Research on Fibroblast Growth Factor 21 and Its Relationship to Patients with Polycystic Ovary Syndrome and Obesity

Introduction

Polycystic ovary syndrome (PCOS) is the most common endocrine disorder affecting women of reproductive age.^1,2 Hormonal imbalances, including elevated luteinizing hormone (LH) secretion, altered gonadotropin-releasing hormone (GnRH) pulsatility, hyperinsulinemia, and increased leptin levels, are frequently implicated in the pathophysiology of PCOS.^1,2

Its diagnosis is based on clinical and/or biochemical androgen excess associated with ovulatory oligo-amenorrhea and polycystic ovarian morphology.^1,3,4 It is closely linked to ectopic fat accumulation and metabolically associated fatty liver disease, which are often associated with insulin resistance and reduced energy expenditure. Consequently, women with PCOS are at increased risk of dyslipidemia and central obesity.^1,4

Organokines are circulating signaling molecules from the liver (hepatokines), adipose tissue (adipokines), and osteocytes.^1,4 Dysregulation of these molecules often leads to alterations in hepatic function and metabolic imbalances, particularly in PCOS patients.^1,4–6

Fibroblast growth factor 21 (FGF21), a hepatokine predominantly expressed in the liver, pancreas, and white adipose tissue, is critical in regulating glucose and lipid metabolism, insulin sensitivity, energy balance, and adipose tissue function.^7,8 Studies have highlighted its potential as an insulin-sensitizing and lipolysis-suppressing molecule, protecting against systemic metabolic dysregulation, particularly in patients with fatty liver.^1,2,4 Elevated FGF21 levels have been associated with obesity, type II diabetes mellitus, and hepatic steatosis.^4,7 Notably, FGF21 levels correlate proportionally with body mass index (BMI) and are closely linked to visceral obesity.^7,9

While the findings above suggest that FGF21 may serve as a reliable biomarker for PCOS, particularly in patients with obesity, several studies have reported conflicting results regarding FGF21 levels in women with PCOS.^2,3,10,11 These discrepancies may be attributed to differences in study populations, BMI distribution, or underlying metabolic phenotypes. In addition to the potential interplay between FGF21 levels, nutritional status, and metabolic disturbances in women with PCOS, the association between FGF21 and hormonal imbalances remains a subject of speculation.^7,10,11 Given that LH and androgens are central to the hormonal diagnostic criteria of PCOS, such possible associations suggest that FGF21 plays a role in the hormonal pathophysiology of the disorder, warranting further investigation.^1,2,11

In this context, this study aims to clarify this discrepancy by evaluating FGF21 levels in women with PCOS compared to healthy control subjects and investigating the relationship between FGF21 and obesity.

Materials and Methods

Study Design

This study was designed as a prospective, non-randomized, controlled study. The local ethics committee approved the study protocol (Approval Date: 03.10.2017, Approval Number: 2017–14/47). The study was conducted with the ethical considerations outlined in the Declaration of the Helsinki.

Population and Sample

The study population comprised consecutive female patients aged 18–55 years who presented to the outpatient clinics of the Department of Endocrinology and Metabolism, Yüksek Ihtisas Hospital, Bursa, Turkey, and were diagnosed with PCOS by satisfying at least two of the three parameters of the Rotterdam Consensus Criteria¹² between October 2017 and February 2018. Pregnant or lactating patients, patients with acute or chronic infection, malignancy, comorbidities, particularly diabetes mellitus, endocrine disorders, and renal failure, patients whose alanine aminotransferase and aspartate aminotransferase levels were 2.5 times higher than normal levels and who have received oral contraceptives, insulin-sensitizing drugs, and statins in the last three months were excluded from the study. The control group consisted of age-matched women who presented to the same outpatient clinics during the study period but did not meet the Rotterdam criteria for PCOS. These women had regular menstrual cycles (21–35 days) and no clinical or biochemical evidence of hyperandrogenism (acne, hirsutism, or alopecia).

Variables and Data Collection

Participants’ demographic characteristics were prospectively recorded. Anthropometric characteristics, including weight (kg), height (m), and waist circumference (cm), were measured and then recorded. Body mass index (BMI) values were calculated by dividing their weight (kg) by the square of their height (m2) and recorded. Waist circumferences (cm), ie, the diameter at the midpoint between the lower margin of the last palpable rib and the anterior superior iliac spine,¹³ were measured and recorded. Body fat percentages were measured using a Tanita Body Composition Analyzer (Model TBF-300, Tanita Corporation, Itabashi-ku, Tokyo, Japan) after ≥ 8 hours of fasting with sufficient fluid intake¹⁴ and recorded.

