Exenatide, Dapagliflozin, or Phentermine/Topiramate Differentially Affect Metabolic Profiles in Polycystic Ovary Syndrome

Karen E. Elkind-Hirsch; N. Chappell; Ericka Seidemann; John Storment; Drake Bellanger


J Clin Endocrinol Metab. 2021;106(10):3019-3033. 

In This Article

Materials and Methods


A randomized, single-blind, comparative 24-week study in women with PCOS to compare the therapeutic effects of 5 drug treatments on metabolic, anthropometric, and hormonal parameters was conducted in the Woman's Hospital Endocrine and Weight Loss Clinic. All participants were nondiabetic, obese, premenopausal women, aged 18 to 45 years and diagnosed with PCOS. PCOS was defined according to modified National Institutes of Health 1990 criteria.[11] Eligible participants were required to have the combination of irregular periods (cycle length outside 21–35 days or < 8 cycles/year) together with biochemical evidence of hyperandrogenism (total testosterone [TT] > 50 ng/dL, or free androgen index [FAI] > 3.87)[16] and exclusion of known disorders for bleeding irregularities and androgen excess. The main inclusion criteria also included obese class I, II, and III (BMI > 30 < 45) and agreement to use effective contraception consistently during therapy. Adequate contraception was defined as an intrauterine device, tubal sterilization, or combination of 2 barrier methods with 1 being a male condom. The study excluded diabetic individuals, current smokers, suspected pregnancy, desiring pregnancy, or injectable hormonal contraceptive use within 6 months and use of oral contraceptives, other steroid hormones, drugs that affect gastrointestinal motility or carbohydrate metabolism, and/or antiobesity drugs within 3 months prior to study entry. The institutional review board of the Woman's Hospital Foundation approved this study.

A total of 130 participants provided written informed consent to enroll in the study. Following written consent, all study participants underwent clinical, anthropometrical, and biochemical assessments at baseline, 12 weeks, and 24 weeks. Eleven individuals were excluded from the study after consenting because they were found to be ineligible based on findings at their laboratory screening. All participants were screened at the initial lab evaluation with a complete metabolic profile and calculated estimated glomerular filtration rate, thyrotropin (TSH), prolactin, TT and FAI, qualitative β human chorionic gonadotropin level, and history of menstrual cycles, and for study inclusion and use of adequate contraception. A total of 119 obese nondiabetic women with PCOS who met the criteria were randomly assigned by computer-generated block allocation sequence to 1 of 5 treatment groups. To ensure that participants remained anonymous, all study patients were assigned an individual study identifier.

Anthropometric Measurements

Trained personnel using standardized protocols at the baseline and follow-up examination obtained anthropometric measurements. At the initial clinic visit, a full physical examination was performed with BW, height, WC, and hip size measured and vital signs (BP, respiration, and temperature) collected. BW was measured to the nearest 0.1 kg using a calibrated digital scale with participants in light clothing and no shoes. Height was measured to the nearest centimeter. BMI was calculated as body weight in kilograms divided by height in meters squared. WC was measured at the narrowest level midway between the rib cage and the iliac crest and hip circumference measured at the widest level over the buttocks while the participants were in the standing position using a flexible measuring tape. The waist-to-hip ratio (WHR) and waist-to-height ratio (WHtR) were calculated and used to estimate abdominal adiposity.[16] BP was measured using a sphygmomanometer and measured in millimeters of mercury (mm Hg).

