Research was carried out at the Endocrinology Unit of Piekary Medical Centre, located within St. Luke’s Municipal Hospital in Piekary Śląskie, Poland. The study cohort included women with diagnosis of PCOS who met the 2004 Rotterdam criteria. These criteria required the presence of at least two of the following three abnormalities: 1. Polycystic ovarian morphology on ultrasound, defined as a ≥12 follicles measuring 2–9 mm in each ovary and/or an increased ovarian volume exceeding 10 mL, 2. Clinical and/or biochemical evidence of hyperandrogenism (serum testosterone concentration >70 ng/dL, 3. Oligomenorrhea or amenorrhea (17).

Hyperandrogenism was diagnosed based on elevated androgen levels (testosterone and/or androstenedione) and/or the presence of clinical features such as excessive body and facial hair (hirsutism) and acne.

Transvaginal ultrasonography was performed using a 6–12 MHz transducer, which provides optimal resolution for detailed visualization of ovarian architecture in the assessment of polycystic ovarian morphology.

A total of 101 women were initially screened for eligibility. Following the exclusion of individuals who did not fulfil the study criteria or presented additional disqualifying factors, 49 participants were ultimately enrolled. The recruitment flow, with the inclusion and exclusion criteria are presents on Figure 1. .

Diagnostic criteria for carbohydrate metabolism disturbances were defined as follows: impaired fasting glucose (100–125 mg/dL), impaired glucose tolerance (plasma glucose concentration of 140–199 mg/dL at 120. min of the OGTT) (26, 27) and insulin resistance, determined by a threshold HOMA-IR value ≥2.1 (27). Diabetes mellitus was diagnosed if at least one of the following criteria was met: the presence of symptoms of hyperglycaemia with a random plasma glucose ≥200 mg/dL, two separate fasting plasma glucose measurements ≥126 mg/dL, plasma glucose concentration ≥200 mg/dL at 120. min of the OGTT (26).

All procedures were carried out in line with the ethical standards set forth in the Declaration of Helsinki. The research protocol was reviewed and approved by the local bioethics committee (KNW/0022/KB1/143/15), and informed consent for participation and data analysis was obtained from every subject.

Research was carried out at the Endocrinology Unit of Piekary Medical Centre, located within St. Luke’s Municipal Hospital in Piekary Śląskie, Poland. Blood samples were collected during morning hours, prior to any food intake. A one-millilitre aliquot of blood, initially collected for standard clinical evaluations, was retained for subsequent analysis. Following centrifugation, these samples were cryopreserved at -70 °C until the required assays were performed. All laboratory analyses were carried out in accordance with the Standard Work Instructions (SWI) at KORLAB Medical Laboratories. Plasma glucose was determined with a reference enzymatic hexokinase method on the Cobas Integra 400+ automated analyzer. Glycated haemoglobin (HbA1c) was assessed by high-performance liquid chromatography (HPLC) employing reversed-phase cation-exchange chromatography, using the ADAMS A1c HA-8180 analyzer. Fasting insulin levels were measured with the ARCHITECT insulin assay, a one-step immunoassay based on chemiluminescent microparticle technology with a microparticle tracer, enabling sensitive detection of human insulin in plasma or serum. Serum testosterone concentrations were quantified using a chemiluminescent microparticle immunoassay (CMIA) on the Abbott Alinity platform. Total cholesterol concentrations were measured using the Cobas Integra 400+ analyzer, employing an enzymatic colorimetric method. LDL cholesterol (LDL-C) was directly quantified with the MULTIGENT Direct LDL assay, which enables accurate measurement of LDL-C in serum or plasma without prior sample pretreatment or centrifugation, using the Architect c8000 analyzer. HDL cholesterol (HDL-C) concentrations were determined via a homogeneous enzymatic colorimetric technique on the Cobas Integra 400+ analyzer. Triglyceride levels were assessed using an enzymatic colorimetric method based on glycerol phosphate oxidase and 4-aminophenazone, also on the Cobas Integra 400+ platform.

The following biochemical parameters were evaluated: fasting plasma glucose, insulin, glycated haemoglobin A1c (HbA1c), total cholesterol (TC), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglycerides (TG). These measurements were subsequently used for the calculation of relevant indices (Table 1).

