Effect of geriatric assessment (GA) and genetic profiling on overall survival (OS) of older adults with acute myeloid leukemia (AML)

7021 Background: GA can predict the risk of toxicities of chemotherapy in older adults. Genetic risk categories correlate with OS in AML. We previously reported a reduction in early mortality in a pre-planned interim analysis of a phase II trial with the use of GA and genetic profiling to personalize therapy selection (NCT03226418) (Blood 2019; 134(s1):120). Here, we present the results of a propensity score matched analysis demonstrating an improvement in OS over a historical control. Methods: Patients ≥60 years with a new diagnosis of AML underwent GA. Patients were considered fit for intensive chemotherapy if they had robust physical function [normal activities of daily living (ADL) and instrumental ADL, and short physical performance battery score of ≥10 out of 12], normal cognitive function (Montreal Cognitive Assessment score of ≥26 out of 30), and hematopoietic cell transplantation comorbidity index (HCT CI) of 0-2 (except for treatment related AML, where a score of 0-2 in addition to the prior history of malignancy was acceptable). Genetic profiling for therapy selection relied on karyotyping and followed the 2017 ELN criteria. Fit patients with good or intermediate-risk AML received intensive chemotherapy. Patients with high-risk AML received low-intensity chemotherapy, or CPX 351 if they were fit and met the FDA-approved indications. Pragmatic aspects of the trial included broad eligibility criteria (e.g. patients on treatment for other malignancy were enrolled) and co-management of patients with community oncologists. Mortality was compared with a historical control treated during the years 2004-2016 (after approval of HMA) and matched on gender, age, Karnofsky Performance Status (KPS), HCT CI and ELN risk category. Results: Treatment group (n = 27) vs. historical controls (n = 32) were matched in terms of age (median age, 70 vs. 68.5 years), ELN risk category (adverse risk 59% vs. 53%), HCT CI (median score of 2), KPS (median 80 vs. 85), and gender (male 44% vs. 50%). In the treatment group, 3 patients received intensive chemotherapy: CPX 351 (n = 2) or 7+3+ gemtuzumab (n = 1). Other patients received HMA alone (n = 16), decitabine and midostaurin (n = 3), or azacitidine and venetoclax after the approval of venetoclax (n = 5). Treatment in the historical control included intensive chemotherapy (n = 20) such as 7+3, or mostly HMA based low intensity chemotherapy (n = 12). OS was significantly higher in the treatment group over historical control with 1-year OS of 66% (95% CI 60-87%) vs. 16% (95% CI 7-35%). Conclusions: Our model to personalize AML therapy selection represents an innovative approach to precision medicine that incorporates both GA for patient profiling and genetic profiling of leukemia cells. Our results appear promising with superior OS (an absolute difference of 50% in 1-year OS) compared to a matched historical control. Clinical trial information: NCT03226418.