Discrimination and stratification tests of cardiovascular disease risk assessment models against ultrasound detection of carotid plaques in type 2 diabetics

Abstract

Atherosclerosis is the early stage of arterial disease, and underlies development of cardiovascular disease (CVD) and stroke. Although sophisticated models for assessing CVD and stroke risk have been derived based on large- scale prospective studies, their abilities in detect ing the presence or absence of atherosclerotic plaque have not been investigated. This study aimed to evaluate and compare discriminatory and risk stratifying abilities of 13 CVD risk assessment models against the ultrasound detection of carotid plaques in type 2 diabetes mellitus (T2DM) patients. Forty-nine T2DM subjects were recruited with informed consent, and major anthropomorphic and biomarker data for these models were collected. The model risk scores were evaluated against the carotid plaques detected by Doppler ultrasound. Only the FHS-Lpts-CHD-10Y model, which is a variant of the Framingham model, revealed an area under the receiver operating curve (AUROC) that was significantly different from a random scoring approach (AUROC: 0.681, p < 0.05), and was able to stratify the risk levels of carotid plaque presence (Chi-Square statistic: 5.99, p < 0.05).