Standard anthropometric measurements, including body weight [kg], height [cm], waist circumference [cm] (WC), and hip circumference [cm] (HC), were obtained in the morning, following an overnight fast. Participants were lightly clothed and barefoot during all assessments. All procedures adhered to standardized protocols, and calibrated, validated instruments were employed. Body weight was recorded to the nearest 0.01 kg using the Tanita BC 420 S MA Body Composition Analyzer (MDD93/42 EEC certified; Tanita Corp., Tokyo, Japan). Height was measured to the nearest 0.05 cm using the Tanita HR100 Stadiometer (Tanita Corp., Tokyo, Japan). Circumferences of the waist, hips, calves, and upper arms were determined to the nearest 1 mm with a SECA 203 measuring tape (Seca, Hamburg, Germany).

From the acquired measurements, individual anthropometric indices were computed and are summarized in Table 1.

Homeostatic model assessment (HOMA-IR) index was calculated using the formula: HOMA-IR = fasting insulinemia (mU/mL) x fasting glycemia (mmol/L)/22.5 (27). Insulin resistance was defined as HOMA-IR ≥2.1 (27), consistent with the criteria applied in the Study participant’s section.

Cut-off values and standards of the tested measurements and anthropometric measurements are presented in Table 2.

Among the studied women, 26 had a WC greater than 88 cm, which may lead to significantly increased risk of metabolic complications. Excess body mass, defined as a BMI >25 was observed in 25 women. Moreover, 22 women with PCOS presented an increased risk of metabolic complication based on WHR index ≥0, 85. A total of 30 participants had a WHtR above 0.50, a threshold that correlates with elevated cardiometabolic risk. Excess body fat assessed using the BAI index was identified in 25 women. VAI >1.675 had 17 participants, which is associated with a risk of metabolic diseases. Based on the LAP index (<41.30), 27 women were classified as having prediabetes or diabetes. 28 participants had a BRI below 4.910. Finally, 21 women had an ABSI <0.076, which is linked to an elevated CVR and diabetes.

Statistical analyses were performed using PS IMAGO PRO 10 (IBM SPSS Statistics 29) and MedCalc^® version 23.3.7. Data distribution was assessed with the Shapiro-Wilk test. Continuous variables are presented as means ± standard deviations for normally distributed data or as medians with interquartile ranges (Q1–Q3) for non-normally distributed data. Group comparisons were conducted using Student’s t-test for normally distributed variables and the Mann-Whitney U test for non-normal variables. Correlations between continuous variables were evaluated using Pearson’s correlation coefficient for normally distributed data and Spearman’s rank correlation coefficient for non-normally distributed data.

The ability of selected anthropometric measures to predict insulin resistance risk (as assessed by HOMA-IR) was evaluated using Receiver Operating Characteristic (ROC) curve analysis. The area under the ROC curve (AUC) was calculated to quantify overall diagnostic performance, and the Youden index was applied to determine optimal cut-off values. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were also calculated. Logistic regression models were employed to examine the association between individual anthropometric indices and the presence of insulin resistance. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported. ROC analysis provided information on the discriminative ability of anthropometric parameters and their optimal thresholds for identifying individuals at risk of insulin resistance, whereas logistic regression quantified the strength of the association between these indices and insulin resistance risk. A p-value < 0.05 was considered statistically significant in all analyses.

Table 3 presents the anthropometric, metabolic, and hormonal parameters of women with PCOS. The median age was 25 years (22–29). The mean BMI was 28.3 ± 7.1 kg/m², BAI was 33 ± 5.7%. The mean fasting glucose was 5 ± 0.7 mmol/L, fasting insulin was 102.1 ± 64.1 pmol/L, median HOMA-IR was 2.7 (1.4–4.1). Insulin after 60 min in OGTT was 703.2 pmol/L (325–835.2), and mean glucose after 120 min in OGTT was 6.8 ± 2.1 mmol/L.

Women with excess weight (overweight and obesity, based on BMI) showed significantly higher fasting insulin levels compared to women with normal weight 96.86 (69.60–137.40) vs. 48.07 (35.88–73.19) pmol/L, p<0.01. Fasting glucose was also higher in the overweight group 5.27 (4.83–5.47 vs. 4.66 (4.44–4.83) mmol/L, p<0.01. During the oral glucose tolerance test, women with excess weight had higher glucose levels after 60 minutes 9.00 (6.97–9.57) vs. 5.77 (4.61–8.38) mmol/L, p<0.01. The HOMA-IR index was significantly higher in the excess weight group 3.40 (2.20–5.15) vs. 1.24 (0.90–2.06), p<0.01, indicating greater insulin resistance.

