Abstract
The tobacco use disorders are the largest preventable cause of morbidity and mortality in the world. A substantial barrier to the development of better intervention and screening measures is the lack of clinically employable biomarkers to detect the existence and extent of tobacco consumption. In prior work, we and others have shown that array based assessment of DNA methylation status at cg05575921 is a sensitive and quantitative method for assessing cigarette consumption. Unfortunately, in general, arrays are not practical clinical tools. Herein, we detail the prediction performance metrics and dose dependency of a clinically implementable droplet digital PCR (ddPCR) assay for cigarette consumption in adults. First, we demonstrate that measurements of cg05575921 as determined by Illumina array and ddPCR are highly correlated (R2 = 0.98, n = 92). Second, using clinical data and biomaterial from 177 subjects ranging from 18 to 78 years of age, we show that the Receiver Operating Characteristic (ROC) area under the curve (AUC) for classifying smoking status using methylation status at cg05575921 is 0.99. Finally, we conduct modeling analyses of cigarette consumption over discrete time periods to show that methylation status is best correlated with mean cigarette consumption over the past year (R2 = 0.5) and that demethylation at cg05575921 is dose dependent with a demethylation (delta beta) of 1% being equivalent to 1.2 cigarettes per day. But we do not find a relationship between Fagerstrom score and DNA methylation. We conclude that ddPCR assessment of cg05575921 methylation is an accurate method for assessing the presence and extent of cigarette consumption in adult subjects. We suggest that skillful clinical implementation of this approach alone or in combination with other assessment methods could lead to substantial reduction of cigarette consumption related morbidity and mortality.