecal chemotherapy; 26% whole brain radiotherapy and 11% alloHSCT. A 2nd CR was obtained in 39% of 1^st CNS+Rel. The median OS from relapse of CNS+Rel was 4.8 months vs. 7.0 months in CNS-Rel (P=0.046), with a 2- and 5-year OS of 15 and 11% vs. 23 and 14%.

Predictive factors of CNS relapse: in univariate analyses, age <45 years, EMI at diagnosis, pulmonary leukostasis, LDH ≥500 UI/L, ferritin ≥900 UI/L, FAB M5 classification, KMT2Ar, trisomy 8, CD56≥20% and FLT3-TKD mutation were significantly associated to an increased risk of CNS+Rel. In multivariate analyses, female sex (HR, 0.42; P=0.024), EMI with lymphadenopathies (HR, 2.9; P=0.008), pulmonary leukostasis (HR4.1; P=0.001), KMT2Ar (HR, 5.2; P =0.003) and CD56 ≥20% (HR, 2.4; P=0.015) were independent predictive factors for CNS+Rel. There was no impact of treatment scheme, including the addition of midostaurin or consolidation intensity (HDAC vs. IDAC or SDAC).

Conclusions: This work describes predictive factors of CNS relapse (EMI, pulmonary leukostasis, KMT2Ar and CD56), that have to be taken into account for CNS prophylaxis in newly-diagnosed AML treated with intensive chemotherapy. We did not find an increased risk of CNS+Rel with IDAC or SDAC consolidation respective to HDAC.