Prediction model for different progressions of Atherosclerosis in ApoE mice based on lipidomics.

Atherosclerosis (AS) is a progressive disease with a complex pathogenesis which is characterized by dyslipidemia and changes in the vascular wall composition. According to the degree of lesions, atherosclerosis can be divided into four stages: hyperlipidemia, lipid stria, fiber plaque, and atherosclerotic plaque. The present study aimed to establish a prediction model for the different pathological stages of AS based on lipidomics. ApoE mice and C57BL/6 mice fed a normal diet were divided into seven groups according to the feeding time (8, 12, 16, 20, 24, 28, and 32 weeks). The changes in the lipid composition and serum content were detected using ultra-performance liquid chromatography coupled with quadrupole time-of-flight high-definition mass spectrometry (UPLC-Q-TOF/MS). Through the results of serum total cholesterol, triglyceridelow density lipoprotein at each time and HE staining of the head and arm artery, the seven time points of the model group were corresponding to the four courses of atherosclerosis. In accordance with the lipid data of each course of AS and mathematical modeling, this study established a multi-index prediction model of the different processes of AS. Notably, while establishing the model, several indicators were combined with one of four dimension reduction methods, such as principal component logistics regression method, cumulative logistics regression method, Partial least squares-discriminant analysis(PLS-DA), and canonical discriminant analysis (CDA). The error rate of the four methods were 28.5%, 16.22%, 18.24%, and 14.86%, respectively. CDA had the lowest error rate and the best prediction accuracy of the AS different courses for the training and verification sets after 5-fold cross-validation of this model. This study showed that lipidomics combined with mathematical methods could establish a non-invasive and accurate model for the prediction of AS.