Prognostic impact of cognitive and other impairments in older adults with acute myeloid leukemia (AML) participating in a geriatric assessment-guided treatment selection trial

Abstract Background: Impairments on geriatric assessment (GA) are common in older adults with AML; hence we designed a trial that used GA guided selection of chemotherapy intensity (ClinicalTrials.gov ID: NCT03226418). In this analysis, we hypothesized that i) specific impairments in GA are independently associated with an increased risk of death, and ii) a greater number of GA impairments increases the risk of death. The novelty of this analysis includes the unique trial design and the concept of demonstrating increased mortality risks associated with the greater burden of impairments in a prospective AML trial. Such data could demonstrate the value of multidimensional assessments, as compared to using a screening tool. Methods We enrolled older adults (≥60 years) with a new diagnosis of AML in a phase II trial and performed multidimensional GA before treatment initiation. GA included measures of physical function [Activities of Daily Living (ADL), Instrumental ADL (IADL), short physical performance battery (SPPB)], cognition [Montreal Cognitive Assessment (MoCA)], comorbidities [Hematopoietic Cell Transplantation Comorbidity Index (HCT CI)], depression [Patient Health Questionnaire-9 (PHQ-9)], and nutrition [Mini Nutritional Assessment-Short Form (MNA)]. A log-rank test examined an association between mortality and impairments in GA. Cox model analyses for overall survival (OS) were performed after adjusting for co-variates such as age, WBC count, and AML risk categories. Results Baseline characteristics of the trial participants (n=73) included a median age of 69 years (range 60-87 years), 49% female, 92% white, and a median KPS of 80 (range 60-100). ELN 2017 risk categories included adverse (60%), intermediate (22%), and good-risk AML (18%). Broad eligibility criteria allowed enrolling patients representative of those treated in real-world practices: 45% of patients were ≥70 years, 57% had ≥2 comorbidities, and 27% had a history of solid malignancies. Thirty-two percent of patients resided in rural areas, and 45% were comanaged with community oncologists. Most had an HCT CI of ≥3 (55%), impaired objective physical function (SPPB) (70% had a score of ≤9), or cognitive impairment (64% had a score of ≤25 on MOCA). ADL and IADL were impaired in 29% and 22%, respectively. Depression screen (PHQ-9 score of ≥10) and nutritional screen (MNA score of ≤11) were abnormal in 33% and 56%. Impairments were also common in patients with KPS score of 80-100. Based on the protocol, eight patients (11%) received intensive chemotherapy; others received low-intensity chemotherapy: azacitidine or decitabine alone (prior to approval of venetoclax) or in combination with venetoclax. The median survival for all patients was 355 days. In log-rank testing, cognitive impairment (MOCA score of ≤25 vs. 26-30) (p=0.049) was associated with poorer OS whereas impairments in ADL (≤5 vs. 6 out of 6) (p=0.050), and SPPB (≤9 vs. 10-12) (p=0.062) showed a trend towards an association with worse OS. Younger patients (60-69 years) were more likely to score well on SPPB (p=0.030) but their OS was not different from those 70 years and older (p=0.79). The number of impairments in GA domains included 0-1 (16.4%), 2-3 (38.4%) and 4-7 (45.2%). Patients with 0-1 vs. 2-3 GA impairments did not have any differences in OS, hence were merged into one group. In adjusted cox model, patients without vs. with cognitive impairment (HR 0.38, 95% CI 0.18-0.81, p=0.012) had lower risk of mortality, and independence in ADLs showed a trend towards lower mortality (HR 0.54, 95% CI 0.29-1.01, p=0.055). Patients with fewer GA impairments (0-3 vs. 4 or more) had lower risk of mortality (HR 0.42, 95% CI 0.23-0.77, p=0.0049). Conclusion Older adults with newly diagnosed AML and normal cognitive function have lower risk of mortality. Patients with 0-3 vs. 4 or more impairments in GA domains have lower risk of mortality. Multidimensional assessments including cognitive testing at AML diagnosis can be useful for prognostication. Supportive care interventions to address geriatric impairments should be tested as a strategy to improve OS.