Efficacy of Bariatric Surgery in the Treatment of Women With Obesity and Polycystic Ovary Syndrome.

CONTEXT: The comparative effectiveness of drugs and surgical therapy for women with obesity and polycystic ovary syndrome (PCOS) has not been systematically compared.
OBJECTIVE: We aimed to determine the difference in efficacy between drug and bariatric surgery therapy for women with obesity and PCOS.
METHODS: This prospective nonrandomized trial enrolled 90 women aged 18 to 40 years with body mass index (BMI) ≥ 27.5 kg/m2 and waist circumference ≥ 85 cm and fulfilling the 2011 Chinese diagnostic criteria for PCOS; 81 subjects completed the study. In the drug group, patients were administered metformin and an oral contraceptive containing ethinyl-estradiol and cyproterone acetate for the first 6 months, and metformin alone for the second 6 months. In the surgical group, patients underwent laparoscopic sleeve gastrectomies. The follow-up period was 12 months. The main outcome was the complete remission of PCOS, requiring 6 consecutive regular menstruation cycles or spontaneous pregnancy.
RESULTS: Median BMI at endpoint was 30.1 kg/m2 in the drug group and 23.7 kg/m2 in the surgical group; complete remission rate was 15% and 78%, respectively. Except endpoint BMI, no difference was observed in free androgen index, ovarian morphology, homeostasis model assessment for insulin resistance, and total weight loss between remission and nonremission patients. Logistic regression analyses also revealed that the final BMI was the major factor influencing the remission of PCOS. The cutoff points for the final BMI were 27.5 kg/m2 for the drug group and 26 kg/m2 for the surgical group. Overall, nearly 95% of patients with an endpoint BMI below the cutoff values achieved complete remission.
CONCLUSION: Complete remission of PCOS in patients with obesity depends on the final BMI after weight loss. Thus, bariatric surgery should be prioritized for these patients.

Polycystic ovary syndrome (PCOS) is a complex endocrine disorder that is diagnosed in reproductive-age women, accounting for nearly 80% of anovulatory infertility cases (1, 2). Typically characterized by irregular menstruation, chronic oligo- or anovulation, and hyperandrogenism, PCOS exhibits heterogeneous features. PCOS is also strongly associated with a wide range of metabolic disorders, such as abdominal adiposity, insulin resistance, obesity, and dyslipidemia (3-6). Currently, more than 60% of patients with PCOS are overweight or obese (7). Metabolic and reproductive outcomes are worsened by an increased body mass index (BMI) in women with PCOS, reflecting a higher prevalence of type 2 diabetes, hypertension, infertility, and other pregnancy-related complications (8-12). Thus, there is a relationship between PCOS and obesity, but the mechanism is not yet fully understood (13, 14).

Due to insufficient understanding of its etiology and its heterogeneous nature, the treatment of PCOS remains symptom-oriented rather than mechanism-based (3). Oral contraceptives have become the mainstay medication for PCOS because of their benefits in ameliorating the symptoms of hyperandrogenism, regular withdrawal bleeding, and prevention of endometrial hyperplasia (15, 16). However, in women with obesity, PCOS seems to be reversible with weight loss (15). Therefore, in the treatment of these patients, weight loss is an urgent issue to be considered, especially given the significant contribution of obesity to the development of PCOS.

Metformin is recommended by an evidence-based guideline for weight loss and improvement of endocrine-metabolic disturbances (17). In patients with PCOS, the combination of metformin and oral contraceptives is plausible, even in patients of normal weight, to avoid the endocrine-metabolic consequences of gaining weight (18-20). In terms of weight loss therapy, bariatric surgery is a first-line treatment for morbid obesity (21). For women with obesity and PCOS, many retrospective studies, and very few prospective studies, have proven the considerable effectiveness of bariatric surgery in weight loss, menstrual recovery, decreased serum androgens, and metabolic benefits (22-25). However, bariatric surgery is still recommended with low priority in weight loss interventions for patients with obesity and PCOS (26). Furthermore, the effectiveness of drugs and surgical therapy has not been systematically compared. To address this issue, we conducted a prospective study of women with obesity and PCOS.

Methods

Ethics and Trial Registration

This study was approved by the Ethics Committee of Shanghai Sixth People’s Hospital on September 28, 2017 (approval number 2017-121). This study was registered at the Chinese Clinical Trial Registry (ChiCTR, number ChiCTR-IOR-17013169).

Study Population

Ninety reproductive-age women (18-40 years of age) meeting the following criteria were prospectively recruited for this study: 1) BMI ≥ 27.5 kg/m2; 2) waist circumference ≥ 85 cm; 3) diagnosed with PCOS. All the subjects were eligible for the indication of bariatric surgery according to the guideline of the Chinese Society of Metabolic and Bariatric Surgery (27). According to the 2011 Chinese criteria based on the principles of the Rotterdam criteria, the presence of oligo-ovulation or anovulation is essential for diagnosis. Satisfying one of the conditions of either clinical (or biochemical) hyperandrogenism or ultrasound inspection of polycystic ovarian morphology (PCOM) and excluding other known diseases causing hyperandrogenism and anovulation disorders was also required (28). PCOM is defined as more than 12 follicles sized 2-9 mm in either or both ovaries, or an ovarian volume ≥ 10 mL in either ovary (ovarian volume is calculated as 0.5 × ovarian length × ovarian width × ovarian height), as determined by ultrasound detection. The exclusion criteria included pregnancy, conditions that made it difficult to assess menstruation or ovulation (eg, a history of hysterectomy, prior bilateral oophorectomy), allergy or contraindications to metformin or oral contraceptives (eg, history or risk factors related to thrombosis, hepatic, or renal dysfunction), or use of any hormonal medication within 3 months before the evaluation. Thyroid hormones, 17-hydroxyprogesterone, prolactin, and glycosylated hemoglobin (HbA1c) levels in all subjects were normal.

