Analysis of acute myeloid leukemia incidence and geographic distribution in Canada from 1992 to 2010 reveals disease clusters in Sarnia and other industrial US border cities in Ontario

Abstract

Background Several risk factors have been implicated in acute myeloid leukemia (AML) leukemogenesis. However, the epidemiologic distribution and precise triggers for AML in Canada remain poorly understood. Methods In this study, demographic data for AML patients in Canada from 1992 to 2010 were analyzed using 3 independent population‐based cancer registries. The AML incidence and mortality rates were examined at the levels of province/territory, city, and forward sortation area (FSA) postal code. Results In total, 18,085 patients were identified. AML incidence was documented to be 30.61 cases per million individuals per year (95% confidence interval [CI], 30.17‐31.06) from 1992 to 2010. Five industrial cities in Ontario were identified where incidence rates were significantly higher than the national average: Sarnia, Sault Ste. Marie, Thunder Bay, St. Catharines, and Hamilton. Analysis at the FSA postal code level identified significant patient clusters of AML in these cities. Specifically, FSA N7V in Sarnia, Ontario had an incidence of 106.81 (95% CI, 70.96‐161.86) cases per million individuals per year, which is >3 times higher than the national average. The pollution from local oil refineries and chemical plants in Sarnia may be implicated as a risk factor for AML in that city. Analysis of mortality rates at the province and city levels corroborated the findings from the incidence data. Conclusion These results provide a comprehensive analysis of AML burden in Canada and reveal striking geographic case clustering in industrial Ontario cities and potentially implicate exposure to materials/pollution from these plants as an important risk factor for developing AML in Canada.

Incidence of acute myeloid leukemia: A regional analysis of Canada

“…The identification of case clustering patterns has been the root cause of the identification of external triggers for several diseases, such as cholera, mesothelioma, lung cancers, and select lymphoproliferative diseases . In our recent study, we investigated in detail the crude incidence and mortality trends and geographic distribution of acute myeloid leukemia (AML) across Canada. Before our report, there was a significant knowledge gap on the epidemiology, distribution, and populations at risk for AML in this country.…”

“…The original article was prepared according to strict journal guidelines on word limits and the number of display figures and tables, and it should be viewed as the first step in this line of investigation. The initial manuscript underwent 2 rounds of peer review by 5 experts in the field from May 10 to September 8, 2018, at which point the article was endorsed for publication.…”

Geographically distributed data should be analyzed with spatial epidemiologic methods

“…Observational data can be complex, and simple data analysis approaches are often not sufficient to account for the study design and likely confounders. Ghazawi et al reported unadjusted analyses of the incidence and mortality rates of acute myeloid leukemia (AML) in Canada and compared rates across different communities. Although the unadjusted analyses themselves may be of interest for understanding differences in the distribution of cases across Canada, comparing unadjusted rates can lead to the incorrect conclusion that certain regions may be “cancer clusters.” A cancer cluster is defined as an area with a greater than expected number of cancer cases among a group of people in a defined geographic area over a specific time period .…”

“…In addition, spatial epidemiology methods allow for the inclusion of confounders such as age. In the data analyzed by Ghazawi et al, spatial models could adjust for not only age but also obesity and smoking distributions across regions. Models can be developed for disease incidence mapping and also to identify cancer clusters.…”

“…Using a spatial modeling approach would have also served to mitigate multiple testing for Ghazawi et al, who tested the incidence and mortality rates for each region against the national rate. This led to hundreds of hypothesis tests and raised concerns over spurious associations.…”