Uncovering Ethnic Genomic Disparities Associated with Inferior Outcome in African Americans Diagnosed with Acute Myelogenous Leukemia

Background: African Americans (AA) have higher cancer-associated mortality than other ethnicities. AA with AML exhibit inferior overall survival (OS) to non-AA subgroups (OS=162 d vs 709 d, respectively, p=0.0001, HR=7.8) Fig 1A. In other cancers, higher genomic instability (GI) and differential oncogenic addiction contribute to AA worse outcomes; however, in AML, genetic causes of inferior outcomes are uncharacterized. In this study, we seek to investigate differential AML mut and other factors that contribute to inferior AA OS. Methods: After IRB approval, 92/250 (36.8%) available AML patients (pt) had next generation sequencing (NGS) data and were included. Mut frequencies were annotated by ethnicity [White (W) vs AA vs Hispanic (H) vs Asians (A)] and prioritized by European Leukemia Network (ELN) 2017 classification. Numeric chromosomic alterations (NCA) were extracted from metaphase karyotyping at diagnosis. Deleteriousness of mutations to protein function was calculated with sorting intolerant from tolerant (SIFT) prediction method. Chi-square and t-test evaluated differences in categorical and continuous variables respectively. Logistic binary regression was used to detect predictors for racial induced survival. SAS software was used. Results: Whites (W), AA, Hispanics (H) and Asians (A) represented 57/91 (62%), 14/91 (15%), 20/91 and (22%), 5/91 (5.4%) of the cohort, respectively. Median age was 63 (W) , 57.5 (AA) , 53 (H), 62 (A), p=NS. 60/92 (65%) were male. Median number of mut were 2.04 (W), 1.8 (AA), 2.2 (H) and 1.8 (A), p=NS. Distribution for adverse ELN-2017 defects were equivalent among AA vs non-AA except RUNX1, p=0.05. Lack of favorable mut was observed among AA. Only AA demonstrated CUX1, previously associated with DNA repair defects, and FBXW7 mut. A non-significant trend for higher NCA in AA (2.7 [W], 6.4 [AA], 3 [H] and 0.8 [A], p=0.09) was observed. Age correlated with NCA and number of mut in AML genome, p=0.02 and p=0.004. Interestingly, correlation effect was mostly driven only by AA, p=0.02 and p=0.003, respectively. SIFT score was 0.01 (W), 0.09 (AA), 0.02 (H), and 0.01 (A). These SIFT results suggest that AA mut are functionally tolerated. After adjusting for multiple factors, only canonical ELN-2017 mut, p=0.02 (OR 0.37, CI 0.16-0.8), and AA race, p=0.005 (OR 0.25, CI 0.1-0.67) increased risk of death (c-statistics=0.66) Fig.1B. Conclusions: In AA, frequency of Adverse ELN-2017 mut is similar to other races, but favorable mutations were lacking. NCA and number of AML genome mut strongly correlate with aging in AA, but not other ethnic groups. Surprisingly, SIFT score failed to demonstrate functional deleteriousness for AA mut, which suggests alternative genomic defects as a culprit for poor outcome. Acknowledging the small number of mut in our cohort, our study advances our understanding of racial disparity in AML survival.

No relevant conflicts of interest to declare.