Association of Chemerin, follicle stimulating hormone, lipid profiles, and cardiometabolic indices with the premature ovarian insufficiency.

This case control study screened 535 women under 40 years old from medical centers affiliated with Kermanshah University of Medical Sciences (January 2023-May 2025). POI cases had prior specialist diagnosis; menstrual disturbance ≥ 4 months was confirmed, and a single basal FSH value (interpreted using a 25 IU/L cut-off) was measured at sampling 9 . Exclusion criteria included conditions known to affect menstruation; genetically confirmed disorders associated with POI-related symptoms; and previous surgeries that could alter menstrual function. Participants were also excluded if they had acute or chronic diseases—such as renal or hepatic failure, autoimmune disorders, cardiovascular disease, cancer, uncontrolled hypertension, dyslipidemia, or diabetes—or if they were following prescribed diets, taking medications or nutritional supplements within the previous three months, or had a BMI greater than 35. Pregnant or breastfeeding women, professional athletes, and those unwilling to participate were also excluded. Medication use was recorded. ‘Using hormones’ referred to active clinically relevant exogenous hormonal therapy; however, such use was rare and therefore not included as a covariate. Lipid-lowering therapy was reported by six participants and was likewise not modeled separately. Out of the initial participants, 497 women were excluded, leaving 38 eligible volunteers who completed the assessment. All participants had been newly diagnosed within the same year as sample collection. A control group was chosen from the Persian Youth Cohort (PYC), which is part of the Prospective Epidemiological Research Studies in IRAN (PERSIAN) Cohort 34 , 35 . The Ravansar cohort is a part of the PYC and is located in western Iran, north of Kermanshah City. The controls for this study included women with regular menstrual cycles (21–35 days), no menopause before the age of 40, no use hormonal medications and no confirmed diseases or syndromes 17 , 35 (Fig.  1 ). Controls were matched to the cases based on age and BMI. The study was approved by the Ethics Committee of Iran University of Medical Sciences (IUMS). (IR.IUMS.REC.1401.599). and was conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all participants. Clinical trial number: not applicable. Fig. 1 Flow chart of participant selection in the case–control study. Flow chart of participant selection in the case–control study. Trained interviewers collected demographic data (age, smoking habits, educational level, medical history, medication use) using a pretested questionnaire. They also measured weight, height, BMI, waist and hip circumferences. Weight was recorded to 0.1 kg (Inbody 770), height to 0.1 cm (BSM 370; Biospace) and BMI was calculated as kg/m 2 . Waist circumference was measured at the midpoint between the lowest rib and iliac crest using an unstretched tape 36 . Fasting blood sugar (FBS) and lipid profile (TG, TC, LDL-C, HDL-C) were measured using enzymatic colorimetric methods (Pars Azmoon, Iran). For hormonal analysis serum levels of Chemerin and FSH were measured by enzyme-linked immunosorbent assay (ELISA) using commercial kits (ZellBio for chemerin and Diaplus for FSH). The Atherogenic Index of Plasma (AIP) is calculated as: Log (TG/ HDL-C). The Lipid Accumulation Product (LAP) is determined as: (WC-58) × TG. The Lipoprotein Combine Index (LCI) is calculated as TC × TG × LDL/HDL-C. Dietary intake was assessed using a validated semi-quantitative Food Frequency Questionnaire (FFQ) administered by trained interviewers, and nutrient intakes were calculated using Nutritionist IV software (N4, First DataBank Inc., USA) with a database adapted for Iranian foods. All analyses were performed using the Statistical Package for the Social Sciences (SPSS, version 26.0; IBM Corp., Chicago, IL, USA). Normality was assessed through visual inspection of histograms and the Kolmogorov–Smirnov test. Continuous variables are reported as mean ± standard deviation (SD), and categorical variables as number (percentage). Continuous variables were modeled linearly, and ordinal or sparse categorical variables were modeled as continuous to enhance model stability. A two-sided P value < 0.05 was considered statistically significant. Logistic regression analyses were first conducted to estimate crude and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) for the association between the main exposures (serum Chemerin and FSH levels, as well as FBS, lipid profile and cardiometabolic indices) and POI (Table 3 ). Adjusted models accounted for potential confounding factors. Receiver operating characteristic (ROC) curve analysis was subsequently performed for each variable to evaluate its discriminative ability for distinguishing individuals with POI from those without. The area under the ROC curve (AUC), sensitivity, specificity, and optimal cut-off values are presented in Table 6 and Fig.  2 . Fig. 2 Area Under the ROC Curve of the Demographic, Anthropometric, Biochemical, and Hormonal Indices for Predicting Premature Ovarian Insufficiency. ( a ) BMI, WC, HC and WHR ( b ) FBS, TG, TC, HDL-C and LDL-C ( c ) LAP, LCI and AIP ( d ) Chemerin and FSH and ( e ) ROC curve of the final multivariable logistic regression model based on variables FBS, age, HDL-C and Chemerin for discriminating between POI cases and controls. Area Under the ROC Curve of the Demographic, Anthropometric, Biochemical, and Hormonal Indices for Predicting Premature Ovarian Insufficiency. ( a ) BMI, WC, HC and WHR ( b ) FBS, TG, TC, HDL-C and LDL-C ( c ) LAP, LCI and AIP ( d ) Chemerin and FSH and ( e ) ROC curve of the final multivariable logistic regression model based on variables FBS, age, HDL-C and Chemerin for discriminating between POI cases and controls. A multivariable logistic regression model was then constructed by entering the selected variables simultaneously. Variables that improved overall model fit were retained in the final model to estimate the probability of case status (Z). The discriminative performance of the final model was assessed using the AUC with 95% CIs. Estimated probabilities from the final model were used to define probability-based strata—low ( P  < 0.31), intermediate (0.31 ≤  P  < 0.76), and high likelihood ( P  ≥ 0.76)—to illustrate the model’s capacity to distinguish cases from controls. Model calibration was evaluated using the Hosmer–Lemeshow goodness-of-fit test. To address the risk of overfitting related to the modest sample size, internal validation was performed using bootstrap resampling with 1,000 iterations. A post-hoc power analysis was conducted using G*Power version 3.1.

