Abstract: [Objective] The best cut-off value of biomarkers for the diagnosis of fatty liver was screened by the receiver operating characteristic (ROC) curve. [Method] Serum was randomly collected from 77 Chinese Holstein dairy cows 7 days postpartum to determine physiological and biochemical indicators. According to the Y value calculated by the Reid equation, the cows were divided into normal group (Y>1) and fatty liver group (Y<1), and the differences in serum indexes of cows in two groups were compared, and the correlation analysis with Y value was performed to screen the group. The indicators with significant differences and correlations are used as biomarkers, and the receiver operating characteristic (ROC) curve was used to evaluate the AUC value, diagnostic sensitivity and specificity, and the best cut-off value of the biomarker. Finally, the accuracy of the Raid equation in diagnosing fatty liver was used as the standard, and the best cut-off value of NEFA and AST was used to detect the fatty liver of 461 dairy cows. [Result] Physiological and biochemical test results showed that NEFA, glucose(Glu), AST, alanine amino transferase(ALT) between two groups had extremely significant difference (P<0.01), total protein(TP) had significant difference (P<0.05), and other indicators had no significant difference (P>0.05). NEFA, Glu, AST, ALT, TP, triglycerides(TG), low-density lipoprotein cholesterol(LDL-C), insulin(Ins), glucagon(Gln) were significantly correlated with Y value and could be used as biomarkers. The AUC value of NEFA, AST, ALT and Gln could be more than 0.5. Accurately diagnose fatty liver, the diagnostic sensitivity were 81.8%, 72.7%, 63.6% and 54.5%, the specificity were 92.8%, 89.4%, 84.8% and 69.7%, the best cut-off value were 471.74 μmol/L, 95.50 U/L, 46.50 U/L and 64.60 ng/mL, respectively. The accuracy of using NEFA, AST and NEFA+AST best cut-off values to diagnose fatty liver were 89.96%, 73.49% and 69.08%, respectively. The incidence rate diagnosed by the Y value was 54.01%, and the incidence rates diagnosed by NEFA, AST and NEFA+AST were 48.59%, 39.70% and 37.31%, respectively. [Conclusion] NEFA, AST, ALT and Gln could be used to quickly and accurately diagnose fatty liver.

Key words: perinatal dairy cows; fatty liver; diagnosis; biomarker; receiver operating characteristic curve