Using Charlson comorbidity index (CCI) to predict outcomes of older patients with acute promyelocytic leukemia (APL).

Abstract

e18751 Background: A large-scale study specifically analyzing an association between comorbidity burden and outcomes is lacking in APL. We hypothesized that a greater comorbidity burden independently predicts worse outcomes in patients ≥60 years with APL. Methods: We utilized the National Cancer Database to identify patients ≥60 years diagnosed with APL between 2004 and 2015. CCI scores were used to evaluate comorbidity burden, and patients were divided into 3 groups with CCI of 0, 1, and ≥2. We used chi-square test to determine association between CCI and different variables. Multiple logistic regression analyses and Cox regression models evaluated the effects of CCI on one-month mortality and overall survival (OS), respectively. Results: Of 2221 patients, 59% had CCI of 0, 27% had CCI 1, and 14% had CCI ≥2. Compared to patients with CCI 1 or CCI ≥2, a higher proportion of patients with CCI 0 had private insurance, and higher education and income status, were treated at academic centers, and received multiagent therapy. One-month mortality was 16%, 24%, and 32% for patients with CCI 0, 1, and ≥2, respectively. After adjusting for other co-variates, patients with CCI 0 had lower one-month mortality compared to CCI 1 (Odds ratio 1.67, 95% confidence interval [CI] 1.29-2.16, p < 0.001) and CCI ≥ 2 (Odds ratio 2.31, 95% CI 1.70-3.13, p < 0.001). Three-year OS was 61%, 53%, and 38% for patients with CCI 0, 1, and ≥2, respectively. After adjusting for other co-variates, OS was worse among patients with CCI 1 (Hazard ratio 1.27, 95% CI 1.10-1.46, p-value < 0.001) and CCI ≥2 (Hazard ratio 1.74, 95% CI 1.48-2.06, p-value < 0.001), compared to patients with CCI 0. Conclusions: Our study is among the first and the largest to examine an association between comorbidity burden and outcomes in older adults with APL. Greater comorbidity burden, indicated by higher CCI, predicted worse one-month mortality and OS, after adjusting for other co-variates. Thus, our study results establish CCI as an important and independent predictor of outcomes in APL. The study results can inform personalized estimates of mortality and OS based on comorbidity burden and facilitate treatment decision-making. Given the significant differences in outcomes of older adults based on comorbidity burden, future trials in APL should present comorbidity data and consider utilizing CCI to risk-stratify patients.