The role of body composition, irisin, and FABP4 levels in adolescent girls with polycystic ovary syndrome

Abstract

Objective

This study aimed to investigate the relationship between body composition parameters and serum irisin and FABP4 levels in adolescent girls with polycystic ovary syndrome (PCOS).

Methods

A total of 123 participants (54 with PCOS and 69 controls) were included in this cross-sectional study. Biochemical and hormonal parameters were assessed, and body composition was evaluated using bioelectrical impedance analysis (BIA). Multivariable logistic regression analysis was performed to identify factors associated with PCOS.

Results

Compared with controls, the PCOS group showed significantly higher levels of total testosterone and luteinizing hormone. Muscle mass index was also higher in the PCOS group, whereas other body composition parameters did not differ significantly between groups. In multivariable logistic regression analysis, total testosterone and Luteinizing hormone emerged as independent predictor of PCOS (OR = 1.089, 95% CI: 1.050–1.128, p < 0.001), (OR = 1.084, 95% CI: 1.003–1.172, p = 0.041) relatively. While BMI Serum irisin and FABP4 was not independently associated with PCOS.

Conclusion

In adolescent PCOS, hyperandrogenemia, particularly total testosterone, remains the strongest independent correlate of disease status. In contrast, circulating irisin and FABP4 do not appear to provide additional independent discriminatory value. These findings should be interpreted within the limitations of a cross-sectional design, and further prospective studies are warranted.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12902-026-02396-7.

Introduction

Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders in women, affecting approximately 6–15% of women of reproductive age [1]. PCOS is characterized by clinical and biochemical features such as anovulation, hyperandrogenism, and polycystic ovarian morphology [2, 3]. The disorder is also closely associated with metabolic issues, including insulin resistance, obesity, dyslipidemia, and an increased risk of cardiovascular disease [4]. Obesity, particularly abdominal fat accumulation, plays a critical role in the pathophysiology of PCOS. Increased adipose tissue can lead to inflammation and insulin resistance, thereby exacerbating hormonal imbalances [5]. Waist circumference measurement is considered an important parameter for assessing increased adiposity and metabolic risks [6]. Moreover, not only body mass index (BMI) but also a more detailed assessment of body composition—including fat and muscle mass parameters—has become important for understanding the metabolic effects of PCOS [7].

Bioelectrical impedance analysis (BIA) is a method that allows for the non-invasive, rapid, and safe assessment of body composition. It is particularly commonly used to determine muscle mass, total body water, and fat percentage [8]. It is widely applied in younger age groups and contributes to the early identification of metabolic risks in adolescents [9].

Parameters assessed using bioelectrical impedance analysis include:

Fat Mass (FM) (kg): The portion of total body mass that is composed of fat. Fat mass is regarded as a key indicator of metabolic risk [10].

Percent Body Fat (PBF) (%): Refers to the proportion of total body weight that is composed of fat. PBF has been shown to provide moderate accuracy in predicting cardiometabolic risk factors, including dyslipidemia, hyperglycemia, and elevated blood pressure [11].

Skeletal Muscle Mass (SMM) (kg): Refers to the appendicular muscle mass [12].

Muscle Mass Index (MMI) (kg/m²): An indicator calculated by dividing total muscle mass by the square of height [13].

Trunk Fat Mass Index (Trunk FMI) (kg/m²): An index calculated by dividing trunk fat mass by the square of height. Elevated trunk fat mass is associated with insulin resistance, hypertension, and inflammation [14].

In recent years, the myokine irisin, secreted by skeletal muscle in response to exercise, has attracted attention due to its association with metabolic diseases. First described in 2012, this molecule is produced through the cleavage of the FNDC5 protein and promotes the browning of white adipose tissue, thereby enhancing energy expenditure [15]. Furthermore, multiple studies have demonstrated that irisin enhances glucose metabolism, decreases insulin resistance, and alleviates metabolic complications associated with obesity [16]. In recent years, irisin has been implicated in the pathophysiology of PCOS; however, existing findings remain inconsistent and controversial [17, 18]. Irisin, released from skeletal muscle through the exercise-induced Ca²⁺–AMPK–PGC-1α–FNDC5 pathway, provides positive effects on muscle mass by promoting muscle growth, myocyte differentiation, and protection against atrophy. The literature reports that irisin levels are positively associated with skeletal muscle mass and muscle performance [16].

