Comparison of clinical outcomes of several risk stratification tools in newly diagnosed AML patients: A real-world evidence in our current therapeutic era.

BACKGROUND OF THE STUDY: AML classification tools have been developed to stratify the risk at AML diagnosis. There is a need to evaluate these tools in the current therapeutic era.

COHORT CHARACTERISTICS: In this retrospective study, we compared five classifiers: ELN 2017, ELN 2022, ALFA classifier, Papaemmanuil et al. classifier, and Lindsley et al. classifier, in a real-life cohort of 281 patients newly diagnosed for AML in Nice University Hospital. In our cohort median age was 68 years old, sex ratio was M/F 56%/44%, performance status was lower than 2 in 73.1% of patients, AML subtype was "De novo" in 71.5%, "secondary" in 22.4%, and "therapy-related" in 6.0% of patients. Intensive chemotherapy was used in 53.0% of patients, and non-intensive chemotherapy in 40.6% of patients. Molecular analysis was available in a large majority of patients and the main mutations found were NPM1 (22.7%), DNMT3A (17.4%), TP53 (13.1%), TET2 (12.4%), and FLT3-ITD (12.4%).

RESULTS: In our findings, the comparison of overall survival between the three prognostic groups in the global cohort was statistically significant in all classifiers: ELN 2017 p < 0.0001, ELN 2022 p < 0.0001, ALFA classifier p < 0.0001, Papaemmanuil classifier p < 0.0001, Lindsley classifier p = 0.001. ELN 2017, ELN 2022, ALFA classifier, Papaemmanuil classifier, and Lindsley classifier were calculated respectively in 99%, 99%, 89%, 90%, and 89% of patients.

CONCLUSIONS: Using Akaike's information criteria (AIC) to compare all five classifiers, ELN 2022 is the best classifier into younger and older patients and for prognosis.