Background

Polycystic ovary syndrome (PCOS) is a common female endocrinology disorder with a prevalence between 5 and 26% depending on the population and diagnostic criteria applied [1, 2]. Three criteria were utilized for the diagnosis of PCOS. The National Institutes of Health (NIH) criteria introduced only oligo-anovulation and hyperandrogenism in PCOS diagnosis [3]. Another criterion is Androgen Excess and PCOS Society (AE-PCOS) criteria, which hyperandrogenism is central in diagnosis with combination with oligo-anovulation and/or PCO [4]. Finally, Rotterdam criteria define PCOS by the presence of at least two out of three criteria: oligo- or anovulation, clinical and/or biochemical hyperandrogenism, and polycystic ovaries (PCO) based on Rotterdam criteria [5]. PCOS remains a syndrome which means that no single diagnostic criterion (such as PCO and hyperandrogenism) is enough for clinical diagnosis [5]. The recent international PCOS guideline suggested the Rotterdam criteria for diagnosing PCOS [6].

Recent meta-analysis reports indicate the prevalence of PCOS 11.04% (Rotterdam), 3.39% (NIH), and 8.03% (AE-PCOS) [7]. According to another meta-analysis among different ethnicity, the prevalence of 5.6% was in Chinese women (Rotterdam), for Caucasians (NIH) 5.5%, Middle Eastern (NIH) 6.1%, (Rotterdam) 16.0%, (AE-PCOS) 12.6%, and Black women (NIH) 6.1% [8].

According to the Rotterdam criteria, a prevalence of 6% in South India was reported in 2017 [9]. Furthermore, the prevalence of 35.3% (Rotterdam), 34.3% (NIH), and 28.9% (AE-PCOS) in Kashmir valley has been reported [10].

PCOS women have multiple risk factors such as obesity, central obesity, and insulin resistance, which increase the risk of type II diabetes (3–7 times) according to the Rotterdam criteria [5]. Some studies reported that women with PCOS are exposed to other long-term risk factors such as cardiovascular disease [11,12,13] and endometrial cancer [14]. Metabolic syndrome, a combination of central obesity, dyslipidemia, hypertension, elevated fasting glucose levels, and insulin resistance, is observed in 30–40% of women with PCOS [15].

South Asians showed higher metabolic consequences of PCOS arising from a younger age with a greater degree of insulin resistance related to central adiposity, in contrast to White Caucasians with this condition [16]. Obesity is an important factor in the etiology of PCOS [17, 18]. Since both obesity and PCOS are highly hereditary, there is a strong possibility for a shared genetic predisposition leading to their co-occurrence [19].

In 2007, a single nucleotide polymorphism (SNP) of fat mass and obesity-associated (FTO) gene was reported by a genome-wide association study which was associated with increased body mass index (BMI) in both children and adults [20, 21]. The FTO gene is located on chromosome 16q12.2 [20] and encodes alpha-ketoglutarate-dependent dioxygenase protein expressed in a wide range of tissues, including adipose tissue and specific parts of the brain. However, the exact function of this protein is unclear. Most studies have shown that FTO is a regulator of fat mass, energy homeostasis, and adipogenesis [22]. Genetic variants of FTO and altered expression have been found to be linked with many obesity-related chronic diseases such as type-II diabetes [23] and cancer [24].

The FTO variant (rs9939609) is the most common SNP, consistently associated with obesity and type II diabetes in multiple populations [25]. The rs9939609 polymorphism is located in intron 1 of the FTO, which has two alleles, A and T. A allele is reported as a risk allele for obesity and type II diabetes [26].

An initial report on a large population in the UK revealed a strong association of SNP rs9939609 with obesity [20]. Consequently, several studies have been undertaken to investigate the relationship between the FTO gene and the risk of obesity. A recent meta-analysis based on eight Indian studies demonstrated that FTO variant rs9939609 increased the risk of obesity by 1.15 times [27]. Obesity is a frequent symptom among PCOS patients. It is reported that over 50% of PCOS subjects are overweight/obese [28]. The frequent co-occurrence of PCOS and obesity raises the possibility of shared genetic predispositions to both traits, and therefore, variants shown to influence BMI and risk factors for obesity are also possible candidates for PCOS. Therefore, the FTO gene may have a role in PCOS pathogenesis via BMI and/or obesity [29]. In European women with PCOS, rs9939609 SNP in the FTO gene modifies anthropometric measurements, as well as insulin sensitivity [30].

