# Comparison of the triglyceride-glucose index and triglyceride-glucose-body mass index for predicting non-alcoholic fatty liver disease in elderly diabetic patients

## TL;DR

The study compares two simple blood-based metrics to detect fatty liver disease in elderly diabetic patients, finding that one is more accurate.

## Contribution

The paper introduces and validates TyG-BMI as a more accurate non-invasive predictor of NAFLD in elderly diabetic patients compared to TyG.

## Key findings

- TyG-BMI showed higher predictive accuracy for NAFLD detection in elderly diabetic patients compared to TyG.
- A TyG-BMI cut-off of 212.886 can rule out NAFLD with high sensitivity and negative predictive value.
- A TyG-BMI cut-off of 251.741 can rule in NAFLD with high specificity and positive predictive value.

## Abstract

Non-alcoholic fatty liver disease (NAFLD) is posing a challenge to global health systems. Developing an effective, simple, noninvasive method to identify NAFLD in elderly diabetic patients, track disease progression, and monitor treatment effects is importance. This study aims to assess the value of triglyceride-glucose index (TyG) and triglyceride-glucose-body mass index (TyG-BMI) in detecting NAFLD elderly patients with diabetes mellitus (DM).

This study enrolled 6,882 individuals aged 60 years or older with DM who underwent liver ultrasonography in this cross-sectional study at Zhaobaoshan Residential District Community Health Service Center, from Jan. 1, 2015, to Oct. 19, 2023. And data was accessed for research purposes after Oct. 1,2024. Participants were randomly divided into a training group and a validation group in a 7:3 ratio. The diagnostic values of TyG and TyG-BMI were assessed using the area under the receiver-operating characteristic curve (AUROC) and Decision Curve Analysis (DCA). Two cut-off points were selected to rule out or rule in NAFLD, and we explored their specificity, sensitivity, negative predictive value, and positive predictive value.

There were 2,210 and 927 participants with NAFLD in the training and validation groups. In a fully adjusted model, TyG and TyG-BMI were correlated with an increased risk of NAFLD in the training group (TyG: OR=3.920, P < 0.001; TyG-BMI: OR=1.032, P < 0.001). These results were consistent in the validation group. The AUCs of TyG and TyG-BMI indicated that both had predictive value for NAFLD, with TyG-BMI showing the higher predictive accuracy. DCA suggested that TyG-BMI is preferable in clinical settings for both groups. In the training group, with a TyG-BMI cut-off of 212.886, the sensitivity was 80.6%, specificity 57.5%. With a cut-off of 251.741, the sensitivity was 32.6%, specificity 90.7%. Thus, a TyG-BMI < 212.886 could rule out NAFLD (SE = 80.6%, NPV = 77.8%), while a TyG-BMI ≥ 251.741 could rule in NAFLD (SP = 90.7%, PPV = 74.8%). These findings were similar in the validation group, with a TyG-BMI < 212.886 ruling out NAFLD (SE = 80.0%, NPV = 77.3%) and a TyG-BMI ≥ 251.741 ruling in NAFLD (SP = 91.5%, PPV = 76.4%).

In conclusion, TyG-BMI is more accurate than TyG in predicting NAFLD in elderly participants with diabetes. This simple, non-invasive, and cost-effective tool effectively classifies elderly diabetic patients with and without NAFLD.