Impact of Comorbidities at Diagnosis of Acute Myeloid Leukemia on One-Year Mortality

The hematopoietic cell transplantation comorbidity index (HCT-CI) was specifically developed to assign weighted scores to comorbidities existing prior to allogeneic HCT; thus stratifying post-HCT mortality risks. The utility of comorbidities assessed prior to treatment for AML is unknown. Here, we a) investigated the impact of each comorbidity on 1-year overall mortality of patients (pts) with newly diagnosed AML, b) designed and validated a new comorbidity score (AML-CI) comparing its performance to that of the HCT-CI, and c) identified other relevant risk factors for AML outcomes.

We retrospectively collected comorbidities and laboratorydata from 1079 pts with newly diagnosed AML who received therapy at 5 institutions from 2008- 2012. Pts were aged ≤49 (29%), 50-59 (25%), 60-69 (26%), and ≥70 (20%) years old. Cytogenetic-risks were favorable (21%), intermediate (36%), or unfavorable (43%). Regimen intensity was low in 18%, intermediate in 63%, and high in 19%. HCT-CI comorbidities were evaluated per HCT-CI standard comorbidity definitions with the exception that renal comorbidity was defined per serum creatinine and/or creatinine clearance. Newly evaluated comorbiditiesincluded including hyperlipidemia, hypertension, deep venous thrombosis, gastroesophageal reflux disease, hypothyroidism, hypoalbuminemia, thrombocytopenia, neutropenia, anemia, elevated lactate dehydrogenase (LDH), smoking, and alcohol intake. Pts were randomly divided into a training (n=710) and a testing set (n=369). In the training set, the unadjusted hazard ratios (HRs) for 1-year overall mortality were calculated for each comorbidity as well as all adjustment factors: gender, age, race, cytogenetic-risks, regimen intensity, WBC, blast count, and marrow blast percentages. Only factors that were associated with overall mortality at a significance level of P <.10 proceeded to the multivariate model. Each comorbidity that entered the multivariate model was then adjusted for the effect of other comorbidities as well as gender, age, cytogenetic-risks, and regimen intensity (Table 1). The adjusted HRs were employedas weights for individual comorbidities. In the validation set, the new AML-CI incorporating comorbidities had comparable power of prediction for 1-year overall mortality as the HCT-CI (c-statistic estimates of 0.6 for each). Augmenting the HCT-CI with the three new covariates: platelets, albumin, and LDH yielded c-statistic estimate of 0.61. Other than comorbidities, age (c-statistic of 0.65) and cytogenetic-risks (c-statistic of 0.62) independently predicted overall mortality (Table 2).

Comorbidities at diagnosis of AML heavily influenced the survival of pts over the year following diagnosis. The new AML-CI has similar predictive power as the HCT-CI suggesting appropriateness of using the HCT-CI at diagnosis of AML given its familiarity to physicians. The augmented HCT-CI, age, and cytogenetic-risks independently stratified for risks of 1-year mortality. In the future, studying physical, cognitive, and social health might further clarify the prognostic role of increasing age. Targeting health limitations with interventions alongside specific AML-therapy might improve survival of patients.