Introduction

Polycystic ovary syndrome (PCOS) is a consequence of a diverse diet and hormonal imbalances, significantly impacting the reproductive health of women globally. This endocrine disorder is characterized by polycystic ovarian morphology, abnormal menstrual cycles, and elevated androgen levels. Its clinical presentation is highly heterogeneous, predominantly causing anovulatory infertility1. Epidemiological studies indicate that PCOS affects 6–15% of women of reproductive age worldwide, a figure expected to rise annually2,3. Mechanistic research has identified numerous factors contributing to the onset and progression of PCOS. Genetic and postnatal developmental processes play pivotal roles in its manifestation. Genetic factors encompass abnormal DNA methylation, hyperandrogenism, and insulin gene mutations, while postnatal factors include androgen exposure, disrupted insulin signaling, hypothalamic-pituitary–gonadal axis dysfunction, intestinal flora imbalance, and unfavorable lifestyle choices such as sedentary behavior and a high-fat, high-sugar diet4,5,6,7,8,9. Despite the involvement of multiple factors in PCOS, the precise reasons for its clinical heterogeneity remain elusive.

Beyond oligomenorrhea, polycystic ovarian morphology, and elevated androgen levels, PCOS patients frequently experience multi-system complications, including obesity, abnormal glucose metabolism, depression, and hypertension. Among these comorbidities, up to 50–70% PCOS patients exhibit insulin resistance (IR)10. PCOS patients with IR face significantly elevated risks of anovulatory dysfunction, anovulation, miscarriage, reduced pregnancy rates, and complications such as gestational diabetes mellitus and preeclampsia11,12. Additionally, studies indicate that premenopausal women with PCOS have an increased incidence of adverse cardiovascular events, such as atherosclerosis, coronary heart disease, and cerebrovascular disease after menopause13,14. These findings suggest that IR not only impacts the short-term pregnancy risks for PCOS patients but might also be associated with long-term metabolic risks in women with a history of PCOS after menopause. Therefore, timely identification and monitoring of IR risk in PCOS patients are crucial.

Growth differentiation factor 15 (GDF15), also known as macrophage inhibitory factor, is a cytokine produced by activated macrophages during stress conditions and belongs to the transforming growth factor β superfamily. Under physiological conditions, GDF15 is predominantly expressed in placental trophoblast cells and minimally in the liver, lung, kidney, and other tissues15. Early research linked GDF15 to the proliferation and migration of prostate cancer cells and myocardial cell damage16,17. Recent studies highlight GDF15’s pivotal role in metabolic diseases, including IR. Hong et al.18 reported elevated GDF15 levels in individuals with impaired fasting glucose and diabetes, independent of risk factors like age and BMI. In animal studies, Patel et al.19 demonstrated that a high-fat diet increases GDF15 levels in mice. Given that long-term high-fat diet is a known risk factor for IR, this suggests a potential association between GDF15 and IR. Mechanistic studies further revealed that a high-fat diet triggers the cellular integrated stress response, leading to phosphorylation of eukaryotic translation initiation factor-2α (EIF-2α) and upregulation of activating transcription factor 4 (ATF4) and C/EBP homologous protein (CHOP), subsequently promoting GDF15 expression19. However, PCOS involves intricate hormonal regulation and metabolic disorders, and it is currently unclear whether GDF15 levels are elevated in IR-PCOS. Therefore, the identification of GDF15 levels in PCOS patients is crucial, as it may serve as a potential biomarker for insulin resistance (IR), a common and significant complication in PCOS. This study aims to investigate whether GDF15 levels are associated with the occurrence of insulin resistance in patients with PCOS. Understanding this relationship can help clinicians identify patients at higher risk of metabolic complications, facilitating early intervention and personalized treatment strategies.

