Venetoclax–Azacitidine Versus Azacitidine for the Treatment of Primary Refractory or First Relapsed Acute Myeloid Leukemia. An IPC-DATAML-MSKCC Retrospective Study

Abstract

Relapsed or refractory (R/R) AML have a poor outcome when treated with intensive chemotherapy (≈30% response rate) [1], or azacitidine (AZA, ≈20% response rate) [2-5]. Both strategies are associated with a poor overall survival (OS) ranging between 6 and 9 months. Venetoclax (VEN)–AZA has been approved for newly-diagnosed AML patients ineligible for IC [6]. Off-label VEN–AZA is frequently used in patients with R/R AML as an alternative to classical intensive salvage regimens or to single agent AZA. However, comparative studies are scarce and patients cohorts are often monocentric and limited in number [7, 8].

We conducted a multicentric retrospective analysis comparing VEN–AZA versus AZA in R/R AML. We extracted data from two French registries (Institut Paoli Calmettes [IPC, Marseille] and DATAML [Toulouse-Bordeaux]), and from one US registry (MSKCC, [New York]) comprising patients treated with VEN–AZA in first morphological relapse, or primary refractory to one or two cycles of IC, between 09-2017 and 12-2023. We compared this dataset with a historical cohort of AML treated with AZA between 2010 and 2022 [5, 7]. All registries have been locally approved (IPC IRB n° 2003-010; DATAML registration number at the CNIL (N°915285) and CCTIRS (N°15.319), MSKCC, protocol #18-258). We used ELN 2022 criteria to assess response [9]. We measured Odds Ratio (OR) to compare complete response (CR) and CR with incomplete hematological recovery (CRi) rates and Hazard Ratio (HR) to compare OS, in selected subsets of AML. COX regression was performed, taking factors significantly associated with OS in univariate analyses.

Among the 344 patients included, 134 were treated with VEN–AZA and 210 with AZA monotherapy. The median age at AML diagnosis was 66.5 (ranges, 20–80) in the VEN–AZA group and 65 (ranges, 18–80) in the AZA group (p = 0.31). Approximately 80% have a PS score equal to one or two. Cytogenetics at diagnosis was adverse, intermediate and favorable in 41 (32%), 83 (64.8%), and 4 (3.1%) patients in the VEN–AZA group versus 75 (35.9%), 133 (63.6%), and 1 (0.5%) in the AZA group (p = 0.12), respectively. Full molecular assessment was available for 109 (81.3%) patients in the VEN–AZA cohort and 65 patients (31.0%) in the AZA cohort. NPM1 mutations were found in 17.4% of the VEN–AZA group vs. 13.8% in the AZA group (p = 0.42). AML in the VEN–AZA group had significantly less FLT3-ITD (6.5% vs. 15.5%, p = 0.02), and more frequent IDH1/2 (27.7% vs. 14.7%, p = 0.02), TP53 (18.2% vs. 7.4%, p = 0.04), N/KRAS genes (18.3% vs. 4.6%, p = 0.01) mutation (mut), or for any gene classified as myelodysplasia related genes (MRG, 41.3% vs. 6.2%, p < 0.01) [9].

Induction consisted of cytarabine plus anthracycline (n = 311, 90%) or etoposide (n = 7, 2%), or CPX-351 (n = 26, 8%). About 33% versus 35% patients were primary refractory to induction whereas 67.2% and 64.3% relapsed (p = 0.92), in the two groups, after a median CR duration of 13 months (ranges, 4–228) versus 11 months, (ranges, 2–84, p = 0.094) in the VEN–AZA and AZA groups, respectively. Thirty-one patients (23.1%) previously underwent an allogenic stem-cell transplantation (allo-SCT), in the VEN–AZA group versus 60 (28.6%) in the AZA group (p = 0.26). The total median number cycles of VEN–AZA or AZA salvage therapy was 2 (ranges, 1–21) versus 4 (ranges, 1–40, p = 0.049).

Thirty-five patients were not evaluable, mainly in the AZA group, because of early premature death (n = 31) or loss of follow up (n = 4). At last follow-up, 79 patients (59%) and 184 (87.6%) have died in the VEN–AZA and AZA groups, respectively (p < 0.001). Day-30 and Day-60 mortality was 5.9% and 7.5% in the VEN–AZA group versus 9.0% and 15.7% in the AZA group (p = 0.411 and 0.029, respectively).

Regarding the 309 evaluable patients, composite complete remission rate (cCR) was 55% (CR = 42%, CRi = 13%) versus 21.9% (CR = 16.3%, CRi = 5.6%, p < 0.001) and overall response rate (ORR) was 70.2% versus 32.6% in the VEN–AZA and the AZA groups of patients (p < 0.001). Time to achieve best response was 31.5 days (ranges, 18–396) in the VEN–AZA group versus 89 days (ranges, 23–1168) in the AZA group, p < 0.001. An allo-SCT was performed in 31 patients (23.1%) versus 24 patients (11.7%, p = 0.005) and median number of cycles prior allo-SCT was 2 (ranges, 1–6) in the VEN–AZA group versus 5 (ranges, 1–15) in the AZA group (p < 0.001). Taking only the patients achieving CR/CRi, allo-SCT was performed in 25/72 patients (34.7%) in the VEN–AZA group versus 11/39 patients (28.2%) in the AZA group (p = 0.67).