Protocol

This was a single-center, prospective, nonrandomized trial. All subjects were recruited at Shanghai Jiao Tong University Affiliated Sixth People’s Hospital from September 2017 to July 2020. Informed consent forms were signed before all subjects were enrolled. Basic information, including the ages of all participants, was recorded. Participants were divided into drug and surgical groups according to the patients’ intentions, with a 12-month follow-up period for all women. Patients in the drug group received an oral contraceptive containing 35 μg ethinyl-estradiol plus 2 mg cyproterone acetate (Bayer, Germany) as 1 tablet daily for the first 6 months and metformin (Bristol-Myers Squibb, China) at 2000 mg/day for the entire 12 months. Laparoscopic sleeve gastrectomies were conducted in patients in the surgical group. Patient characteristics, including clinical and biochemical parameters, were evaluated 3 to 7 days before treatment and at 12 months after treatment. Data were collected each month regarding the number of subjects menstruating during the entire follow-up for both groups.

Body weight and height were measured while the subjects were barefoot and only wearing light clothing. BMI was defined as weight (kg) divided by the squared body height (m2). Waist circumference and hip circumference were measured and used to calculate the waist to hip ratio (W/H). Baseline blood samples, including fasting and postprandial samples, were collected randomly for amenorrheic subjects, or during days 3 to 5 of a spontaneous menstrual cycle. Serum levels of luteinizing hormone (RRID: AB_2800498, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2800498/resolver?q=AB_2800498%2C&l=AB_2800498%2C&i=rrid:ab_2800498-143:11732234; the intra- and interassay coefficients of variation (CV) were 0.8% and 2.0%, respectively), follicle-stimulating hormone (RRID: AB_2800499, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2800499/resolver?q=AB_2800499%2C&l=AB_2800499%2C&i=rrid:ab_2800499-143:11775863; intra-assay CV = 1.8%, interassay CV = 5.3%), total testosterone (RRID: AB_2783736, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2783736/resolver?q=AB_2783736%2C&l=AB_2783736%2C&i=rrid:ab_2783736-2719192; intra-assay CV = 2.1%, interassay CV = 3.2%), sex hormone–binding globulin (SHBG, RRID: AB_2891222, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2891222/resolver?q=AB_2891222%2C&l=AB_2891222%2C&i=rrid:ab_2891222-2826678; intra-assay CV = 2.4%, interassay CV = 2.8%), and dehydroepiandrosterone sulfate (DHEAS, RRID: AB_2909490, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2909490/resolver?q=AB_2909490%2C&l=AB_2909490%2C&i=rrid:ab_2909490-2844946; intra-assay CV = 2.4%, interassay CV = 4.7%) were measured with a Cobas 8000 automatic analyzer using an electrochemiluminescence immunoassay (Hitachi, Roche, Tokyo, Japan). The free androgen index (FAI) was evaluated as total testosterone (nmol/L)/SHBG (nmol/L) × 100%. Serum 17-α-hydroxyprogesterone was measured by luminescence immunoassays (LDN Labor Diagnostika Nord, Nordhorn, Germany, RRID: AB_2909471, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2909471/resolver?q=AB_2909471.&l=AB_2909471.&i=rrid:ab_2909471-2844927) with intra- and interassay CV of 4.0% and 2.0%, respectively, and anti-Müllerian hormone (AMH) levels were determined using an enzyme-linked immunosorbent assay kit (picoAMH, ANSH Labs, Webster, TX, USA, RRID: AB_2783659, https://scicrunch.org/resources/data/record/nif-0000-07730-1/AB_2783659/resolver?q=AB_2783659%2C&l=AB_2783659%2C&i=rrid:ab_2783659-2719115) with intra- and interassay CV of 5.0% and 6.0%, respectively. The triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) levels were determined using an automated biochemical detector (7600-120; Hitachi, Tokyo, Japan). Fasting plasma glucose (FPG) levels were determined with the glucose oxidase method (Beckman Coulter, CA, USA). Fasting serum insulin was detected using a Cobas e411 analyzer (Roche, Mannheim, Germany). HbA1c levels were determined using high-performance liquid chromatography (Bio-Rad Inc., Hercules, CA, USA). Homeostasis model assessment for insulin resistance (HOMA-IR) was calculated to evaluate insulin sensitivity as follows: HOMA-IR = [fasting insulin (mIU/L) × FPG (mmol/L)]/22.5 (29). Ovarian ultrasonography was conducted by scanning the pelvis using an 8 MHz transducer. Each ultrasonographic detection was performed by a radiologist who was blinded to the intervention.

Efficacy Endpoints

The complete PCOS remission rate was set as the primary efficacy endpoint. Complete PCOS remission was defined as satisfying one of the following criteria: (i) regular consecutive menstrual cycles (cycles between 21 and 35 days, including 21 and 35 days) from the seventh to the twelfth month of the follow-up period, or (ii) spontaneous pregnancy (pregnant without any assisted reproductive technology) during seventh to the twelfth month of the follow-up period and keeping regular menstrual cycles before pregnancy in the second half of follow-up. Otherwise, if one of the criteria was not satisfied, the status was defined as nonremission. Secondary endpoints included the free testosterone index and ovary morphology at the end of the study.

Sample Size

The sample size was calculated by PASS 11.0.7 statistical software. Based on the literature and preliminary results, we assumed complete remission rates in the drug and surgical groups of 48.8% and 83%, respectively (25, 30, 31). Based on this assumption, we set α to 0.05, the testing power to 0.8, and the expected dropout rate to 20%; therefore, the required sample size for this experiment was 90, with 45 cases each in the drug and bariatric surgery treatment groups.

Statistical Analysis

SPSS v22.0 (IBM Corp., Armonk, NY, USA) was used for statistical analysis. The normal distribution of continuous variables was assessed using the Kolmogorov-Smirnov test. Continuous variables obeying a normal distribution in all groups are presented as the mean ± standard deviation, or as the median with the interquartile range. Student’s t test was performed to compare continuous variables with a normal distribution. The Mann–Whitney test was performed to compare nonnormally distributed continuous variables. Logistic regression analysis, including univariate and multivariate models, was performed to screen for potential determinants of PCOS remission. Statistical significance was set as P < 0.05. Based on the results of logistic regression analysis, receiver operating characteristic (ROC) analysis was conducted for both study groups, with the area under the curve (AUC) calculated. The odds ratio (OR) was calculated, with the confidence interval set to 95%. Youden’s index was calculated to ascertain the cutoff point.

