Several scoring systems have been developed to assess suitability of individual patients for intensive acute myeloid leukemia (AML) therapy. We sought to compare the performance of these scores in a cohort of 428 consecutive adults with AML who received conventional induction chemotherapy in five academic centers in France. All scoring systems identified a subset of patients with increased 28 and 56-day mortality although the prediction accuracy was overall limited with C-statistics of ranging from 0.61 to 0.71 Overall survival (OS) prediction was more limited and restricted to scoring systems that include AML-related parameters. The outcome of 104 patients (24%) considered unsuitable for intensive chemotherapy based on criteria used in recent randomized trials was similar to that of the other 324 patients (28-day mortality, odds ratio [OR] = 1.88, P = 0.2; 56-day mortality, OR = 1.71, P = 0.21; event-free survival, hazard ratio [HR] = 1.08, P = 0.6; OS, HR = 1.25, P = 0.14) with low discrimination (C-statistic: 0.57, 0.56, 0.50, and 0.52 for 28-day, 56-day mortality, EFS, and OS, respectively). Together, our findings indicate that the accuracy of currently available approaches to identify patients at increased risk of early mortality and shortened survival after intensive AML therapy is relatively limited. Caution regarding the use of available scoring systems should be warranted in clinical decision-making.
© 2022. The Author(s), under exclusive licence to Springer Nature Limited.