Diagnostic performance of the triglyceride–glucose index and its derivatives for insulin resistance in women with polycystic ovary syndrome

Background Polycystic ovary syndrome (PCOS), one of the most common endocrine disorders in women of reproductive age, is frequently accompanied by insulin resistance (IR) and increased cardiometabolic risk. The triglyceride–glucose (TyG) index has recently gained attention as a simple surrogate marker of IR derived from routinely available biochemical parameters. We aimed to evaluate the diagnostic utility of the TyG index and its derivatives for identifying IR in women with PCOS. Methods In this cross-sectional study, we included 136 women with PCOS and 94 age-matched healthy controls. Anthropometric measurements (body mass index [BMI], waist circumference) and fasting biochemical parameters (glucose, insulin, lipid profile), as well as reproductive hormones, were obtained from medical records. IR and β-cell function were assessed using HOMA-IR, HOMA-β, the Quantitative Insulin Sensitivity Check Index (QUICKI) and the fasting glucose/insulin ratio (FG-IR). The TyG index, TyG-BMI and triglyceride/high-density lipoprotein cholesterol (TG/HDL-C) ratio were calculated. Group comparisons were performed using appropriate parametric or non-parametric tests. Correlations between TyG-based indices and conventional IR markers were evaluated using Spearman’s rho. The diagnostic performance of TyG, TyG-BMI and TG/HDL-C for detecting IR (defined by HOMA-IR ≥ 2.5) was analysed using receiver operating characteristic (ROC) curves and Youden’s index to determine optimal cut-off values. Results Women with PCOS had significantly higher BMI, waist circumference, fasting glucose, insulin, HOMA-IR, TyG and TyG-BMI, and lower QUICKI and FG-IR compared with controls (all p < 0.05). TyG and TyG-BMI showed moderate to strong positive correlations with HOMA-IR and TG/HDL-C, whereas QUICKI and FG-IR were inversely correlated with these indices. In ROC analysis among women with PCOS, the TyG index demonstrated good ability to identify IR (AUC ≈ 0.75), with an optimal cut-off around 8.2 providing a balanced sensitivity and specificity. TyG-BMI and TG/HDL-C also showed statistically significant, but comparatively lower, discriminative performance. Conclusions The TyG index and TyG-BMI are significantly elevated in women with PCOS, correlate with established IR markers and provide moderate diagnostic accuracy for identifying IR using only fasting glucose and lipid values. Although they do not outperform HOMA-IR, TyG-based indices may serve as practical, low-cost complementary tools for metabolic risk assessment in PCOS, particularly in settings where insulin measurements are not routinely available.