Thus far, researches have shown contradictory findings. Some studies confirmed an association with PCOS [19, 31,32,33], whereas others reported no such association [15, 34, 35]. These contradictory results may be due to different factors such as ethnicity, diagnosis criteria, or whether BMI was adjusted in each study. A meta-analysis in 2014 has suggested an effect of ethnicity on the association between FTO polymorphism and PCOS, which was significantly associated with East Asians but not Caucasians [36]. Therefore, there is a need for further studies to investigate the association of FTO gene polymorphism with PCOS in other populations, as there is no such report from the Indian population to the best of our knowledge.

This study aimed to investigate the impact of common polymorphism (rs9939609) of FTO gene on South Indian women with PCOS and healthy controls and to assess the association of this polymorphism with metabolic and endocrine parameters.

Methods

Participants

The study was conducted on 100 South Indian PCOS women and 70 controls in the reproductive age range of 18–45 years in December 2017 to October 2019 as a case–control association study. The study was approved by the Institutional Human Ethical Committee, University of Mysore, Manasagangothri, Mysore (No. 151/PhD/2017-18) , and written informed consent was obtained from all the subjects. PCOS samples were collected from private hospitals in Mysuru, Karnataka, India, namely Santasa IVF Centre, and Ashwini hospital.

According to the recent study [9], we assumed a PCOS prevalence of 6% in the South India. The following formula was applied to calculate the sample size in the current study [37]. Based on the following formula, the minimum sample size with a 95% confidence level should be 87 individuals, but we increase the sample size to 100 for the PCOS group to achieve more precise results.

$$n = Z2P \, \left( {1 - P} \right)/d2$$

n: sample size; Z: statistic corresponding to the level of confidence; P: expected prevalence; d: precision (corresponding to impact size).

PCOS samples were collected from private hospitals of xxxx, xxxx, India, namely xxxx, and xxxx hospital. They were diagnosed as PCOS according to the consensus of Rotterdam criteria satisfying at least two of the three criteria—chronic anovulation (oligo‑ and/or anovulation) was present, clinical (hirsutism was determined by Ferriman-Gallwey score ≥ 8) [38,39,40] and/or biochemical hyperandrogenism (serum testosterone more than 0.73 ng/ml or 1.82 nmol/L), and PCO (PCO is characterized by ≥ 12 follicles with the measurement of 2–9 mm in diameter, or ovarian volume should be more than 10 ml in at least one ovary) [5]. Women with Cushing's syndrome, adrenal 21-hydroxylase deficiency, androgen-secreting tumors, thyroid dysfunction, hyperprolactinemia, and other pituitary, adrenal or ovarian diseases were excluded. General information, reproductive information (menstruation, fertility), clinical features of PCOS, drug history, family history of PCOS, diabetes, heart disease, hypertension, and daily diet and physical activity were collected from each patient.

The control group was selected from healthy volunteer women aged between 18 and 45 years with normal ovulating cycles as well as normal ultrasound appearance of the ovaries, and without any history of thyroid disorder, dyslipidemia, insulin resistance, hypertension, diabetes, hyperandrogenism. Investigations were conducted by collecting each person's health history and documents, filling the genetic registry, and confirming the information by clinical and hormonal analysis.

BMI as a weight (kg) divided by the square of height (m) and waist circumference (WC) (at the midpoint between the lowest rib and iliac crest) were measured. Obese and overweight were defined according to the World Health Organization [41]. The waist-to-hip ratio (WHR) was calculated by the waist circumference divided by hip circumference. Resting blood pressure (BP) was assessed.

Metabolic analysis

During the 2nd–5th day of the menstrual cycle or any day in amenorrheic patients, the blood samples were collected from each subject after a 10–12 h overnight fast.

