Associations of four triglyceride-related metabolic surrogate indices with heart rate variability and confirmed cardiovascular autonomic neuropathy in type 2 diabetes: a cross-sectional study

Background. Cardiovascular autonomic neuropathy (CAN) is an underdiagnosed complication of type 2 diabetes, and reduced vagally mediated heart rate variability (HRV) is an early functional manifestation. We compared four routinely derived insulin resistance-related indices—the triglyceride-glucose (TyG) index, triglyceride-to-HDL-cholesterol ratio (TG/HDL-C), C-reactive protein-triglyceride-glucose index (CTI), and waist circumference-adjusted TyG index (TyG-WC)—in relation to cardiac autonomic function. Methods. This cross-sectional study evaluated resting vagally mediated HRV, expressed as log-transformed root mean square of successive differences (log-RMSSD) and high-frequency power (log-HF), and confirmed CAN, defined as at least two abnormal cardiovagal reflex tests using age-stratified reference limits within the Toronto Consensus framework. Multivariable linear and logistic regression assessed associations, and areas under the receiver operating characteristic curve (AUCs) compared discrimination for confirmed CAN. An exploratory analysis examined whether adding urinary albumin-to-creatinine ratio (UACR) to CTI improved cross-sectional stratification. Results. Confirmed CAN was present in 235 of 1,427 evaluable participants (16.5%). After full adjustment, TyG, TG/HDL-C, and CTI were inversely associated with both HRV outcomes (all P < 0.05). TyG-WC was associated with log-HF (β − 0.074; P = 0.037), but not log-RMSSD (P = 0.213). CTI (odds ratio 1.29 per 1-SD, 95% CI 1.06–1.57; P = 0.010) and TyG-WC (1.25, 1.05–1.48; P = 0.012) were associated with confirmed CAN, whereas TyG and TG/HDL-C were not. CTI had the highest AUC (0.641, 95% CI 0.598–0.683), although pairwise differences were not significant. The CTI–CAN association persisted after expanded clinical and medication adjustment, whereas the TyG-WC association attenuated. In exploratory analysis, CTI plus ln(UACR + 1) improved discrimination over CTI alone (AUC 0.692 vs 0.647; DeLong P = 0.001), with observed CAN prevalence increasing from 7.7% to 33.0% across ascending predicted-probability bands. Conclusions. Insulin resistance-related indices were associated with reduced vagally mediated HRV, while CTI showed the most robust association with confirmed CAN. CTI combined with UACR may improve cross-sectional identification, but these associational findings require external, preferably prospective, validation.