Limitations of Performance Status (PS) Assignment in Predicting Outcomes of Acute Myeloid Leukemia (AML) and the Role of Objective Parameters in Predicting PS Assignment,

Performance status is traditionally used to judge fitness in cancer patients. It is often a key factor in determining appropriateness for chemotherapy and entry into clinical trials. Most research on PS has been in solid tumor patients, but it is generally assumed to be valid in hematologic malignancies as well. However, PS assessment can be problematic, especially in patients who have become ill quickly or where there are conflicting parameters; e.g. patients whose function is limited but who are still actively working. We explored the role of PS in patients with AML to determine if alternative, more objectively determined parameters, could predict PS. We evaluated the relationship between PS and rates of complete remission (CR) and duration of overall survival (OS) as well as the relationship between objectively determined pretreatment parameters and these outcomes.

Approval for retrospective data collection was obtained by the institutional IRB. A randomly selected cohort of 145 patients with newly diagnosed AML was identified using pharmacy records of treatment with induction agents cytarabine, daunorubicin or azacitidine. We excluded patients with acute promyelocytic leukemia. Pretreatment data was collected from the electronic medical record of each subject, including AML subtype, age, gender, PS, presence of infection at diagnosis, peripheral blood white blood cell count, peripheral blood blast percentage, hemoglobin, lactate dehydrogenase, aspartate transaminase, alanine transaminase, creatinine, albumin, left ventricular ejection fraction (LVEF), bone marrow cellularity, bone marrow blast percentage and cytogenetics. Achievement of complete remission was assessed and overall survival was calculated in months from diagnosis to either death or date of censorship. PS was obtained from physician notes and documented as a Zubrod score. In the absence of documentation, the authors reviewed medical records for presenting symptoms, mobility and functional status and made an assignment. Retrospective assignment of PS was reviewed and confirmed by at least two of the authors. Logistic regression was performed to examine the relationship between pretreatment characteristics and CR, and the effect of pretreatment characteristics and the assignment of PS, dichotomized as good (0–1) vs. poor (2–4). Linear regression analysis was used to investigate the relationship between pretreatment characteristics and OS.

A total of 120/145 (83%) were assigned a PS of 0, 1. Twenty patients were considered to have a PS of 2 and five patients had a PS of 3. There was a statistically significant relationship between a good PS assignment and a higher albumin level (OR 5.32, p=0.000, 95% CI 2.32, 12.22). There was no significant relationship between the remaining clinical parameters and PS assignment, including age at diagnosis. Ninety-three patients (64%) achieved a CR. There was no relationship between PS, assigned either contemporarily or retrospectively, with the achievement of CR. The median overall survival was 15 months (range 0 to 92 months). PS did not predict the duration of OS. As is well reported, CR was significantly more likely in younger (< 60 yrs) patients and patients with favorable cytogenetics. Median OS was also prolonged in these patients. No additional pretreatment characteristics were significantly related to CR or OS except LVEF. Patients with an LVEF greater than 50% were more likely to achieve a CR (P=0.0001) and had a longer OS.

We found that, among pretreatment characteristics, a higher albumin level was associated with a good performance status. Age was not a predictor of PS. Contrary to expectations, in this population of treated AML patients, PS did not predict either CR or OS. In addition to the association of younger age and favorable cytogenetics with better outcomes, we found that a higher LVEF is also associated with higher CR rates, but this is likely explained by choice of chemotherapeutic agents.

Our data calls into question the use of PS as a tool to guide therapy in routine clinical management of acute myeloid leukemia. Prospective studies should evaluate the utility of more objective and more reliably documented clinical parameters than PS for predicting patient outcome.

No relevant conflicts of interest to declare.