Results

Participants

The participant flowchart is shown in Fig. 1. A total of 158 women were screened for eligibility, 90 of whom were ultimately enrolled in this study, and 81 patients completed the study.

Figure 1.

Flowchart of the study participants.

Overall Comparison of Drug Group and Surgical Group at Baseline and Endpoint

All clinical characteristics of the subjects at baseline and endpoint in both study groups are presented in Table 1. At baseline, body weight, BMI, waist circumference, hip circumference, free androgen index and DHEAS were significantly higher in the surgical group than in the drug group, while SHBG and HDL-C were markedly lower in the surgical group than in the drug group. After the treatment, subjects in the surgical group showed remarkable decreases in BMI, waist circumference, hip circumference, W/H, total testosterone, free androgen index (P = 0.003), AMH (P = 0.006), androstenedione (P = 0.005), FPG, fasting insulin, HOMA-IR, HbA1c, TC, and HDL-C (P = 0.003), as well as a significant increase in SHBG (all P < 0.01). Patients in the drug group only showed decreased BMI (P = 0.018), lower HbA1c (P = 0.023), and increased SHBG (P = 0.037). The percentage of patients with > 12 follicles on one or both sides dropped by nearly one-third in the surgical group (P = 0.002) but not in the drug group (P = 0.479) (Table 1). Additionally, the decrease in weight, BMI, FAI, and the percentage of subjects with multiple follicles (>12 on one or both ovaries) as well as the increase in SHBG (all P < 0.001) were all much larger in the surgical group than in the drug group. However, the decrease in maximum ovarian volume (MaxOV) (P = 0.184) showed no significant difference between the 2 groups (Fig. 2).

Table 1.

Characteristics of subjects at baseline and endpoint with drug or surgery treatment

	Drug group (n = 40)			Surgical group (n = 41)
	Baseline	Endpoint	P value	Baseline	Endpoint	P value
Age (years)	27.6 ± 0.8			28.7 ± 0.7
Weight (kg)	86.3 ± 11.7	81.9 ± 11.1	0.092	99.4 ± 17.8*	65.3 ± 9.2	<0.001
BMI (kg/m2)	31.2 (29.5–34.4)	30.1 (28.6–32.3)	0.018	35.6 (32.8–38.8)***	23.7 (21.9–25.9)	<0.001
Waist (cm)	98.5 (95.0–102.8)	98.0 (92.0–102.0)	0.208	110.0 (103.0–119.8)***	84.0 (76.0–90.0)	<0.001
Hip (cm)	105.5 (99.0–111.5)	106.5 (102.0–110.0)	0.754	112.0 (108.5–123.5)***	95.0 (90.0–101.0)	<0.001
W/H	0.93 ± 0.07	0.91 ± 0.07	0.146	0.96 ± 0.06	0.88 ± 0.08	<0.001
17-α-hydroxyprogesterone (ng/ml)	1.3 (1.0–1.8)			1.3 (0.9–2.2)
Total testosterone(nmol/L)	1.9 (1.4–2.4)	2.0 (1.5–2.6)	0.704	2.0 (1.4–2.7)	1.3 (1.0–2.0)	0.005
SHBG (nmol/L)	21.9 (16.0–29.8)	22.7 (19.5–41.0)	0.037	15.1 (11.5–19.4)***	62.3 (49.4–72.1)	<0.001
FAI (%)	9.4(6.5–11.9)	7.4 (3.9–10.7)	0.084	12.9 (9.2–20.6)**	2.4 (1.7–3.9)	<0.001
Androstenedione (ng/mL)	4.6 (3.1–5.6)	4.2 (3.9–4.6)	0.255	3.5 (2.9–4.5)	2.7 (1.7–3.6)	0.005
DHEAS (ug/dL)	319.8 ± 107.8	358.6 ± 98.0	0.139	326.8 ± 141.4*	276.3 ± 118.6	0.083
AMH (ng/mL)	9.3 (5.1–16.6)	8.8 (6.1–13.9)	0.718	9.3 (5.7–22.1)	6.0 (3.6–13.7)	0.006
MaxOV (mL)	9.7 (7.7–12.9)	10.6 (7.5–11.9)	0.927	10.0 (6.0–11.4)	7.5 (5.2–8.7)	0.091
Follicle number ≥ 12 (%)	92.5	85	0.479	90.2	61.0	0.002
FPG (mmol/L)	5.2 (4.9–5.7)	5.2 (4.9–5.5)	0.384	5.6 (4.8–7.0)	4.4 (4.2–4.8)	<0.001
Fasting insulin (μIU/mL)	23.0 (18.4–35.4)	22.9 (15.0–36.4)	0.900	24.9 (18.7–34.0)	6.7 (4.7–8.9)	<0.001
HOMA-IR	5.6 (4.2–9.0)	5.3 (3.2–11.1)	0.651	6.8 (4.9–9.1)	1.3 (0.9–1.8)	<0.001
HbAlc (%)	5.7 (5.5–6.0)	5.5 (5.2–5.7)	0.023	5.8 (5.5–6.7)	5.1 (4.9–5.5)	<0.001
Triglycerides (mmol/L)	1.4 (0.9–2.0)	1.5 (0.9–2.0)	0.612	1.5 (1.2–2.0)	0.8 (0.6–1.00)	<0.001
Total cholesterol (mmol/L)	5.0 (4.3–5.6)	5.1 (4.3–6.0)	0.819	4.9 (4.3–5.5)	4.8 (4.1–5.1)	0.210
HDL-C (mmol/L)	1.2 (1.1–1.5)	1.3 (1.1–1.4)	0.262	1.1 (0.9–1.3)**	1.3 (1.1–1.5)	0.003
LDL-C (mmol/L)	3.1 ± 0.1	3.1 ± 0.2	0.759	3.1 ± 0.1	2.8 ± 0.1	0.115

Data are presented as median (interquartile range) or mean ± standard error of the mean.

P value, endpoint vs baseline in the same treatment group; P < 0.05, P < 0.01, P < 0.001, baseline of surgical group vs baseline of drug group.