Materials and methods

Subjects and group

Between December 2022 and February 2024, 233 PCOS patients were recruited from the Second Xiangya hospital. Clinical data, including sociodemographic information, height, weight, and family history. The patients were divided into two groups: the IR-PCOS group (n = 124) and the non-IR-PCOS group (n = 109). Furthermore, based on the distribution of GDF15 levels, PCOS patients were divided at the median into the normal GDF15 group (n = 116) and the high GDF15 group (n = 117). Exclusion criteria comprised (shown as Fig. 1): (1) other causes of hyperandrogenism, such as Cushing’s syndrome, interstitial theca cell hyperplasia, atypical congenital adrenal hyperplasia, and androgen-secreting tumors of ovarian or adrenal origin; (2) conditions leading to ovulation disorders, such as thyroid dysfunction, premature ovarian failure, hypothalamic-pituitary amenorrhea, and hyperprolactinemia; (3) severe infections, malignant tumors, immunodeficiency, and the use of corticosteroids or other immunosuppressants; (4) recent history (within the past month) of androgen or estrogen supplementation, oral antidiabetic drugs (e.g., biguanides, sulfonylureas, α-glucosidase inhibitors, DPP-4 inhibitors), and injectable antidiabetic drugs (e.g., insulin and GLP-1 receptor agonists); (5) autoimmune diseases (e.g., Hashimoto’s thyroiditis, Sjogren’s syndrome, systemic lupus erythematosus), genetic disorders (e.g., 3β-hydroxysteroid dehydrogenase deficiency, true hermaphroditism), and cardiac insufficiency. This study received approval from the ethical review committee of the Second Xiangya Hospital, Central South University, and all procedures complied with the Declaration of Helsinki. The informed consents were obtained from all participants.

Fig. 1
figure 1

Flowchart of inclusion and exclusion criteria of participants.

Full size image

PCOS diagnosis and insulin resistance evaluation

All patients suspected of having PCOS were diagnosed based on the Rotterdam criteria established by the European Society of Human Reproduction and Embryology and the American Society for Reproductive Medicine (ESHRE/ASRM)20. These criteria include: (1) ultrasound evidence of polycystic ovarian morphology; (2) abnormal menstruation, such as oligo-ovulation or anovulation; and (3) hyperandrogenism or clinical signs of hyperandrogenism, such as hirsutism, acne, or alopecia. The diagnosis of insulin resistance was made by an endocrinologist and assessment was used the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) and the Quantitative Insulin Sensitivity Check Index (QUICKI). The formulas for HOMA-IR and QUICKI are: HOMA-IR = (Fasting Insulin (μU/ml) x Fasting Blood Glucose (mmol/L))/22.5, and QUICKI = 1/(Log(Fasting Plasma Glucose (mg/dl)) + Log(Fasting Insulin (μU/ml))). An HOMA-IR value of 2.69 or higher is considered indicative of IR21,22.

Sample collection and preparation

All patients were fasted after 8:00 pm the night before, with blood samples collected the following morning. Blood samples were collected between 7:00 and 11:00 am following the patient visits. Prior to blood collection, all patients were in the early follicular phase of their menstrual cycle. A 3 ml blood sample was drawn from the median cubital vein into a yellow-top tube containing a separation gel. This sample was then centrifuged at 3,500 rpm for 15 min to obtain serum. Concurrently, a 2 ml blood sample was collected in a purple-top tube containing EDTA-K2 anticoagulant for plasma collection. Plasma was obtained through high-speed centrifugation for subsequent hormone analysis. Both serum and plasma samples were stored in a medical freezer at − 80 °C until analysis.

Laboratory measurement

Serum and plasma samples stored at − 80 °C were thawed to room temperature before analysis. The inter-assay and intra-assay variability for GDF15 were < 6% and 2.8%, respectively, when using kits within their validity period for GDF15 detection. GDF15 levels were determined using the enzyme-linked immunosorbent assay (ELISA) method. Specifically, 100 µl serum was added to a microplate pre-coated with a specific antibody. After subsequent steps of washing, antibody binding detection, additional washing, binding with streptavidin–horseradish peroxidase, and final washing, a substrate was added for color development. The optical density (OD) value was then measured using a microplate reader (Autobio Instruments, China).

Serum levels of low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), total cholesterol (TC), triglycerides (TG), alanine aminotransferase (ALT), aspartate aminotransferase (AST), and fasting blood glucose (FBG) were assessed using the automatic biochemical immunoassay Cobas 8000 (Roche, Germany). Blood glucose was measured via the hexokinase method, while fasting insulin (Fins) levels were determined using the chemiluminescence method. Additionally, follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradiol (E2), prolactin (PRL) and testosterone (TST) levels were analyzed using the fully automated chemiluminescent immunoanalyzer i4000SR (Abbott, USA).

