Are Racial and Ethnic Disparities In Acute Myeloid Leukemia Due To Differences In Treatment?

In our previous work using the Surveillance Epidemiology and End Results (SEER) database, we demonstrated that despite younger age at presentation and a higher prevalence of favorable cytogenetic factors, black and Hispanic patients have increased mortality from Acute Myeloid Leukemia (AML) compared with non-Hispanic whites (NHW). The role of treatment has not been studied on a population level due to the limitations of SEER data with respect to treatment variables. The purpose of this study is to explore explanations for disparities in AML using a novel database containing both demographic and clinical variables. We will evaluate the relationship between the quality of AML care and outcomes. The hypothesis is that outcome disparities from AML may be explained by differences in receipt of treatment by race/ethnicity.

All patients with AML were identified in the California Cancer Registry (CCR) database linked to the hospital discharge abstracts from the Office of Statewide Health Planning and Development (OSHPD) during the years 1998-2008. Kaplan Meier (KM) survival curves were generated to predict survival probabilities by race/ethnicity. These were stratified by age based on our prior findings. Logistic regression models estimated the odds of treatment defined as chemotherapy and/or hematopoietic stem cell transplant by race/ethnicity. Cox proportional hazard models estimated the hazard of mortality by race with adjustment for age, gender, year of diagnosis, co-morbidities, and presence of the t(8;21), APL, and 11q23 subtypes. Models were further adjusted for receipt of treatment.

A total of 11,084 patients were included in the study. Black and Hispanic patients were diagnosed at younger ages (<61 years) and had higher rates of APL subtype compared to NHWs. Hispanic and Asian/Pacific Islanders (API) patients had higher rates of t(8;21) subtypes compared to NHW. API and NHW had the highest rates of 11q23 subtype. Logistic regression models showed decreased odds of chemotherapy and hematopoietic stem cell transplant for black patients compared to NHW (0.74 95% CI (0.61-0.91); 0.62 95% CI (0.45-0.85), respectively) which persisted after adjustment for t(8;21), APL, and 11q23 subtypes. Odds of hematopoietic stem cell transplant were also decreased for Hispanic patients compared to NHW (0.68 95% CI (0.58-0.82)) despite adjustment for subtypes. Multivariable models adjusted for gender, age, year of diagnosis and comorbidities demonstrated that compared to NHW, blacks had an increased risk of death (1.15 95% CI (1.05-1.26)) whereas APIs had a decreased risk of death (0.84 95% CI (0.84-0.96)). Adjustment for t(8;21), APL, and 11q23 subtypes did not attenuate the disparity for blacks. Adjustment for treatment (chemotherapy and/or transplant) slightly moderated the risk of death (HR 1.10 95% CI (1.01-1.22)) for black patients.

Our work suggests that treatment differences may play a role in survival disparities from AML; however these differences do not completely explain the differences in survival. Socioeconomic status factors or unmeasured genetic factors may explain the observed differences. Future studies aimed at addressing disparities in AML should assess mortality with attention to these factors.

No relevant conflicts of interest to declare.