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


To the best of our knowledge, this is the first randomized, controlled trial comparing the clinical efficacy of the GLP-1RA EQW and SGLT2i DAPA, in combination and alone, DAPA/MET, and the weight loss medication PHEN/TPM on metabolic, anthropometric, and hormonal parameters in obese nondiabetic women with PCOS. Our data showed that dual treatment with EQW and DAPA was clinically superior compared to the other therapies in the management of this patient population. Greater improvements in participants with EQW plus DAPA may be explained in part by their different, and potentially complementary, mechanisms of action[30] and confirm other studies showing that combining these 2 agents may exert stronger beneficial effects than each drug alone.[14,31,32]

With regard to the FBG, our results showed that all therapies consistently lowered blood sugar. In addition, all treatments resulted in improved insulin sensitivity, estimated both by the HOMA-IR and the SIOGTT. Modest weight loss is well known to reduce the risk of future diabetes in individuals with prediabetes. Although mechanisms are not completely understood, weight loss leads to improvement in insulin resistance, which can reduce insulin secretory demands and potentially ameliorate glucotoxic and lipotoxic effects on pancreatic β cells. Therefore, it cannot be excluded that the modest reduction in BW contributed in part to the improvement in insulin sensitivity in all groups. Participants on all therapies showed significantly improved glucose excursion during an OGTT after 24 weeks of treatment, with individuals on EQW/DAPA showing a significantly greater drop in MBG levels compared with DAPA, DAPA/MET, and PHEN/TPM. While no consistent changes in corrected first-phase insulin secretion with treatment were demonstrable, a significant increase was seen in the EQW and EQW/DAPA groups as compared with DAPA, DAPA/MET, and PHEN/TRM. Overall carbohydrate metabolism as estimated by patients' IS-SI improved steadily over 24 weeks in all groups, with EQW and EQW/DAPA showing significantly greater improvement than the other groups, indicative of improved β-cell function. Whereas PHEN/TPM results in consistent significant changes in BMI and WC only, EQW/DAPA and EQW alone resulted in significant improvements in insulin secretion as well as MBG over an OGTT. There is available evidence that GLP-1RAs act directly on pancreatic β cells to stimulate insulin secretion.[7] This finding confirms the beneficial effect of GLP-1 agonists on β-cell function in this obese, prediabetic population.

Combination EQW/DAPA and PHEN/TPM resulted in the greatest loss of weight, TBF with DXA, and WC compared with DAPA alone or the DAPA/MET combination. Controlled PHEN/TPM, as an adjunct to lifestyle modification, has previously shown significant weight loss compared with placebo in overweight and obese individuals with prediabetes or cardiometabolic disease.[33,34] In agreement with other trials in T2DM patients, the concomitant addition of EQW and DAPA resulted in significantly greater weight loss than either individual therapy, with a larger proportion of patients experiencing weight loss of at least 5%.[31,32] GLP-1 RAs not only have a significant effect on weight loss in overweight T2DM patients but also in nondiabetic overweight individuals, reducing subcutaneous fat areas in particular.[35] In this study, the values of BMI in this population positively predicted whole-body fat percentage by DXA among obese women with PCOS.

Body composition has a significant influence on metabolic activity. At-risk obese women with PCOS have a higher WC, WHR, and WHtR than their obese counterparts. The improvement in central adiposity was most dramatic with EQW/DAPA and PHEN/TPM therapy, where mean WC and mean WHtR measurements significantly decreased over 24 weeks, whereas central adiposity was not as reduced at 24 weeks on DAPA/MET. This finding is in agreement with prior studies in nondiabetic participants with PCOS that showed that MET did not consistently have an effect on decreasing abdominal adiposity.[36] There is now considerable evidence that cardiometabolic complications from adult obesity are closely related to body fat distribution, with a sex dimorphism. Imaging techniques with DXA enable us to evaluate patterns of fat deposition, including levels of abdominal adiposity based on the AGR.[18,19] Similar to WHR, where body shape is described using a ratio, the body trunk/torso and leg fat are used instead of WC and hip circumferences. It has been shown that an increase in the android fat distribution (with values of AGR > 1) is associated with conditions such as dyslipidemia and insulin resistance, as well as other cardiovascular risk factors such as impaired glucose tolerance, hypercholesterolemia, hypertriglyceridemia, and hypertension.[19] Using DXA analyses, a decrease in the AGR was found with all treatments, with EQW/DAPA showing the most significant improvement in AGR as compared with other DAPA groups. Furthermore, a lower TLR measured by DXA generally means a lower risk for certain health issues, whereas a high TLR showed a strong association to diabetes and mortality that was independent of total and regional fat distributions.[20] We found that TLR was significantly lower with EQW/DAPA compared with DAPA and DAPA/MET treatment. The most improvement in body fat distribution was observed with EQW/DAPA therapy (redistribution of central fat) as reported for GLP1 agonist therapy previously.[10,35] We further examined the relationship between insulin sensitivity and AGR in these obese hyperandrogenic women. Fasting insulin resistance as measured with HOMA-IR was significantly positively associated with increased abdominal adiposity (AGR). There was an even stronger negative correlation between OGTT-derived insulin sensitivity (SIOGTT) and central adiposity as defined by the AGR. Our findings corroborate the observations of previous researchers showing that insulin sensitivity was negatively correlated with AGR in obese children and adolescents.[19]