Abbreviations: AMH, anti-Müllerian hormone; BMI, body mass index; DHEAS, dehydroepiandrosterone sulfate; FAI, free androgen index; FPG, fasting plasma glucose; HbAlc, glycosylated hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostatic model of insulin resistance; LDL-C, low-density lipoprotein cholesterol; MaxOV, maximum ovarian volume; SHBG, sex hormone–binding globulin; TC, total cholesterol; TG, triglycerides; W/H, ratio of waist circumference to hip circumference.

Figure 2.

Changes in clinical and hormonal indicators. (A) Weight, (B) BMI, (C) SHBG, (D) FAI, (E) MaxOV, and (F) the percentage of patients with over 12 follicles. Abbreviations: BMI, body mass index; FAI, free androgen index; MaxOV, maximum ovarian volume; SHBG, sex hormone–binding globulin.

Subgroup Analysis Based on the Outcome

The complete remission rate of the surgical group was 78%, which was more than 5 times greater than that of the drug group (Fig. 3A). At baseline, there was no significant difference in all parameters between patients with complete remission and without remission in drug group; however, in the surgical group, weight, BMI, and androstenedione were markedly lower in patients with complete remission than those without remission. Compared with the status before treatment, patients at the endpoint presented with lower BMIs in both the drug and surgical groups, regardless of remission status (Table 2). Meanwhile, the BMI of subjects with complete remission was much lower than that of subjects without remission in both groups (all P < 0.05) (Fig. 3B). The FAI and MaxOV were similar between patients with and without remission in both groups (Fig. 3C and 3D). The percentage of patients with > 12 follicles in patients with complete remission was less than one-fifth of those without remission in the drug group, but it was similar between surgical patients with and without remission (Fig. 3E). Patients both with and without complete remission in the surgical group experienced significant decreases in body weight, waist and hip circumference, W/H, FAI, FPG, fasting insulin, HbA1c, and TG, as well as a marked increase in SHBG (all P < 0.05), but patients in the drug group did not show the same changes. Subjects who achieved complete remission in the surgical group also had lower total testosterone (P = 0.011), lower androstenedione (P < 0.001), lower DHEAS (P = 0.028), and higher HDL-C (P = 0.011) levels, whereas subjects in the nonremission group had lower LDL-C levels (P = 0.036). Furthermore, the total weight loss (TWL) (P = 0.003) of patients with complete remission was significantly higher and waist circumference (P = 0.017) was significantly lower than those in subjects without remission in the drug group. In the surgical group, the waist circumference of patients with complete remission was lower than that of the others, but TWL showed no significant difference (Table 2).

Figure 3.

Subgroup analysis of the endpoint. (A) The proportion of patients in remission or nonremission. Comparison of patients in or out of remission regarding (B) BMI, (C) FAI, (D) MaxOV, and (E) the percentage of patients with over 12 follicles. (F) The proportion of patients menstruating each month during the follow-up. (G) Kaplan-Meier curves for the cumulative proportion of patients with regular menstruation. Abbreviations: BMI, body mass index; FAI, free androgen index; MaxOV, maximum ovarian volume.

Table 2.

Characteristics of subjects with different outcomes at baseline and endpoint with drug or surgery treatment

	Drug group				Surgery group
	Nonremission (n = 34)		Complete remission (n = 6)		Nonremission (n = 9)		Complete remission (n = 32)
	Baseline	Endpoint	Baseline	Endpoint	Baseline	Endpoint	Baseline	Endpoint
Age (years)	28.3 ± 4.5		24.5 ± 6.0		27.3 ± 2.8		29.0 ± 5.2
Weight (kg)	87.7 ± 11.6	84.0 ± 10.4*	78.3 ± 9.3	70.2 ± 6.8	110.5 ± 19.8	76.3 ± 5.6***	96.3 ± 16.1†	62.9 ± 7.1***
BMI (kg/m2)	31.3 (30.1–35.1)	30.4 (29.1–32.4)*	29.5 (28.3–31.7)	26.5 (25.5–27.8) *, ###	39.5 (37.1–43.3)	28.7 (26.5–28.9)***	35.0 (32.8–37.6)††	23.0 (21.5–24.4)***, ###
TWL (%)		4.5 (2.3–6.1)		8.3 (6.0–16.3)		30.1 (26.3–38.7)		34.3 (27.5–39.8)
Waist (cm)	99.0 (94.0–105.0)	98.0 (92.0–103.0)	96.5 (94.3–99.3)	91.0 (82.8–95.0) *, #	113.5 (09.5–124.3)	91.0 (86.3–95)***	110.0 (102.5–119.8)	82.0 (76.0–88.0)***, #
Hip (cm)	105.0 (102.0–110.0)	107.0 (102.0–110.0)	104.5 (92.0–115.5)	103.0 (92.8–111.8)	120.0 (112.0–129.5)	101.5 (95.8–105.0)***	112.0 (107.0–121.5)	95.0 (89.0–100.0)***,#
W/H	0.9 ± 0.06	0.9 ± 0.05	0.9 ± 0.1	0.9 ± 0.1	1.0 ± 0.06	0.9 ± 0.07*	1.0 ± 0.06	0.9 ± 0.09***
Total testosterone (nmol/L)	1.9 (1.4–2.4)	1.8 (1.4–2.4)	2.0 (1.2–2.9)	2.3 (1.7–2.8)	2.1 (1.1–3.3)	1.4 (1.1–2.0)	1.9 (1.4–2.7)	1.3 (1.0–2.2)**
SHBG (nmol/L)	21.9 (16.9–29.7)	29.5 (19.8–40.4)	22.6 (13.2–34.2)	46.0 (22.3–69.0)	12.5 (9.4–17.2)	68.2 (50.3–96.6) ***	16.5 (12.6–19.6)	60.9 (44.9–71.1)***
FAI (%)	9.7 (6.5–12.1)	7.6 (4.9–10.8)	9.2 (7.3–11.6)	4.0 (3.2–9.8)	17.2 (5.8–28.4)	2.0 (1.5–3.6) ***	12.5 (9.2–19.5)	2.4 (1.7–3.9)***
Androstenedione (ng/mL)	4.6 (3.1–5.7)	4.1 (3.9–4.6)	4.4 (3.2–5.5)	4.4 (3.7–4.9)	3.1 (2.2–4.1)	3.0 (1.4–4.6)	3.9 (3.2–4.6)†	2.5 (1.7–3.2)**
DHEAS (ug/dL)	323.3 ± 17.0	367.7 ± 20.3	297.9 ± 145.8	306.1 ± 95.2	295.5 ± 169.6	310.8 ± 114.6	335.6 ± 134.3	263.7 ± 120.2*
MaxOV (mL)	9.4 (7.3–11.9)	10.9 (8.2–12.0)	12.4 (8.8–15.8)	7.5 (5.5–10.2)	9.2 (5.4–11.6)	11.6 (7.0–13.4)	10.2 (6.1–11.3)	7.6 (5.6–8.8)
Follicle number ≥ 12 (%)	97.1	97.1	66.7	16.7	77.8	77.8	93.8	56.3***
FPG (mmol/L)	5.2 (4.9–5.7)	5.2 (4.9–5.6)	5.2 (5.0–5.6)	5.4 (4.3–5.7)	5.3 (4.6–6.7)	4.2 (3.8–4.5)*	5.6 (4.9–6.9)	4.9 (5.2–5.6)***
Fasting insulin (μIU/ml)	24.3 (17.8–38.0)	21.4 (13.4–48.2)	23.6 (22.6–27.3)	27.6 (17.5–46.2)	32.2 (25.1–88.3)	7.6 (3.8–12.6)***	21.3 (18.0–30.9)††	6.7 (4.4–9.0)***
HOMA-IR	5.5 (4.1–9.2)	5.3 (2.9–9.5)	5.6 (5.0–6.7)	5.1 (3.3–6.9)	8.2 (5.4–25.6)	1.3 (0.7–2.5)***	6.6 (4.4–9.0)	1.4 (0.9–2.0)***
HbAlc (%)	5.7 (5.5–6.0)	5.5 (5.2–5.7)	5.8 (5.3–6.0)	5.4 (5.2–5.8)	5.8 (5.4–6.0)	5.0 (4.9–5.1)***	5.8 (5.5–6.0)	5.2 (4.9–5.6)***
Triglycerides (mmol/L)	1.5 (0.8–2.1)	1.5 (0.8–1.9)	1.2 (1.1–1.8)	1.3 (1.0–1.8)	1.5 (1.1–1.7)	0.8 (0.7–1.0) *	1.5 (1.2–2.2)	0.8 (0.6–1.0)***
Total cholesterol (mmol/L)	5.0 (4.3–5.5)	5.0 (4.4–6.0)	4.9 (3.9–5.7)	5.0 (4.2–5.7)	5.2 (4.2–5.5)	4.9 (4.3–5.5)	4.5 (4.4–5.2)	4.2 (3.7–4.8) #
HDL (mmol/L)	1.2 (1.1–1.5)	1.3 (1.1–1.4)	1.1 (1.0–1.3)	1.7 (1.1–1.9)	1.0 (0.8–1.3)	1.3 (1.1–1.4)	1.1 (0.9–1.2)	1.3 (1.1–1.5) **
LDL (mmol/L)	3.1 ± 0.8	3.0 ± 0.9	3.1 ± 0.9	3.1 ± 0.5	2.9 ± 0.6	2.3 ± 0.5*	3.1 ± 0.8	2.5 ± 0.8