Plasma glucose levels were determined by the glucose oxidase–peroxidase method (ARKRAY Inc., Japan). Cholesterol, triglycerides, high-density lipoprotein (HDL) were assessed by an enzymatic method (Meril diagnostics, India), and low-density lipoprotein (LDL) was calculated by Friedewald equation. Fasting insulin levels were estimated by ELISA (CALBIOTECH, USA). Insulin resistance was assessed by homeostasis model assessment (HOMA-IR) Index = (fasting plasma insulin [in microunits per milliliter] × fasting plasma glucose [in milligrams per deciliter])/405 [42] and insulin sensitivity was assessed by the quantitative insulin sensitivity check index (QUICKI) = 1/log fasting insulin (in microunits per milliliter) + log fasting glucose (in milligrams per deciliter) [43]. Luteinizing hormone (LH) and follicle-stimulating hormone (FSH) were estimated by commercial kits (Meril kit, India), and testosterone levels were estimated with an enzyme-linked fluorescent immunoassay method using mini VIDAS instrument (the Compact Automated Immunoanalyzer) and BioMérieux testosterone kit (Biomerieux, France). Five samples with high and five samples with low testosterone concentration were analyzed four times. The intra-assay coefficient of variation was 4.6%.

SNP genotyping

Peripheral blood samples were obtained from all participants and transferred into anticoagulant tubes. Genomic DNA extraction was done using a QIAamp DNA mini kit (QIAGEN, Germany) according to the manufacturer's protocol. Polymerase chain reaction (PCR) was performed with the primers for FTO rs9939609 (forward: 5′-GCAAAATGGCAACACACACT-3′; reverse: 5′-ACCCCCAGTACTACATCTACC-3′) with an initial denaturation 95 °C, 5′, 35 cycles of amplification; denaturation (95 °C, 45′′), annealing(50 °C, 45′′), elongation (72 °C, 30′′), and extension (72 °C, 5′). The PCR products were purified by Qiagen Gel purification kit and sequenced using the ABI 3130xl genetic analyzer (Applied Biosystems, USA).

Statistical analysis

All values were expressed as mean ± standard deviation (SD). For the evaluation of the normality of data distribution, the Shapiro–Wilks test was used. Independent t test and one-way analysis of variance (ANOVA) were applied for comparing the differences between groups. Leven's test was used to test the homogeneity of variance, and to recognize any heterogeneity of variance, Games–Howell test was performed. Post hoc was executed for multiple comparisons. Analysis of covariance (ANCOVA) was applied to exclude the possible confounding effects of age and BMI.

For comparison of genotype and allele frequency between groups, Chi-square test was applied. We used the binary logistic regression (forward LR) method to identify the PCOS development risk factors. All biochemical and hormonal variables and FTO (rs9939609) gene were analyzed by logistic regression to identify risk factors for PCOS.

Additionally, logistic regression was performed to analyze the association of metabolic variables with FTO polymorphism. In the regression model, FTO polymorphism as a dependent variable was categorized into two groups: AA + AT [mutant allele] and TT [wild allele]. Variables which were found to have an individually significant interaction were retained in the model. A p value < 0.05 was considered statistically significant. All analyses were performed on SPSS software (SPSS, version 23.0; IBM). SNPSTATS program [44] was applied to calculate the Hardy–Weinberg equilibrium, genotype, and allele frequency differences.

Results

Metabolic and endocrine characteristics of PCOS and control subjects

One hundred PCOS patients and 70 control subjects participated in the study. Clinical and hormonal measurements are presented in Table 1. In the PCOS group, 54% were obese and overweight, while the percentage was only 34% in the control group.

Table 1 Comparison of metabolic and endocrine characteristic between PCOS and controls
Full size table

As expected, BMI and WC, LH, LH/FSH ratio, and testosterone levels were significantly higher in the PCOS group. Higher levels of other metabolic parameters such as BP ≥ 130/85 mm Hg, cholesterol, LDL, fasting glucose levels, insulin, HOMA-IR, QUICKI were observed, whereas there was a significant reduction in HDL (HDL < 50 mg/dL) when compared with the control group (Table 1).

Fasting glucose levels showed no significant differences between PCOS and control groups after age and BMI were adjusted by ANCOVA (Table 1).

Increased WC, diastolic blood pressure (DBP), LH/FSH ratio, and testosterone levels are significantly associated with PCOS

All variables and FTO rs9939609 polymorphism (categorized to TT and AA + AT) were applied for the logistic regression analysis to determine PCOS risk factors.