Table 1 summarizes demographic, anthropometric, reproductive, biochemical and hormonal characteristics. The mean age was 36.95 ± 5.36 years in the POI group and 34.11 ± 5.25 years in controls. Significant differences were observed for HC, education, age at first pregnancy, FBS, TC, TG, LDL-C, LAP and FSH. Chemerin was higher in the POI group (25.30 ± 9.29 ng/L) compared to the control group (22.50 ± 4.75 ng/L), but not significant ( P  = 0.12) (Table 1 ). Table 1 Anthropometric, reproductive, biochemical and hormonal parameters of POI and control groups. Variable POI group (n = 38) Healthy control group (n = 38) P value Demographic and anthropometric data  Age (years) 36.95 ± 5.36 34.11 ± 5.25 < 0.001**  BMI (kg/m 2 ) 26.44 ± 3.80 26.91 ± 4.06 0.603*  WC (cm) 95.02 ± 12.78 89.84 ± 9.74 0.051*  HC (cm) 109.60 ± 7.95 103.95 ± 8.12 0.005**  WHR 0.84 ± 0.13 0.85 ± 0.05 0.909**  Education (Years) 12.21 ± 3.14 9.21 ± 4.41 0.001**  Smoking status Non smoker 32 (84.2%) 33 (86.8%) 1.000****  Current Smoker 1 (2.6%) 0 (0.0%)  Passive Smoker 5 (13.2%) 5 (13.2%)  Physical activity (MET*min/week) 5363.50 ± 14,050.78 4749.42 ± 1877.85 0.790 Reproductive Factors  Age at menarche (years) 13.55 ± 1.30 13.16 ± 1.66 0.340**  Age at first pregnancy (years) 25.14 ± 5.37 21.35 ± 4.28 0.008**  Number of pregnancy (n = 31), (n = 28) 2.00 ± 1.15 2.11 ± 1.03 0.567**  Abortion No 0.32 ± 0.67 0.35 ± 0.62 0.720**  Age at first abortion (n = 26), (n = 26) 22.75 ± 15.00 23.25 ± 11.36 0.712**  Duration of breast feeding (months) (n = 26), (n = 23) 30.08 ± 17.17 34.61 ± 20.93 0.736** Biochemical data  Fasting blood glucose (mg/dl) 102.97 ± 12.64 90.81 ± 7.23 < 0.001*  TC (mg/dl) 192.57 ± 38.74 170.73 ± 26.89 0.006*  TG (mg/dl) 146.28 ± 79.95 104.34 ± 58.97 0.005**  HDL-C (mg/dl) 48.07 ± 10.34 43.92 ± 6.65 0.053**  LDL-C (mg/dl) 97.57 ± 25.97 119.58 ± 29.75 0.003**  LAP 62.09 ± 51.24 39.26 ± 31.09 0.022  AIP 0.43 ± 0.26 0.33 ± 0.19 0.056  LCI 20.34 ± 17.90 15.40 ± 12.93 0.172  Serum chemerin level (ng/L) 25.30 ± 9.29 22.50 ± 4.75 0.120**  Serum FSH level (mIU/mL) 19.99 ± 2.59 5.47 ± 3.55 < 0.001** Data are presented as mean ± SD. *Obtained from t test for normal distribution, Mann Whitney for non-normal, and chi-squared for categorical variables. *t test. **Mann–Whitney U. ***Chi-Square. *****Fisher-Freeman-Halton Exact Test. BMI, Body Mass Index; WC, Waist Circumference; HC, Hip Circumference; WHR, waist-to-hip ratio; FBS, Fasting Blood Sugar; TG, triglyceride; TG, Triglyceride; TC, Total Cholesterol; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol AIP, Atherogenic index of plasma; LAP, Lipid Accumulation Product; LCI, Lipoprotein Combine Index; FSH, Follicle Stimulating Hormone. Anthropometric, reproductive, biochemical and hormonal parameters of POI and control groups. Data are presented as mean ± SD. *Obtained from t test for normal distribution, Mann Whitney for non-normal, and chi-squared for categorical variables. *t test. **Mann–Whitney U. ***Chi-Square. *****Fisher-Freeman-Halton Exact Test. BMI, Body Mass Index; WC, Waist Circumference; HC, Hip Circumference; WHR, waist-to-hip ratio; FBS, Fasting Blood Sugar; TG, triglyceride; TG, Triglyceride; TC, Total Cholesterol; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol AIP, Atherogenic index of plasma; LAP, Lipid Accumulation Product; LCI, Lipoprotein Combine Index; FSH, Follicle Stimulating Hormone. Table 2 shows that no significant differences were observed between the POI and control groups for energy, carbohydrates, fats, protein, calcium, iron, zinc, selenium, and vitamins C and E ( P  > 0.05). Table 2 Nutrient intakes of premature ovarian insufficiency and control groups. Variable POI group (n = 38) Healthy control group (n = 38) P value* Energy (Kcal/d) 2240.71 ± 141.34 2329.23 ± 101.37 0.329 Carbohydrate (g/d) 311.99 ± 19.28 330.98 ± 18.06 0.350 Protein (g/d) 77.68 ± 6.31 80.79 ± 5.00 0.506 Fat (g/d) 82.63 ± 6.47 82.02 ± 5.48 0.959 Calcium (mg/d) 1134.35 ± 110.20 1329.19 ± 173.69 0.625 Iron (mg/d) 36.87 ± 5.56 48.22 ± 9.95 0.400 Zinc (mg/d) 12.40 ± 1.09 13.67 ± 0.92 0.114 Selenium (µg/d) 97.20 ± 7.19 79.17 ± 5.18 0.155 Vitamin C (mg/d) 135.79 ± 12.73 140.19 ± 14.85 0.852 Vitamin E (mg/d) 13.48 ± 1.17 13.73 ± 0.96 0.540 Data are mean ± SE. *Obtained from Mann Whitney U for non-normal variables. Nutrient intakes of premature ovarian insufficiency and control groups. Data are mean ± SE. *Obtained from Mann Whitney U for non-normal variables. Table 3 displays odds ratios (95% confidence intervals) for POI versus controls across three logistic regression models: unadjusted (Model 1), age-adjusted (Model 2), and