BACKGROUND

Information and tools that help individualize assessments of risk are important to counsel older patients with acute myeloid leukemia (AML) about their treatment options. Incorporating comorbidity with other risk factors into a model to assess chances of survival may help us estimate with greater certainty whether older patients are medically fit to tolerate intensive chemotherapy as induction therapy and/or allogeneic hematopoietic cell transplantation (HCT) as postinduction potentially curative therapy, thereby fine-tuning their treatment choices.

OBJECTIVES

(1) Develop a risk-stratification model, including comorbidities, for outcomes of patients with newly diagnosed AML; (2) assess predictors of likelihood of receiving allogeneic HCT in older and medically infirm patients with AML; and (3) compare outcomes and quality of life in patients treated with HCT vs chemotherapy.

METHODS

We retrospectively collected comorbidities, laboratory, and outcome data from 1100 patients with newly diagnosed AML who received therapy at 5 institutions from 2008 to 2012. Data were randomly divided into a training (n = 733) and a validation (n = 367) set for development and validation of the model, respectively. We developed and tested a comorbidity index for prediction of 1-year mortality. We then integrated the comorbidity index with age groups and cytogenetic/molecular risk groups into a global AML-composite model (AML-CM) for prediction of 1-year mortality. The performance of both the comorbidity index and the AML-CM were validated in the validation set using Kaplan-Meier survival curves and area under the receiver operating characteristic curves (AUROCs). We designed a prospective observational longitudinal study to enroll patients with AML from time of first presentation at 13 collaborating centers to answer questions about the second and third aims of the study.

RESULTS

Comorbidity scores of 0 to 2, 3 to 4, 5 to 6, and ≥7 were associated with overall survival rates of 73%, 65%, 57%, and 20%, respectively. The comorbidity index had an AUROC value of 0.687 for 1-year mortality. The comorbidity index, age (hazard ratio [HR], 1.8 for ages 50-59 years and HR, 2.0-2.5 for ages >60 years compared with age <50 years; < .0001), and cytogenetic risks per the European LeukemiaNet classification (HR, 1.8 for intermediate risk and HR, 2.8 for adverse risk compared with favorable risk; < .0001) independently stratified the population for risks of 1-year mortality. Weighted scores, per the multivariate model, of the comorbidity index, age groups, and cytogenetic risk groups were summed to form the AML-CM that had an AUROC of 0.758 (SD, 0.030) for prediction of 1-year mortality after intensive or less-intensive induction therapies (ie, therapies that achieve, if successful, complete remission after 1 cycle vs those that achieve lesser responses and require >1 cycle to achieve complete remission, if possible); patients with AML-CM scores of 4 to 6, 7 to 9, and ≥10 had 2-year survival rates of 48% (intensive treatment) vs 33% (less intensive), 37% vs 12%, and 22% vs 11%, respectively. Overall, 42% of all patients in the prospective cohort received allogeneic HCT. Factors associated with a lower likelihood of receiving transplant included age ≥70 years (HR, 0.40; = .0001); low cytogenetic risk (HR, 0.28; < .0001); less-intensive induction therapy (HR, 0.56; = .02); poor performance status (HR, 0.49; = .0005); and relapse after initial complete remission (HR, 0.41; = .001), impaired cognition (HR, 0.45; = .007), and impaired hearing (HR, 0.71; = .009). In the prospective cohort, unadjusted univariate models showed statistically significantly lower risks of mortality in association with receiving allogeneic HCT ( = .0001). However, in multivariate models adjusted for other risk factors including the AML-CM, the benefit of allogeneic HCT was negated ( = .91 in the overall population). Function, as measured by the 4-meter walk test, steadily improved in the first 2 years after diagnosis among those not receiving HCT, whereas it remained stable for observations after HCT ( value for no HCT vs HCT = .02).

CONCLUSIONS

(1) The AML-CM provides important prognostic information on treatment outcomes. Patients with higher AML-CM scores are good future candidates for a randomized clinical trial evaluating the benefit of induction regimens with different intensities. (2) Modifiable patient-specific barriers to allogeneic HCT (eg, comorbidities, limited physical strength, cognitive impairment) could be identified and targeted by interventions in future trials. (3) Randomized studies are needed to define the role of allogeneic HCT for treatment of older and medically infirm patients with AML.

LIMITATIONS

Retrospective and observational studies cannot be certain that they capture all confounding factors. The external validity of observational studies could be limited by the characteristics of patients who agreed to be enrolled.