The 10-minute targeted geriatric assessment (10-TaGA), a 10-domain-to-one-score tool, predicts early death in intensively treated older AML patients: First results from a multicenter study

Abstract

Abstract Background: Older patients with acute myeloid leukemia (AML) are a particularly difficult-to-treat population, owing to the higher incidence of adverse risk disease and decreased tolerance to therapy. Treatment options are usually split according to the patient's fitness for intensive treatment, although an evidence-based definition of fitness for traditional chemotherapy is still lacking. Geriatric assessment has proved to be prognostic and can aid the treatment decision process. A comprehensive but easy-to-use tool, however, is needed to overcome barriers to the wide implementation of geriatric assessment, including in resource-constrained settings. The 10-minute Targeted Geriatric Assessment (10-TaGA) is a 10-domain screening tool previously proven to predict overall survival in older adults from a general hospital (Aliberti MJR et al 2019). It provides a score between zero to one accounting to ten evaluated domains. Methods: This is a prospective multicenter observational study designed to evaluate 10-TaGA as a predictor of early death (within 30 days of treatment initiation) in older AML patients treated with intensive therapy from four tertiary oncohematology centers in Brazil. Patients 60 years old or older with newly diagnosed AML were initially included. BCR::ABL positive cases were excluded. A sample of 60 patients was calculated based on previous data from the participating centers. One year after the study starting, the protocol was amended, due to the low enrollment rate, to include patients aged 50 years or more. Enrollment started in April/2023. Data were updated until 25-Jun-25 for this report. 10-TaGA was applied by two trained geriatricians. The study was approved by the research ethics committee, and all participants signed the informed consent form. Results: A total of 40 patients were included so far. Three patients ultimately did not receive intensive chemotherapy and were excluded. Thirty-six patients with more than 30 days of follow-up were included in this interim analysis. Median age at diagnosis was 61 (IQR 59-64) years old, with ten (27.8%) aged less than 60 years. There was a male predominance (55.6%). AML was secondary to cancer therapy in 3 (8.3%) cases. Two (5.6%) patients had a myeloid neoplasm before AML diagnosis. Disease risk was classified as favorable, intermediate or adverse in 7 (19.4%), 15 (41.7%), and 7 (19.4%) cases, respectively, according to Adapted Genetic Risk (Silveira D et al 2015). Proper classification was not possible in 6 (16.7%) patients. 10-TaGA was successfully applied in all patients. The median score adjusted for education level (10-TaGA score) was 0.275 (IQR 0.163 - 0.350). Anthracycline plus continuous infusion cytarabine (7+3) was the induction therapy for 34 (94.4%), while two (0.6%) received FLAG ± anthracycline. Eleven (30.6%) patients died within 30 days of induction initiation. Median 10-TaGA score was higher in the early death group (0.300 vs. 0.200; p=0.015). Guided by the highest Youden's index, patients with 10-TaGA score <0.280 were considered fit and those ≥0.280, unfit. Risk of early death was considerably higher in the unfit group (HR 8.0; 95%CI: 1.4-45.4; p=0.027), with an early death rate of 50% versus 11.1% in the unfit versus fit group, respectively. Major or minor impairments were detected at least in one patient for each domain. The most frequently impaired domains were nutrition status (66.6%), previous admission in a health care service (63.9%), gait speed (61.1%), self-rated health (47.2%), and depressive symptoms (33.3%). Conclusions: We report for the first time that 10-TaGA predicts early death in a homogeneously treated older AML cohort. Previous studies have shown the value of some domains of a comprehensive geriatric assessment, such as cognition, depressive symptoms, and physical performance in predicting outcomes in AML. 10-TaGA has the benefit of evaluating ten domains and producing a single score in a short application time. We have also shown that impairments were spread across all domains, suggesting that the overall results, and not the impairments in one or two domains, are responsible for the prognostic value of the score.