multivariable-adjusted (Model 3). In the unadjusted model, POI was associated with higher hip circumference, education, physical activity, FBS, TC, TG, LAP, FSH and Chemerin ( P  < 0.05). In the age-adjusted model (Model 2), the results were largely consistent with the unadjusted models. Women with higher HC, education, FBS, TC, TG, LDL-C, FSH and Chemerin had significantly elevated odds of POI, while low to moderate physical activity was also associated with higher odds of POI ( P  < 0.05). In the fully adjusted model (Model 3), higher education, physical activity, TC, TG, and Chemerin levels remained significant, even after adjusting for age, BMI, physical activity, and total energy intake ( P  < 0.05). Table 3 Odds ratios for the association between Premature Ovarian Insufficiency and demographic, anthropometric, and biochemical parameters based on logistic regression models. Un-adjusted a Model 1 OR 95% CI P value Age-adjusted b Model 2 OR 95% CI P value Multivariable-adjusted c Model 3 OR 95% CI P value Age (years)  < 30 years 1.00 (reference)  ≤ 30 years 2.63 (0.62–11.07) 0.186 2.03 (0.62–11.07) 0.186 2.97 (0.51–17.18) 0.223 BMI (kg/m)  Normal (18.5–24.9) 1.00 (reference)  Overweight (25–29.9) 1.25 (0.44–3.49) 0.670 1.05 (0.36–3.06) 0.923 1.267 (0.350–4.586) 0.718  Obesity (≥ 30.0) 0.70 (0.21–2.29) 0.556 0.63 (0.18–2.12) 0.457 1.65 (0.38–7.19) 0.500 WC (cm)  Low < 80 1.00 (reference)  High ≥ 80 1.23 (0.34–4.46) 0.745 1.17 (0.32–4.33) 0.805 1.23 (0.18–8.24) 0.826 HC (cm)  Low  103.5 4.16 (1.52–11.40) 0.005 3.88 (1.40–10.76) 0.009 NA NA WHR  Low  0.85 1.70 (0.68–4.23) 0.251 1.61 (0.64–4.06) 0.308 1.43 (0.42–4.82) 0.563 Educational Years 1.21 (1.07–1.38) 0.002 1.22 (1.07–1.38) 0.002 1.22 (1.42–1.06) 0.006 Smoking status  Never 1.00 (reference)  Current Smoker + Passive 1.23(0.34–4.46) 0.745 1.31 (0.35–4.88) 0.678 1.72 (0.36–8.24) 0.496 Physical activity (MET*min/week)  High > 3000 1.00 (reference)  Low and moderate < 600–3000 13.03 (3.82–44.29) < 0.001 13.67(3.67–48.02) < 0.001 16.20 (4.20–62.5) < 0.001 FBS (mg/dl)  < 100 (mg/dl) 1.00 (reference) ≤ 100 (mg/dl) 11.68 (3.45–39.58) < 0.001 12.37 (3.52–43.43) < 0.001 NA NA TC (mg/dl)  < 200 (mg/dl) 1.00 (reference)  ≤ 200 (mg/dl) 4.80 (1.63–14.13 0.004 4.46 (1.49–13.28) 0.007 5.32 (1.40–20.16) 0.014 TG (mg/dl)  < 150 (mg/dl) 1.00 (reference)  ≤ 150 (mg/dl) 5.54 (1.63–18.83) 0.006 5.21 (1.52–17.85) 0.004 5.39 (1.21–23.86) 0.026 HDL-C (mg/dl)  < 45 (mg/dl) 1.00 (reference)  ≤ 45 (mg/dl) 1.37 (0.55–3.39) 0.490 0.77 (0.30–1.93) 0.581 0.68 (0.22–2.08) 0.507 LDL-C (mg/dl)  < 100 (mg/dl) 1.00 (reference)  ≤ 100 (mg/dl) 0.40 (0.15–1.05) 0.063 0.36 (0.13–0.98) 0.046 0.32 (0.10–1.03) 0.058 LAP 1.02 (1.00–1.04) 0.033 1.01 (1.00–1.03) 0.051 1.02 (0.99–1.04) 0.081 AIP 7.39 (0.91–59.81) 0.061 6.09 (0.73–50.98) 0.095 3.42 (0.25–46.8) 0.357 LCI 1.02 (0.99–1.05) 0.184 1.01 (0.98–1.05) 0.259 1.01 (0.98–1.05) 0.319 Serum FSH level (mIU/mL) 2.03 (1.29–3.18) 0.002 2.03 (1.31–3.15) 0.002 NA NA Serum Chemerin level (ng/L)  < 21.88 (mg/dl) 1.00 (reference)  ≥ 21.88 (mg/dl) 2.72 (1.05–7.03) 0.038 3.00 (1.13–7.94) 0.027 4.02 (1.18–13.67) 0.026 a Model 1: un adjusted (Crude model). b Model 2: adjusted for age. c Model 3: adjusted for model 2 and BMI, physical activity and energy. d Waist circumference categories according to World Health Organization cut-off points 37 . OR, Odds Ratio; CI, Confidence Interval; BMI, Body Mass Index; WC, Waist Circumference; HC, Hip Circumference; WHR, waist-to-hip ratio; FBS, Fasting Blood Sugar; TG, triglyceride; TC, Total Cholestrol; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol; LAP, Lipid Accumulation Product; AIP, Atherogenic index of plasma; LCI, Lipoprotein Combine Index; FSH, Follicle Stimulating Hormone; NA, not applicable; Odds ratios for the association between Premature Ovarian Insufficiency and demographic, anthropometric, and biochemical parameters based on logistic regression models. a Model 1: un adjusted (Crude model). b Model 2: adjusted for age. c Model 3: adjusted for model 2 and BMI, physical activity and energy. d Waist circumference categories according to World Health Organization cut-off points 37 . OR, Odds Ratio; CI, Confidence Interval; BMI, Body Mass Index; WC, Waist Circumference; HC, Hip Circumference; WHR, waist-to-hip ratio; FBS, Fasting Blood Sugar; TG, triglyceride; TC, Total Cholestrol; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol; LAP, Lipid Accumulation Product; AIP, Atherogenic index of plasma; LCI, Lipoprotein Combine Index; FSH, Follicle Stimulating Hormone; NA, not applicable; Bootstrap analysis confirmed the stability of variables