Comprehensive Geriatric Assessment Does Not Predict Overall Survival in Older Patients with Acute Myeloid Leukemia (AML)

Background

Age and cytogenetics are the strongest predictors of overall survival (OS) in older patients with acute myeloid leukemia (AML), but practitioners recognize that outcomes are also affected by medical comorbidities, physical function and a variety of psychosocial factors. Recent data suggest that geriatric assessment, including measures of physical, cognitive and psychological function, may be predictive of OS and helpful for risk stratification in AML (Klepin 2013). We evaluated the ability of a comprehensive geriatric assessment (CGA) to predict overall survival in newly diagnosed older patients with AML.

Patients and Methods

All newly diagnosed AML patients age ≥60 years treated at Weill Cornell Medical Center and The New York Presbyterian Hospital completed a CGA including the OARS Physical Health Section, Mental Health Inventory (MHI-17), MOS Social Activity Survey, Activities of Daily Living (ADL) subscale of the MOS Physical Health, OARS Instrumental ADL subscale, Timed Up & Go, Blessed-Orientation-Memory-Concentration Test, and Karnofsky Performance Status (KPS). Laboratory data, medications, transfusions and days of hospitalization were also collected. Comorbidities were assessed using the Hematopoietic Cell Transplantation Comorbidity Index (HCT-CI). OS was estimated by the Kaplan-Meier method and the log-rank test was used to compare survival profiles. The independent effects of the CGA and clinical risk factor variables on OS were estimated using hazard ratios in an adjusted Cox regression model.

Results

In total, 126 patients were evaluated (median age 74, range 60-90). Fifty-one percent of patients had a HCT-CI score >1, with the most common comorbidities being a history of cancer (28.57%), cardiac disease (20.6%), and psychiatric disturbances (17.5%). Half of the patients had prior hematologic disorders and 29% had poor-risk cytogenetics. Most patients (84.9%) received decitabine-based induction strategies; 44 of these patients (34.9%) subsequently received intensive salvage regimens (median age 69). The other 19 patients (15.1%) were treated with standard induction chemotherapy, and 29 patients (21%) underwent allogeneic stem cell transplantation (median age 68). One hundred twenty-three patients (96.6%) completed the CGA with a mean time to completion of 26 minutes (± 8.9 minutes). Thirty-five percent of patients did not complete the entire assessment and only 61.1% completed the follow up CGA. Median OS was 11.1 months (range 0.36-52.64), with 1-year survival of 47.6%, complete remission (CR) rate of 39.8%, and 30-day mortality of 2.4%. In univariate analysis, age (P=0.0289), physician-assigned KPS (P=0.0031), sex (P=0.0074), ELN cytogenetic risk (P=0.0194), creatinine (P=0.027), albumin (P=0.0052), white blood cell (WBC) count (P=0.0135), LDH (P=0.0004), and treatment response (P=0.0001) were significant clinical predictors of OS. Significant CGA variables included Blessed Orientation-Memory-Concentration score (P=0.0035), Bend-Kneel-Stoop (P=0.0239), “someone to prepare your meals” (P=0.0253) and self-reporting of heart disease (P<0.001). In a multivariate analysis controlling for age and cytogenetic risk, physician-assigned KPS (HR, 1.804; 95% CI 1.175 to 2.768), self-reported cardiac history (HR, 2.290; 95% CI 1.383 to 3.794), and WBC count <11.2/ul (HR, 2.360; 95% CI 1.415 to 3.936) were the only independent prognostic factors for overall survival.

Conclusion

In this study, age and cytogenetics remain the strongest predictors of OS in older patients with AML. While completion of the CGA was feasible, only performance status (KPS) was predictive of OS. Many measures previously reported as significant predictors of outcome, including impaired physical function (Klepin 2013), medication intake (Hurria 2011), pain (Sherman 2013), and HCT-CI score (Sorror 2005, 2009) were not predictive in our study population. The role of the CGA as a predictor of OS in AML requires further evaluation. The utility of the CGA in predicting functional performance and/or quality of life for older AML patients throughout treatment should also be investigated.