Women with excess weight (based on BAI) had significantly higher fasting insulin 102.60 (72.11–137.40) vs. 50.94 (34.80–79.11) pmol/L, p<0.01, fasting glucose 5.27 (4.94–5.47) vs. 4.66 (4.44–4.84) mmol/L, p<0.01, and HOMA-IR 4.00 (2.20–5.15) vs. 1.55 (0.90–2.68), p<0.01compared to women with normal weight. Glucose after 60 minutes in OGTT was also higher 9.00 (6.77–9.68) vs. 7.27 (5.38–8.94) mmol/L, p=0.05. These findings are summarized in Table 4.

Women with waist circumference >88 cm had significantly higher fasting insulin 104.04 vs. 48.07 pmol/L, p<0.01, fasting glucose 5.24 vs. 4.72 mmol/L, p=0.01 and HOMA-IR 4.03 vs. 1.40, p < 0.01 compared to women with WC ≤88 cm. During OGTT, glucose at 60 minutes 9.05 vs. 5.88 mmol/L, p<0.01and insulin at 60 minutes 764.86 vs. 325.03 pmol/L, p=0.0225 were also higher in the group with elevated WC (Table 5).

Women with WHR ≥0.85 exhibited significantly elevated fasting insulin 104.04 vs. 52.38 pmol/L, p<0.01, fasting glucose 5.30 vs. 4.77 mmol/L, p=0.02, glucose at 120 minutes during OGTT 6.77 vs. 5.66 mmol/L, p=0.02, glucose at 60 minutes during OGTT 9.16 vs. 6.40 mmol/L, p<0.01, insulin at 60 minutes during OGTT 768.44 vs. 373.46 pmol/L, p=0.04, and HOMA-IR 4.03 vs. 1.70, p<0.01compared to women with WHR<0.85.

Similarly, patients with WHtR≥0.5 showed significantly higher fasting insulin 100.64 vs. 44.49 pmol/L, p<0.01, fasting glucose 5.27 vs. 4.72 mmol/L, p<0.01, glucose at 120 minutes during OGTT 6.73 vs. 5.66 mmol/L, p<0.01, glucose at 60 minutes during OGTT 9.07 vs. 5.55 mmol/L, p<0.01, insulin at 60 minutes during OGTT 764.86 vs. 297.05 pmol/L, p<0.01, and HOMA-IR 3.50 vs. 1.18, p<0.01 compared to women with WHtR<0.5 (Table 6).

Patients with VAI>1.675 showed significantly higher fasting insulin 98.68 vs. 59.73 pmol/L, p<0.01, glucose after 120 minutes during OGTT 6.99 vs. 6.22 mmol/L, p=0.04, glucose after 60 minutes during OGTT 9.05 vs. 7.63 mmol/L, p=0.04, and HOMA-IR index 3.23 vs. 1.77, p<0.01compared to women with VAI ≤ 1.675.

Women with LAP≥41.30 had significantly higher fasting insulin 113.19 vs. 49.51 pmol/L, p<0.01, fasting glucose 5.30 vs. 4.72 mmol/L, p<0.01, glucose after 60 minutes during OGTT 9.10 vs. 6.90 mmol/L, p<0.01, insulin after 60 minutes during OGTT 768.44 vs. 332.20 pmol/L, p=0.0162, and HOMA-IR index 4.03 vs. 1.70, p<0.01 compared to women with LAP<41.30 (Table 7).

Patients with BRI≥4.910 showed significantly higher fasting insulin 123.77 vs. 50.94 pmol/L, p<0.01, fasting glucose 5.33 vs. 4.72 mmol/L, p<0.01, glucose after 120 minutes during OGTT 6.88 vs. 5.72 mmol/L, p=0.01, glucose after 60 minutes during OGTT 9.13 vs. 6.72 mmol/L, p<0.01, insulin after 60 minutes during OGTT 772.03 vs. 340.81 pmol/L, p=0.02, and HOMA-IR index 4.08 vs. 1.70, p<0.01 compared to women with BRI<4.910.