Laboratory Measurements

Oral glucose tolerance tests (OGTTs) were carried out (starting at 6:30–9:30 AM) after an 8- to 12-hour overnight fast. After the collection of a baseline blood sample, a 75-g oral glucose load was administered; additional blood samples were drawn 30, 60, and 120 minutes later for analysis of glucose and insulin levels. Fasting baseline blood specimens were also used for measures of an androgen profile (TT, dehydroepiandrosterone sulfate [DHEA-S], sex hormone– binding globulin [SHBG], FAI), a lipid panel (total cholesterol [CHOL], high-density lipoprotein [HDL-C], low-density lipoprotein [LDL-C], and triglycerides [TRG]). Glucose tolerance was defined as normal, impaired, or diabetic according to the criteria of the American Diabetes Association.[17]

Plasma glucose (PG) levels were determined with a glucose analyzer using the glucose oxidase method (Glucose Reagent Kit, Bayer). Serum insulin was determined in all samples in duplicate by microparticle enzyme immunoassay (Abbott AxSYM System, Abbott Laboratories). Levels of CHOL, HDL-C, and TRG were determined using standard enzymatic colorimetric assays on an automated clinical chemistry analyzer, whereas LDL-C was calculated according to the Friedewald equation. Electrolytes, serum creatinine, and liver enzymes were measured using a standard automated kinetic enzymatic assay. Circulating levels of TSH, prolactin, β human chorionic gonadotropin, SHBG, and DHEA-S were analyzed using competitive binding immunoenzymatic assays with direct chemiluminometric technology (Beckman Coulter Access 2). TT concentration was measured with an automated chemiluminescent microparticle immunoassay, the Architect second-generation testosterone assay (Abbott Diagnostics) with 5 ng/dL as the minimum detectable concentration of TT. The intra-assay and interassay coefficients of variation were estimated at less than 7% and 11%, respectively, over the sample concentration range for all assays.

Assessment of Body Composition

Total and regional body composition was determined using dual-energy x-ray absorptiometry (DXA) (Hologic Discovery A model, Hologic Inc) at the start and completion of the study trial. Participants in hospital gowns were positioned in the supine position on the DXA table, and instructed to keep their arms separated from their trunk, hands placed flat on the table, palms facing down, away from their thighs adjacent to the side of the body, and their legs separated from one another. Average scanning time was approximately 8 minutes. With this method, body composition consisting of body fat (in kilograms) and lean (in kilograms) soft tissue was estimated. Data were analyzed with Hologic QDR Software for Windows (version 12.5), which integrates whole-body measurement and standard body regions, such as the trunk, arms, and legs, delineated by specific anatomical landmarks. For each region of the whole body (head, trunk, arms, and legs), fat and lean body mass were determined, expressed as mass (in grams). The total body fat mass percentage (TBF%) was calculated by dividing the weight of the total fat mass (TFM; kg) by body weight. Lean mass (in kilograms) is the muscle tissue, skeletal tissue, and water in the body. Ratios of regional DXA mass compartments (like android to gynoid fat mass) have also been used to stratify risk for metabolic disease.[18] The android region is typically defined as the region between the last thoracic rib and the upper part of iliac wings. Gynoid fat deposition was assessed by lower limb fat percentage and includes the gluteofemoral region. Android-to-gynoid fat ratio (AGR) was determined by using the fat percentage in the lower limbs and in the abdominal region.[19] Trunk fat was the amount of fat measured by DXA from below the neck to the pelvis excluding limbs. Trunk-to-leg fat ratio (TLR) was calculated with DXA data as trunk fat mass divided by the sum of left and right leg fat mass, multiplied by 100. A ratio of truncal-to-leg fat also has been associated with cardiometabolic disease–related outcomes in adults.[20] This novel body shape measure provides additional information regarding central adiposity to better stratify individuals at risk for diabetes and mortality, even among those with normal BMI.[20]

