During May 2015 and September 2015, eight communities in Sichuan were randomly selected. We only screened all residents aged ≥40 years who lived in the community for more than 6 months. A structured face-to-face questionnaire was used in the initial screening by interviewers. The eight conventional stroke risk factors were assessed in the CNSSS questionnaire, including hypertension, diabetes, dyslipidemia, AF, overweight/obesity, smoking, physical inactivity, and family history of stroke. Subjects were classified as the high-risk population for stroke if they experienced at least three of the aforementioned eight stroke-related risk factors or a history of stroke. Exclusion criteria included: (i) severe cardiovascular, liver or renal disease, (ii) blood system disease and coagulation dysfunction, (iii) acute and chronic inflammation, malignant tumors and immune system diseases, (iv) transient ischemic attack only, and (v) individuals declined to participate in the study. Finally, 16,892 valid individual records were enrolled. Among them, 2,893 participants were the high-risk population. However, DNA and carotid ultrasonography information was obtained from 2,377 participants among the 2,893 high-risk population. A total of 2,377 subjects were classified into carotid stenosis and non-carotid stenosis (<15% stenosis) groups on the basis of their vascular imaging. The associations of gene-gene interactions among the 6 variants with carotid stenosis were performed by the GMDR approach.

The associations of gene-gene interactions among the 6 variants in 3 genes related to inflammation and endothelial function with carotid stenosis were performed by the GMDR approach. The best model for carotid stenosis including ITGA2 rs4865756 and HABP2 rs7923349 scored 8/10 for cross-validation consistency and 10/10 for sign testing.

Microsoft Excel, Adobe AcrobatFunding provided by: Scientific Research Foundation of Sichuan Provincial Health Department*Crossref Funder Registry ID: Award Number: 16ZD046