Diagnostic value of small dense low-density lipoprotein and trace elements in coronary artery disease.

Abstract

Coronary artery disease (CAD) is a worldwide leading cause of death. Considering that 20%-40% of patients with CAD have a long asymptomatic period of atherosclerosis, it has become urgent to explore the feasibility of diagnosing CAD at an early stage. This is an observational, case-control study, a total of 489 consecutive CAD patients and 75 non-CAD controls were recruited. The levels of low-density lipoprotein subfractions (LDLC1-7) in serum were measured by the Quantimetrix Lipoprint LDL system. The levels of 18 trace elements (vanadium, chromium, manganese, cobalt, nickel, copper, zinc, gallium, arsenic, selenium, strontium, cadmium, tin, antimony, barium, mercury, thallium, and lead) were tested using inductively coupled plasma mass spectrometry. Six machine learning algorithms (Logistic Regression, K Neighbors, GaussianNB, Random Forest, Decision Tree and XGBoost) were used to construct CAD diagnostic models. The levels of LDLC-3, LDLC-4, LDLC-5, and lead were significantly higher in CAD patients, while the levels of LDLC-1, chromium, manganese, cobalt, and strontium were lower (p < 0.05 for all). Univariate logistic regression analysis indicates that LDLC-3, LDLC-4, and lead were the risk factors for CAD development (odds ratio >1 and p < 0.05 for all), while LDLC-1, chromium, manganese, cobalt, and strontium were the protective factors for CAD (odds ratio < 1 and p < 0.05 for all). XGBoost had the best overall diagnostic performance among the six algorithms. There are significant differences in the levels of several LDL subfractions and trace elements between non-CAD controls and CAD patients. These biomarkers may help the diagnostic of CAD while applying machine learning algorithms.