Fatty acid-binding protein 4 (FABP4), also known as adipocyte FABP (A-FABP), is a member of the lipocalin family with a molecular weight of approximately 14–15 kDa. It is principally expressed in adipocytes, but is also found in various other cell types, including macrophages, dendritic cells, and placental trophoblasts [19]. FABP4, which binds and transports intracellular fatty acids and plays a key regulatory role in lipid metabolism, is also crucial for fundamental metabolic processes, including adipogenesis and lipolysis. Its expression increases in response to inflammatory stimuli, contributing to the development of metabolic inflammation, and is thereby associated with obesity, type 2 diabetes, cardiovascular diseases, and certain types of cancer [20]. Recent studies have shown that serum FABP4 levels are significantly higher in women with PCOS compared to control groups. Furthermore, FABP4 levels are reported to be positively associated with markers of metabolic dysfunction, including BMI and insulin resistance [21]. However, some studies have reported significant associations between FABP4 levels and parameters reflecting adipose tissue amount, such as total fat mass or visceral fat percentage [22].

This study evaluated serum irisin and FABP4 levels in individuals with PCOS and controls. It also examined the relationships between these biomarkers and body composition parameters obtained via BIA, which is important for elucidating their effects on muscle and adipose tissue in PCOS. In particular, the associations of muscle-related parameters, such as SMM and MMI, with irisin levels, and of fat-related parameters, such as FM and PBF, with FABP4 levels, may help to better understand the roles of these molecules in the metabolic and body composition aspects of PCOS.

This study aimed to evaluate whether serum irisin and FABP4 levels, along with detailed body composition parameters, are independently associated with PCOS in adolescent girls.

Materials and methods

This cross-sectional and prospective study was conducted at two centers: the Department of Pediatric Endocrinology at Erciyes University and Kayseri City Hospital. Adolescent girls diagnosed with polycystic ovary syndrome (PCOS) were compared with healthy controls.

A total of 123 girls aged 11–18 years participated in the study, the diagnosis of PCOS in adolescents was based on the presence of persistent menstrual irregularity (oligomenorrhea) at least two years after menarche, together with clinical and/or biochemical hyperandrogenism, after exclusion of other related disorders. Ovarian morphology assessed by ultrasound was not used for diagnostic purposes, in accordance with the 2023 International Evidence-based Guideline for the Assessment and Management of PCOS in adolescents [2].

PCOS Group (54 cases)

Patients with clinical signs of hyperandrogenism (modified Ferriman–Gallwey Score (mFGS) > 8) and/or laboratory findings (total testosterone (TT) > 55 ng/dL, free androgen index (FAI) > 3.5) presenting with irregular menstrual bleeding at least two years after menarche. In this group, hyperandrogenism due to other endocrinological causes—such as Cushing’s syndrome, non-classic congenital adrenal hyperplasia, and hyperprolactinemia—was excluded using appropriate diagnostic methods [2]. Hirsutism was evaluated using the mFGS [23]. Phenotypic classification for polycystic ovarian morphology was not included in the study, as there are no clearly defined criteria for adolescents [3].

Control Group (69 cases)

Girls whose menstrual cycles were regular at least two years after menarche, with no chronic illness, endocrine disorder, medication use, or pregnancy. Participants had TT < 55 ng/dL and a free androgen index (FAI) < 3.5.

Classification by BMI

The participants were divided into two groups: the obese group (BMI ≥ 2 SD) and the non-obese group (BMI < 2 SD) [24].

Standard deviation scores (SDS) for all anthropometric measurements were calculated using age- and sex-specific reference values for Turkish children [12].

Blood samples were collected after an overnight fast. In regularly menstruating participants, blood sampling was standardized to the early follicular phase (days 2–5 of the menstrual cycle). In participants with oligomenorrhea or irregular menstrual cycles, samples were collected regardless of cycle phase, as precise menstrual staging was not feasible in this group. Fasting glucose and HbA1c were measured as metabolic parameters; however, fasting insulin levels and HOMA-IR were not available in the study dataset. The following parameters were measured: Luteinizing Hormone (LH), Follicle-Stimulating Hormone (FSH), TT, estradiol (E2), Dehydroepiandrosterone sulfate (DHEA-SO4), and Sex Hormone-Binding Globulin (SHBG). The free androgen index (FAI) was calculated using the following formula: FAI = 100 × (total testosterone / SHBG) (nmol/L / nmol/L) [1].

Irisin and FABP4 levels

Plasma levels of irisin and FABP4 were measured using commercial enzyme-linked immunosorbent assay (ELISA) kits (ELK Biotechnology Co. Ltd., Denver, CO 80202, USA). The results were calculated using computerized data reduction against standards, applying a four-parameter polynomial regression model, and expressed in pg/mL.

The performance characteristics of these assays were < 8% and < 8% CV intra-assay and < 10%, and < 10% irisin and FABP4, respectively.

Body composition measurements

Body composition analyses were performed using a multi-frequency bioelectrical impedance analysis (BIA) device, Tanita MC-780 MA (Tanita Corp., Tokyo, Japan). Measurements were conducted with participants wearing light clothing, without shoes, and in a fasting state.

The bioelectrical impedance parameters assessed included: fat mass (FM), percent body fat (PBF), skeletal muscle mass (SMM), muscle mass index (MMI), and trunk fat mass index (Trunk FMI).

