Prediction of survival with intensive chemotherapy in acute myeloid leukemia

Progress with intensive chemotherapy and supportive care measures has improved survival in newly diagnosed acute myeloid leukemia (AML). Predicting outcome helps in treatment decision making. We analyzed survival as the treatment endpoint in 3728 patients with newly diagnosed AML treated with intensive chemotherapy from 1980 to 2021. We divided the total study group (3:1 basis) into a training (n = 2790) and a validation group (n = 938). The associations between survival and 27 characteristics were investigated. In the training cohort, the multivariate analysis identified 12 consistent adverse prognostic variables independently associated with worse survival: older age, therapy-related myeloid neoplasm, worse performance status, cardiac comorbidity, leukocytosis, anemia, thrombocytopenia, elevated creatinine and lactate dehydrogenase, cytogenetic abnormalities, and the presence of infection at diagnosis except fever of unknown origin. We categorized patients into four prognostic groups, favorable (7%), intermediate (43%), poor (39%), and very poor (11%) with estimated 5-year survival rates of 69%, 36%, 13%, and 3% respectively (p < .001). The predictive model was validated in an independent cohort. In a subset of patients with molecular mutation profiles, adding the mutation profiles after accounting for the effects of previous factors identified NPM1 (favorable), PTPN11, and TP53 (both unfavorable) mutations as molecular prognostic factors. The new proposed predictive model for survival with intensive chemotherapy in patients with AML is robust and can be used to advise patients regarding their prognosis, to modify therapy in remission (e.g., proposing allogeneic stem cell transplantation in first remission), and to compare outcome and benefits on future investigational therapies.

Automated summary

This study analyzed the survival of newly diagnosed AML patients treated with intensive chemotherapy and identified adverse prognostic variables that are independently associated with worse survival. A predictive model was developed to estimate the patients' survival outcome, which can be used to guide treatment decisions and compare outcomes of different therapies.