Association between urine retinol-binding protein levels and nonalcoholic fatty liver disease: A cross-sectional study in Chinese population.

Abstract

The prevalence of nonalcoholic fatty liver disease (NAFLD) has been rapidly increased, becoming a public health problem worldwide. Our objective was to investigate the association between urine retinol-binding protein (RBP) and NAFLD in a Chinese population and develop a multivariate logistic regression model for NAFLD prediction. A total of 317 NAFLD patients and 391 healthy controls were enrolled in this cross-sectional study based on inclusion and exclusion criteria, from whom fasting urine and blood were collected for further study. Urine RBP level and other parameters were measured and compared between NAFLD subjects and controls. Urine RBP levels (expressed by RBP/creatinine ratio) in NAFLD patients were significantly higher than controls (median 133.1mg/g vs 110.7mg/g; P<.001). Urine RBP/creatinine ratio was verified as an independent factor for NAFLD prediction after adjustment in multivariate logistic regression. The area under curve (AUC) of receiver operating characteristic (ROC) was 0.889 with the 95% confidence interval from 0.867 to 0.912.With a cutoff point of 0.215, the sensitivity and specificity of urine RBP/creatinine ratio in NAFLD prediction were 81.1% and 84.5%, respectively. Our results demonstrated that urine RBP/creatinine ratio was an independent risk factor for NAFLD while the predictive model for NAFLD diagnosis is noninvasive with high sensitivity and specificity.