Comparison of clinical, metabolic, and hormonal parameters in lean vs. obese women with polycystic ovary syndrome: a single- center study from Bangladesh

A bi-directional relationship exists between obesity and polycystic ovary syndrome (PCOS). Although most of the patients with PCOS are obese, patients with normal body mass index (BMI) still account for a certain proportion of women with PCOS; such non-obese patients with PCOS have lesser degrees of insulin resistance and fewer metabolic risk factors. This study was conducted to compare the clinical, metabolic, and endocrine parameters of the two categories of women with PCOS: leanand obese PCOS. We evaluated 523 women with PCOS attending the Endocrinology outpatient department of a tertiary hospital in Bangladesh. The women with PCOS having BMI <23 kg/m were categorized as lean PCOS and BMI ≥23 kg/m as obese PCOS. Out of 523 women with PCOS studied, the frequencies of lean and obese PCOS were 23.3% and 76.7%, respectively. Age, systolic blood pressure (BP), diastolic BP, fasting plasma glucose, plasma glucose 2-hours after oral glucose tolerance test, serum triglyceride, total cholesterol, low-density lipoprotein cholesterol, total testosterone, and thyroid-stimulating hormone were higher in the obese PCOS group. Serum prolactin was higher in the lean PCOS group. Modified FerrimanGallwey score and high-density lipoprotein cholesterol were similar in the two groups. Acanthosis nigricans (89.3% vs. 45.9%), prediabetes (24.4% vs. 13.1%), diabetes (6.7% vs. 1.6%), prehypertension (22.2% vs. 9.8%), hypertension (8.7% vs. 0.8%), metabolic syndrome (61.3% vs. 14.8%), and biochemical hyperandrogenism (27.7% vs. 11.5%) were more frequent in the obese group than the lean group. The obese PCOS group had an 8.35-fold higher risk of having metabolic syndrome than the lean group. The two groups had similar frequencies of menstrual irregularity, hirsutism, a family member with type 2 diabetes, and dyslipidemia. Though metabolic abnormalities are more frequently observed when obesity is associated with PCOS, lean women with PCOS also have adverse metabolic consequences. Screening for metabolic abnormalities should be considered in all patients with PCOS irrespective of BMI category.


Introduction
Polycystic ovary syndrome (PCOS) is the commonest endocrine disorder in women of reproductive age. The prevalence of PCOS varies from country to country, which is as low as 6-10% in the western world and as high as 22.5% in the Indian subcontinent (1,2) . This heterogeneous androgen-excess disorder presents with different degrees of reproductive and metabolic dysfunctions. PCOS is associated with insulin resistance (IR) and metabolic syndrome (1) . There is a bidirectional relationship between obesity and PCOS; both exacerbate each other in a neverending cyclical manner (3) . Although 50-80% of patients with PCOS are obese, normalweight patients still account for a significant proportion of the PCOS population (4,5) . In clinical practice, patients with PCOS are classified into two broad categories based on their body mass index (BMI), obese PCOS with BMI ≥23 kg/m 2 , and lean PCOS with BMI <23 kg/m 2 (6,7) . Lean patients with PCOS have lesser degrees of insulin resistance and fewer metabolic risk factors, which suggests that the pathogenesis of this group may be different from that of obese PCOS patients (4,5) . In addition to the metabolic features, differences in the reproductive and endocrine parameters between the two groups are also observed (6)(7)(8)(9)(10)(11) .
Data highlighting the reproductive, endocrine, and metabolic characteristics of lean and obese Bangladeshi women with PCOS are scarce. This study was conducted to compare the features of the two categories of women with PCOS in our setting.