Logistic regression analysis showed the association of increased WC, DBP, LH/FSH ratio, and testosterone levels with PCOS. The final model demonstrated the significant associations of increased WC [OR 1.05 (1–1.09)], DBP [OR 1.26 (1.14–1.38)], LH/FSH ratio [OR 7.66 (3.71–15.77)], and testosterone levels [OR 8.76 (3.65–21.01)] with PCOS. Increased LH/FSH ratio and testosterone levels were more significantly associated than increased WC and DBP (Data not shown) (Additional file 1: Table S1).

Allele and genotype frequency

The genotype distributions of rs9939609 were consistent in the Hardy–Weinberg equilibrium across case–control groups. The genotype frequencies were as follows: 28% TT, 57% TA, and 15% AA for PCOS group and 37.1% TT, 51.4% TA, and 11.4% AA for control group.

Association between the FTO polymorphism and PCOS susceptibility was analyzed under additive, dominant, recessive models. The results in Table 2 demonstrated that the genotypes of rs9939609 were not associated statistically with the increased risk of PCOS (under additive, dominant, and recessive models). The allele frequency comparison is presented in Table 2, and no association was observed with PCOS [OR 1.30 (0.83–2.02)].

Table 2 Allele and genotype frequency analysis
Full size table

The samples were further divided into obese/overweight (BMI ≥ 25) and non-obese groups (BMI < 25) to observe the effect of BMI in the association of rs9939609 with both groups. In PCOS and controls, 54% (54/100) and 31% (24/70) were obese/overweight, respectively. The polymorphism was categorized as AA + AT and TT. The Chi-square results showed no significant association (obese/overweight group [OR 2.05(0.66–6.40)]; non-obese group [OR 1(0.43–2.29)]). Similarly, allelic association in obese/overweight and non-obese groups with PCOS was not significant (obese/overweight group [OR 1.03 (0.52–2.04)]; non-obese group [OR 1.21 (0.65–2.26)]) (Data not shown) (Additional file 1: Table S2).

Association of FTO polymorphism with fasting glucose levels, serum testosterone, and BMI in PCOS subjects

Genetic analysis of inheritance models did not show any significant association between the genotypes of rs9939609 and increased risk of PCOS. Furthermore, a genotype–phenotype association analysis was conducted on the PCOS group (Table 3). Most of the metabolic and endocrine characteristics were similar between genotypes. But individuals with at least one risk allele, AT and AA genotypes, had significantly higher BMI, fasting glucose levels, and testosterone levels compared with individuals carrying the wild TT genotype for SNP rs9939609 (Table 3).

Table 3 Metabolic and endocrine characteristics of the PCOS group according to different genotypes for SNP rs9939609
Full size table

Further analysis by logistic regression showed the association of fasting glucose levels and testosterone levels with FTO polymorphism in PCOS subjects (Table 4). Additionally, we applied logistic regression to analyze the association of variables with FTO variant in control subjects. The associated variable in the final model was LDL with OR: 1.01 and P value: 0.028. Although the p value was significant, the odds ratio showed a slight difference with 1, indicating that LDL is associated with the FTO variant but not strong in the control group.

Table 4 Association of FTO rs9939609 polymorphism (AA + AT vs. TT) with different variables in PCOS
Full size table

Discussion

The present study suggests that fasting glucose levels and testosterone levels are associated with FTO variant rs9939609 in South Indian women with PCOS for the first time to the best of our knowledge. However, no association was observed between this polymorphism and the risk of PCOS. Similar PCOS studies across different populations did not show consistent results. Some studies demonstrated that FTO variant rs9939609 is significantly associated with PCOS [19, 31,32,33, 45, 46], while other studies found no association between that variant with PCOS but an association with obesity, hormonal, and metabolic traits in PCOS patients [15, 47,48,49,50].

The meta-analysis in 2014 suggests that there might be an association between FTO polymorphisms and PCOS risk, specifically in an East Asian population, although the association was dependent on adiposity [36]. Two other meta-analysis identified a significant association between FTO gene polymorphism rs9939609 and the risk of PCOS [29, 51], and a recent meta-analysis in 2020 has reported a significant association between FTO rs9939609 polymorphism with PCOS under different inheritance models [52].

Our study found no association between FTO variant rs9939609 and PCOS by genetic inheritance and logistic regression analysis even after subgrouping according to BMI. However, A risk allele was observed as more prevalent in the PCOS population than control but not statistically significant (Table 2).