Statistic analysis

Statistical analysis and data processing were performed using GraphPad Prism 9.0 and R 4.2.1. Prior to analysis, all measurement data were assessed for normality. Data following a normal distribution were presented as mean ± standard deviation (x̅ ± SD) and analyzed using Student-t test. For non-normally distributed data, values were expressed as the interquartile range [M(P25, P75)] and analyzed using the Mann–Whitney U test. Count data were presented as percentages and compared between groups using the chi-square test. Statistical power analysis was performed using SPSSAU software. A power of 0.8 or greater at a significance level of α = 0.05 was considered adequate to detect statistically significant effects. The power values for both the non-IR-PCOS and IR-PCOS group were ≥ 0.8, indicating that the sample size was sufficient for the analyses conducted. Spearman correlation analysis was employed to assess the strength of association between GDF15 and each variable. The relationship between GDF15 and IR in women with PCOS was analyzed using both univariate and multivariate logistic regression. Restricted cubic spline analysis was employed to explore the nonlinear association between GDF15 levels and PCOS. Receiver operating curve analysis (ROC) was used to evaluate the predictive efficacy. A P-value of < 0.05 was considered statistically significant.

Results

Clinical characteristic

Table 1 presents the clinical information and laboratory data of the PCOS patients. Of the 233 PCOS patients, 124 (53.2%) were diagnosed with IR. Compared to the non-IR-PCOS group, the IR-PCOS group exhibited higher BMI, elevated levels of TG, non-HDL-C, ALT, FPG, Fins, and HOMA-IR, and lower levels of HDL-C and QUICKI (all P < 0.05). These findings suggest that patients with IR-PCOS experience more pronounced disturbances in glucose and lipid metabolism. Furthermore, IR-PCOS patients showed a higher prevalence of familial diabetes and elevated levels of LH, LH/FSH, PRL, TST and GDF15 (P < 0.05). However, no significant differences were observed in age, family history of hypertension, AST, FSH, and E2 between the two groups (all P > 0.05).

Table 1 Baseline data between PCOS patients with and without IR.
Full size table

Clinical analysis of high versus normal GDF15 levels in PCOS patients

To investigate the physiological function of GDF15 in PCOS, we divided the subjects into a high GDF15 group and a normal GDF15 group. Due to the lack of established reference ranges for GDF15 in women of reproductive age, we categorized the groups based on the median value of GDF15 distribution (< 666.78 pg/ml vs. ≥ 666.78 pg/ml). The results, as shown in Table 2, indicate that compared to the normal GDF15 group, the high GDF15 group had higher BMI, TC, FPG, Fins, HOMA-IR, QUICKI, TST, and a higher incidence of diabetes in family members, while PRL and HDL-C levels were lower (P < 0.05). These findings suggest that the PCOS population with elevated GDF15 levels may have an increased incidence of metabolic abnormalities such as IR and dyslipidemia.

Table 2 Clinical analysis of high versus normal GDF15 levels in PCOS patients.
Full size table

Spearman analysis between GDF15 and clinical variables

To ascertain the potential association between GDF15 and IR in patients with PCOS, Spearman correlation analysis was employed to investigate the relationship between IR-related clinical parameters and serum GDF15 levels in PCOS individuals. As depicted in Table 3, GDF15 exhibited positive correlations with HOMA-IR, FPG, Fins, BMI, TST, and TG (with r = 0.321, 0.242, 0.308, 0.360, 0.234, and 0.307, respectively; all P < 0.05). Conversely, it demonstrated negative correlations with QUICKI and HDL-C (with r = − 0.321 and − 0.243, respectively; both P < 0.05).

Table 3 Spearman analysis.
Full size table

Logistic regression between circulating GDF15 level and the prevalence of IR in PCOS population

Both univariate and multivariate logistic regression analyses were further employed to assess the impact of clinical variables on IR development in PCOS patients. Clinical variables with a P-value of < 0.1 in the univariate regression analysis were incorporated into the multivariate regression model. The multivariate logistic regression analysis revealed that GDF15 remained a significant predictor of IR in women with PCOS (P < 0.05, OR 3.810 and 95% CI 1.871 to 7.758) (Table 4).