Subjects and methods
This cross-sectional study was conducted among newly diagnosed patients with PCOS attending the Endocrinology outpatient department of Mymensingh Medical College Hospital, Mymensingh, Bangladesh, from January 2017 to December 2019. The study protocol was approved by the institutional review board of Mymensingh Medical College. In adults, the diagnosis of PCOS was made by using the year 2003, revised Rotterdam criteria (12) . The diagnosis of PCOS in adolescent girls was made based on the presence of clinical and/or biochemical evidence of hyperandrogenism (after exclusion of other pathologies) in the presence of persistent menstrual irregularities (13) . Those receiving drug treatment for diabetes, hypertension (HTN), dyslipidemia, and obesity, on hormonal contraceptives, pregnant and lactating women, those having the pelvic inflammatory disease, malignancy, and patients with abnormal serum thyroidstimulating hormone (TSH) or very high serum prolactin (i.e., ≥2000 µIU/mL) were excluded. A semi-structured questionnairebased interview was conducted on a one-toone basis to collect detailed information on clinical presentation and family history. Height (to ±0.1 cm) was measured in all the individuals using wall-mounted stadiometers, and body weight (to ±0.1 Kg) measured using electronic calibrated scales; BMI was calculated from height and weight using the formula: height / weight 2 . These women with PCOS were divided into two groups: overweight or obese ones with BMI ≥23 kg/m 2 (based on the BMI categories applicable to the Asian Indians) (14) were included in the obese PCOS group and those with BMI <23 kg/m 2 (normal and underweight women) were included in the lean PCOS group (6)(7) . Waist circumference (WC) was measured (to ±0.5 cm) at the end of a gentle expiration midway between the lower rib margins and the iliac crests. Blood pressure(BP) was measured twice in each study subject by the auscultatory method, using a standard validated aneroid sphygmomanometer, after at least five minutes of rest; two separate readings were taken at an interval of minimum three minutes, and the average of the two readings was used. Hypertension (HTN) and prehypertension (pre-HTN) were defined according to the Joint National Committee VII criteria (15) . Hirsutism was assessed by the modified Ferriman-Gallwey (F-G) score; a score ≥8 was the cut-point for diagnosis of hirsutism (16) . Oral glucose tolerance test (OGTT) with a 75-gram glucose load was done in all after overnight fasting for at least 8 hours; fasting plasma glucose (FPG) and plasma glucose 2-hour after OGTT (PG 2H-OGTT) were measured by glucose oxidase method using fully automatic biochemistry analyzer (MINDRAY BS-380). Prediabetes and diabetes mellitus (DM) were diagnosed according to criteria described by the American Diabetes Association (17) . The lipid profile was measured in fasting states using the above analyzer. Dyslipidemia was defined according to cutoffs described in the Adult Treatment Panel (ATP) III guideline (18) . Metabolic syndrome was diagnosed using the modified National Cholesterol Education Program (NCEP) ATP III diagnostic criteria (19) . In adults, WC ≥80 cm was considered as the cutoff for abdominal obesity (19) . In the adolescents aged 10-16 years, WC ≥90th percentile (or adult cutoff if lower) for Indians was used to define abdominal obesity (20) . Transvaginal ultrasonography (USG) was preferred in the married patients, whereas transabdominal pelvic USG was done in unmarried ones. Serum thyroidstimulating hormone (TSH), total testosterone (TT), and prolactin were measured using radioimmunoassay (RIA) by automated hormone analyzer LB 2111 Multi Crystal Gamma Counter. A total of 523 cases were included in the final analysis.
Statistical analysis: Statistical analysis was done using Statistical Product and Service Solutions (SPSS) for Windows, version 23.0 software (SPSS Inc; Chicago, IL, USA). The categorical variables were presented as number (%), measurable variables with normal distribution were presented as mean±standard deviation (SD), and those not following normal distribution were presented as median (interquartile range, IQR). Student's t-test, Chi-square test, and Mann-Whitney U tests were performed as applicable for comparing the variables between different groups. p-value ≤0.05 was considered as statistically significant.