physical activity (> 3000 MET-min/week), TC (≥ 200 mg/dL), TG (≥ 150 mg/dL) and Chemerin (≥ 21.88 ng/mL) (bias-corrected ORs: 5.84, 3.52, 2.79, and 2.39, respectively; P  < 0.05). However, age  0.05) (Table 4 ). Post-hoc power analysis revealed high power (> 85%) for PA, TC, and TG, whereas variables education, age, and LDL-C exhibited low power ( 0.05), except Model FBS (Table 4 ). Table 4 Bootstrap results for the Hosmer–Lemeshow statistics and power analysis for Model 3. Variable Conventional risk ratio Bootstrap mean ratio Bootstrap 95% confidence interval Bootstrap significance (Sig.) Hosmer–Lemeshow test (Sig.) Statistical power Age (> 30 years) 2.97 0.91 (0.30–3.14 × 10⁹) 0.204 0.464 45% BMI (kg/m) Overweight (25–29.9) 1.27 0.99 (0.35–11.84) 0.725 0.464 12% BMI (kg/m) Obesity (≥ 30.0) 1.66 1.33 (0.38–18.07) 0.498 0.464 25% WC (cm) High ≥ 80 1.23 1.21 (0.17–12.00) 0.809 0.674 15% WHR High > 0.85 1.43 1.40 (0.36–7.38) 0.542 0.619 25% Educational Years 1.23 1.20 (1.08–1.56) 0.003 0.106 32% Smoking status (Current + Passive Smoker) 1.72 1.48 (0.26–16.48) 0.500 0.914 28% Physical activity (MET*min/week) (> 3000) 16.21 5.84 (5.74–3.70 × 10⁹) 0.001 0.464 99.9% TC (mg/dl) (≥ 200) 5.32 3.52 (1.49–67.70) 0.004 0.682 88% TG (mg/dl) (≥ 150) 5.39 2.79 (1.07–147.06) 0.023 0.933 85% HDL-C  < 45 (mg/dl) 0.69 1.40 (0.14–2.63) 0.540 0.565 18% LDL-C (mg/dl) (≥ 100) 0.32 0.39 (0.05–1.07) 0.051 0.994 38% LAP 1.02 1.03 (1.00–1.10) 0.190 0.376 12% AIP 3.42 4.01 (0.16–270.08) 0.397 0.421 22% LCI 1.02 1.02 (0.97–1.07) 0.359 0.646 10% Chemerin (ng/L) (≥ 21.88) 4.03 2.39 (1.27–46.81) 0.018 0.493 78% Sig., significance level ( p value). Bootstrap estimates were calculated using 1000 bootstrap resamples. Statistical significance was considered at P  < 0.05. Significant values are in bold. Table 5 Interpretation of statistical power levels for the study variables. Variables Interpretation Statistical power (%) Physical activity, TC, TG Excellent ≥ 80% Chemerin (78%) Good 70–80% Age* (45%) Moderate 50–70% LDL-C (38%), Education (32%) Low 30–50% Overweight (12%), Obesity (25%), WC (15%), WHR (25%), Smoking (28%), HDL-C (18%), LAP (12%), AIP (22%), LCI (10%) Very Low  30 years) BMI (kg/m) Overweight (25–29.9) BMI (kg/m) Obesity (≥ 30.0) WC (cm) High ≥ 80 WHR High > 0.85 Smoking status (Current + Passive Smoker) Physical activity (MET*min/week) (> 3000) TC (mg/dl) (≥ 200) TG (mg/dl) (≥ 150) HDL-C < 45 (mg/dl) LDL-C (mg/dl) (≥ 100) Chemerin (ng/L) (≥ 21.88) Sig., significance level ( p value). Bootstrap estimates were calculated using 1000 bootstrap resamples. Statistical significance was considered at P  < 0.05. Significant values are in bold. Interpretation of statistical power levels for the study variables. *Age (45%) is considered near the threshold for the moderate category. ROC analysis evaluated the ability of each parameter to discriminate between individuals with and without POI. The area under the curve (AUC), sensitivity, specificity, and optimal cut-off values are presented in Table 6 (Fig.  2 ). The AUC for FSH reached 0.988 (sensitivity 0.974, specificity 0.974). The AUCs for FBS and LAP were 0.776 and 0.715 respectively. WC, HC, TG, TC, HDL-C and AIP all showed AUC values greater than 0.6, indicating moderate discriminative ability for distinguishing POI cases from controls 38 . The diagnostic value of the Chemerin assay in the POI group was assessed using the ROC curve, resulting in an AUC of 0.604 (0.472–0.735) with the best sensitivity (0.500) and specificity (0.789) at 25.14 ng/L. Table 6 also reports additional details. Table 6 Area under the ROC curve for demographic, anthropometric, biochemical, and hormonal variables discriminating premature ovarian insufficiency. Variable(S) ROC AUC (95% CI) Sensitivity (%) Specificity (%) Cut off point Youden Index (J) P value WC (cm) 0.604 0.56–0.80 23 97 103.50 0.21 0.118 HC (cm) 0.687 0.56–0.80 92 44 100.75 0.36 0.005 FBS (mg/dl) 0.776 0.67–0.88 55 92 100.50 0.47 < 0.001 TG (mg/dl) 0.688 0.56–0.80 63 73 114.50 0.36 0.005 TC (mg/dl) 0.679 0.55–0.80 52 81 196.00 0.34 0.007 HDL-C(mg/dl) 0.629 0.50–0.75 42 81 46.50 0.23 0.053 AIP 0.634 0.50–0.76 55 76 0.45 0.31 0.044 LAP 0.715 0.60–0.82 57 76 46.89 0.34 0.001 LCI 0.562 0.42–0.69 42 89 21.19 0.31 0.355 Chemerin (ng/L) 0.604 0.47–0.73 50 78 25.14 0.28 0.120 FSH (mIU/mL) 0.988 0.96–1.00 97 97 14.37 0.94 < 0.001 Sensitivity and specificity were calculated by receiver operating characteristic (ROC) curve analysis. P  < 0.05 versus AUC = 0.5 (no discrimination). The Youden index, which is the best compromise between sensitivity and specificity (sensitivity + specificity_1). A