Comparative discriminative ability of CVAI and traditional insulin resistance indices for MAFLD in Chinese adults with type 2 diabetes.

BACKGROUND

Metabolic dysfunction-associated fatty liver disease (MAFLD) is highly prevalent in patients with type 2 diabetes mellitus (T2DM), necessitating simple and reliable non-invasive screening tools. The Chinese Visceral Adiposity Index (CVAI) has been developed to assess visceral adipose dysfunction, but its discriminative performance for MAFLD compared with traditional insulin resistance (IR) indices in T2DM populations remains unclear.

METHODS

This retrospective cross-sectional study included 2,945 Chinese adults with T2DM recruited from 45 community health centers. We calculated seven IR indices: CVAI, TyG, TyG-BMI, TyG-WC, TyG-WHtR, METS-IR, and SPISE. Multivariable logistic regression was employed to assess independent associations with MAFLD. Discriminative ability was evaluated using receiver operating characteristic (ROC) curve analysis with Delong's tests. Net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated to assess incremental discriminative value over a baseline clinical model. Restricted cubic spline (RCS) regression, decision curve analysis (DCA), and subgroup analyses were also performed.

RESULTS

Among participants, 1,313 (44.6%) were diagnosed with MAFLD. All IR indices were independently associated with MAFLD after full adjustment (all P < 0.001). CVAI demonstrated the highest discriminative ability (AUC = 0.754, 95% CI: 0.737-0.771), significantly outperforming all other indices (all P < 0.001). Adding CVAI to the baseline model significantly improved risk reclassification (NRI = 0.598, 95% CI: 0.533-0.665; IDI = 0.081, 95% CI: 0.072-0.090) and yielded the broadest clinical utility across threshold probabilities of 10-90% in DCA. The association between CVAI and MAFLD was significantly stronger in females and older adults (both P for interaction < 0.05). RCS analysis revealed a predominantly linear relationship (non-linear P = 0.0581).

CONCLUSION

CVAI is independently associated with MAFLD and exhibits superior discriminative performance compared with traditional IR indices in Chinese adults with T2DM. As a simple, non-invasive index derived from routine clinical parameters, CVAI holds promise as an effective screening tool for MAFLD, particularly in females and older adults. Prospective studies are warranted to validate its diagnostic utility.