Data are presented as median (interquartile range) or mean ± standard error of the mean.

Abbreviations: BMI, body mass index; DHEAS, dehydroepiandrosterone sulfate; FAI, free androgen index; FPG, fasting plasma glucose; HbAlc, glycosylated hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostatic model of insulin resistance; LDL-C, low-density cholesterol lipoprotein; MaxOV, maximum ovarian volume; SHBG, sex hormone–binding globulin; TC, total cholesterol; TG, triglycerides; TWL, total weight loss; W/H, ratio of waist circumference to hip circumference.

*P < 0.05,

** P < 0.01,

*** P < 0.001, endpoint vs baseline in the same outcome of one treatment;

# P < 0.05,

## P < 0.01,

### P < 0.001, endpoint vs endpoint in the same treatment group;

† P < 0.05,

†† P < 0.01,

††† P < 0.001, baseline vs baseline in the same subgroup.

Different Menstruation Patterns Between Study Groups

Menstruation rate analysis of patients with regular menstruation, from the first to the last month of follow-up, presented with very different patterns between the 2 groups. For the first 6 months of observation, all patients in the drug group maintained regular withdrawal bleeding with the help of oral contraceptives. However, the situation worsened during the next 6 months of efficacy evaluation. Interestingly, the surgical group showed a completely different outcome. Most of the patients in the surgical group experienced a single episode of withdrawal bleeding immediately after the surgery. Although their menstruation was initially irregular, it tended to be regular over the following 5 months. Nearly 80% of the patients maintained regular menstruation for the last 6 months (Fig. 3F). The Kaplan-Meier curve revealed a difference in the proportion of patients who retained their regular menstrual cycles over time in both groups (Fig. 3G).

Association Between the Endocrine-Metabolic Indicators and the Outcome of PCOS

Logistic regression analyses, including univariate and multivariate models, were conducted to evaluate the potential risk factors for nonremission of PCOS. BMI, TWL, FAI, SHBG, MaxOV, and HOMA-IR at the endpoint were all significantly associated with the outcome of nonremission in the univariate analysis. After adjustment for age, waist circumference, hip circumference, androstenedione, DHEAS, HOMA-IR, TC, TG, HDL-C, LDL-C, endpoint BMI (OR = 2.515; 95% CI, 1.694-3.734; P < 0.001) were still related to an increased risk of nonremission of PCOS (Table 3).

Table 3.

Univariate and multivariate analyses for determining the key factors of nonremission outcome

Variable	Univariate model		Multivariate model
	OR (95%CI)	P value	OR (95%CI)	P value
BMI (kg/m2)	1.641(1.339–2.011)	<0.001	2.403 (1.475–3.917)	<0.001
TWL (%)	0.899 (0.863–0.938)	<0.001	1.242 (0.946–1.630)	0.834
SHBG (nmol/L)	1.015 (0.982–1.043)	<0.001	1.000 (0.962–1.040)	0.986
FAI (%)	1.812 (1.324–2.446)	<0.001	1.072 (0.791–1.453)	0.655
MaxOV (ml)	1.458 (1.110–1.915)	0.007	1.242 (0.946–1.630)	0.119
HOMA-IR	1.261 (1.078–1.475)	0.004	0.961 (0.759–1.217)	0.741

The multivariate model was adjusted for age, waist circumference, hip circumference, androstenedione, DHEAS, HOMA-IR, TC, TG, HDL-C, and LDL-C. The BMI in the table refers to BMI of patients at the endpoint.