Sample selection and ethical approval

The sample size was not determined by an a priori power calculation. Instead, all eligible adolescents who attended the Pediatric Endocrinology outpatient clinics at Erciyes University and Kayseri City Hospital during the study period and met the inclusion criteria were consecutively enrolled. Therefore, the sample size was based on feasibility and availability of eligible participants (feasibility-based consecutive sampling approach).

This study was approved by the Ethics Committee of Erciyes University Faculty of Medicine (Approval No: 2024/287). The study was conducted in accordance with the principles of the Declaration of Helsinki and relevant institutional guidelines. Written informed consent was obtained from the parents or legal guardians of all participants. In addition, written assent was obtained from all child participants prior to their inclusion in the study.

Statistical analysis

Data analysis was performed using IBM SPSS Statistics version 25.0. The distribution of continuous variables was assessed using the Kolmogorov–Smirnov test. For normally distributed data, comparisons between groups were performed using the Student’s t-test, while the Mann–Whitney U test was applied for non-normally distributed data. Categorical variables were compared using the chi-square (χ²) test.

Multivariate logistic regression analysis was performed to identify independent variables influencing the diagnosis of PCOS. A p-value of < 0.05 was considered statistically significant for all tests.

Results

Baseline characteristics

When participants were divided into PCOS and control groups, no significant differences were observed between the groups in terms of age, age at first menarche, BMI, waist circumference, and glucose metabolism parameters (p > 0.05). The PCOS group had significantly higher mFGS and triglyceride (TG) levels (p < 0.001 and p = 0.048, respectively). LH, TT, and FAI levels were markedly higher in the PCOS group (p < 0.001). No significant differences were observed in FSH, E2, DHEA-SO4, irisin, or FABP4 levels. According to bioelectrical impedance analysis, only the muscle mass index (MMI) was significantly higher in the PCOS group (p = 0.005), while other body composition parameters did not differ between groups (p > 0.05) (Table 1).

Table 1.

The comparison of clinical and biochemical findings

	PCOS (n = 54)	Control (n = 69)	p
Demographic and Anthropometric Data
Age, years	15.48 (14.15–16.43)	15.80 (14.49–17.10)	0.199
Age at first menarche, years	11.79 ± 1.16	12.09 ± 1.20	0.161
BMI, kg/m²	25.91 ± 4.75	25.12 ± 6.0	0.427
BMI, SDS	1.35 ± 1.33	0.99 ± 1.63	0.185
Waist circumference, cm	82.78 ± 12.11	81.48 ± 13.22	0.582
Modified Ferriman–Gallwey score	2 (0–11)	0 (0–0)	< 0.001
Biochemical Parameters
Glucose, mg/dL	84.79 ± 7.08	83.36 ± 6.28	0.245
HbA1c, %	5.31 ± 0.27	5.29 ± 0.21	0.660
TG, mg/dL	100 (72.75–153.25)	89 (63–110.5)	0.048
HDL-C, mg/dL	45.2 (37.5– 52.2)	47.05 (41.57–55.55)	0.183
FSH, IU/L	5.33 (4.27– 6.46)	5.03 (3.31–6.22)	0.213
LH, IU/L	11.75 (7.57– 20.42)	6.28 (3.86–8.97)	< 0.001
Estradiol, pg/mL	50.15 (38.82–68.05)	51.65 (34.3-95.22)	0.178
TT, ng/dL	57.25 (43–67.50)	30.5 (22.2–37.82)	< 0.001
DHEA-SO4, µg/dL	414.5 (264–2405)	324 (175.5–1720)	0.641
FAI, %	7.23 ± 4.23	2.97 ± 1.78	< 0.001
Irisin, pg/mL	275.93 ± 94.54	297.29 ± 100	0.242
FABP4, pg/mL	38.01 ± 14.23	37.47 ± 14.61	0.844
Bioelectrical Impedance Parameters
Fat Mass, kg	20.55 (14.50– 26.15)	18.45 (13.27– 27.57)	0.585
Percent Body Fat, %	31.74 ± 6.79	30.69 ± 7.67	0.443
Appendicular Skeletal Muscle Mass, kg	17.35 ± 2.02	16.94 ± 2.86	0.393
Muscle Mass Index, kg/m²	16.39 ± 1.60	15.53 ± 1.59	0.005
Trunk Fat Mass Index, kg/m²	3.70 (2.39–4.85)	3.30 (2.17–4.94)	0.500

Abbreviations: BMI: Body Mass Index, DHEA-SO4: Dehydroepiandrosterone Sulfate, FAI: Free Androgen Index, HDL-C: High-Density Lipoprotein Cholesterol, SD: Standard Deviation, TG: Triglyceride, TT: Total Testosterone, FM: Fat Mass, PBF: Percent Body Fat, SMMa: Appendicular Skeletal Muscle Mass, SDS: Standard Deviation Score, MMI: Muscle Mass Index, Trunk FMI: Trunk Fat Mass Index