Table 4 Univariate and multivariate analyses for the risk factors of IR in female with PCOS.
Full size table

To examine the relationship between circulating GDF15 levels and IR-PCOS, we conducted a stratified analysis based on GDF15 levels and calculated the odds ratios (ORs) for IR risk, with Q1 (the lowest) as the reference (Table 5). Initially, serum GDF15 concentration showed a positive correlation with IR prevalence in PCOS patients. After adjusting for clinical parameters such as age, BMI, family history of diabetes, and hypertension in Model 1, the results remained consistent. In Model 2, after further adjustment for additional laboratory parameters including TC, TG, HDL-C, LDL-C, non-HDL-C, ALT, AST, FSH, LH, LH/FSH, PRL, E2, and TST alongside clinical parameters, the association remained statistically significant with minimal change. The fully adjusted ORs for IR in Model 2 was 4.490(95% CI 1.541,13.088) in quartile 4 of serum GDF15 levels (the highest) compared to quartile 1 (the lowest).

Table 5 Association of the prevalence of IR with the serum GDF15 concentration in PCOS women.
Full size table

Nonlinear association of GDF15 levels with IR in PCOS female

Figure 2 depicts the correlation between IR prevalence and GDF15 levels in women with PCOS utilizing restricted cubic spline analysis. The results unveiled an S-shaped relationship between GDF15 and IR-PCOS. Notably, the incidence of IR exhibited a gradual escalation with increasing GDF15 concentration, with the most significant rise occurring within the range of 500–1000 pg/ml.

Fig. 2
figure 2

Restricted cubic spline model of the ORs of IR-PCOS with serum GDF15 levels.

Full size image

ROC analysis using GDF15 and other variables

Based on the results of both univariate and multivariate logistic analyses, PRL, HDL-C, and GDF15 exhibited associations with IR. Subsequently, these indicators were subjected to ROC analysis to assess their predictive capacity for IR. The ROC analysis revealed that the area under the curve (AUC) for GDF15, PRL, and HDL-C in predicting IR was 0.706 (95% CI 0.639–0.772), 0.589 (95% CI 0.516–0.662), and 0.657 (95% CI 0.587–0.728), respectively (Fig. 3). The optimal cut-off value for GDF15 was determined to be 594.63 pg/L, yielding a Youden index of 0.315, sensitivity of 0.774, and specificity of 0.541. Furthermore, logistic regression analysis indicated that the combined probability of the aforementioned three indicators in predicting IR was 0.763 (95% CI 0.701–0.826).

Fig. 3
figure 3

ROC curve of GDF15 in predicting the occurrence of IR in PCOS patients.

Full size image

Discussion

To the best of our knowledge, this is the first study to investigate serum GDF15 levels in relation to IR in women with PCOS. In this study, we categorized participants based on their IR status and found that GDF15 levels were significantly higher in IR-PCOS patients compared to non-IR-PCOS patients. After adjusting for confounding risk factors, elevated GDF15 remained significantly associated with IR, suggesting that increased circulating GDF15 levels in reproductive-age women with PCOS may reflect an existing state of IR.

IR is a prevalent metabolic disorder among women with PCOS, characterized by diminished insulin sensitivity and responsiveness across various tissues, including the liver, skeletal muscle, and adipose tissue. This inefficiency in insulin utilization leads to persistent hyperglycemia, establishing a detrimental cycle of “hyperinsulinemia-IR-hyperglycemia”23. In the liver, excessive intrahepatic fat accumulation in PCOS can inhibit insulin receptor function, disrupting insulin signaling and affecting hepatic glucose production24. In skeletal muscle, IR manifests as impaired glucose uptake due to dysfunctional insulin signaling25,26,27. Adipose tissue dysfunction in PCOS may involve factors such as hypoxia, lipolysis, and oxidative stress, contributing to reduced glucose transporter-4 expression and compromised insulin-mediated glucose transport28. Chronic inflammation is also implicated, with macrophages in adipose tissue releasing cytokines that exacerbate IR29,30. These findings highlight the multifaceted nature of IR development in PCOS.