Moreover, WC, DBP, LH/FSH ratio, and testosterone were detected in the logistic regression analysis as PCOS risk factors, especially LH/FSH ratio and testosterone. The FTO variant was not included in the model.

PCOS patients show several metabolic consequences and cardiovascular risk factors. Increased blood pressure is associated with obesity, glucose intolerance, and insulin‐resistant. The logistic regression analysis showed DBP as a PCOS risk factor. It may be due to DBP being a more effective cardiovascular risk factor than systolic blood pressure (SBP) until age 50 [53]. Our findings verified the study, which reports that the LH/FSH ratio can be used to diagnose PCOS [54]. Hyperandrogenemia is a common feature of PCOS [55], and obesity, especially central obesity, is one of the frequent consequences of PCOS.

Furthermore, we found the association of high testosterone levels with the FTO variant rs9939609 (Table 4). However, different studies have shown contradictory results of the association between the FTO gene and hyperandrogenemia. In one investigation on UK PCOS women, no association was found between FTO genotype and androgen level [19]. Another study on the Polish population showed no relationship between testosterone, sex hormone-binding globulin, and free androgen index with different genotypes [30]. However, some studies have reported an association of hyperandrogenemia with FTO variant 9939609 [33, 48]. The association may be mediated via a mechanism that leads to an increase in adiposity [56].

One study was conducted on European ethnicity women with PCOS, and another study on Sothern Brazilian women with PCOS demonstrated the association of fasting glucose levels with FTO [15, 49]. Nevertheless, few studies failed to report any relationship between glucose level and FTO [47, 57]. Furthermore, some studies showed no relationship between FTO and PCOS after adjusting with BMI [19, 34]; however, there was an association between risk allele A of FTO rs9939609 and increased 2-h glucose levels after correction for BMI. A similar higher 2-h glucose levels was observed in association with rs9939609 in women with PCOS in the absence of BMI adjustment [48]. Our study demonstrated an association between fasting glucose levels and FTO rs9939609 in South Indian PCOS women even after BMI adjustment (Tables 3, 4).

Increased BMI in subjects with AA and AT genotypes of variant rs9939609 of FTO (Table 3) is in agreement with previous studies, which report the relationship of the A allele with increased BMI [19, 30, 47, 48, 50].

So far, studies were not successful in identifying the mechanism connecting the FTO gene directly to PCOS. Most investigations in which an association was reported between FTO and PCOS susceptibility also detected an increase in adiposity, which might be related to a genetic interaction of the FTO gene with other susceptibility genes, owing to the polygenic background of PCOS [30]. It is also essential to note that polymorphisms in FTO gene intron 1 are in linkage disequilibrium, posing difficulty in associating the best single SNP within this region with PCOS [15].

In the current study, we found the association between rs9939609 and hormonal and metabolic traits (testosterone levels, fasting glucose levels, and BMI) in the PCOS group similar to several previous studies in different populations but no association through inheritance genetic models [15, 47,48,49,50]. It can be due that race and region are two possible influences on the FTO association studies in PCOS subjects towing to ethnic genetic diversities. Furthermore, the Asian studies are commonly limited to East Asia, especially Chinese [31, 45, 50] and Korean population [47], while the participants in the current were exclusively South Indians. Also within Korean and Chinese populations FTO rs9939609 variant was associated with PCOS through obesity and metabolic disorders [45, 47].

Limitations

Some limitations in this study should be noted. Obesity and metabolic abnormalities are more prevalent in PCOS-affected older women than PCOS-affected younger women [58]. In the current study, PCOS and control age groups are 25.42 ± 2.91 and 23.08 ± 3.25, respectively. Since FTO intronic variants are associated with obesity and metabolic consequences, further investigation in various age groups, especially older ages in certain populations, is desired to generalize the result of the current study. Moreover, this study was limited to 100 samples of PCOS and 70 controls in the South of India. However, our study indicates the significant association of hyperandrogenemia, fasting glucose levels, and BMI with FTO variant rs9939609. Therefore, further studies with larger sample sizes are required to generalize the results in other populations.

Conclusions

The present study was the first to investigate the FTO gene variant rs9939609 association with PCOS susceptibility in South Indian women. Our findings indicated that FTO SNP rs9939609 was not related to PCOS, but suggests a significant association of FTO polymorphism rs9939609 with hyperandrogenemia, fasting glucose levels, and BMI in South Indian women with PCOS.