Normally distributed variables are expressed as mean ± standard deviation (SD), while non-normally distributed variables are presented as median [interquartile range (IQR)]. Comparisons between groups were performed using the independent samples t-test for normally distributed variables and the Mann–Whitney U test for non-normally distributed variables

Obesity subgroup analyses

When subjects were divided into obese and non-obese groups, no significant differences were observed between non-obese PCOS (n = 19) and non-obese control (n = 35) participants in terms of age, age at first menarche, BMI, or waist circumference. Glucose, HbA1c, triglyceride, HDL-cholesterol and DHEA-SO4 levels also did not differ significantly between the groups (p > 0.05). Serum LH, E2, TT, mFGS and FAI levels were significantly higher in the PCOS group compared to controls (p < 0.05). No significant differences were detected in irisin and FABP4 levels (p > 0.05). Similarly, body composition parameters did not show significant differences between non-obese PCOS and control participants (p > 0.05) (Table 2).

Table 2.

Clinical, anthropometric, and biochemical characteristics of the non-obese cohort (non-obese PCOS patients, n = 19; non-obese controls, n = 25; total n = 54)

	PCOS (n = 19)	Control (n = 35)	p
Demographic and Anthropometric Data
Age, years	15.57 (14.06–16.69)	15.47 (13.90–17.10)	0.821
Age at first menarche, years	12 (11–13)	12 (11–13)	0.232
BMI, kg/m²	23.0 ± 2.8	21.9 ± 3.2	0.113
BMI, SDS	0.6 ± 1.0	0.1 ± 1.2	0.092
Waist circumference, cm	73.1 ± 6.7	71.3 ± 5.6	0.295
Modified Ferriman–Gallwey score	2 (0–12)	0 (0–0)	0.002
Biochemical Parameters
Glucose, mg/dL	83.6 ± 7.2	82.9 ± 6.1	0.705
HbA1c, %	5.2 ± 0.2	5.2 ± 0.2	0.814
TG, mg/dL	92.5 (72.7–121)	81 (58–110)	0.228
HDL-C, mg/dL	47.7 ± 8.7	52.7 ± 13.8	0.268
FSH, IU/L	4.75 (3.96– 6.48)	5.06 (3.55–6.22)	0.246
LH, IU/L	11.10 (5.90–22.10)	7.0 (4.40–9.30)	0.004
Estradiol, pg/mL	56.70 (43.55–77.15)	57.40 (35–104.5)	0.019
TT, ng/dL	53.70 (38.05–60.15)	26.90 (17.20–36.15)	< 0.001
DHEA-SO4, µg/dL	420 (222.5–3235)	272 (141.5–1787.5)	0.392
FAI, %	6.28 ± 3.35	2.58 ± 1.56	< 0.001
Irisin, pg/mL	298 ± 104.2	298.7 ± 101.2	0.981
FABP4, pg/mL	39.0 ± 15.3	38.5 ± 14.8	0.869
Bioelectrical Impedance Parameters
Fat Mass, kg	14.50 (12.25–15.70)	13.90 (11.3–16.90)	0.599
Percent Body Fat, %	25.3 ± 4.8	25.3 ± 4.8	0.977
Appendicular Skeletal Muscle Mass, kg	15.8 ± 1.8	15.1 ± 1.7	0.131
Muscle Mass Index, kg/m²	14.8 ± 1	14.4 ± 1	0.201
Trunk Fat Mass Index, kg/m²	2.1 ± 0.7	2.2 ± 0.8	0.470

Abbreviations: BMI: Body Mass Index, DHEA-SO4: Dehydroepiandrosterone Sulfate, FAI: Free Androgen Index, HDL-C: High-Density Lipoprotein Cholesterol, SD: Standard Deviation, TG: Triglyceride, TT: Total Testosterone, FM: Fat Mass, PBF: Percent Body Fat, SMMa: Appendicular Skeletal Muscle Mass, SDS: Standard Deviation Score, MMI: Muscle Mass Index, Trunk FMI: Trunk Fat Mass Index

Normally distributed variables are expressed as mean ± standard deviation (SD), while non-normally distributed variables are presented as median [interquartile range (IQR)]. Comparisons between groups were performed using the independent samples t-test for normally distributed variables and the Mann–Whitney U test for non-normally distributed variables

No significant differences were observed in age, age at first menarche, BMI, or waist circumference between obese PCOS (n = 35) and control participants (n = 34). Serum LH, TT, mFGS and FAI levels were significantly higher in the PCOS group (p < 0.001), whereas no differences were observed for glucose, HbA1c, HDL-C, TG, FSH, E2, DHEA-SO4. Irisin and FABP4 levels were similar between groups (p > 0.05). No significant differences were found in body composition parameters between the groups (Table 3).

