Does the Region of Diagnosis Correlate with Overall Survival in Patients with Acute Myeloid Leukemia (AML) - a SEER Review of AML Diagnosis from 2004 - 2007

Background: Epidemiologic features accounting for poor outcomes in patients with acute myeloid leukemia (AML) have been under-reported in the literature. Although it is generally accepted that AML portends an ominous prognosis, there is a paucity of understanding in the etiology and risk factors attributing to such poor outcomes in this condition. We aimed to study the epidemiology and survival outcome of AML and its correlation with the region of diagnosis by reviewing the Surveillance, Epidemiology, and End Result Program (SEER database).

Methods: A SEER database (version 8.1.5) review was performed for all age groups with an International Classification of Diseases in Oncology (ICD-O-3) diagnosis of histologically confirmed non acute promyelocytic leukemia (APL)-AML from 2004 -2007. Age of included patients ranged from 15 years to 90 years. Collected variables included date of diagnosis, age at diagnosis, gender, ethnicity, location of diagnosis, reported intervals of follow up, and 3- year overall survival (OS). Primary outcome was 3-year OS correlation with the region/state of diagnosis. All SEER Registries were included in the analysis. Data were analyzed using Kaplan Meier and Cox proportional hazard regression model.

Results: A total of 13,238 pts with non APL-AML were identified between 2004 and 2007. The mean age of the overall study population was 61.6 years. The Caucasian to non-Caucasian race ratio was 5:1 and male to female ratio was 7:6. The 3-year OS was 24% in the West region (CA, WA, HA, AL and UT). Patients in East region (NJ, CT) had better 3-year OS (24.5%) when compared with patients in West region. However, patients in Midwest (MI,IW and KY) and South regions (GA, NM and LA) had worse 3-year OS when compared to the West region ( 21.2% and 23.2 % respectively). The 3 year OS was statistically associated with region of diagnosis (P= 0.0003). This conclusion held after adjusting for race, gender, and age with statistical significance (P = 0.0165). Furthermore, the 3 year OS was statistically significant for its association with ethnicity as non-Caucasian groups had better OS compared with Caucasian populations (HR 0.92). However, after adjusting for region, gender and age, the association between OS and ethnicity was not significant (P = 0.588).

Conclusion: Non APL-AML OS was statistically associated with the region of diagnosis and presumed treatment based on analysis of all available data. The patients in East region had better survival compared with patients in West region (HR 0.96). However, the patients in South and Midwest regions had poorer survival when compared with the West region (HR 1.01 and 1.13, respectively). This survival difference was statistically significant after adjusting for other confounding factors including age, race and gender. Also the difference in OS among different patient racial populations was not significant when adjusted for the region of diagnosis and treatment.