GDF15 has been identified as a marker associated with several metabolic conditions, including diabetes and obesity31,32,33,34,35. Clinical studies reveal that circulating GDF15 levels are significantly higher in patients with type 2 diabetes mellitus (T2DM) and obesity compared to healthy individuals. These levels positively correlate with blood glucose control indicators such as HOMA-IR and Fins. After adjusting for potential confounders like BMI, GDF15 concentrations remained elevated in individuals with abnormal glucose metabolism compared to the normal group. Moreover, GDF15 levels were higher in T2DM patients than in those with impaired glucose tolerance35,36. This suggests that more severe IR correlates with higher GDF15 levels.

However, the role of GDF15 in IR-PCOS remains unclear. Current evidence suggests that its elevation might be a consequence rather than a predictor of the condition’s development. In obesity, increased circulating GDF15 levels are observed in both human and animal models, and treatment with GDF15 ameliorates obesity symptoms35,37,38. GDF15 knockout in mice results in reduced exercise capacity and a more obese phenotype compared to wild-type mice39. In diabetic contexts, studies suggest a beneficial role for GDF15, as demonstrated by Chen et al.40, who showed that GDF15 overexpression alleviates diabetic nephropathy, improves lipid homeostasis, glucose tolerance, insulin sensitivity, and reduces renal injury in a mouse model. These findings support the idea that elevated GDF15 levels are a compensatory response to IR in PCOS pathology.

In our study, 53.2% (124/233) of PCOS subjects exhibited IR, aligning with epidemiological data that 50–70% of PCOS individuals experience IR10. Although we have identified elevated levels of GDF15 in this population, the exact mechanism by which GDF15 affects IR as a result of pathological development remains unclear. One potential mechanism involves oxidative stress. GDF15 mediates the activation of the AMPK-p53 pathway, enhancing glucose intolerance, fatty acid oxidation, and reducing inflammation in mice41. Elevated glucose levels in endothelial cells induce GDF15 expression, which inhibits the PI3K/Akt/eNOS pathway, reducing endothelial cell apoptosis42. Inflammation may also play a significant role. Research has shown that GDF15 overexpression inhibits the AGE/RAGE axis and its downstream inflammatory molecules in IR mice, particularly suppressing the TLR4/MyD88/NF-κB pathway, reducing inflammation, and enhancing insulin sensitivity40. Additionally, GDF15’s anti-inflammatory properties are associated with a decrease in pro-inflammatory M1 macrophages and an increase in anti-inflammatory M2 macrophages in adipose tissue, suggesting a potential mechanism for ameliorating IR43. There is also evidence that GDF15 may affect IR through the central nervous system44.GFRAL, a receptor for GDF15, is predominantly expressed in the hindbrain and the nucleus of the solitary tract across various species, including humans45,46. Supplementation with GDF15 in GFRAL-deficient mice led to reduced metabolic improvements, highlighting the importance of the GDF15-GFRAL axis in regulating β-adrenergic signaling and insulin action in the liver and adipose tissue47. In conclusion, the elevation of GDF15 in IR-PCOS patients appears to be a compensatory response to IR. It may influence the progression of IR in women with PCOS by reducing oxidative stress, attenuating inflammatory responses, and modulating central nervous system pathways.

Nevertheless, our study possesses certain limitations: (1) The cross-sectional design exclusively enrolled women with PCOS, omitting an assessment of GDF15 levels across three distinct groups—IR-PCOS, non-IR-PCOS, and healthy controls. Consequently, we cannot ascertain whether PCOS and IR collectively contribute to elevated GDF15 levels. (2) While we incorporated multiple risk factors for IR in women with PCOS, the limited sample size prevented stratification for other potential risk factors. (3) This study remains confined to a clinical perspective, lacking an in-depth mechanistic discussion. (4) Existing literature suggests that metformin can elevate GDF15 levels. Although we excluded patients who had consumed biguanides within the past month, the potential influence of other medications, both biguanides and non-biguanides, on our findings remains unaddressed. Thus, longitudinal follow-up of PCOS patients is essential to acquire more comprehensive data in future studies.

Conclusion

In short, elevated circulating GDF15 levels correlate with the development of IR in women with PCOS. High GDF15 levels may serve as predictors for a heightened risk of IR within the PCOS population. Should future research validate this association and elucidate the underlying mechanisms between GDF15 and IR in women with PCOS, it could empower clinicians to identify IR tendencies in women of childbearing age with PCOS, adjust medication regimens judiciously, and significantly mitigate the complexity of complications in PCOS patients, thereby enhancing global women’s reproductive health.