Trial Design

Randomized, Single-blind, Parallel, Prospective Study Trial. All participants were randomly assigned to 1 of 5 investigation product treatment groups: EQW; 2 mg weekly, DAPA; 10 mg daily, EQW/DAPA; (2 mg weekly/10 mg daily), combined DAPA/MET (DAPA 10 mg/MET 2000 mg XR daily), or the weight loss medication, PHEN/TPM (PHEN 7.5 mg/TPM 46 mg ER daily) and treated for 24 weeks. Exenatide once weekly (Bydureon), dapagliflozin (Farxiga), and combination dapagliflozin/metformin XR (Xigduo) were provided by AstraZeneca Pharmaceuticals, and phentermine/topiramate XR (Qysymia) was provided by Vivus Inc. All participants were assigned to 1 of these 5 groups based on computer-generated random numbers using a block randomization method. If participants dropped out of the study, they were replaced so that an equal number of participants completed the trial (up to 119 participants started medication with the goal of 20 participants per treatment group completing the trial). The pharmacy "single-blinded" all the treatment arms to the primary investigators by filling color-coded bags (A, B, C, D, E) with 1 of 5 medications: A) EQW, B) DAPA, C) EQW/DAPA, D) DAPA/MET, and E) PHEN/TPM and dispensing open-label medications to study participants. All investigators were blinded to drug treatment. The research nurse instructed study participants on how to take their medication and how to inject medications (EQW and EQW/DAPA group) when required. All participants were dispensed medication and home pregnancy tests for 3 months. They were required to perform monthly home pregnancy tests until the midstudy visit. The participants were instructed to bring back their bag of medication at each visit.

All participants received the same counseling concerning the benefits of lifestyle modification through diet and exercise. The study participants were also encouraged to increase daily exercise (eg, walking, using stairs), although this was not formally assessed. The participants received further encouragement during follow-up phone calls to adhere to the regimen. Side effects of the treatment and reason for any withdrawals from the study were recorded.

Participants were seen at 10 to 12 weeks for clinical and laboratory evaluation including a pregnancy test. BW, BMI, waist (WC) and hip circumference, and BP were measured at every visit. All side effects of treatments and reason for any withdrawal from the study were recorded. Adherence to medical treatment was also evaluated. Participants were dispensed another 3 months of medication and home pregnancy tests and required to perform monthly home pregnancy tests until the final visit.

At study completion (24 weeks of treatment), all clinical and laboratory tests were repeated. All anthropometric parameters and physical measurements, including vital signs and DXA, were again performed and calculations repeated for posttreatment effects. During the whole study period, compliance with the treatment was documented. Questioning regarding the occurrence of adverse events (AEs) and use of concomitant medication took place throughout the trial. Adherence to treatment and AEs and secondary events were recorded from baseline and throughout the follow-up visits. Each AE was evaluated by all the researchers and when necessary (eg, pregnancy) immediately reported to the institutional review board of the Woman's Hospital, the company, and to the US Food and Drug Administration for follow-up and recommendations.

Assessment of Insulin Sensitivity and Secretion

To evaluate insulin sensitivity and secretion, indexes derived from either fasting or OGTT-stimulated concentrations of glucose and insulin were used:

The homeostatic model assessment of insulin resistance index (HOMA-IR) was calculated using the equation: fasting insulin concentration (μIU/mL) × fasting glucose concentration (mmol/L) 22.5.[21]

Matsuda's insulin sensitivity index (SIOGTT) was calculated according to the formula:

10 000/√ [fasting glucose (mg/dL) × fasting insulin (μU/L) × [MPG × MSI during OGTT] where MPG (mg/dL) is mean PG OGTT, and MSI (μIU/mL) is mean serum insulin during OGTT.[22]

Early pancreatic β-cell response was estimated as the insulinogenic index (IGI) derived from the ratio of the increment of insulin to that of glucose 30 minutes after a glucose load (insulin 30 minutes − insulin 0 minutes/glucose 30 minutes − glucose 0 minutes) corrected for by the relative level of insulin resistance (IGI/HOMA-IR).[23,24]

An estimation of β-cell compensatory function, the insulin secretion–sensitivity index (IS-SI) was derived by applying the concept of the disposition index to measurements obtained during the 2-hour OGTT and calculated as the index of insulin secretion factored by insulin sensitivity (ΔINS/ΔPG 30 × Matsuda SIOGTT) during the OGTT.[25,26]