Geriatric assessment and survival among older adults receiving postremission therapy for acute myeloid leukemia

TO THE EDITOR:

Geriatric assessment is feasible and can predict survival in older adults starting therapy for acute myeloid leukemia (AML).^1-4  Among older adults receiving intensive therapy, impaired objectively measured physical function (Short Physical Performance Battery; SPPB) and cognition are associated with worse survival.³  In the nonintensive setting, a frailty score inclusive of impairment in activities of daily living (ADL) is associated with worse survival.¹  Studies investigating patient characteristics, such as comorbidity and polypharmacy, further support evaluation of patient-specific vulnerabilities to inform mortality risk.^5-11 

Most studies, however, have focused on the use of geriatric assessment at the time of diagnosis. Geriatric assessment–detected vulnerabilities are dynamic and may improve or worsen during therapy. Short-term declines in self-reported and objective physical function occur after intensive induction therapy.^4,12  However, the use of repeated geriatric assessment at the time of postremission therapy to predict clinically important outcomes is unknown. The primary objective of this study was to evaluate the association between geriatric assessment measures obtained at the time of postremission evaluation and overall survival (OS).

Our analysis cohort was derived from a prospective observational study of older adults (N = 74) with newly diagnosed AML that investigated geriatric assessment during induction chemotherapy.^2,3  The predictive utility of geriatric assessment at the time of diagnosis was published previously.³  Here, we assess the predictive utility of the geriatric assessment performed at the time of postremission evaluation among the 40 patients who achieved complete remission or complete remission with incomplete count recovery after intensive induction chemotherapy and were deemed eligible to receive postremission (consolidation) chemotherapy. This study (ClinicalTrials.gov: NCT02662933) was approved by the Institutional Review Board of Wake Forest University Health Sciences, and all participants provided written informed consent in accordance with the Declaration of Helsinki.

Geriatric assessment data were collected during the postinduction outpatient follow-up visit at the time of postremission therapy evaluation (∼8 weeks after induction hospitalization discharge). Measures included Modified Mini-Mental State Exam, Center for Epidemiologic Studies Depression Scale (CES-D), Distress Thermometer, Hematopoietic Cell Transplantation Comorbidity Index score (HCT-CI), polypharmacy (≥5 prescription medications), Pepper Assessment Tool for Disability survey (including subscales assessing mobility, as well as basic and instrumental ADL), SPPB (4-m walk, repeated chair stands, and balance tests), and hand grip strength.

OS was the primary outcome, defined from the date of the postremission evaluation visit to the date of death or last follow-up. Data collected from the medical record included demographics, Eastern Cooperative Oncology Group (ECOG) performance status, laboratory studies, tumor characteristics, treatment information (type, dose), and number of days hospitalized within 6 months of postremission therapy. Dose delay during postremission therapy was defined as a delay in intended treatment for >1 week for medical reasons, per physician documentation in the medical record.

Medians and interquartile ranges and frequencies describe the baseline characteristics of participants and performance on geriatric assessment measures. OS was estimated using the Kaplan-Meier method. The log-rank test was used to compare survival by SPPB score (<9) and CES-D (≥16). Cox proportional-hazards models were fit for clinical, demographic, and geriatric assessment measures as predictors of OS, in unadjusted models and in a backward regression model, entering all variables with P values < .20 and retaining those with P values < .05. A 2-sided α level of 0.05 was used to indicate statistical significance. All statistical analyses were conducted using SAS (v 9.4; SAS Institute, Cary, NC).

Table 1 describes the patient population (median age, 68.7 years). Most patients (82.5%) received ≥2 cycles of postremission chemotherapy; 42.5% had ≥1 dose delay. At postremission evaluation, between 20.0% and 82.5% of patients had impairment on individual geriatric assessment measures (Table 1).