Table 3.

Clinical, anthropometric, and biochemical characteristics of the obese cohort (obese PCOS patients, n = 35; obese controls, n = 34; total n = 69)

	PCOS (n = 35)	Control (n = 34)	p
Demographic and Anthropometric Data
Age, years	15.34 (14.15–16.32)	16.21 (15.14–17.12)	0.110
Age at first menarche, years	12 (11–13)	12 (12–13)	0.293
BMI, kg/m²	30.8 ± 3.05	32.0 ± 4.6	0.317
BMI, SDS	2.6 ± 0.5	2.7 ± 0.7	0.393
Waist circumference, cm	88.3 ± 11	91.6 ± 10.4	0.212
Modified Ferriman–Gallwey score	0 (0–9)	0 (0–5)	0.002
Biochemical Parameters
Glucose, mg/dL	85.3 ± 11	91.6 ± 10.4	0.336
HbA1c, %	5.3 ± 0.2	5.3 ± 0.3	0.563
TG, mg/dL	110.5 (67–175.75)	93 (68.25–117.5)	0.177
HDL-C, mg/dL	44.1 ± 9	46.7 ± 9	0.270
FSH, IU/L	5.72 (4.46–6.37)	4.81 (3.03–6.17)	0.492
LH, IU/L	12.30 (8.97–19.42)	5.31 (3.62–7.10)	< 0.001
Estradiol, pg/mL	47.1 (37.70–65.30)	46.7 (31.35– 80.45)	0.966
TT, ng/dL	63.30 (48.75–72.47)	27.1 (21.55– 35.60)	< 0.001
DHEA-SO4, µg/dL	423 (282–1980)	464 (218–1710)	0.798
FAI, %	8.35 ± 4.29	4.03 ± 2.06	< 0.001
Irisin, pg/mL	264.5 ± 88.6	295.7 ± 100.2	0.182
FABP4, pg/mL	34.6 ± 12.3	39.8 ± 15.9	0.255
Bioelectrical Impedance Parameters
Fat Mass, kg	26 (22.52– 28.62)	27.9 (25.65–32.81)	0.071
Percent Body Fat, %	35.1 ± 4.6	35.9 ± 3.8	0.442
Appendicular Skeletal Muscle Mass, kg	18.1 ± 1.6	19 ± 2.4	0.063
Muscle Mass Index, kg/m²	17.2 ± 1.1	16.8 ± 0.9	0.239
Trunk Fat Mass Index, kg/m²	4.5 ± 1.1	4.9 ± 1.1	0.164

Abbreviations: BMI: Body Mass Index; DHEA-SO4: Dehydroepiandrosterone Sulfate, FAI: Free Androgen Index; HDL-C: High-Density Lipoprotein Cholesterol, SD: Standard Deviation, TG: Triglyceride, TT: Total Testosterone, FM: Fat Mass, PBF: Percent Body Fat, SMMa: Appendicular Skeletal Muscle Mass, SDS: Standard Deviation Score, MMI: Muscle Mass Index, Trunk FMI: Trunk Fat Mass Index

Normally distributed variables are expressed as mean ± standard deviation (SD), while non-normally distributed variables are presented as median [interquartile range (IQR)]. Comparisons between groups were performed using the independent samples t-test for normally distributed variables and the Mann–Whitney U test for non-normally distributed variables

Correlation analyses

No significant correlations were observed between irisin levels and age, BMI, waist circumference, biochemical parameters, or body composition measures; however, a weak negative correlation was found with the mFGS (r = − 0.187, p = 0.026). FABP4 levels showed a significant negative correlation with FSH (r = − 0.182, p = 0.040) and DHEA-SO4 (r = − 0.336, p < 0.001). A near-significant positive correlation was observed between irisin and FABP4 levels (r = 0.155, p = 0.075) (Table 4).

Table 4.

Correlations of Irisin and FABP4 Levels with Demographic, Biochemical, and Body Composition Parameters