value closer to 1 indicates a greater diagnostic significance 39 – 41 . AUC, Area Under the ROC Curves; CI, Confidence Interval; ROC, Receiver Operating Characteristic; WC, Waist Circumference; HC, Hip Circumference; FBS, Fasting Blood Sugar; TG, Triglyceride; TC, Total Cholesterol; HDL-C, High-Density Lipoprotein Cholesterol; AIP, Atherogenic index of plasma; LAP, Lipid Accumulation Product; LCI, Lipoprotein Combine Index; FSH, Follicle Stimulating Hormone. Area under the ROC curve for demographic, anthropometric, biochemical, and hormonal variables discriminating premature ovarian insufficiency. Sensitivity and specificity were calculated by receiver operating characteristic (ROC) curve analysis. P  < 0.05 versus AUC = 0.5 (no discrimination). The Youden index, which is the best compromise between sensitivity and specificity (sensitivity + specificity_1). A value closer to 1 indicates a greater diagnostic significance 39 – 41 . AUC, Area Under the ROC Curves; CI, Confidence Interval; ROC, Receiver Operating Characteristic; WC, Waist Circumference; HC, Hip Circumference; FBS, Fasting Blood Sugar; TG, Triglyceride; TC, Total Cholesterol; HDL-C, High-Density Lipoprotein Cholesterol; AIP, Atherogenic index of plasma; LAP, Lipid Accumulation Product; LCI, Lipoprotein Combine Index; FSH, Follicle Stimulating Hormone. The final multivariable logistic regression model, which included variables FBS, age, HDL-C, LDL-C and Chemerin, demonstrated excellent discriminative ability (AUC = 0.929; 95% CI 0.876–0.981; Fig.  2 ). The estimated probability of case status was defined as: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} Z & = - 25.867 + (0.160 \times {\mathrm{FBS}}) + (0.168 \times {\mathrm{Age}}) + (0.120 \times {\mathrm{Chemerin}}) \, \\ & \quad + { (}0.129 \times {\mathrm{HDL-C)}} - {(}0.039 \times {\mathrm{LDL-C)}} \\ \end{aligned}$$\end{document} The probability of POI (P) was computed using the logistic transformation \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{P}} = {\mathrm{e}}^{{\mathrm{Z}}} /(1 + {\mathrm{e}}^{{\mathrm{Z}}} )$$\end{document} For practical application, rounded coefficients may be used: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {\mathrm{Z}} & \approx - 26 + (0.16 \times {\mathrm{FBS}}) + (0.17 \times {\mathrm{Age}}) + (0.12 \times {\mathrm{Chemerin}}) \, \\ & \quad + { (}0.13 \times {\mathrm{HDL-C)}} - (0.04 \times {\mathrm{LDL-C}}) \\ \end{aligned}$$\end{document} A probability threshold of P  = 0.30 P  = 0.30 P  = 0.30 yielded the best overall classification performance (sensitivity 94.7%, specificity 76.3%, Youden index = 0.710). Individuals with probabilities ≥ 0.30 were classified as cases, whereas those with probabilities < 0.30 were classified as controls. This stratification reflects the model’s ability to distinguish case status within the study dataset.

POI is a common reproductive endocrine disorder influenced by several etiological factors. Serum Chemerin shows moderate diagnostic potential. Additionally, in the current study TC, TG, AIP, LAP, FBS, FSH, WC and HC may serve as potential diagnostic value for POI. The proposed multivariable model provided strong discriminative performance. These findings suggest that combining multiple variables into a single multivariable model substantially enhances discriminative accuracy compared with individual variables alone. Nonetheless, these results should be interpreted as preliminary. External validation and larger prospective studies are needed to confirm their robustness and determine whether they can be translated into clinical practice.

Despite the growing data, its multifactorial etiology and pathogenesis on POI remain poorly understood, and there is limited public awareness of its prevention and management 42 . This study aimed to evaluate serum Chemerin, follicle-stimulating hormone, lipid profiles, and cardiometabolic indices associated with the POI. To our knowledge, this is the first investigation to assess serum Chemerin levels in individuals with POI. As a novel marker of metabolic and cardiovascular dysfunction, could Chemerin be a valuable biomarker for distinguishing individuals with POI? ROC analysis revealed that the diagnostic performance of serum Chemerin in patients with POI. The AUC of the serum Chemerin assay reached a value of 0.604 (sensitivity 50.0%, specificity 78.6%) at a cutoff of 25.14 ng/L, indicating moderate potential as a diagnostic biomarker for POI. To our knowledge, no previous studies have reported cutoff points for serum Chemerin in POI. Multivariable logistic regression indicated that higher serum