The role of 3D body scan derived anthropometric measurements in predicting subclinical atherosclerosis in the REACT cohort

Background Cardiovascular disease (CVD) is the leading cause of death worldwide. The most common underlying factor, atherosclerosis, develops silently over decades and has been extensively linked to obesity. Body mass index and waist circumference are traditionally used as measures of central obesity but do not consider detailed body shape and are not a part of current CVD risk predictors. Purpose This study aimed to investigate the ability of a range of 3D body scan derived anthropometric measurements to predict the presence of subclinical atherosclerosis as well as the total volume of atherosclerotic plaque. Methods This is a preliminary analysis of a subset of Danish individuals in the REACT Phase I (DETECT) study which aims to include 16,000 Danish and Spanish individuals aged 18-70 years, not diagnosed with atherosclerotic cardiovascular disease. The presence and volume of atherosclerotic plaque in the carotid and femoral arteries bilaterally was determined using 3D vascular ultrasound. Full body optic 3D scans of participants in an upright position were used to calculate a range of body measurements. These parameters were used along with age and sex in two random forest analyses to determine their ability to predict subclinical atherosclerosis, both as a binary outcome (any/none) and according to burden of atherosclerosis. The second analysis included only participants who were shown to have subclinical atherosclerosis. Variable importance was assessed by ranking variables by mean decrease in Gini impurity index for the classification model and by percentage increase in mean squared error for the regression model. The models were evaluated using accuracy, R2 and root mean squared error. Results For this preliminary subanalysis, we included 381 participants with visualised atherosclerotic plaque and 258 without. The median age was 49 years [IQR: 40 - 59] and 48% were male. Mean height was 172 cm [SD: 9] and mean waist circumference was 93 cm [SD: 14]. The classification model indicated that age was the most important variable followed by thigh circumference and waist circumference (Figure 1). Accuracy of prediction was 0.74. Among participants shown to have plaque, age was again found to contribute the most to the prediction of the total volume of plaque, followed by bust circumference, wrist circumference and abdominal circumference (Figure 1). R2 was 0.17 and root mean squared error was 1.33. Conclusion While age was the strongest predictor of subclinical atherosclerosis, several non-traditional measures showed a notable contribution to the models. Body measures predictive of the presence of subclinical atherosclerosis seemed to differ substantially from those predictive of the burden of plaque.