The aim of this study was to evaluate how comorbidities and molecular landscape relate to outcome in patients with acute myeloid leukemia (AML) aged 60 years or older who received intensive induction therapy. In 91 patients, 323 mutations were identified in 77 genes by next-generation sequencing, with a median of four mutations per patient, with NPM1, FLT3, TET2, and DNMT3A being the most frequently mutated genes. A multistate model identified FLT3, IDH2, RUNX1, and TET2 mutations as associated with a higher likelihood of achieving complete remission while STAG2 mutations were associated with primary refractory disease, and DNMT3A, FLT3, IDH2, and TP53 mutations with mortality after relapse. Ferrara unfitness criteria and performance status were the best predictors of short-term outcome (area under the curve = 82 for 2-month survival for both parameters), whereas genomic classifications better predicted long-term outcome, with the Patel risk stratification performing the best over the 5-year follow-up period (C-index = 0.63 for event-free and overall survival). We show that most genomic prognostic classifications, mainly used in younger patients, are useful for classifying older patients, but to a lesser extent, because of different mutational profiles. Specific prognostic classifications, incorporating performance status, comorbidities, and cytogenetic/molecular data, should be specifically designed for patients over 60 years.
© 2023 Wiley Periodicals LLC.