With a median follow up of 32.4 months (ranges, 29.5–37.4) in the VEN–AZA group and 101 months (ranges, 56.8-non-reached [NR]) in the AZA group, median OS was 10.8 (ranges, 8.1–14.5) versus 6.8 months (ranges, 5.4–8.4, p = 0.013, Figure 1). Median OS from AML diagnosis date was 24.4 (ranges, 21.2–32.6) versus 19.2 months (ranges, 16.1–22.5, p = 0.001). Median OS for AML patients who undergone an allo-SCT following VEN–AZA was NR (ranges, 11.9-NR) in the VEN–AZA group versus 17.5 (ranges, 4–22.4) in the AZA group, whereas non-transplanted AML had a 9.2 (ranges, 6.5–11.9) versus 6.6 months (ranges, 5.1–8.3) median OS, respectively, p < 0.001, Figure 1.

FIGURE 1

Open in figure viewerPowerPoint

Left panel. Overall survival of patients treated with venetoclax (VEN) combined with azacitidine (AZA) or AZA monotherapy. Right panel. Overall survival of patients treated with VEN–AZA and AZA +/− allogeneic stem cell transplantation (allo-SCT).

Taking AML subsets separately, primary refractory patients had a 45% cCR rate when treated with VEN–AZA versus 22% with AZA (p = 0.01) whereas first relapsed AML had a 60% versus 22% cCR (p < 0.001). Response rates for patients relapsing after an allo-SCT were 43% versus 16% (p = 0.01), in the VEN–AZA and AZA groups, respectively, compared to 58% versus 24% in the non-transplanted group (p < 0.001). Poor-risk cytogenetics was associated with a 26% versus 11% cCR (p = 0.06) compared to 67% versus 27% in the intermediate/low-risk cytogenetics group (p < 0.01) in the VEN–AZA and AZA groups, respectively. Mutations in the TP53 gene were associated with a low response rate, in the two groups (11% and 25%, p = 0.3). On the other hand, IDHmut patients had a 76% versus 29% (p = 0.004) and NPM1mut patients had a 90% versus 47% response rate (p = 0.007), in the VEN–AZA and AZA groups, respectively.

Survival in primary refractory AML was 8.1 months (95% CI, 6.2–11), versus 7.13 months (95% CI, 5.8–11.4, p = 0.55), in the VEN–AZA and AZA groups, respectively, compared with 13.1 (95% CI, 9.2-NR) versus 6.8 months OS (95% CI, 4.8–10.8, p = 0.001), in the first relapsed AML group of patients. OS for patients relapsing after allo-SCT was 6.5 (95% CI, 2.5–11.9) versus 4.7 months (95% CI, 3.9–6.8, p = 0.58) in the VEN–AZA versus AZA groups, respectively. Poor-risk cytogenetics and TP53mut AML patients had a 4.1 and 1.9 months OS, in the VEN–AZA and AZA group, respectively, that was non-statistically different from OS in the AZA-treated patients (4.8 and 1.6 months, respectively). Consistently, IDHmut and NPM1mut patients had a longer OS when treated with VEN–AZA (18 months in both groups) and a lower OS when treated with AZA monotherapy (13.6 and 5.1 months, p = 0.27 and 0.02, respectively).

Finally, we performed a multivariate analysis, taking VEN–AZA salvage treatment, a PS score ≥ 2, an adverse cytogenetics, a TP53 mutations, and the realization of an allo-SCT consolidation (in a time-dependent manner) in the model. Factors associated with a significantly lower OS were PS score ≥ 2 (HR = 2.81, [95% CI, 0.28, 0.47], p < 0.001), adverse cytogenetics (HR = 3.01, [95% CI, 1.87, 4.85] p < 0.001), and TP53 mutation (HR = 1.93, [95% CI, 1.08, 3.45], p = 0.027). On the contrary, factors independently associated with a longer OS were VEN–AZA treatment (HR, 0.57 [95% CI, 0.36, 0.88], p = 0.012) and allo-SCT consolidation (HR, 0.48 [95% CI, 0.24, 0.92], p = 0.028).

The results from this large retrospective study shows that VEN–AZA salvage therapy is associated with a low early death rate (< 10%) and a high response rate (55%). It appears to be an efficient bridge to transplant strategy in approximately 25% of the patients. This translated into a long median OS in the VEN–AZA after allo-SCT (2-year OS, > 50%). VEN–AZA salvage treatment might be comparable to IC salvage therapy, even though response rated appeared lower (around 30% ORR), in recent phase 3 trials [1, 10]. Some attempts to combine VEN plus IC in the R/R setting has yielded interesting results that deserve to be confirmed in bigger cohorts.

Although outcomes are encouraging, not all the subsets of AML benefit from the addition of VEN to AZA in R/R AML. Primary refractory, prior allo-SCT, adverse cytogenetics and/or TP53mut AML have the same poor outcome when treated with VEN–AZA or AZA monotherapy. In these subsets, adding VEN to AZA must be discussed in a case-by-case manner. Indeed, the rapidity of achieving response with VEN–AZA may be taken into consideration to perform an allo-SCT rapidly. On the other hand, AZA can be a preferred choice for patients with a poor general status, and for whom quality of life and best supportive cares are the main objectives. This study has some limitations due to the low number of patients in the molecular subgroups and the locally assessed response evaluations. Moreover, no data are available on toxicities and VEN dose and number of days as well as the absence of minimal residual disease evaluations. Further prospective studies will help to better assess the benefits of VEN–AZA on R/R AML.

To conclude, our data shows that VEN–AZA is an effective salvage treatment with a low early mortality and a high response rates, prolonging OS for some subsets of AML failing intensive chemotherapy. This study paves the way for future prospective trials using VEN–AZA salvage backbone in R/R AML.