The median OS from postremission evaluation was 21.9 months. Age, sex, ECOG performance status, and clinical characteristics were not associated with OS (Table 2). Impairment in objectively measured physical function (SPPB < 9) and depressive symptoms (CES-D > 16) were associated with worse OS. Compared with those without impairment in SPPB, patients with SPPB score < 9 had an OS of 16.5 months vs 53.3 months (P = .03). Similarly, compared with those with CES-D score < 16, patients with CES-D score ≥ 16 had a median OS of 16.9 months vs 25.5 months (P = .02). Cognitive impairment, distress, comorbidity, polypharmacy, and self-reported physical function were not associated with survival. In multivariate analyses, SPPB (<9) and depressive symptoms (CES-D ≥ 16) assessed at the time of postremission evaluation were independently associated with an increased risk for death (SPPB: hazard ratio [HR], 2.66; 95% confidence interval [CI], 1.19-5.97; CES-D: HR, 2.46; 95% CI, 1.11-5.44) (Table 2). In exploratory analyses, there was minimal attenuation of effect when including comorbidity (HCT-CI) in the model (SPPB: HR, 2.49; 95% CI. 1.07-5.80; CES-D: HR, 2.34; 95% CI, 1.04-5.28,) and there was no evidence of effect modification of comorbidity on these relationships.

In this study, we demonstrated that impairment in objective measures of physical function and self-reported depressive symptoms assessed at postremission follow-up were associated with an increased risk for mortality in older patients with AML, whereas chronologic age and performance status were not. Our findings illustrate the value of repeat geriatric assessment to optimize risk stratification and better capture the dynamic vulnerabilities associated with AML and its treatment among older adults. This approach may guide management during the course of therapy and survivorship. The postinduction period is a critical time for treatment decision-making because patients and physicians are often considering consolidation therapies, such as cytarabine, bone marrow transplantation, or lower-intensity treatments, to prolong remission. Therefore, geriatric assessment measures add value by providing the opportunity to intervene with supportive care interventions and to calibrate prediction of treatment tolerance and benefit, which may yield quality-of-life and survival benefits.^13,14 

This study adds to the literature regarding the value of measuring objective physical function using the SPPB in this patient population. We have previously demonstrated that SPPB at diagnosis predicts survival and that SPPB declines post intensive induction in an older adult population.^3,12  We now show that lower SPPB at the time of postremission evaluation is associated with shorter survival. This finding is consistent with the data from other investigators who have shown that gait speed (included in the SPPB) is predictive of mortality in older adults with hematologic disorders.¹⁵  Most importantly, recognizing prevalent or acquired physical frailty is so important because it can be intervened upon. Interventions targeted toward improving physical function may be instituted prior to or during treatment.^16-19  Results of this analysis raise the possibility that intervening on physical function in this setting could improve functional outcomes, as well as survival.

Our study also highlights the prognostic significance of depressive symptoms, consistent with prior work in various cancer types.^20,21  Interestingly, in the upfront induction setting, we did not find an association between depression and mortality.³  We hypothesize that depression at baseline may reflect a reaction to AML diagnosis, whereas persistent depressive symptoms at remission may reflect coping and psychological resilience. Alternatively, a patient’s psychological health may be driven by the symptom burden, which could influence survival. Like physical function, depressive symptoms are potentially modifiable.²²  In our cohort, participants with depressive symptoms at postremission were more likely to have been on an antidepressant at diagnosis (40% vs 10%) or to have depressive symptoms at baseline, representing a higher-risk group that could be targeted earlier for intervention.

It is notable that cognitive impairment at postremission evaluation was not associated with worse survival in this analysis. Interestingly, an exploratory analysis suggests that cognitive impairment assessed at diagnosis, regardless of whether it improved postinduction, remains a potential risk for mortality (HR, 1.99; 95% CI, 0.91-4.34). These results did not achieve statistical significance and are hypothesis generating, but they might be explained by baseline cognitive impairment representing delirium for many patients. Delirium, regardless of cause, is associated with mortality.

Strengths of our study include a homogenous well-characterized older AML cohort who received intensive induction. We used validated geriatric assessment measures. Limitations include the small sample size from a single academic institution. The small sample size limits precision in estimating effects, may miss small effects, and may explain why some known prognostic factors are not associated with survival in this analysis. The geriatric assessment did not include evaluation of social support or a formal assessment of delirium. It is not known whether SPPB and depressive symptoms are associated with survival in the nonintensive setting.

In conclusion, we found that impaired objectively measured physical function and depressive symptoms were associated with worse survival in older patients with AML undergoing postremission therapy. Our data support the study of repeat geriatric assessment measures during AML treatment to personalize treatment decision making and the potential value of designing interventions targeting geriatric assessment vulnerabilities to improve outcomes for older adults with AML.