The comparison of clinical, metabolic, and hormonal parameters between lean and obese PCOS groups is given in Tables 1 and  2. Age, BMI, WC, systolic BP, diastolic BP, FPG, PG 2H-OGTT, serum triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), TT, and TSH were higher in obese PCOS group; serum prolactin was higher in the lean PCOS group; modified F-G score and high-density lipoprotein cholesterol (HDL-C) levels were similar in the two groups. The frequencies of acanthosis nigricans, prediabetes, diabetes, pre-HTN, HTN, metabolic syndrome, and biochemical hyperandrogenism were higher in the obese group. The two groups had statistically similar frequencies of menstrual irregularity, hirsutism, type 2 diabetes in the first-degree family members, and dyslipidemia.
Though polycystic ovarian morphology (enlarged ovary and polycystic appearance) was higher in the obese group, the difference was not statistically significant.   Table 3.  Metabolic syndrome and dysglycemia were more common among the lean PCOS subjects aged ≥25 years than those aged <25 years in the group. Elevated BP (HTN and pre-HTN) and dysglycemia were more common among the obese PCOS subjects aged ≥25 years than those aged <25 years.
In binary logistic regression analysis, study subjects in the obese PCOS group and age group ≥25 years had significantly higher metabolic syndrome risks than their counterparts [ Table 4].

Discussion
In this study, the frequencies of lean-and obese PCOS were 23.3% and 76.7%, respectively. The obese PCOS subjects had higher metabolic risk factors and a higher degree of biochemical hyperandrogenism.
Insulin resistance (IR) is generally agreed to be the underlying cause of PCOS (1,12) . Obesity and higher BMI further add to the risk of IR and metabolic syndrome (6) . Though the prevalence of IR is reported to be 6-22% in lean PCOS, there is disagreement whether these thin women with PCOS suffer from IR to the same degree as their obese counterparts (21) . A study done by Yildirim et al. has demonstrated the presence of a higher proportion of preperitoneal and visceral adiposity among lean PCOS patients compared to weight-matched controls (22) . They concluded that lean patients with PCOS have intrinsic IR, whereas obese patients with PCOS have intrinsic resistance as a part of the disease and extrinsic resistance due to obesity (22) . Another study done by Bozkirli et al. has demonstrated both lean and obese PCOS phenotypes have insulin resistance compared to controls, indicating that factors other than obesity may be involved in IR seen in PCOS (23) . Faloia et al., in a study conducted to evaluate metabolic characters in lean and obese patients with PCOS, did not find significant metabolic alterations in lean women with PCOS (24) . The observation of Faloia et al. indicates that obesity underpins the metabolic alterations exhibited by overweight or obese patients (24) .
A study done by Sachdeva et al. in India observed that lean-and obese PCOS prevalence were 24.39% and 75.61%, respectively (6) . However, the prevalence of lean PCOS was found higher in other Indian studies done by Majumdar et al. (33.3% lean and 66.7% obese) and Akshaya et al. (44% lean and 56% obese) (7,9) . Nahar et al., in their study, found 52% of Bangladeshi women diagnosed with PCOS to have BMI ≥25 kg/m 2 ; the rest (44%) of them had BMI <25 kg/m 2(25) . In the current study, the frequencies of lean and obese PCOS were 23.3% and 76.7%, respectively, which are almost similar to the observation of Sachdeva et al. (6) . The higher frequency of obese PCOS in this study than another study conducted by Nahar et al. in Bangladesh is probably due to lower BMI cutoffs used to define obese and overweight women in this study (25) .
