Genetic risk model to predict outcomes for AML treated with HMA and venetoclax

Abstract Introduction Existing genetic risk models in acute myeloid leukemia (AML) were developed primarily in patients treated with intensive chemotherapy. Hypomethylating agents (HMAs) combined with venetoclax have become the standard of care for older or unfit patients, and data in younger populations are emerging. However, a genetic risk model specific to patients treated with HMA and venetoclax is lacking. Methods We retrospectively analyzed adult AML patients (≥18 years) diagnosed and treated at the University of Kentucky Markey Cancer Center between January 2015 and June 2025. Eligibility required receipt of HMA and venetoclax induction. Demographic information, genetic profiles, and treatment details were abstracted from medical records. AML was defined per WHO 2016 criteria. NGS was performed using a CLIA-certified 97-gene panel with 5% VAF detection threshold. Responses were assessed per ELN 2022 criteria. Outcomes included composite remission (CR/CRi), relapse, relapse-free survival (RFS), and overall survival (OS). RFS and OS were estimated by Kaplan–Meier. Cox proportional hazards models identified predictor variables for RFS and OS. Variables with significant associations were used to construct an additive genetic risk score, assigning points based on the scaled logarithm of hazard ratios (log[HR]). Patients were stratified into low-, intermediate-, and high-risk groups. Statistical analysis was performed using R version 4.4.1. Results A total of 107 patients were included, with a mean age at diagnosis of 67 years (range: 22–89). The cohort was primarily non-Hispanic white (92%), and 56% were male. 70 patients (66%) had de novo AML, 25 (23%) had antecedent MDS/MPN, and 12 (11%) had therapy-related AML. Cytogenetically, 45 patients (43%) had a normal karyotype. 25 patients (23%) had a complex karyotype, 13 patients (12%) had trisomy 8, and 9 (8%) had core-binding factor AML. Mutations were present in 89% of patients. The most common mutations were TET2 (25%), TP53 (21%), NPM1 (20%), SRSF2 (19%), RUNX1 (16%), DNMT3A (16%), IDH2 (13%), FLT3 (12%), ASXL1 (10%), BCOR (10%), and NRAS (10%). Among FLT3 mutations, 7% were FLT3-TKD and 5% were FLT3-ITD. CR/CRi was achieved in 71% of patients, 5 of whom eventually underwent stem cell transplantation. The median time to transplant was 4.9 months, following a median of 4 cycles of HMA and venetoclax. Over a median follow-up of 21.7 months, 43% of responders relapsed. The median RFS was 12.2 months (95% CI: 11.1–19.3) among responders. The median OS for the entire cohort was 12.2 months (95% CI: 10.2–20.9), with 40 patients (37%) alive at the time of analysis. Using Cox proportional hazards models, NPM1 mutation was the only mutation associated with improved RFS (HR 0.17; p=0.004), while ASXL1 (HR 2.61; p=0.01), SRSF2 (HR 2.37; p=0.04), and TET2 (HR 2.36; p=0.03) predicted reduced RFS. Complex karyotype was the only chromosomal profile significantly associated with RFS (HR 2.76; p=0.02). For OS, NPM1 mutation (HR 0.36; p=0.03) and IDH2 mutation (HR 0.74; p=0.05) predicted prolonged OS, as did normal karyotype (HR 0.54; p=0.02). Conversely, complex karyotype (HR 2.56; p=0.001), TP53 mutation (HR 2.03; p=0.02), and TET2 mutation (HR 2.06; p=0.01) were identified as significant negative predictors of OS. The variables that demonstrated significant association with RFS and OS were confirmed through multivariate analysis. A genetic risk score was developed using scaled log(HR) values for these variables. Points were assigned as follows: complex karyotype (+3), TET2 mutation (+2), TP53 mutation (+2), ASXL1 mutation (+1), SRSF2 mutation (+1), IDH2 mutation (-1), normal karyotype (-2), and NPM1 mutation (-4). Patients were stratified into low-risk (<0), intermediate-risk (0–3), and high-risk (>3) groups, each of which contained 35-36 patients. This model showed significant survival discrimination (Chi-square test, p=0.004). Median OS for the high-risk group was 5.9 months, 11.8 months for the intermediate-risk group, and 19.2 months for the low-risk group. Conclusion In patients with AML treated with HMA and venetoclax, co-genetic profiles strongly influence survival outcomes. NPM1 and IDH2 mutations, along with normal karyotype, were favorable, while TET2, ASXL1, SRSF2, and TP53 mutations, along with a complex karyotype, conferred adverse risk. Our novel genetic risk model effectively stratified patients into prognostically distinct groups and may inform individualized treatment strategies.