A simpler diagnostic formula for screening nonalcoholic fatty liver disease.

OBJECTIVE

To increase the accuracy of non-invasive diagnosis of nonalcoholic fatty liver disease (NAFLD), clinical and laboratory NAFLD indicators were integrated into a diagnostic formula.

METHODS

A total of 141 patients with clinically diagnosed NAFLD and 30 healthy controls were enrolled. We collected case history, body weight, height and mass index (BMI), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma glutamyl transpeptidase, blood urea nitrogen and blood uric acid (UA), serum creatinine, plasma total cholesterol, triglyceride, low density lipoprotein, glycosylated hemoglobin, fasting plasma glucose, fasting insulin, ultrasonic tests, Fibroscans, and other data. Linear correlation, multiple linear regressions, and receiver operating characteristic (ROC) curve methods were used to process and analyze the collected data. The performance of Fibroscan and our diagnostic formula was compared in reference to the findings of liver biopsy.

RESULTS

The identified NAFLD diagnostic indices consisted of BMI, ALT, AST and UA. A regression formula was proposed as: CAP = 113.163 + 0.252 * ALT + 6.316 * BMI. Diagnosis of the area under the ROC curve was 0.927, the sensitivity was 87.68%, and specificity was 90%. The cutoff was 277.67 (p < 0.01). The accuracy of the NAFLD diagnosis with the proposed formula was significantly higher than FibroScan (82.6% vs 69.6%; p = 0.005).

CONCLUSIONS

NAFLD diagnosis with the proposed formula demonstrated both high sensitivity and specificity, and its accuracy was significantly higher than FibroScan. This formula only utilized non-invasive clinical and laboratory findings and the calculation was simple. It can be conveniently used for clinical diagnosis of NAFLD.