Laboratory Measurements

Blood samples were collected from the participants after an overnight fast in the early follicular phase on the third day of the menstrual cycle. These samples were tested for hemoglobin, fasting blood glucose and insulin, total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), LH, follicle-stimulating hormone (FSH), estradiol (E2), total testosterone, DHEA-S, androstenedione, and 17-hydroxyprogesterone levels. Low-density lipoprotein cholesterol (LDL-C) values were calculated using the following formula:¹⁵ TC-HDL-C- TG/5. Additionally, homeostasis model assessment insulin resistance (HOMA-IR) values were calculated using the following formula: fasting insulin (μIU/mL) x fasting blood glucose/405 2.

An enzyme-linked immunosorbent assay (ELISA) kit (Code: EKO994) (Human FGF21 PicoKine™ ELISA Kit, Sunred Biotechnology Company, Shanghai, China) was used to measure serum FGF21 levels. The standard provided with the kit, initially at a concentration of 960 ng/L, was diluted using the standard dilution buffer and then serially diluted to obtain concentrations of 480 ng/L, 240 ng/L, 120 ng/L, 60 ng/L, and zero ng/L. Before the analysis, serum samples stored at −80°C were thawed at room temperature. The prepared standards and samples were added to antibody-coated wells, followed by enzyme-conjugated antibodies to each well. The plates were sealed with a transparent cover and incubated at 37°C for 1 hour. Following incubation, the contents of the wells were aspirated, and each well was washed five times with 0.35 mL of washing solution. After washing, Chromogen solutions A and B were added to each well. The plates were covered again and incubated at 37°C for 10 minutes. To terminate the enzymatic reaction, 100 µL of stop solution containing 0.5 M H₂SO₄ was added to each well. The optical density of each well was measured at 450 nm using an ELISA plate reader. According to the kit specifications, the assay range was 6–1800 ng/L with a 5.5 ng/L sensitivity.

Study Groups

The patient and control groups were further categorized based on BMI values:³

1. PCOS groups (Group PCOS, n=62)

a. Non-obese PCOS group (BMI <30 kg/m²) (n=36)

b. Obese PCOS group (BMI ≥30 kg/m²) (n=26)

2. Control Groups (Group Control, n=30)

a. Non-obese control group (BMI <30 kg/m²) (n=19)

b. Obese control group (BMI ≥30 kg/m²) (n=11)

Statistical Analysis

The study’s primary outcome was the difference in serum FGF21 levels between the groups, and the secondary outcome was the impact of obesity on FGF21 levels.

The predicted mean matching (PMM) method, which is used for variables with a missing data rate below 20%, was used for FGF21 (ng/mL) with a missing data rate of 11.96% due to the continuous nature of the data and the necessity to preserve sensitivity to outliers. PMM preserves data distribution by estimating missing values through closest matches to observed values, does not require a normality assumption, and is resistant to outliers. Five FGF21 (ng/mL) datasets were generated using PMM. A single complete dataset containing the averages of these five data sets was used for analyses. R-project version 4.4.2 (R: A Language and Environment for Statistical Computing, R Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2023, retrieved from https://www.R-project.org), and multiple imputation by chained equations (MICE)¹⁶ software packages were used for imputation of missing data.

The normal distribution characteristics of numerical variables were analyzed using the Shapiro–Wilk test and quantile-quantile (Q-Q) plots. Normally distributed continuous variables were expressed using mean ± standard deviation values, nonnormally distributed continuous variables were expressed as median with minimum and maximum values, and categorical variables were expressed as numbers (n) and percentage (%) values.

In comparing the differences in numerical variables between the two independent groups, independent samples t-test was used for numerical variables determined to conform to the normal distribution, including height, HDL-C, LDL-C, TC, fasting glucose, and FGF21 levels and Mann–Whitney U-test was used for numerical variables determined to not conform to the normal distribution, including age, weight, BMI, very low-density lipoprotein cholesterol (VLDL-C), TG, insulin, HOMA-IR, FSH, LH, E2, testosterone, DHEAS, and 17-OH progesterone levels. Specifically, serum FGF21 levels were compared between the Non-obese Control, Non-obese PCOS, Obese Control, and Obese PCOS groups using the Kruskal–Wallis test.