	Irisin		FABP4
	r	p	r	p
Demographic Parameters
Age, years	0.030	0.725	0.117	0.179
Age at first menarche, years†	0.085	0.315	0.057	0.517
BMI, kg/m²†	-0.073	0.315	-0.064	0.473
BMI, SD†	-0.077	0.365	-0.038	0.665
Waist circumference, cm†	-0.133	0.121	-0.057	0.516
Modified Ferriman–Gallwey score†	-0.226**	0.007	-0.126	0.148
Biochemical Parameters
Glucose, mg/dL	0.012	0.887	0.102	0.252
HbA1c, %	0.044	0.659	0.069	0.502
Triglycerides, mg/dL†	0.132	0.173	-0.030	0.766
HDL-C, mg/dL	0.030	0.761	0.059	0.561
FSH, IU/L†	-0.035	0.686	-0.202*	0.021
LH, IU/L†	-0.120	0.158	-0.091	0.307
Estradiol, pg/mL†	0.050	0.557	0.107	0.227
Total testosterone, ng/dL†	-0.028	0.753	-0.078	0.400
DHEA-SO4, µg/dL†	0.001	0.992	-0.318**	< 0.001
FAI, %†	-0.094	0.448	0.100	0.438
Irisin, pg/mL	NA	NA	0.155	0.075
FABP4, pg/mL	0.155	0.075	NA	NA
Bioelectrical Impedance Parameters
Fat Mass, kg†	-0.090	0.290	-0.030	0.735
Percent Body Fat, %†	-0.106	0.209	-0.029	0.740
Appendicular Skeletal Muscle Mass, kg†	0.035	0.687	0.097	0.271
Muscle Mass Index, kg/m²	-0.017	0.844	0.082	0.351
Trunk Fat Mass Index, kg/m²	-0.060	0.483	0.059	0.508

Abbreviations: BMI: Body Mass Index; DHEA-SO4: Dehydroepiandrosterone Sulfate; FAI: Free Androgen Index; HDL-C: High-Density Lipoprotein Cholesterol; SD: Standard Deviation; r: Correlation coefficient; p: significance. Self-correlations are not applicable (NA). Normality of variables was assessed prior to correlation analysis. Pearson correlation was applied for normally distributed variables, whereas Spearman rank correlation was used for non-normally distributed variables. Variables analyzed with Spearman correlation are indicated with (†). Statistical significance was defined as * p < 0.05, with ** indicating p < 0.01

Logistic regression analyse

Multivariable logistic regression analysis was performed to identify independent predictors of PCOS. The dependent variable was coded as 0 = control and 1 = PCOS. Variables were selected based on their clinical and biological relevance to PCOS pathophysiology. Collinearity diagnostics were assessed prior to logistic regression analysis, and results are presented in Supplementary Table 1.

The overall model was statistically significant (Omnibus test: χ² = 63.522, df = 5, p < 0.001), indicating that the included variables significantly improved model fit compared with the null model. The model demonstrated acceptable explanatory power, with a Cox & Snell R² of 0.411 and a Nagelkerke R² of 0.554. The Hosmer–Lemeshow test indicated an acceptable model fit (p > 0.05).Among the variables included in the model, total testosterone was identified as the strongest independent predictor of PCOS (B = 0.085, OR = 1.089, 95% CI: 1.050–1.128, p < 0.001). LH was also independently associated with PCOS (B = 0.081, OR = 1.084, 95% CI: 1.003–1.172, p = 0.041). In contrast, BMI-SDS (p = 0.703), irisin (p = 0.105), and FABP4 (p = 0.703) were not independently associated with PCOS after adjustment for other variables (Table 5).

Table 5.

Multivariable logistic regression analysis of independent predictors of PCOS

Variable	Beta (B)	S.E.	Wald	OR (95% CI)	p-value
BMI-SDS	0.059	0.156	0.145	1.061 (0.782–1.440)	0.703
LH (IU/L)	0.081	0.040	4.163	1.084 (1.003–1.172)	0.041
Total testosterone (ng/dL)	0.085	0.018	21.746	1.089 (1.050–1.128)	< 0.001
Irisin (pg/mL)	-0.005	0.003	2.629	0.995 (0.989–1.001)	0.105
FABP4 (pg/mL)	0.006	0.016	0.145	1.006 (0.975–1.039)	0.703
Constant	-3.593	1.188		-	0.002

Logistic regression analysis was performed to identify independent predictors of PCOS. The dependent variable was coded as 0 = control and 1 = PCOS. Variables were selected based on their clinical and biological relevance to PCOS pathophysiology. Omnibus test of model coefficients: χ² = 63.522, df = 5, p < 0.001. -2 Log likelihood: 98.777. Cox & Snell R²: 0.411. Nagelkerke R²: 0.554. Hosmer–Lemeshow test indicated an acceptable model fit (p > 0.05)

ROC analysis based on predicted probabilities from the multivariable logistic regression model demonstrated the ability of the model to discriminate between PCOS patients and controls (Fig. 1).

Fig. 1.

Receiver operating characteristic (ROC) curve analysis of the multivariable logistic regression model for predicting PCOS. The area under the curve (AUC) was 0.861 (SE = 0.040, 95% CI: 0.781–0.940, p < 0.001), indicating good discriminative performance

Discussion

In this study, the relationships between PCOS and various hormonal, metabolic, and body composition parameters were examined in detail, with a particular focus on the potential roles of emerging biomarkers such as irisin and FABP4 in the pathophysiology of PCOS.