Abbreviations: BMI, body mass index; FAI, free androgen index; FPG, fasting plasma glucose; MaxOV, maximum ovarian volume; OR, odds ratio; SHBG, sex hormone–binding globulin; TWL, total weight loss.

The Cutoff Points of Final BMI Based on Outcome

ROC analysis was conducted to find a meaningful cutoff point to provide clinical guidance. The AUC of the final BMI to predict PCOS in the drug group was 0.94 (95% CI, 0.82-1.00; P < 0.001). Moreover, when calculating the maximum Youden’s index, the cutoff value of the final BMI fell to 27.5 kg/m2, with a sensitivity of 100% and specificity of 83%. In the surgical group, the AUC was 0.89 (95% CI, 0.72-1.00; P < 0.001) and the cutoff point was 25.7 kg/m2, with 89% sensitivity and 91% specificity (Fig. 4). Subgroup analyses were conducted based on the above cutoff values (Table 4). After making a comparison of baseline parameters at different BMI subgroups within the same treatment, we found that in the drug group, baseline BMI of subjects with a final BMI < 27.5 kg/m2 was significantly lower than those with a final BMI ≥ 27.5 kg/m2; while in the surgical group, the values for weight, BMI, and waist circumference at baseline were all lower in the patients with a final BMI < 26 kg/m2 than those with a final BMI ≥ 26 kg/m2. The complete remission rate of the drug group was 100% in subjects with final BMI < 27.5 kg/m2 and only 2.9% in subjects with final BMI ≥ 27.5 kg/m2. Additionally, the TWL was remarkably increased and the percentage of patients with > 12 follicles was decreased in the patients with complete remission in drug group, while the FAI and MaxOV showed no difference (Fig. 5A-5E). Similarly, in the surgical group, the complete remission rates were 93.8% and 22.2%, corresponding to subjects with final BMI < 26 kg/m2 and ≥ 26 kg/m2, respectively. However, the TWL, FAI, MaxOV, and the percentage of patients with over 12 follicles showed no significant difference between the subjects who achieved complete remission and those who did not (Fig. 5F-5J).

Figure 4.

ROC curve for the drug and surgical groups. (A) Drug group, (B) Surgical group. As a reference, a curve with an AUC of 0.5 has been plotted (dashed line). Abbreviations: AUC, area under the curve; ROC, receiver operating characteristic.

Table 4.

Characteristics of subjects with different BMI stratification at baseline and endpoint with drug or surgery treatment

	Drug group				Surgery group
	BMI < 27.5 (n = 5)		BMI ≥ 27.5 (n = 35)		BMI < 26 (n = 32)		BMI ≥ 26 (n = 9)
	Baseline	Endpoint	Baseline	Endpoint	Baseline	Endpoint	Baseline	Endpoint
Complete remission rate (%)		100		2.9		93.8		22.2
Age (yr)	24.8 ± 3.0		28.0 ± 0.8		28.5 ± 0.9		29.1 ± 1.4
Weight (kg)	78.2 ± 4.6	69.2 ± 3.2*	87.2 ± 1.9	83.7 ± 1.8*	95.1 ± 2.7	62.0 ± 1.3***	108.8 ± 3.0##	77.0 ± 1.7***
BMI (kg/m2)	29.3 (28.0–30.4)	25.9 (25.4–27.1)*	31.3 (30.1–34.5)#	30.4 (29.1–32.4) *	34.7 (31.8–37.6)	22.9 (21.5–24.4) ***	40.4 (37.1–46.6)###	28.7 (27.5–28.9) ***
TWL (%)		9.3 (6.7-16.4)		4.5(2.4-6.0)		34.2 (28.7-39.2)		29.1 (24.9-40.0)
Waist (cm)	95.0 (93.5–100.5)	92.0 (82.5–95.0) *	99.0 (95.0–105.0)	98.0 (92.0–103.0) *	109.0 (102.3–118.6.0)	81.5 (75.3–85.8) ***	115.0 (111.5–134.0)#	92.0 (90.0–95.0) ***
Hip (cm)	102.0 (90.0–112.0)	102.0 (90.5–112.0)	106.0 (102.0–112.0)	108.0 (104.0–110.0) *	112.0 (106.5–121.0)	93.5 (88.5–100.0) ***	113.0 (112.0–136.0)	102.0 (98.5–103.5) *
W/H	1.0 ± 0.05	0.9 ± 0.06	0.9 ± 0.01	0.9 ± 0.09	1.0 ± 0.01	0.9 ± 0.01	1.0 ± 0.01	0.90 ± 0.02
Total testosterone(nmol/L)	2.2 (1.4–3.2)	2.4 (2.1–2.9)	1.9 (1.4–2.3)	1.8 (1.4–2.4) *	1.9 (1.4–2.8)	1.3 (1.1–1.8) ***	2.1 (1.2–2.6)	1.4 (1.0–2.2) ***
SHBG (nmol/L)	29.1 (13.4–37.2)	65.1 (24.8–71.8) *	21.9 (16.1–29.7)	22.0 (18.3–33.6)	15.7 (11.1–19.5)	58.1 (44.8–68.6) ***	14.2 (10.6–21.1)	74.7 (60.3–119.6) ***
FAI (%)	9.1 (6.5–11.8)	4.4 (3.0–10.7) **	9.5 (6.5–11.8)	7.2 (4.6–11.6)*	12.8 (9.3–20.3)	2.8 (2.0–4.0)***	14.2 (5.8–26.9)	1.6 (1.1–3.2)***
MaxOV (ml)	10.2 (6.1–11.2)	7.6 (5.1–8.7)**	9.2 (5.4–11.8)	6.1 (5.0–9.6)**	10.6 (8.5–16.8)	8.1 (5.7–11.8)*	9.9 (7.7–12.1)	10.7 (7.6–11.7)
Follicle number ≥ 12 (%)	60	20	97.1	97.1	93.8	56.3***	77.8	77.8
FPG (mmol/L)	5.1 (5.0–5.4)	5.3 (4.2–5.7)	5.2 (4.9–5.7)	5.2 (4.9–5.6)	5.3 (4.9–6.9)	4.4 (4.2–4.9)*	6.3 (4.7–6.9)	4.4 (3.7–4.8)*
Fasting insulin (μIU/ml)	23.2 (22.5–30.3)	25.7 (17.4–47.2)	22.8 (17.8–35.6)	24.4 (14.1–47.6)	23.4(18.0–30.9)	6.7 (5.1–9.0)***	32.2 (20.9–88.3)#	5.7 (2.2–9.4)***
HOMA-IR	5.2 (5.0–7.2)	6.1 (3.3–12.0)	5.5 (4.1–9.0)	4.6 (2.6–9.7)	6.4 (4.3–8.9)	1.4 (1.0–1.9)***	8.2 (5.5–25.6)	1.3 (0.4–1.8)***
HbAlc (%)	5.7 (5.3–5.9)	5.3 (5.1–5.6)	5.7 (5.5–6.0)	5.5 (3.2–9.5)	5.8 (5.5–6.4)	5.2 (4.9–5.5)*	5.8 (5.4–6.4)	5.0 (4.8–5.3)*
Triglycerides (mmol/L)	1.3 (1.1–1.9)	1.4 (1.2–2.3)	1.5 (0.8–2.0)	1.4 (0.8–1.9)	1.5 (1.2–2.0)	0.8 (0.6–1.0)*	1.55 (1.1–2.7)	0.9 (0.7–1.1)*
Total cholesterol (mmol/L)	5.5 (3.8–5.7)	6.0 (5.5–6.2)	4.9 (4.3–5.5)	5.0 (4.2–5.9)	5.2 (4.2–5.5)	4.8 (4.3–5.5)*	4.5 (4.4–5.4)	4.1 (3.7–4.9)
HDL-C (mmol/L)	1.1 (1.0–1.4)	1.7 (1.7–1.9)*	1.1 (1.2–1.5)	1.2 (1.1–1.4)	1.1 (0.9–1.3)	1.2 (1.0–1.5)	1.0 (0.8–1.2)	1.3 (1.2–1.4)
LDL-C (mmol/L)	3.2 ± 0.5	3.3 ± 0.3	3.1 ± 0.1	3.1 ± 0.2	3.1 ± 0.2	2.8 ± 0.2*	3.1 ± 0.1	2.3 ± 0.2*