Levels of CHOL, HDL-C, LDL-C, and TRG/HDL ratio were not consistently altered after 24 weeks with any drug treatment, although TRG concentrations were significantly reduced with EQW/DAPA compared to PHEN/TPM. Recently, a study demonstrated beneficial effects of a combination of the SGLT2i DAPA with the GLP1-RA EQW with greater reductions in TRG concentrations vs either EQW or DAPA in participants with T2DM inadequately controlled on MET therapy.[31]

Hyperandrogenemia is another potential adverse cardiometabolic feature of PCOS. The reduced secretion of SHBG in participants with PCOS is believed to result from the combined inhibitory influence of portal insulin levels,[37] proinflammatory mediators mostly secreted by liver adipose tissue,[38] and excessive circulating androgens[39] in association with insulin resistance, visceral adiposity, and hyperandrogenism. Even a small weight loss with PCOS is associated with improvements in hormonal levels and beneficial effects on ovulation and menstrual frequency.[5] Significant reductions in TT levels and FAI were seen in all treatment groups consistent with weight loss observed with all drugs in this trial. Although not directly evaluated, 4 participants conceived during the trial demonstrating return of ovulatory function.

A statistically significant reduction both in SBP and DBP was observed in all groups consistent with weight loss with all drug treatments in this study. Weight loss has been postulated as a contributing mechanism for lowering BP.

A major weakness of the present study is the complex study design and relatively small size of the treatment groups studied. Additionally, the small sample size in some groups was due to lack of completion of the whole study protocol as caused by drug side effects (in particular with the weight loss drug). Additionally, concentrations of TT were obtained by second-generation immunoassay so they may not be as sensitive or specific to detect differences between drug treatment groups as if they had been measured using liquid chromatography–tandem mass spectrometry assays. Another inadequacy of this study is the lack of a placebo-only arm. Because we were interested in the effects of different treatment regimens for obese women with PCOS, this study was designed to provide all participants with drug therapy. Limitations of the present study further include the absence of gold-standard measures of insulin sensitivity such as the euglycemic-hyperinsulinemic clamp. While the hyperinsulinemic euglycemic clamp is the gold-standard technique to measure whole-body insulin sensitivity, it is a complex and time-consuming procedure requiring numerous frequent blood samples. The OGTT-based indices are validated measures that have been widely used in previous studies. In this study, we used the Matsuda index to measure insulin sensitivity, which, as previously mentioned, has high reproducibility and is strongly correlated with the clamp. Last, serial assessments were made over only 24 weeks of treatment.

Our study has several strengths, including the homogeneity of the population of obese women with PCOS (National Institutes of Health criteria) and inclusion of a weight loss–only medication. A point of concern was the high rate of dropouts in the weight loss–only medication PHEN/TPM group compared to EQW and EQW plus DAPA, where compliance was very high. The percentages of participants that completed the trial were EQW 87%; EQW/DAPA 91%; DAPA 74%; DAPA/MET 73%, and PHEN/TPM-64%. Attempts were made to limit and assess the effects of dropouts.

Most of the medications were well tolerated. As previously found with EQW, the most frequently reported side effects of EQW and EQW/DAPA were injection site reactions and gastrointestinal, with nausea the most common AE. The rates of clinically diagnosed urinary and genital tract infections were higher in participants receiving DAPA, as expected. Combination EQW/DAPA therapy did not lead to an increase in the incidence of infections. The main side effect of MET added to DAPA was stomach upset. Adherence to drug was lowest in the PHEN/TPM group, whose members experienced common AEs such as insomnia, light-headedness, fatigue, kidney stones, and headache. PHEN/TPM has been reported to have primarily neurologic side effects such as dizziness and mood changes.[33,34] The greatest loss rate was on PHEN/TPM because of unpleasant side effects.

The potential therapeutic effects of EQW and DAPA (alone or in combination) or combination DAPA/MET have never been evaluated in the subpopulation of obese women with PCOS. When administered in combination, the complementary mechanisms of action of a SGLT2i (increasing urinary caloric loss) and a GLP-1RA (suppressing appetite and caloric intake) may represent a novel treatment approach for obesity, with the combination lowering BW more than either component alone.[30] In conclusion, the present results show that combined treatment with EQW and DAPA is superior to EQW and DAPA alone, DAPA/MET, and PHEN/TPM in terms of clinical and metabolic benefits in this patient population. MET is currently the only recommended therapy for the management of individuals with prediabetes and PCOS. These findings, together with the convenience of once-daily oral dosing and once-weekly injection, support further large, double-blinded, randomized trials of longer duration in obese prediabetic populations to assess dual GLP1-RA and SGLT2i therapy for BW loss and cardiometabolic disease prevention.