The Pearson’s chi-square test was used to compare BMI groups (≥30 kg/m² and <30 kg/m²). Two-way Analysis of variance (ANOVA) was employed to test the hypothesis that FGF21 levels may be affected by PCOS and obesity by assessing the primary effects of PCOS and BMI on FGF21 levels and possible interactions between them. The Levene’s test (p=0.145), used for assumption checks, confirmed the homogeneity of variances between the groups. In addition, the approximate linear alignment of points along the 45-degree line in the Q-Q plot demonstrated that the normality assumption was met.

Due to significant differences between the groups in age (p<0.001) and HOMA-IR (p=0.020), Analysis of covariance (ANCOVA) was performed to control the potential confounding effects of these variables on FGF21 levels. The linearity and homogeneity of regression slope assumptions in the ANCOVA model were assessed graphically and found satisfactory.

Pearson’s correlation coefficients were used to evaluate the relationship between FGF21 levels and normally distributed variables, including waist circumference and body fat percentage. Spearman’s rho correlation coefficients were used for non-normally distributed variables, including age, BMI, and HOMA-IR.

All statistical analyses were performed using the Jamovi project 2.3.28 (Jamovi, version 2.3.28.0, 2023, retrieved from https://www.jamovi.org) and JASP 0.19.2 (Jeffreys’ Amazing Statistics Program, version 0.19.2, 2024, retrieved from https://jasp-stats.org) software packages. Probability (p) statistics of ≤ 0.05 were deemed to indicate statistical significance.

Results

The distribution of participants’ demographic and anthropometric characteristics by the patient and control groups is shown in Table 1. Accordingly, there was a significant difference between the groups regarding age (p<0.001). The median age of the patient group was significantly lower than the control group (31.5 years vs 24.0 years). There was no significant difference between the groups in other demographic and anthropometric characteristics (p>0.05), including the rate of patients with obesity (BMI>30 kg/m²) (p=0.798).

Table 1 Demographic and Anthropometric Characteristics of Patients with PCOS and the Control Group

The distribution of participants’ laboratory characteristics by the patient and control groups is shown in Table 2. Accordingly, there were significant differences between the groups in metabolic markers (p<0.05). Insulin (11.7 vs 9.1 U/L, p=0.006) and HOMA-IR (2.5 vs 2.0, p=0.020) values were significantly higher in the patient group than in the control group, while TC (173.5 vs 190.8 mg/dL, p=0.031) and LDL-C (99.2 vs 120.1 mg/dL, p=0.003) levels were significantly lower. Hyperandrogenism markers, including testosterone (54.3 vs 38.1 ng/dL, p=0.001), DHEAS (245.5 vs 154.0 ug/dL, p=0.001), and 17-OH progesterone (0.6 vs 0.3 ng/mL, p=0.001) levels were significantly higher in the patient group than in the control group. On the other hand, there was no significant difference between the patient and control groups in FGF21 level (150.1±48.2 vs 154.7±47.4 ng/mL, respectively, p=0.669).

Table 2 Laboratory Characteristics and Biochemical Parameters of Patients with PCOS and the Control Group

When age was added to ANCOVA as a covariate, it was determined that there was no significant difference between the patient and control groups in FGF21 level (p=0.298) and that age had a marginal but nonsignificant effect on FGF21 levels (p=0.100). Similarly, when HOMA-IR was added to ANCOVA as a covariate, it was determined that there was no significant difference between the patient and control groups in FGF21 level (p=0.298) and that HOMA-IR did not have any effect on FGF21 levels (p=0.649). These results suggest that FGF21 levels were similar between the groups, independent of age and insulin resistance (Figure 1).

Figure 1 Box plot illustrating the distribution of FGF21 levels (ng/mL) among groups categorized by patient groups (PCOS and control) and obesity status (Non-Obese and Obese), with individual data points overlaid for detailed visualization.

Two-way ANOVA revealed no significant difference in FGF21 levels between the groups independent of BMI (p=0.852). On the other hand, it was found that BMI had a marginal, yet not significant, effect on FGF21 levels independent of PCOS status (p=0.078). There was no significant relationship between the PCOS status and BMI (p=0.526) (Table 3).