Women with ABSI≥0.076 had significantly higher fasting insulin 83.95 vs. 49.51 pmol/L, p=0.0102, and glucose after 60 minutes during OGTT 9.13 vs. 5.99 mmol/L, p<0.01compared to women with ABSI<0.076 (Table 8).

Patients with BRI≥4.910 showed significantly higher fasting insulin 123.77 vs. 50.94 pmol/L, p<0.01, fasting glucose 5.33 vs. 4.72 mmol/L, p<0.01, glucose after 120 minutes during OGTT 6.88 vs. 5.72 mmol/L, p=0.01, glucose after 60 minutes during OGTT 9.13 vs. 6.72 mmol/L, p<0.01, insulin after 60 minutes during OGTT 772.03 vs. 340.81 pmol/L, p=0.02, and HOMA-IR index 4.08 vs. 1.70, p<0.01 compared to women with BRI<4.910.

Women with ABSI≥0.076 had significantly higher fasting insulin 83.95 vs. 49.51 pmol/L, p=0.0102, and glucose after 60 minutes during OGTT 9.13 vs. 5.99 mmol/L, p<0.01compared to women with ABSI<0.076 (Table 8).

Table 9 shows the diagnostic performance of selected anthropometric and metabolic indices in predicting glucose and insulin dysregulation in women with PCOS. The analysis of anthropometric parameters and metabolic indices in relation to HOMA-IR values showed that most of the tested indicators had high predictive ability, with notable differences in performance. The VAI yielded the best results, with the highest area under the ROC curve (AUC = 0.903; 95% CI: 0.78–0.97), indicating excellent discriminatory ability. The optimal cut-off value was 0.99, with a sensitivity of 85.7%, specificity of 90.5%, positive predictive value (PPV) of 92.3%, and negative predictive value (NPV) of 82.6%, suggesting VAI is a particularly effective indicator for identifying elevated HOMA-IR levels. LAP was the second most effective indicator, with an AUC of 0.883 (95% CI: 0.76–0.96) and a cut-off of 27.9, demonstrating a sensitivity of 82.1%, PPV of 92%, and NPV of 79.2%. BMI also showed strong predictive value (AUC = 0.852; 95% CI: 0.72–0.94), with a cut-off of 27 kg/m², sensitivity of 85.7%, specificity of 90.5%, and PPV of 92.3%, highlighting its utility as a predictor of insulin resistance. Both waist-to-height ratio (WHtR) and waist circumference (WC) exhibited high AUCs (0.821 and 0.831, respectively) with sensitivities of 85.7% and 89.3%, reflecting good predictive performance. WHR achieved an AUC of 0.791, with a sensitivity of 82.1%, specificity of 71.4%, and NPV of 75%, indicating moderate effectiveness. The Body Adiposity Index (BAI) also performed well (AUC = 0.781), with a cut-off of 31.5%, sensitivity of 85.7%, and specificity of 66.7%. Body Roundness Index (BRI) had an AUC of 0.821, comparable to WHtR, whereas A Body Shape Index (ABSI) showed limited utility with a lower AUC of 0.643. Receiver-operating characteristic (ROC) curves identifying the threshold for HOMA-IR for VAI, LAP, BMI, WC, BRI, WHtR, WHR, BAI are presented on Figure 2.

In the studied group most of the examined anthropometric and metabolic indices were significantly associated with insulin resistance (Table 10). The strongest associations were observed for body mass index (BMI) and visceral adiposity index (VAI), both showing a 57-fold higher risk of insulin resistance (BMI≥27.25kg/m²: OR = 57.0; 95%CI:9.41–345.15; p<0.001; VAI≥1.07: OR = 57.0; 95%CI:9.41–345.15; p<0.001). Waist circumference (WC≥77.25cm: OR = 20.8; 95%CI:4.53–95.89; p<0.001), waist-to-height ratio (WHtR≥0.51: OR = 19.2; 95%CI:4.46–82.60; p<0.001), and lipid accumulation product (LAP≥32.04: OR = 43.7; 95%CI:7.60–251.17; p<0.001) also showed significant predictive value. Other significant predictors included waist-to-hip ratio (WHR≥0.79: OR = 11.5; 95%CI:2.97–44.51; p<0.001), body adiposity index (BAI≥31.5%: OR = 11.1; 95%CI:2.74–45.26; p=0.001), and body roundness index (BRI≥3.48: OR = 19.2; 95%CI:4.46–82.60; p<0.001.