Performance of atherogenic index of plasma combined with anthropometric indices as a novel predictor for non-alcoholic fatty liver disease in overweight and obese individuals

BACKGROUND: Non-alcoholic fatty liver disease (NAFLD) is an emerging public health concern, including in Bangladesh, where cases are increasingly being recognized among overweight and obese individuals. It is often linked to lipid abnormalities such as raised triglycerides (TG) and reduced high-density lipoprotein cholesterol (HDL-C). Since diagnostic imaging is not always feasible, simple biochemical and anthropometric measures may help identify at-risk individuals. The Atherogenic Index of Plasma (AIP), calculated as the logarithmic ratio of TG to HDL-C, and measures like waist circumference (WC) and Body Mass Index (BMI) may be useful for this purpose. This study aimed to evaluate the performance of AIP alone and in combination with anthropometric indices (AIP-WC-BMI; AWB Model) for predicting NAFLD in overweight and obese adults.METHODS: This cross-sectional analytical study was conducted in the Department of Biochemistry and Molecular Biology, BSMMU, Dhaka. A total of 218 overweight and obese individuals were recruited by non-probability sampling from the Hepatology outpatient department. After informed consent, detailed clinical and anthropometric data (height, weight, WC, BMI) were collected. BMI was calculated as weight (kg)/height (m2). Abdominal ultrasonography was performed to diagnose NAFLD. Fasting blood samples were analyzed for lipid profile and alanine transaminase (ALT). Among the participants, 110 were classified as NAFLD and 108 as non-NAFLD. AIP was calculated as log(TG/HDL-C), and the AWB Model was generated through regression combining AIP, WC, and BMI. Data were analyzed using SPSS 27.0. Continuous variables were compared using Student’s t-test or Mann-Whitney U Test. ROC analysis determined the area under the curve (AUC), sensitivity, specificity, predictive values, accuracy, and Youden’s J statistic. Statistical significance was set at P<0.05.RESULTS: Most participants were female (58.3%). NAFLD subjects showed significantly higher TG, total cholesterol, ALT, AIP, WC, and BMI, and lower HDL-C than non-NAFLD (P<0.001). The frequency of NAFLD increased across higher AIP quartiles in both sexes. ROC analysis revealed all parameters were significant predictors: AUCs were 0.843 (AIP), 0.790 (WC), 0.730 (BMI), and 0.890 (AWB Model), with the combined model performing best. Optimal cut-offs were 0.49 (AIP), 86.5 cm (WC), 25.9 kg/m2 (BMI), and -0.02 (AWB Model). AIP showed highest sensitivity (95.5%) but moderate specificity (62%), while the AWB Model achieved balanced sensitivity (86.4%) and specificity (81.2%) with the highest accuracy (83.9%) and validity (Youden’s J = 67.6%). WC and BMI showed lower validity and predictive power compared to the combined model.CONCLUSIONS: All studied parameters were effective predictors of NAFLD, but the AWB Model demonstrated superior diagnostic performance, with high sensitivity, specificity, and validity. AIP alone was highly sensitive but less specific, while WC and BMI were less reliable. The AWB Model, integrating biochemical and anthropometric indices, offers a cost-effective, simple, and noninvasive tool for early detection of NAFLD in overweight and obese individuals, potentially aiding in timely intervention and prevention of disease progression.