Data are presented as median (interquartile range) or mean ± standard error of the mean.

Abbreviations: BMI, body mass index; DHEAS, dehydroepiandrosterone sulfate; FAI, free androgen index; FPG, fasting plasma glucose; HbAlc, glycosylated hemoglobin A1c; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostatic model of insulin resistance; LDL-C, low-density lipoprotein cholesterol; MaxOV, maximum ovarian volume; SHBG, sex hormone–binding globulin; TC, total cholesterol; TG, triglycerides; W/H, ratio of waist circumference to hip circumference.

*P < 0.05,

** P < 0.01,

*** P < 0.001, endpoint vs baseline in the same subgroup;

# P < 0.05,

## P < 0.01,

### P < 0.001, baseline vs baseline in the same treatment group.

Figure 5.

Subgroup analysis based on the cutoff values. Comparison of patients over the cutoff value of BMI in the drug group for (A) complete remission rate, (B) TWL, (C) FAI, (D) MaxOV, and (E) the percentage of patients with over 12 follicles, and in the surgical group for (F) complete remission rate, (G) TWL, (H) FAI, (I) MaxOV, and (J) the percentage of patients with over 12 follicles. Abbreviations: BMI, body mass index; FAI, free androgen index; MaxOV, maximum ovarian volume; TWL, total weight loss.

Discussion

The complex interactions of multiple genetic and environmental factors account for the complex traits of PCOS, which adversely affect further exploration of the potential mechanisms underlying the syndrome and its effective treatment options (32-34). Currently, PCOS cannot be reversed by available therapies, except for sustained and steady weight loss in some patients with obesity (15). In this prospective study, we assessed the effects of first-line pharmacologic therapy and bariatric surgical treatment on the endocrine-metabolic profiles and clinical manifestations in women with obesity and PCOS. No previous studies, to our knowledge, have made such a comparison. This study showed that surgical treatment is far more effective than drug therapy. The benefits of bariatric surgery for PCOS include steady and reliable weight loss, recovery from irregular menstrual cycles, and improvement in both hormonal and metabolic profiles. In light of the superior efficacy of bariatric surgery, surgical treatment should be prioritized for treating patients with obesity and PCOS.

Although the interactions and internal associations between obesity and PCOS remain controversial (35-37), this study further demonstrated a strong correlation between BMI and remission in subjects with obesity and PCOS. Additionally, the cutoff values in this study may provide guidance for surgical and drug treatment in patients with obesity and PCOS. That is, to ensure efficacy, patients in the drug group are advised to lose weight to have a BMI below 27.5, and in the surgical group, the BMI needs to be less than 26. There is a difference of 1.5 between these 2 cutoff values, which may represent the ability of the drug’s effect to compensate for the increase in BMI. Previous studies have indicated that even a minor weight loss of 5% improves reproductive and metabolic outcomes (38-40). Since these studies on lifestyle intervention or drug treatment only achieved slight weight loss (38, 41), their conclusion tended to emphasize the importance of percentage of weight loss on the remission of PCOS and the association between final BMI and the remission of PCOS was not explored. Similarly, in this trial, we also observed that almost all patients with a weight loss of more than 5% achieved complete remission. However, patients who experienced a weight loss of 26% in the surgical group still failed to achieve remission. Previous studies on bariatric surgery did not investigate the relationship between biochemical or physical parameters and clinical outcomes of PCOS (24, 25, 42, 43). For the first time, we conducted logistics regression analyses and comprehensively assessed the correlation between each parameter and the outcome of PCOS, and our results imply that for patients with PCOS and obesity, the final BMI after weight loss is far more important than the other factors.

