Alterations and diagnosis potential of serum lipid profiles in rheumatoid arthritis patients

The aim of the study is to compare the serum lipid profiles of rheumatoid arthritis (RA) patients and healthy volunteers and find the association between the serum lipid profiles and RA in Chinese Han population. Blood samples were collected from 2001 patients with early diagnostic RA and 3883 matched healthy controls for their serum lipid profiles including total-cholesterol (T-CHOL), triglyceride (TG), high density lipoprotein-cholesterol (HDL-C) and low density lipoprotein-cholesterol (LDL-C). Associations between RA and alterations of the lipid profiles were estimated with univariable and multivariable logistic regression. T-CHOL (OR = 0.54, 95% CI: 0.50-0.58), TG (OR = 0.4, 95% CI: 0.37-0.44) and HDL-C (OR = 0.01, 95% CI: 0.01-0.01) were significant protective factors, while LDL-C (OR = 5.58, 95% CI: 4.95-6.29) was a significant risk factor to RA. The results from Random Forest revealed that serum lipid profiles have powerful prediction ability to distinguish RA from normal controls with sensitivity of 94.7%, specificity of 90.4% and accuracy of 93.2%, respectively. Risk discriminatory model based on T-CHOL, TG, HDL-C, LDL-C, gender and age in logistic regression further showed strong prediction ability of serum lipid profiles in RA with the area under the curve (AUC) of receiver operating characteristic (ROC) = 0.972 (95% CI: 0.968-0.976). Serum lipid profiles were significantly associated with rheumatoid arthritis and have important implication in RA auxiliary diagnosis.