Table 3 Distribution of FGF21 Levels According to BMI Groups in Patients with PCOS and the Control Group

FGF21 levels showed no significant differences across the four study groups, based on obesity and PCOS (p=0.284) (Table 4).

Table 4 Comparison of FGF21 Levels-Based Obesity and PCOS

The results of the correlation analyses are shown in Table 5. Accordingly, FGF21 level had a weak negative correlation with age (r=−0.178, p=0.166), BMI (r=−0.178, p=0.165), waist circumference (r=−0.128, p=0.323), body fat percentage (r=−0.102, p=0.429), and HOMA-IR (r=−0.173, p=0.179) (p>0.05).

Table 5 Correlation Analysis Results Between FGF21 and Other Variables

Table 6 presents the results of the correlation analyses between FGF21 levels and sex hormones in different patient subgroups. Analysis of the entire study population revealed a statistically significant positive correlation between FGF21 and DHEAS (rho=0.282, p=0.007). This relationship remained robust and even strengthened when exclusively analyzing the PCOS patient cohort (rho=0.364, p=0.004). However, this correlation was not maintained in the obesity-only subgroup (rho=0.152, p=0.369).

Table 6 Spearman Correlation Coefficients Between FGF21 and Hormonal Parameters

No statistically significant correlations were identified between FGF21 and other hormonal parameters, including FSH, LH, estradiol, testosterone, and 17-OH progesterone (p>0.05).

Discussion

The study findings showed that FGF21 was not a significant factor in developing PCOS and that age, BMI, waist circumference, body fat percentage, and HOMA-IR were not correlated with FGF21 levels.

An experimental study demonstrated that adolescent prenatally androgenized sheep exhibited decreased FGF21 signaling in subcutaneous adipose tissues and reduced hepatic expression and circulating concentrations of FGF21.⁸ However, the study found that circulating FGF21 levels, which were significantly low during adolescence and early adulthood, returned to normal in adulthood by 30 months.

Although the study’s authors suggested that prenatal androgen overexposure is a realistic clinical model for PCOS, we hypothesized that complex interactions exist between hyperandrogenism and its metabolic consequences on FGF21 levels. There are conflicting findings regarding FGF21 levels in patients with PCOS.^{1–4,7,10,17,18} Garcia-Beltran et al¹ demonstrated significantly higher levels of FGF21 in adolescents with PCOS than in control girls. Several other studies have reported similar findings.^3,11,18 On the other hand, in their study, they found that nonalcoholic fatty liver disease was significantly associated with FGF21 levels in PCOS patients. They stated that a one pg/mL increase in FGF21 concentration increased the likelihood of fatty liver disease by 1%. Therefore, FGF21 may be a potential biomarker for hepatic disease in adolescents with PCOS. Giannouli et al found no significant difference in FGF21 concentrations between PCOS patients and healthy control subjects,⁴ which aligns with our findings and several other studies.^2,10,17

These inconsistencies highlighted the heterogeneity of PCOS and the complex metabolic and hepatic factors influencing FGF21 levels. The lack of significant differences in FGF21 levels between PCOS patients and healthy control subjects may be attributable to variations in the characteristics of the studies’ populations, such as the degree of hyperandrogenism, hepatic steatosis, and insulin resistance, as well as methodological differences, including the timing of sample collection relative to the menstrual cycle.^3,19 Our study’s absence of fatty liver assessment contributed to the lack of significant associations, as FGF21 is closely linked to hepatic lipid metabolism, which is an important limitation in interpreting the role of FGF21.

The impact of obesity on FGF21 levels in patients with and without PCOS is still a matter of debate. Several studies suggested that the increase in FGF21 levels is unrelated to PCOS and is proportional to BMI.⁷ These studies also demonstrated positive correlations between FGF21 levels and body fat mass and its percentage, waist circumference, and HOMA-IR. In parallel, the relationship between obesity and FGF21 was explained by the increase in FGF21 expression in the visceral fat of obese individuals.^7,20 Along these lines, Gorar et al¹¹ suggested that increased FGF21 levels in patients with PCOS were associated with increased BMI values. In contrast, we, as Sahin et al¹⁰ did not find a significant correlation between these variables. Sahin et al concluded that FGF21 was not a helpful marker for metabolic abnormalities, including insulin resistance, dyslipidemia, obesity, and hypertension in PCOS patients.¹⁰