There was a significant difference in baseline BMI between patients with complete remission and nonremission in the surgery group, but not in the drug group. For those patients with a higher BMI at baseline, the gap between their baseline and target BMIs was greater, making it much tougher to reach the target. However, patients with a higher baseline BMI are also able to achieve complete remission if they try to get the target. In our study, 1 patient in the surgical group with a baseline BMI of 54.9 kg/m2 lost weight to final BMI of 24.8 kg/m2 and achieved complete remission at the endpoint. Thus, the most important factor of PCOS remission is the final BMI. For patients with a higher baseline, a more aggressive operation is recommended to achieve their desired weight loss. Furthermore, if a physician is concerned that a patient’s baseline BMI is too high to reach the target BMI, other treatments, including dietary adjustment, exercise, and medication should be supplemented as early as possible for greater weight loss.

Hyperandrogenism and insulin resistance commonly coexist in PCOS patients, even in those who are not obese, leading to PCOM and aggravation of PCOS (32, 44). Pharmacological treatment aiming at hyperandrogenism and insulin resistance was reported to improve menstrual and ovulatory function (41, 45). However, in this trial, we conducted a detailed subgroup analysis based on the endpoint and discovered that the endocrine-metabolic profiles were similar regardless of remission in both drug and surgery groups, except for the final BMI. Furthermore, from the univariate and multivariate logistic regression analyses, we found that SHBG, FAI, MaxOV, and HOMA-IR may all play roles in the remission of PCOS, only the final BMI is the true determinant, indicating that insulin resistance, androgen excess, and ovarian morphology may be consequences of obesity. Thus, excess energy may be the root cause of obesity combined with PCOS. Our study recommends that the treatment target of PCOS with obesity should focus on weight loss.

Notably, a single subject in the surgical group who lost too much weight, with a BMI dropping below 20, failed to achieve complete remission. Since her serum androgen level and ovarian morphology detected by ultrasound were normal, she could no longer be diagnosed with PCOS. She was the only patient who did not conform to both the criteria for complete remission and the diagnostic criteria for PCOS. Therefore, patients with obesity and PCOS undergoing bariatric surgery should be monitored for potential amenorrhea due to excessive weight loss.

Additionally, we observed that the improvement of PCOM was not as pronounced as the other biochemical parameters in response to treatments. Several studies also found that the ovarian morphology of patients with PCOS did not exhibit significant improvement or only showed slight improvement after lifestyle intervention or drug therapy, while the other parameters of PCOS ameliorated significantly (46-48). The underlying mechanism is still unknown, and the possible reasons we speculate are as follows: first, the change of ovarian morphology is more difficult than the change of biochemical indexes of endocrine and metabolism, and the ovarian enlargement as well as the number of small antral follicular increased shown as PCOM are relatively hard to reverse. Second, PCOM is not the same as PCOS since PCOM occurs in many adolescents and healthy individuals without PCOS, indicating that there may not be clear causal relationship between PCOM and PCOS.

The very different patterns of change in menstrual cycles between groups also aroused our interest. In the drug group, the menstruation of most patients became irregular again after the discontinuation of oral contraception, which indicates that a single medication, namely metformin, was unable to maintain efficacy. Although previous studies demonstrated that metformin alone was effective in inducing the ovulation rate in women with PCOS, this effect was observed in nonobese subjects (49-52). In patients with obesity and PCOS of this study, the efficacy of metformin alone seems to be limited, with the complete remission rate being below 20%. However, in the surgical group, the menstruation of the subjects gradually returned to a regular pattern, with the most significant improvement observed in the seventh month. This is the first study to collect data each month regarding the number of subjects menstruating after bariatric surgery. Our data showed that although 70% of patients had a withdrawal bleeding within 1 week immediately after bariatric surgery, this percentage dropped dramatically in the second month. It indicated that the bleeding immediately after bariatric surgery is likely a withdrawal bleeding, rather than an indicator of menstrual improvement. The actual remission rate was no more than 30% from the second month, and then increased monthly until it reached its maximum of approximately 80% in the seventh month.

Bariatric surgery in the treatment for obesity complicated with PCOS is still underemphasized in the existing guidelines. As stated by the Guidelines for Diagnosis and Treatment of Obesity and Type 2 Diabetes Mellitus from the Chinese Society of Metabolic and Bariatric Surgery (CSMBS) (53), bariatric surgery may be considered for patients (27.5 kg/m2  ≤ BMI < 32.5 kg/m2) whose obesity cannot be readily controlled with lifestyle changes and medication, and if the candidate has at least 2 components of metabolic syndrome. In addition, for men with waist circumference ≥ 90 cm and women with waist circumference ≥ 85 cm, the priority of surgical treatment can be increased as appropriate. The clinical practice guidelines published by the European Association for Endoscopic Surgery (EAES) and American Association of Clinical Endocrinologists (AACE) (54, 55), respectively, both suggest that bariatric surgery should be considered for patients (30 kg/m2 ≤ BMI < 35 kg/m2) with poorly controlled type 2 diabetes and/or arterial hypertension despite optimal medical therapy. However, PCOS is not a recommended indication for bariatric surgery in patients with relatively mild obesity. Therefore, we recommend that bariatric surgery may be considered for patients with obesity (27.5 kg/m2 ≤ BMI < 32.5 kg/m2 for Asian and 30 kg/m2 ≤ BMI < 35 kg/m2 for American and European) and PCOS who have poor response to lifestyle intervention and medical therapy.

The major limitation of our study is that it was originally designed as a randomized trial but failed to be conducted as such because surgical treatment is invasive and differs greatly from drug treatment; thus, patients showed a strong personal preference in the choice of treatment. In addition, the relatively small sample size and inclusion of a single racial/ethnic population are also limitations. Furthermore, it is possible that statistical power was inadequate for some comparisons with the small number of patients in the subgroup analysis. Additionally, a longer follow-up period is needed to assess the long-term outcomes and complications of bariatric surgery.

Conclusion

Our study confirms that BMI, rather than TWL, is highly correlated with PCOS remission. Compared with drug therapy, bariatric surgery is far more effective; thus, we suggest that bariatric surgery should be considered as the first-line treatment for patients with PCOS and obesity. Furthermore, our findings recommend target BMI for these patients, eg, 27.5 kg/m2 for patients taking medicine and 26.0 kg/m2 for those undergoing bariatric surgery.