In a meta-analysis done by Chowdhury et al., the weighted pooled prevalence of metabolic syndrome in Bangladeshi women using any criteria was 32%, with a higher prevalence in older age (26) . A study done by Jesmin et al. to assess the prevalence of metabolic syndrome in 1485 rural women of Bangladesh using NCEP ATP III modified criteria demonstrated that the prevalence of metabolic syndrome was 31.25%, and the prevalence increased with age; 6.40% in <25 years, 20.07% in 25-34 years, 31.98% in 35-44 years, and 46.02% in the age group 55-64 years (27) . In this study, 50.5% of the women with PCOS had metabolic syndrome; the frequency was 14.8% in lean PCOS and 61.3% in obese PCOS; the obese group had more than 8-fold higher risk of metabolic syndrome compared to the lean group. Among the lean PCOS subjects, we found metabolic syndrome in 11.7% in the age group <25 years and 31.6% in those aged ≥25 years; in the obese PCOS, the frequencies were 58.5% and 66.9% in the age groups, respectively. So, it may be said that even lean PCOS women have a higher prevalence of metabolic syndrome than their non-PCOS otherwise healthy counterparts of the same age in our setting. Sachdeva et al. observed a lower prevalence of metabolic syndrome (24.39%) among Indian women with PCOS; though, like our study, the prevalence was higher in the obese PCOS compared to the lean PCOS (29.03% vs. 10%) in their study (6) .
In Indian women with PCOS, HTN was more common in the obese group as compared to lean PCOS in the studies done by Sachdeva (28) . We observed higher frequencies of HTN (8.7% vs. 0.8%) and pre-HTN (22.2% vs. 9.8%) in the obese PCOS subjects compared to the lean ones (p<0.001). Both the systolic and the diastolic BP levels were higher in the obese group. The frequency of hypertension was higher in the higher age group.
In India, Majumdar et al. observed a higher prevalence of both prediabetes (25% vs. 10%, p=0.000) and diabetes (11.7% vs. 6%, p=0.000) in obese compared to the lean POCS patients (7) . On the contrary, Sachdeva et al. and Saxena et al. observed no differences in FPG and PG 2H-OGTT between lean and obese groups (6,28) . Indian obese PCOS women had a higher frequency of family history of T2DM than their lean counterparts (9,28) . Though the frequency of type 2 DM in the first degree relatives was identical in the lean-and obese PCOS women in our study ( (6,7,9,28) . We observed no differences in the lean and obese groups regarding menstrual irregularity (93.4% vs. 90.5%, p=0.366).
The frequency of hirsutism was similar in both lean and obese PCOS groups in the studies done by Akshaya et al. and common in the obese PCOS women according to Sachdeva et al.and Majumder et al. (6,7,9,28) . Sachdeva et al. also observed higher F-G scores in obese patients (6) . The mean F-G score and frequency of hirsutism in our study were similar in the two groups ( .5%) demonstrated higher frequencies of acanthosis nigricans in obese PCOS though the differences were not statistically significant (9,28) . However, acanthosis nigricans was more common in the obese subjects than the lean ones (89.3% vs. 45.9%, p<0.001) in this study.
Both lean and obese groups had similar TSH in the studies done by Thathapudi et al. and Saxena et al (10,28) . However, we observed significantly higher TSH (1.37 vs. 1.03 µIU/mL, p=0.001) in the obese PCOS than lean PCOS. Serum prolactin was higher in the lean PCOS (276.1 vs. 241.2 µIU/mL, p=0.016) in our study. Saxena et al. also had similar observations (13.6±8.2 vs. 11.9±4.6 ng/mL), though the difference observed by them was not statistically significant (28) . This study has some limitations. It was a single-center study; the sample may not represent the whole country. No healthy control group was included limiting the comparison with the lean and obese PCOS. hirsutism was more Nevertheless, this one is the first study in Bangladesh comparing the clinical, metabolic, and hormonal parameters in lean and obese PCOS and may serve as the basis for further large-scale, multi-center study in this subject.

Conclusion
Lean and obese women have similar menstrual irregularities and hirsutism, but biochemical hyperandrogenism is more prevalent in the obese PCOS. Both lean and obese women with PCOS have adverse metabolic consequences; metabolic abnormalities are more frequently observed when obesity is associated with PCOS. All patients with PCOS should be considered for screening for metabolic abnormalities irrespective of BMI category.