Diagnosing PCOS during adolescence is more challenging than in adults due to the physiological hormonal changes characteristic of this developmental period. While different phenotypes—such as ovulatory PCOS and non-hyperandrogenic PCOS—can be identified in adults based on the Rotterdam criteria, polycystic ovarian morphology detected by pelvic ultrasonography is generally considered physiological in adolescents is therefore not recommended for diagnostic purposes [3]. Consequently, PCOS in adolescents is primarily evaluated based on the classic phenotype, characterized by oligomenorrhea together with clinical or biochemical hyperandrogenism [25]. Participants exhibiting the classical PCOS phenotype were likewise enrolled in our study. Consequently, the PCOS group was found to have significantly higher mFGS, LH, TT, and free androgen index (FAI) levels compared with the control group. This finding is consistent with previous studies demonstrating that hyperandrogenism remains the predominant clinical and hormonal feature of PCOS during adolescence [26, 27].

Lipid metabolism in individuals with PCOS has been reported to be affected through various pathways, including steroid hormone biosynthesis, sphingolipid metabolism, and fatty acid metabolism. It has also been demonstrated that triglyceride levels show a positive correlation with BMI and insulin resistance in individuals with adolesant PCOS [28, 29]. In our study, despite having similar BMIs, triglyceride levels were found to be significantly higher in the PCOS group compared with the control group. In contrast, the meta-analysis by Badaracco et al. reported no significant difference in triglyceride levels between women with adult PCOS and healthy controls [30]. This discrepancy may be attributed to heterogeneity in sample characteristics, as well as differences in dietary habits and lifestyle factors across studies. On the other hand, the absence of significant differences in other metabolic indicators such as fasting glucose, HbA1c, and HDL-C suggests that overt metabolic complications may not yet be apparent in the majority of participants in our study.

Although increased fat mass in women with PCOS has been consistently reported in numerous studies, findings regarding lean mass—which is predominantly composed of muscle mass—have been inconsistent in the literature [31, 32]. Fighera et al. reported that women with PCOS phenotypes A and B had higher skeletal muscle mass compared with other phenotypes. These groups were also described as being more obese and having higher rates of metabolic syndrome. In multivariate regression analysis, increased muscle mass was found to be positively associated with higher androgen levels, as well as increased fat and bone mass [31]. On the other hand, in a meta-analysis by Kazemi et al., reported to total lean mass was associated with obesity rather than hyperandrogenism [32]. In our study, the increased muscle mass index (MMI) observed in the PCOS group is consistent with the reported increases in trunk and total lean mass in clinical conditions previously associated with hyperandrogenism [31]. Although BMI was similar between the PCOS and control groups, the increased MMI appears to be explained by elevated androgen levels.

Findings regarding irisin levels in the literature are inconsistent: studies have reported higher, lower, or similar concentrations in the PCOS group compared with controls [17, 18, 33, 34]. In the study by Paczkowska et al., irisin levels were similar between the PCOS and control groups, which had comparable BMI and androgen levels. Irisin levels were found to be negatively correlated with BMI, fat mass, and insulin resistance [35]. In this study, irisin levels were not significantly associated with metabolic parameters such as BMI, waist circumference, or lipid profile, but were negatively correlated with clinical hyperandrogenism. Differences between our findings and previous reports may be related to population characteristics, as most existing studies have been performed in adult women rather than adolescents.

Studies on serum FABP4 levels have yielded conflicting results. Hu et al. reported significantly higher FABP4 mRNA levels in ovarian granulosa cells compared with controls. They also demonstrated that increased serum FABP4 was associated with HOMA-IR, BMI, and testosterone levels, suggesting a potential role for FABP4 in the pathophysiology of PCOS [36]. Möhling et al. reported a positive correlation between FABP4 levels and both BMI and fat mass and lean mass measured by DEXA in individuals with PCOS. They noted that, as a molecule secreted from adipose tissue, FABP4 showed a particularly strong association with fat mass. However, no significant relationship was observed between FABP4 levels and either testosterone or insulin resistance [22]. In the present study, no significant differences in FABP4 levels were detected between the PCOS and control groups. A potential explanation for the absence of significant differences in FABP4 levels between groups may be the similar body fat distribution between the PCOS and control groups, in addition to its stronger association with visceral adiposity rather than total or subcutaneous fat mass, which was not directly assessed in the present study [37].

Polycystic ovary syndrome (PCOS) is increasingly recognized as a multifactorial disorder in which genetic, metabolic, and endocrine factors interact in a complex manner. In this context, the lack of association between RETN gene polymorphisms and PCOS susceptibility in previous studies further supports the heterogeneous genetic background of the disease, suggesting that single genetic variants are insufficient to explain its development [38]. Similarly, metabolic biomarkers such as irisin and FABP4 reflect different aspects of muscle–adipose tissue communication and energy homeostasis rather than isolated pathogenic pathways. Emerging evidence from other adipokine systems, including CTRP family proteins, also highlights the integrated nature of metabolic signaling networks involved in insulin resistance and adiposity [39]. Collectively, these findings suggest that PCOS pathophysiology is driven by a coordinated dysregulation of multiple genetic and metabolic pathways rather than a single determinant.

