Impact of G-CSF in AML patients treated with venetoclax and azacitidine: A dataml registry study

Abstract Introduction The venetoclax-azacitidine (VEN-AZA) combination has become a standard treatment for AML in older patients (pts) or those ineligible for intensive chemotherapy (IC). While the VIALE-A registration study showed a median survival of 14.7 months (m), real-life studies did not to reproduce this result. In this study, we analyzed treatment patterns, adverse events, responses and outcomes of pts treated with VEN-AZA in the DATAML registry.MethodsMain inclusion criteria: age ≥ 18 y; newly diagnosed AML, de novo or secondary to cytotoxic therapy, MDS, CMML or MPN; IC ineligible; at least 1 cycle of VEN-AZA between 01/01/2020 and 12/31/2022. Exclusion criteria: previous exposure to VEN; participation in a clinical trial within 30 days prior to VEN.Pts were treated in 2 university and 21 regional hospitals. VEN-AZA was recommended according to the label, but bone marrow blast clearance at D21C1 allowed VEN to be stopped before D28. Tumor lysis syndrome, antimicrobial prophylaxis and G-CSF use were carried out according to local practices. Intervals between each cycle, duration of VEN treatment, AZA dose modification, use of antifungal and/or antibiotic prophylaxis, use of G-CSF, transfusion needs, site of treatment and hospitalizations were analyzed during the first 6 cycles.Results199 pts were included: median age, 75.6 y; male sex, 56%; PS 0-1, 72%; secondary AML, 49% (including 11% post-MPN); other cancer <5y, 12%; infection at diagnosis, 18%; WBC, 4.6 G/L, WBC≥25 G/L, 20%; adverse cytogenetics, 44%; mutations: TP53 (26%), NPM1 (17%), IDH1 (13%), NRAS (12%), IDH2 (11%), KRAS (7%), FLT3-ITD (6%);refined ELN2024 risk, favorable (29%), intermediate (18%), adverse (53%). 18 pts (9%) had received AZA for prior MDS.The first cycle was performed on an outpatient basis in 40% of pts. Antifungal or antibiotic prophylaxis was given in 79% and 32% of pts, respectively. G-CSF was given in 33% of pts at C1 then increasingly, up to 71% at C6. Pts received a median of 4 cycles (8 if CR/CRi, 2 if treatment failure), and 37% received > 6 cycles. The median number of venetoclax days per cycle was 24, 24, 21, 21, 14, 14 in C1 to 6. Median interval between 2 cycles was 33 days. Beyond C6, dose/duration reductions concerned 55% of pts for VEN and 43% for AZA.The rate of febrile neutropenia was 44%, 32%, 9%, 14%, 10% and 5% during C1 to 6. Death rate at D30 and D60 was 3% and 14%. Death rate from infection at D30 and D60 was1% and 6%.CR+CRi rate was 58% (CR, 32%, CRi, 26%). At D21C1, CR, CRi and MLFS were observed in 6%, 10% and 36% of pts. In pts with D21C1 BM blasts<5%, the CR/CRi rate was 86% after 2 cycles. Only 11% of pts received a second line after failure or relapse.With a median follow-up of 27 m, median OS was 8.9 m (IQR, 3.5-22.5), with significant variations according to mutations (NPM1, 28 m; IDH1/2, 20 m; TP53, 4.6 m; secondary-type mutations, 10 m) or refined ELN2024 risk (fav, 20.7 m; int, 11.9 m; adv, 5.1 m). In multivariate analyses (MV), secondary AML (HR 1.67), adverse cytogenetics (HR 1.51), IDH1mut (HR 0.54) and TP53mut (HR 2.34) were significantly associated with OS. For CR/CRi, only adverse cytogenetics (HR 0.36) was significant.Since a recent study showed a prolonged OS in CR/CRi pts who received G-CSF (DiNardo C, Am J Hematol 2024), we focused on this population. 17 CR/CRi pts did not and 90 did receive G-CSF during the first 6 cycles (n=45 at C1 and n=45 > C1). Median OS was 10.1 m (2.8-13.8) in pts without G-CSF vs 20.3m (8.4-39.5) in pts with G-CSF (p=0.001). G-CSF use was significantly associated with better OS in MV (HR, 0.41; 95% CI 0.23-0.73; p=0.003). Moreover, the number of G-CSF cycles was also significantly associated with better OS in MV (HR for each cycle from 0 to 6, 0.77; 95% CI 0.67-0.88; p<0.001). G-CSF use from C1 onwards was also associated with a higher rate of MRD negativity by flow cytometry in CR/CRi pts (63% vs 18%, p=0.017).ConclusionThis study confirms a markedly improved response rate compared to historical experience with AZA alone and confirms a major OS benefit in some subgroups (NPM1, IDH1/2). However, real-life experience has led to the selection of a more severe patient population negatively affecting OS compare to VIALE-A. Although the mechanisms by which G-CSF could improve treatment efficacy have not been established and a randomized study would be necessary, our study strongly suggests that G-CSF may improve response and survival in patients treated with VEN-AZA.