Does venetoclax-based therapy beat hypomethylating agents in unfit newly diagnosed AML?

A real-world test of the VIALE-A survival benefit

The trial that established venetoclax plus azacitidine for unfit AML, VIALE-A, excluded patients with ECOG performance status 3 or higher, significant organ dysfunction, and prior hypomethylating agent exposure. Whether its survival benefit reproduces in an unselected population has been uncertain: real-world series have reported lower response rates, and a meta-analysis of 24 studies could not show a statistically significant survival benefit. This study asks two questions: does the VIALE-A benefit reappear in a large real-world cohort, and which clinical and molecular subgroups actually gain from adding venetoclax?

The authors analyzed the PETHEMA registry, which collects consecutive patient data across affiliated institutions in Spain, Portugal, Argentina, Colombia, and Chile. They included 2,610 adults with newly diagnosed non-promyelocytic AML who were judged ineligible for intensive chemotherapy by their treating physician and received first-line HMA monotherapy (1,681 patients) or a venetoclax-based combination (929 patients). Patients treated in front-line clinical trials were excluded.

The two groups differ in ways that matter for interpretation. Venetoclax-treated patients were slightly younger (median 74 vs 75 years), more often had ECOG 0–2 (95.3% vs 90.2%), and more often had white blood cell counts of at least 15 × 10⁹/L (48.6% vs 28.2%). Secondary AML (42.5% vs 41.3%), prior HMA exposure (3.5% vs 2.8%), cytogenetic risk, and the prevalence of NPM1, FLT3-ITD, IDH1, IDH2, and TP53 alterations were balanced. The largest imbalance was calendar time: 99.4% of venetoclax patients were diagnosed from 2019 onward versus 37.0% of HMA patients.

Composite remission counts three response states

The primary endpoint was overall survival, measured from treatment initiation to death from any cause, with living patients censored at last follow-up. Response was assessed locally at each center using modified 2022 European LeukemiaNet recommendations, without central adjudication, which reflects routine practice but means response calls were not independently verified.

The key response measure is composite complete remission, or CRc, which combines three states: complete remission (CR), complete remission with incomplete hematologic recovery (CRi), and morphological leukemia-free status (MLFS). CRc is therefore broader than true CR and should not be read as equivalent to it. The paper reports the components separately, which allows readers to see where the difference came from.

Response data were available for 2,145 of 2,610 patients (82.2%), so response comparisons rest on a smaller denominator than the survival analyses. Survival analyses used the full cohort, with propensity score matching, era-restricted analysis, and multivariable Cox regression as checks on the main comparison.

Remission more than tripled and early deaths fell

CRc was 64.4% with venetoclax-based therapy (513/797 evaluable) versus 20.0% with HMA alone (270/1,348; P < 0.001). The gap came mainly from true CR, which rose from 15.1% to 48.4%, with smaller contributions from CRi (11.8% vs 4.7%) and MLFS (4.1% vs 0.1%). Refractory disease was less frequent with venetoclax (14.2% vs 46.6%), while partial remission was similar (10.4% vs 12.6%; P = 0.127). Death before response assessment occurred in 11.0% of venetoclax patients versus 20.8% of HMA patients.

Early mortality moved in the same direction. Death within 30 days occurred in 4.8% of venetoclax patients (45/929) versus 7.5% of HMA patients (126/1,681; P = 0.011), and death within 60 days in 9.7% (90/929) versus 15.7% (264/1,681; P < 0.001). This is notable because venetoclax is associated with hematologic toxicity; the authors interpret the lower early mortality as evidence of an early benefit driven by disease control rather than by tolerability.

Within venetoclax regimens, azacitidine plus venetoclax produced the highest CRc (65.8%), followed by decitabine plus venetoclax (58.1%) and low-dose cytarabine plus venetoclax (30.8%; P < 0.001). That ordering is confounded by indication: the decitabine and low-dose cytarabine subgroups contained far more secondary AML (70.3% and 96.1% vs 39.4%) and prior HMA exposure (31.6% and 50.0% vs 2.0%). Among HMA monotherapy, azacitidine and decitabine did not differ significantly in CRc (20.6% vs 17.0%; P = 0.552).

The survival hazard ratio matches VIALE-A

Median overall survival was 11.4 months with venetoclax-based therapy (95% CI 10.3–13.4) versus 8.2 months with HMA alone (95% CI 7.4–9.1), giving a hazard ratio of 0.66 (95% CI 0.60–0.73; P < 0.001). One-, two-, and three-year survival rates were 48.7%, 31.4%, and 21.4% versus 37.9%, 15.5%, and 8.7%. Median follow-up was 25.3 months.

The hazard ratio of 0.66 is identical to the one reported in VIALE-A, even though this cohort included patients VIALE-A excluded: ECOG 3–4 (4.7%), prior HMA exposure (3.5%), adverse-risk cytogenetics (36.5%), and secondary AML (42.5%). The absolute median survival, however, was lower than VIALE-A's 14.7 months, a gap the authors attribute to the broader, less selected population.

Three checks support the main comparison. Propensity score matching produced the same hazard ratio of 0.66. An era-restricted analysis comparing venetoclax patients diagnosed 2021–2025 with HMA patients diagnosed 2010–2020 gave a hazard ratio of 0.71 (95% CI 0.63–0.80). Censoring patients at the date of allogeneic transplant did not materially change the benefit (HR 0.67; 95% CI 0.61–0.75). These adjustments reduce, but cannot eliminate, the concern that the calendar-time imbalance between groups drives part of the difference.