The association between FGF21 and hormonal imbalances remains a subject of speculation.^7,10,11 Given that LH and androgens are central to the hormonal diagnostic criteria of PCOS, such possible associations suggest that FGF21 plays a role in the hormonal pathophysiology of the disorder, warranting further investigation.^1,2,11 Olszanecka-Glinianowicz et al⁷ no association between elevated FGF21 levels and sex hormone disturbances in women with PCOS. Conversely, other studies have demonstrated positive correlations between FGF21 and LH or testosterone levels,¹¹ while a negative correlation between FGF21 and DHEA-S has also been observed.¹⁰ Given that LH and testosterone are central to the hormonal diagnostic criteria of PCOS, these associations suggest that FGF21 may play a role in the hormonal pathophysiology of the disorder, warranting further investigation.^1,2,11 The correlation analysis in this study showed a statistically significant positive correlation between FGF21 and DHEAS in the entire study population, contrary to Sahin’s study.¹⁰ As an interesting finding, the degree of this correlation was relatively increased in all PCOS patients. However, lack of significance between FGF21 and DHEAS might indicate that the observed relationship may be linked explicitly to PCOS pathophysiology rather than obesity status alone. The positive correlation between FGF21 and DHEAS specifically in PCOS patients represents a novel finding within our study cohort that warrants consideration in the context of PCOS’s complex endocrine and metabolic interactions. When stratifying by obesity status, the absence of significant correlations between FGF21 and hormonal parameters in the obese subgroup suggests differential regulatory mechanisms may be at play depending on the underlying pathophysiological condition.

The inconsistency between our findings and previous studies may reflect sample size constraints and differences in study populations. Specifically, the small number of participants in each subgroup (obese/non-obese PCOS and control groups) may have limited our statistical power to detect subtle but potentially meaningful differences. Furthermore, the unequal group sizes may have affected the robustness of intergroup comparisons. These conflicting findings may be attributed to the complex interplay between FGF21 and metabolic processes. Therefore, further studies are needed to elucidate the underlying pathophysiological mechanisms of PCOS. Future research should also consider stratification by liver fat content and other metabolic comorbidities to more accurately isolate the effects of FGF21.

Limitations of the Study

This study has several limitations worth mentioning. First, the relatively small sample size may have limited the generalizability of the findings. The limited number of participants in each may have exceptionally constrained the detection of statistically significant differences in FGF21 levels between PCOS phenotypes and control groups. Therefore, more extensive cohort studies are needed to validate the findings. Secondly, considering the study’s design and methodological variations, such as differences in diagnostic criteria for PCOS, sample size, and the assays used for FGF21 measurement, longitudinal studies may be more helpful in investigating the temporal relationships between FGF21 and PCOS. Thirdly, the potential impact of confounding factors, such as dietary habits, physical activity levels, and genetic predispositions, was not comprehensively evaluated. Lastly, the lack of advanced imaging techniques or liver biopsies to assess hepatic steatosis and fibrosis may have limited the ability to explore the role of FGF21 in hepatic manifestations of PCOS. Since FGF21 is known to increase in response to hepatic stress and steatosis, this omission may have obscured important underlying relationships. Addressing these limitations in future studies will provide a more comprehensive understanding of the role of FGF21 in PCOS.

Conclusions

In conclusion, FGF21 does not appear to play a significant role in the pathophysiology of PCOS or its associated metabolic abnormalities due to the lack of significant differences between women with PCOS and healthy controls, regardless of their obesity status. We found no significant correlation between FGF21 levels and key metabolic parameters, including age, BMI, waist circumference, body fat percentage, or HOMA-IR. However, the interpretation of these findings is limited by the small sample size, absence of hepatic steatosis assessment, and potential residual confounding. Additionally, the absence of data on hepatic steatosis—despite FGF21 being a recognized biomarker for fatty liver—may have led to an underestimation of its role in metabolically active PCOS phenotypes, particularly in obese patients So, these findings suggest that FGF21 may not serve as a distinct biomarker for PCOS independent of metabolic factors such as insulin resistance and body composition.

This study’s findings will likely contribute to the growing evidence on FGF21 and its role in metabolic disorders. Future studies with a longitudinal design that includes more extensive and diverse populations and comprehensively evaluates metabolic and hormonal factors are needed to clarify the complex interactions between FGF21 and hyperandrogenism, obesity, and insulin resistance in PCOS.