Chemerin levels significantly increased the odds of POI (4.02; 95% CI 1.18, 13.67). Logistic regression revealed significant associations between serum Chemerin, lipid parameters, and selected cardiometabolic indices with POI. ROC-based evaluation indicated that the model could discriminate between individuals with and without POI, with lower probability thresholds favoring sensitivity and higher thresholds favoring specificity. These probability-based categories offer a classification framework appropriate for a case–control design, rather than one intended for prediction. These findings should be interpreted cautiously and require external validation in independent populations. Due to limited evidence on Chemerin levels in POI, relevant studies were referenced to support our interpretations. Manzel et al. investigated serum Chemerin levels in relation to bone health and reported higher levels in postmenopausal women compared with peri- and premenopausal women 43 . Shi et al. reported significantly lower serum Chemerin levels in women with postmenopausal osteoporosis compared with controls 44 . However, studies examining Chemerin in PCOS have reported higher serum Chemerin levels in women with PCOS 45 . Elevated Chemerin has been shown to regulate ovarian steroidogenesis 15 , 25 – 28 . It reduces steroidogenesis and cholesterol synthesis after IGF-1 or FSH induction 16 , 29 by downregulating P450 levels and MAPK/ERK1/2 phosphorylation 26 . Chemerin also inhibits FSH-induced steroidogenesis via prohibitin regulation 29 and decreases aromatase expression 23 . It reduces granulosa cell proliferation through IGF-1R signaling 30 and suppresses FSH receptor activity 16 . Thus, Chemerin-related mechanisms appear to align with hormonal alterations, particularly in POI. Although serum Chemerin did not show a significant difference between groups when evaluated as a continuous variable, the dichotomized analysis using the cutoff of 21.88 ng/L revealed a significant association with the outcome. This discrepancy may suggest a threshold-dependent relationship, whereby the biological effects of Chemerin become more prominent above a certain concentration. Supporting this interpretation, the ROC analysis demonstrated a modest discriminative ability (AUC = 0.604), indicating that Chemerin may still provide clinically meaningful information when interpreted in conjunction with other markers. Furthermore, categorical analysis was not feasible for several variables due to insufficient numbers in certain subgroups, which resulted in model non-convergence. To maintain statistical robustness, these variables were analyzed in continuous form. This methodological consideration may explain some of the differences observed between continuous and categorical analyses across variables. Patients with POI often exhibit hyperlipidemia, raising cardiovascular risk 46 . We observed elevated TC and TG in these patients, showing discriminative performance (AUC: 0.679 and 0.688; cutoffs: 196.00 mg/dl and 114.50 mg/dl). Multivariable logistic regression revealed that TC and TG above the cutoffs were associated with higher odds of POI, with ORs of 5.32 (95% CI 1.40, 20.16) and 5.39 (95% CI 1.21, 23.86), respectively. Consistente with our findings, two recent meta-analysis reported higher TC and TG in POI cases, while HDL-C levels remained unchanged 47 , 48 . Some studies have found higher TC in women with POI, while TG showed no difference 8 , 49 . Other studies found higher TG in women with POI 50 whereas Daan, et al., observed no significant differences between the two groups 32 . There is a critical knowledge gap regarding the temporal relationship between dyslipidemia and POI. Specifically, it is unclear whether dyslipidemia precedes POI or emerges as a concurrent manifestation. Mitochondria-mediated lipid peroxidation has been implicated in ferroptosis, a novel iron-dependent cell death driven by excessive oxidative lipid damage, which may contribute to POI in Basonuclin 1 mutations. This is supported by cisplatin-induced ovarian toxicity models, where granulosa cell lipid peroxidation triggers ferroptosis, leading to follicular arrest, ovarian fibrosis, and functional decline 51 , 52 . Abnormal lipid metabolism in POI may result from estrogen deficiency, causing nuclear receptor dysregulation, hepatic oxidative stress, and immune senescence 46 , 53 . The beneficial effects of intact ovarian function on lipid profiles are attributed to estradiol. As an endogenous antioxidant, estradiol provides vasodilatory and cardiovascular protection while promoting lipid degradation and