NPM1 and IDH1/2 mutations show limited prognostic impact in relapsed AML: Evidence from the avalon cohort

Abstract Background: NPM1 and IDH1/2 mutations are established prognostic biomarkers in newly diagnosed (ND) acute myeloid leukemia (AML) treated with venetoclax (VEN) and hypomethylating agents (HMAs). However, their role in relapsed/refractory (R/R) AML treated with VEN+HMA remains unclear. The AVALON study is a multicenter real-world cohort including both ND and R/R AML patients. We aimed to investigate the prognostic impact of NPM1, IDH1, and IDH2 mutations in both disease settings. Methods: The study population included AML patients from AVALON, both R/R and ND cases, with available molecular testing for IDH1, IDH2, or NPM1 mutations. Overall response rate (ORR, calculated in patients with available response assessment), event-free survival (EFS), duration of response (DOR), and overall survival (OS) were compared between mutated versus wild-type patients. An additional analysis was performed in patients from both settings who were tested for all three genes. Results: Among 190 AML patients, 147 had R/R disease and 43 were ND. In the R/R cohort, 114 (78%) underwent molecular profiling for at least one mutation: IDH1 (n=50), IDH2 (n=56), NPM1 (n=91). IDH1 mutations were found in 3/50 patients (6%). ORR was 2/3 (67%) in mutated versus 20/41 (49%) in wild-type (p=1.00); median EFS was 2.4 vs 6.9 months (p=0.37), DOR 1.0 vs 4.7 months (p=0.24), and OS 2.4 vs 8.9 months (p=0.18).IDH2 mutations were identified in 10/56 patients (18%); ORR was 5/8 (63%) vs 18/41 (44%) in wild-type (p=0.45); median EFS was 11.3 vs 5.4 months (p=0.51), DOR 6.5 vs 4.7 months (p=0.87), and OS 11.3 vs 6.2 months (p=0.56). NPM1 mutations were present in 12/91 patients (13%); ORR was 6/11 (55%) vs 37/73 (51%) in wild-type (p=1.00); median EFS was 5.4 vs 6.3 months (p=0.77), DOR 6.8 vs 8.3 months (p=0.13), and OS 6.0 vs 9.1 months (p=0.72). In the subset of 55 R/R patients with complete profiling for NPM1, IDH1, and IDH2, patients with at least one mutation (n=25) had an ORR of 13/22 (59%), median DOR of 6.5 months, EFS of 8.9 months, and OS of 8.9 months. Triple-wild-type patients (n=30) had an ORR of 12/27 (44%), DOR of 6.5 months, EFS of 8.0 months (p=0.84), and OS of 9.7 months (p=0.99), with no statistically significant differences between groups. In the 43 ND AML patients, NPM1 or IDH1/2 mutated patients (n=9) showed higher response rates and longer survival compared to triple-wild-type cases (n=7). ORR was 7/8 (88%) vs. 3/6 (50%) (p=0.12). Median DOR was 13.1 vs 3.2 months (p=0.01), EFS 20.1 vs 5.1 months (p=0.006), and OS 25.6 vs 6.3 months (p<0.001). Conclusions: In this real-world cohort of AML patients treated with VEN+HMA, NPM1, IDH1, and IDH2 mutations retained their favorable prognostic value in the newly diagnosed setting but were not predictive of improved outcomes in the relapsed/refractory setting. In R/R AML, response rates and survival did not significantly differ between mutated and wild-type patients. These findings suggest that the prognostic value of commonly favorable mutations is context-dependent and may diminish after disease relapse, highlighting the need for dynamic and phase-specific prognostic models in AML.