Negative correlations were observed between FABP4 and both DHEA-SO₄ and FSH levels. The similarity in BMI, fat mass, and fat mass percentage between the groups likely contributed to these findings. Notably, the significant negative correlation observed with DHEA-SO₄ may warrant further investigation into potential interactions between adrenal-derived androgen production and FABP4. In the multivariable analysis, no independent association was identified between irisin or FABP4 and PCOS. Although a weak inverse relationship was observed in the regression analyses, the effect sizes were very small and were not considered clinically meaningful for either biomarker.

Several biochemical and anthropometric parameters associated with polycystic ovary syndrome (PCOS) were analyzed using logistic regression. The identification of TT level as significant risk factor highlights the importance of these parameter for the diagnosis of PCOS during adolescence [2]. Other variables included in the model, such as BMI, LH, irisin and FABP4 showed values approaching statistical significance but did not reach the significance threshold. These findings support the central role of androgen excess in adolescent PCOS, while suggesting that novel biomarkers provide limited independent predictive value in this study population.

In the present study, body composition was assessed using BIA. Although BIA is a practical, non-invasive, and widely used method in clinical settings, it is not considered a reference-standard technique for detailed body composition assessment. This limitation is particularly relevant in adolescent populations, where physiological growth and developmental changes may influence body composition measurements. In addition, several factors such as hydration status, timing of the measurement, recent physical activity, and food intake can affect the accuracy and reliability of BIA-derived estimates. Therefore, findings related to MMI and other body composition parameters should be interpreted with caution. More precise methods, such as dual-energy X-ray absorptiometry (DXA), would provide more accurate and reproducible assessments. Accordingly, the body composition results of the present study should be considered exploratory and warrant confirmation in future studies using more robust methodologies.

Limitations

The limitations of this study include the absence of an a priori sample size calculation; however, all eligible adolescents were consecutively enrolled during the study period based on feasibility. The sample size may still be considered relatively limited, particularly in the subgroup analyses stratified by obesity status, which may have reduced the ability to detect modest differences in irisin and FABP4 levels. The relatively small sample size, particularly in subgroup analyses, and the lack of adjustment for multiple comparisons may limit the statistical power of the study and increase the risk of both type I and type II errors. Therefore, negative findings for irisin, FABP4, and body composition parameters should not be interpreted as definitive absence of association.

In addition, physical activity levels, which are known to influence circulating irisin concentrations, were not assessed or controlled for and may have acted as a potential confounding factor. Another limitation of this study is that pubertal status, physical activity, and dietary intake were not systematically assessed and therefore could not be controlled for in the analyses. Due to the cross-sectional design of the study, causal relationships between irisin, FABP4, body composition, and PCOS cannot be established. These findings should therefore be interpreted with caution and confirmed in larger, well-controlled prospective studies.

In addition, fasting insulin levels and HOMA-IR were not available in the dataset, which limits a more comprehensive assessment of insulin resistance and its potential relationship with irisin and FABP4.

In conclusion, this study highlights the importance of classical androgenic parameters in adolescent PCOS, while suggesting that irisin and FABP4 do not provide additional independent diagnostic value in this cohort. However, these findings should be interpreted in light of the study limitations, including the relatively small sample size and the lack of assessment of physical activity levels. Larger, prospective studies are needed to further elucidate the potential metabolic and clinical roles of these biomarkers.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

L.K. and U.G.S. wrote the main manuscript text and revised the manuscript for intellectual content. S.M. and N.S. performed the irisin and FABP4 biochemical assays and contributed to data analysis. N.H. and U.G.S. provided guidance on the study design and methodology, particularly regarding the irisin and FABP4 analyses, and critically reviewed the manuscript. L.K., U.G.S., and N.H. contributed to the interpretation of the irisin and FABP4 biochemical assays and prepared the figures illustrating the findings. All authors read and approved the final manuscript.

Funding

This study was financially supported by the Erciyes University Scientific Research Projects (BAP) Program (Project No: TSA-2023-13015).

Data availability

All raw data generated and analysed during the current study is already included in this published article. The data analysis files (statistical scripts and processed data) are not publicly available as they contain intermediate identifiers or institutional information but are available from the corresponding author on reasonable request. The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request. The data are not publicly available due to ethical and institutional restrictions but can be shared upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Erciyes University Faculty of Medicine (Approval No: 2024/287). The study was conducted in accordance with the principles of the Declaration of Helsinki and relevant institutional guidelines. Written informed consent was obtained from the parents or legal guardians of all participants. In addition, written assent was obtained from all child participants prior to their inclusion in the study.

Consent for publication

Not required as this study did not include any identifiable individual data, images, or personal information of the participants.

Competing interests

The authors declare no competing interests.