excretion. It lowers TC and LDL-C, modifies their distribution in the body, and limits cholesterol deposition in arterial walls. Estradiol suppresses hepatic lipase, slows HDL breakdown, accelerates LDL-C clearance, and enhances LDL receptor activity, thereby reducing serum TC and LDL-C levels and overall improving lipid metabolism 48 . These findings suggest early dyslipidemia screening and targeted lipid-lowering interventions, may delay POI progression and reduce systemic complications 51 . The paradoxical LDL-C pattern may relate to sparse subgroup counts, limited use of lipid-lowering medications, and early metabolic changes in POI that diverge from typical dyslipidemia profiles. Furthermore, AIP, LAP, FBS, FSH, WC and HC showed AUCs > 0.6, indicating their potential discriminative ability for POI. However, multivariable logistic regression indicated that they were not statistically significant ( P  > 0.05), warranting further study. Nevertheless, the findings provide preliminary evidence of an association between these variables and POI, highlighting the need for additional studies in this area. Additionally, this study found significantly higher POI among individuals with low-to-moderate physical activity (OR = 13.03; 95% CI 3.82–44.29). The wide confidence interval likely reflects the limited sample size and sparse data in some activity categories, and therefore this result should be interpreted cautiously. Our findings align with previous reports showing a protective role of regular physical exercise against POI 54 , 55 . Moderate activity helps maintain ovarian function by supporting HPO-axis regulation and physiological hormone levels 56 , whereas excessive exercise may disrupt hormonal balance and accelerate dysfunction 57 . At the molecular level, PTEN loss activates the PI3K/Akt pathway and prematurely recruits primordial follicles 58 , while TSC/mTOR overactivation impairs follicle growth 59 . Regular physical activity may partly preserve ovarian function through modulation of these pathways 54 . The present study found a significant increase in POI prevalence among highly educated women (OR 1.2; 95% CI 1.07, 1.38). Some studies report higher educational attainment in women with POI compared to controls 32 , while others found no difference 60 . Conversely several studies have reported an elevated prevalence of POI among women with lower education levels. Although the underlying mechanisms remain unclear, further research is warranted to elucidate this association 61 – 63 . Several variables with potential discriminative value for POI were not found to be significant, likely due to limited variability between groups. For instance, smoking was absent in both groups, precluding assessment of its association with POI, despite consistent links reported in previous studies. However, it is worth noting that smoking has been consistently linked to POI in numerous studies. Long-term chemical exposures, including smoking, may disrupt PI3K/PTEN/Akt and TSC/mTOR signaling pathways, affecting cell proliferation, differentiation, or apoptosis, and consequently ovarian function. To the best of our knowledge, this is the first study to investigate the associations of serum Chemerin, follicle stimulating hormone, lipid profiles, cardiometabolic indices with the POI, providing novel insights into its potential underlying pathways. However, several limitations should be considered. First, the case–control design precludes the establishment of causal relationships and carries an inherent risk of reverse causality. Second, the relatively small sample size may have limited the statistical power—potentially increasing the risk of type II error for certain variables (e.g., education and LDL-C)—and may have contributed to model overfitting. While bootstrap resampling suggested internal stability for the main predictors, external validation in larger, more powered cohorts remains necessary. A larger sample size and more data may yield more robust and significant findings. Third, all participants were in overt stages of POI, which might not fully reflect the dynamic changes occurring across the full spectrum of disease progression, particularly in occult or biochemical stages. Finally, residual confounding and selection bias cannot be entirely ruled out. For instance, educational level may influence reproductive behaviors such as delayed childbearing, which could indirectly affect the observed associations. Despite these limitations, our findings provide a valuable foundation for developing future diagnostic criteria and understanding the pathophysiology of POI.