Benefit is largest in IDH2, NPM1, and favorable cytogenetics

The survival benefit was not uniform. The largest relative reductions in death risk appeared in IDH2-mutated AML (median 24.3 vs 8.0 months; HR 0.43; 95% CI 0.28–0.65), NPM1-mutated AML (15.4 vs 6.2 months; HR 0.46), FLT3-ITD-positive AML (11.1 vs 4.9 months; HR 0.48), and IDH1-mutated AML (21.0 vs 11.9 months; HR 0.55). By cytogenetic risk, the benefit was greatest in favorable-risk disease (median not reached vs 7.7 months; HR 0.25), then intermediate (15.4 vs 11.6 months; HR 0.65) and adverse risk (8.7 vs 5.8 months; HR 0.72).

The FLT3-TKD subgroup did not reach significance (HR 0.57; 95% CI 0.29–1.12; P = 0.098). The authors note that the significant FLT3-ITD result contrasts with the non-significant subgroup effect in VIALE-A and attribute the difference to greater statistical power in this larger sample.

TP53-altered AML showed a significant but modest survival benefit (7.0 vs 3.7 months; HR 0.70; 95% CI 0.54–0.91; P = 0.007), and survival remained poor in absolute terms. The benefit was concentrated in NGS-detected TP53 mutations (HR 0.67; 95% CI 0.49–0.92), while FISH- or karyotype-only alterations such as del(17p) showed no significant difference (HR 0.80; P = 0.337). This distinction matters because TP53 positivity in the study was defined as either an NGS mutation or a FISH/karyotype abnormality, and the two detection routes did not behave the same way.

No survival gain in AML arising from MDS or MPN

The benefit was present in de novo AML (median 13.8 vs 9.0 months; HR 0.60; 95% CI 0.52–0.70) and in secondary AML overall (HR 0.77; 95% CI 0.66–0.90), but the secondary AML result was not uniform across etiologies. When split by origin, the benefit was confined to therapy-related AML (HR 0.60; 95% CI 0.45–0.80). There was no significant improvement in AML arising from MDS (HR 0.88; P = 0.202), MDS/MPN overlap (HR 1.10; P = 0.723), or MPN (HR 0.86; P = 0.561).

This is the paper's most consequential negative finding. The authors describe MDS-AML and MPN-AML as the most difficult-to-treat population in this setting and highlight the lack of substantial benefit there. The CRc comparison points the same way: the smallest absolute remission differences were in MDS/MPN-AML (53.8% vs 24.0%) and TP53-mutated AML (50.4% vs 18.7%), even though both comparisons were statistically significant.

Prior HMA exposure also marked a boundary. Venetoclax-based therapy conferred a clear survival advantage in HMA-naïve patients (HR 0.64; 95% CI 0.56–0.74) but not in the 49 patients with prior HMA exposure (HR 0.78; P = 0.472), a small subgroup that limits how firmly this null result can be interpreted.

Longer survival came from more remissions, not longer remissions

Disease-free survival, measured from first CRc to relapse or death, was comparable between groups: 10.3 months with venetoclax-based therapy versus 11.5 months with HMA alone (HR 0.88; 95% CI 0.74–1.05; P = 0.148). The only significant disease-free survival benefit appeared in MRC favorable-risk patients (HR 0.10; 95% CI 0.02–0.38).

Read together with the remission and mortality results, this shapes how the survival gain should be understood. The authors argue that the overall survival advantage is primarily driven by the higher CRc rate and lower early mortality, coupled with equal disease-free survival among responders. In other words, more patients reached remission and fewer died early, rather than responders staying in remission longer.

Transplant access differed: among 622 CRc patients with transplant data, 8.5% proceeded to allogeneic transplant in first CRc, 11.1% in the venetoclax group versus 4.9% in the HMA group (P = 0.010). Among transplanted patients, overall survival did not differ significantly between groups (HR 1.83; P = 0.341), a small and underpowered comparison.

Molecular profile should guide who gets venetoclax

The practical message is a selection rule rather than a blanket endorsement. The largest relative survival gains in this cohort were in IDH2-mutated, NPM1-mutated, FLT3-ITD-positive, IDH1-mutated, and favorable-risk cytogenetic AML, where hazard ratios ranged from 0.25 to 0.55. TP53-altered disease gained significantly but remained at 7.0 months median survival, and AML arising from antecedent MDS or MPN showed no significant survival benefit at all.

Two boundaries should travel with those numbers. First, the comparison is retrospective and non-randomized, and the calendar-time imbalance was large; the era-restricted hazard ratio of 0.71 is the more conservative estimate of the treatment difference. Second, response was adjudicated locally without central review, and molecular data were available for roughly 60% of patients, so subgroup hazard ratios rest on smaller and unevenly tested groups.

Within venetoclax-based regimens, azacitidine plus venetoclax was the dominant backbone (93.0% of the venetoclax cohort) and was associated with better survival than decitabine plus venetoclax (HR 1.48; 95% CI 1.01–2.19) or low-dose cytarabine plus venetoclax (HR 1.98; 95% CI 1.29–3.04), though the non-azacitidine subgroups were enriched for secondary AML and prior HMA exposure. The paper's own priority is therefore to refine patient selection: expect the most from venetoclax in IDH2-, NPM1-, FLT3-ITD-, and IDH1-mutated or favorable-risk disease, and do not assume a survival gain in AML evolving from MDS or MPN.

Sources used in this answer