Real-World Multicenter Comparison of Venetoclax and Azacitidine Versus Intermediate or High Dose Cytarabine-Based Salvage for Patients with AML in First Relapse after Front Line Intensive Chemotherapy. a Dataml-IPC-Auraml Consortium Study

Introduction: Prognosis in relapsed acute myeloid leukemia (AML) patients (pts) is dismal, with no standard chemotherapy. Intermediate (I) or high (H)-dose cytarabine (DAC)-based regimens are common salvage treatments (Döhner, Blood 2022) leading to around 50% CR/CRi (Megías-Vericat, AJH 2018). Venetoclax (VEN)-azacitidine (AZA) is a front-line standard of care for unfit pts but is also widely used in pts with molecular relapse, primary induction failure (PIF) or morphologic relapse (Rel). In this study, we report the largest cohort of pts with AML treated with VEN-AZA compared to salvage I/HDAC-based intensive chemotherapy (IC) in first Rel.

Methods: Inclusion criteria were 18-75y old AML pts, first Rel after front line IC, treated by I/HDAC-based IC or VEN-AZA between January 2015 and September 2023. Exclusion criteria were 3^rd drug added to VEN-AZA, molecular Rel, CR1/CRi1 after 2 induction courses.

Results: 347 pts from 9 centers treated with VEN/AZA (n=125, 36%) or I/HDAC (n=222, 64%) were included. 191 (55%) were male, median age was 63y (IQR 54.4-70; range 20-75) and 57y (IQR 45-63; range 19-73) in VEN-AZA and I/HDAC groups (p<0.0001), respectively; 29 (23.4%) and 15 (6.8%) pts had secondary AML, cytogenetic risk was intermediate in 83 (68%) and 172 (77.5%) (p=0.03); 17 (14.3%) and 82 (41%) pts had a NPM1 mutation (p<0.0001), in VEN-AZA and I/HDAC groups, respectively. Induction chemotherapy was 3+7 including 15 (4.3%) CPX-351. All pts were in CR1/CRi1 and post remission strategy was I/HDAC-based for 245 (74.7%) pts whereas other pts received non-IC. Finally, 49 (39.2%) and 39 (17.5%) pts underwent an allogeneic stem cell transplantation (HSCT) in CR1/CRi1 (p<0.0001), in VEN-AZA and I/HDAC groups, respectively.

All pts relapsed after a median delay of 12 months (IQR 7-22; range 1-226), 73 (21%) pts ≤6 months and 274 (79.0%) pts >6 months. Performans status (PS) at relapse was 0-1 in 247 (86.7%) pts and ≥2 in 38 (13.3%) pts. Median WBC at relapse were 3x10⁹/l (IQR 1.9-7.1; range 0.3-265.7). There was no significant difference between the 2 groups for these characteristics.

Finally, at relapse, 76 (60.8%) and 147 (66.2%) pts obtained a CR2/CRi2 in VEN-AZA and I/HDAC-based salvage groups, respectively (p=0.35). Pts received a median of 2 cycles of VEN-AZA (IQR 2-5; range 1-21), best response has been considered whatever the time. Early death during first month occurred in 18 (14.4%) and 20 (9.0%) pts in VEN-AZA and I/HDAC groups, respectively (p=0.15). Bridge to transplant has been obtain for 31 (24.8%) and 120 (54.1%) pts in VEN-AZA and I/HDAC group, respectively (p<0.0001). Multivariate logistic regression for factors associated with CR2/CRi2 included gender, sAML, cytogenetic, FLT3-ITD and NPM1, consolidation in CR1/CRi1, previous HSCT, age at Rel, VEN-AZA or I/HDAC, PS at relapse. Independent factors associated with CR2/CRi2 were female (OR 0.55, CI95% 0.32-0.94, p=0.03) and PS 0-1 at Rel (OR 0.35, CI95% 0.17-0.76, p=0.008).

After a median FU for alive pts from Rel of 21 months, median OS was 15 and 16 months; 1y-OS was at 51.7% and 55.3%, in VEN-AZA and I/HDAC groups (p=0.19), respectively. Cox model for factors associated with OS included also time to relapse and HSCT in CR2/CRi2 and found secondary AML (aHR 1.73, CI95% 1.18-2.56, p=0.006), adverse cytogenetic risk (aHR 1.52, CI95% 1.07-2.16, p=0.02), time to relapse >6 months (aHR 0.38, CI95% 0.28-0.52, p<0.0001) and HSCT in CR2/CRi2 (aHR 0.28, CI95% 0.21-0.39, p<0.0001). Median LFS was 14 months in both groups; 1y-LFS was at 55.5% and 52.9%, in VEN-AZA and I/HDAC groups (p=0.95), respectively. Cox model for factors associated with LFS were time to relapse >6 months (aHR 0.37, CI95% 0.23-0.59, p<0.0001) and HSCT in CR2/CRi2 (aHR 0.39, CI95% 0.27-0.57, p<0.0001).

Conclusion: This cohort is characterized by first relapse including three quarters of relapse > 6 months, no PIF, no previous exposure to HMA for AML and mostly intermediate risk cytogenetics. Salvage treatment allowed high CR2/CRi2 rates, close to those observed in front-line for VEN-AZA and at the upper limit for salvage treatment based on I/HDAC. These high response led to interesting bridge to transplant rates in both groups and long LFS and OS. These results raise the question of prospective evaluation of non-intensive chemotherapy in first relapse for AML patients.