Cluster Analysis Evaluating PM2.5, Occupation Risk and Mode of Transportation as Surrogates for Air-pollution and the Impact on Lung Cancer Diagnosis and 1-Year Mortality

Abstract

Objective: Epidemiological studies have reported the close relationship between risk for lung cancers and air pollution<br />in particular, for non-smoking related lung cancers. However, most studies used residential address as proxies which may<br />not estimate accurately an individual's air pollution exposure. Therefore, the aim of this study was to identify risk factors<br />such as occupation and mode of transportation associated with lung cancer diagnosis and death. Methods: Subjects<br />with lung cancer (n=514) were evaluated both by chart reviews for clinical data and interviews to determine residential<br />address for ten years, main occupation and main mode of transportation. Annual particulate matter with diameter size<br />less than 2.5 micrometre (PM2.5) concentration were calculated based on particulate matter with diameter size less than<br />10 micrometre (PM10) data recorded by Malaysian Department of Environment. Logistic regression analysis, cluster<br />analysis and the Cox regression analysis were performed to the studied variables. Results: This study concurred with<br />previous studies that lung adenocarcinoma were diagnosed in predominantly younger, female non-smokers compared<br />to the other types of lung cancers. Lung adenocarcinoma subjects had annual PM2.5 that was almost twice higher than<br />squamous cell carcinoma, small cell carcinoma and other histological subtypes (p=0.024). Independent of smoking,<br />the κ -means cluster analysis revealed two clusters in which the high risk cluster involves occupation risk with air<br />pollution of more than four hours per day, main transportation involving motorcycle and trucks and mean annual PM2.5<br />concentration of more than 30 based on residential address for more than ten years. The increased risk for the high-risk<br />cluster was more than five times for the diagnosis of lung adenocarcinoma (OR=5.69, 95% CI=3.14-7.21, p<0.001).<br />The hazard ratio for the high-risk cluster was 3.89 (95% CI=2.12-4.56, p=0.02) for lung adenocarcinoma mortality at<br />1 year. Conclusion: High-risk cluster including PM2.5, occupation risk and mode of transportation as surrogates for<br />air-pollution exposure was identified and highly associated with lung adenocarcinoma diagnosis and 1-year mortality.