A functional reproductive system relies on proper endocrine, metabolic, and immune factors 1 . In humans, dysfunction in reproductive systems or disruption of natural processes may result in reproductive disorders including premature ovarian failure, Polycystic Ovary Syndrome (PCOS), and endometriosis 1 , 2 . Premature ovarian insufficiency (POI) or premature menopause is an emerging public health concern, with a reported prevalence of 1–5.5% 3 , 4 . A study in the Iranian population reported that 3.5% of women experienced POI 5 . Clinically, POI is defined as the loss of ovarian activity before the age of 40, occurring approximately a decade before natural menopause, which typically happens at an average age of 50 to 53 years 6 , 7 . It is characterized by amenorrhea, elevated levels of luteinizing hormone (LH) and follicle-stimulating hormone (FSH), and reduced estradiol levels 7 , 8 . The diagnosis of POI is a serious event for a woman, as it is a condition with medical, psychological, and reproductive implications 9 , including increased risks of mortality, cardiovascular disorders, hypertension, type 2 diabetes, bone fractures and neuropsychological conditions 4 . It also significantly impacts family planning, especially as women increasingly delay childbearing 10 . POI may be idiopathic or associated with autoimmune and genetic abnormalities, infections, chemotherapy, radiotherapy, surgery, and certain medications 9 . Modifiable lifestyle and environmental factors may also influence the risk of POI 9 – 11 . The association between body mass index (BMI) and menopause remains controversial, with some studies reporting earlier menopause in both higher and lower BMI, while others show no association 10 , 12 . Adipose tissue serves as an energy reservoir and is also the largest endocrine organ, secreting hormones known as adipokines. Dysfunction or abnormal levels of adipokines, due to underweight or obesity may contribute to reproductive disorders 1 , 11 , 13 . As a multifaceted adipokine, Chemerin, also referred to as tazarotene-induced gene 2 (TIG-2) or retinoic acid receptor responder 2 (RARRES2) 5 , is a multifaceted adipokine expressed in various organs, including white adipose tissue and the ovaries 13 – 15 . It binds to three receptors—chemokine-like receptor1 (CMKLR1), G-protein-coupled receptor 1(GPR1) and C–C motif chemokine receptor-like 2 (CCRL2)—to mediate diverse functions and signaling pathways 1 , 16 – 18 . Chemerin is involved in inflammation, adipogenesis, metabolism, angiogenesis, and insulin secretion in adipose cells 1 , 19 and is associated with conditions such as obesity, diabetes, and cancer 14 , 20 – 23 . On the other hand, Chemerin and its receptors are expressed in the ovary and have a functional role, reduces steroidogenesis and cholesterol synthesis after IGF-1 or FSH stimulation 13 , 15 , 24 – 28 by decreasing the levels of several proteins, including P450 and MAPK phosphorylation ERK1/2 16 , 25 , 29 . In addition, chemerin regulates prohibitin expression, inhibiting FSH-induced steroidogenesis through suppression of the PI3K/Akt pathway and also downregulation of cyp19, NR5a 1/2 29 . Chemerin also lowers aromatase expression, a key enzyme in sex hormone metabolism that catalyzes the conversion of androgens to estrogens in adipose tissue and ovarian granulosa cells 23 and granulosa cell proliferation by reducing IGF-1R signaling pathways 30 . In addition, Chemerin has been shown to inhibit the FSH receptor activity 16 . These mechanisms are consistent with the hormonal alterations observed in POI. In particular, the decline in estrogen and androgens has been associated with metabolic dysregulation, contributing to the development of metabolic syndrome and an increased risk of cardiovascular disease. These biological and metabolic alterations may highlight the potential role of metabolic disturbances in the pathophysiology of POI 8 , 31 – 33 . The Chemerin system is implicated in normal reproduction and reproductive disorders such as PCOS and gynecological cancer 1 , 2 , but its role in women with POI remains unexplored. This study aimed to investigate the association between serum Chemerin levels, follicle-stimulating hormone, lipid profiles, and cardiometabolic indices with the POI in Iranian women.