Effects of Empagliflozin on Metabolic Parameters in Polycystic Ovary Syndrome

A Randomized Controlled Study

Zeeshan Javed; Maria Papageorgiou; Harshal Deshmukh; Alan S. Rigby; Unaiza Qamar; Jehangir Abbas; Amer Y. Khan; Eric S. Kilpatrick; Stephen L. Atkin; Thozhukat Sathyapalan


Clin Endocrinol. 2019;90(6):805-813. 

In This Article


In this first study on the comparative effects of the SGLT2 inhibitor, empagliflozin and metformin in overweight and obese women with PCOS, we demonstrated that treatment with empagliflozin over a 12-week period, resulted in significant reductions in weight, BMI, waist and hip circumference, total body fat mass and BMR, compared to treatment with metformin, but did not differentially affect hormonal or metabolic parameters.

Obesity is prevalent among women with PCOS and has been associated directly and/or indirectly with negative metabolic, cardiovascular, endocrine, reproductive and mental health outcomes.[1,2] Weight reduction exerts positive effects on PCOS-related outcomes; therefore, it is a primary goal of the management of the condition.[3] In the present study, women with PCOS in the empagliflozin group had a mean weight loss of 1.5 kg, which is similar to the weight loss reported in previous short-term trials.[12] Mechanistically, initial weight loss in empagliflozin studies is attributed to the calorie loss (approximately 200-300 kcal/d) associated with glucose excretion, but also to the mild diuretic effects of the drug.[9,12] Conversely, the steady-state weight loss associated with SGLT2 inhibitors treatment may result from fat loss.[22,23] In animal models, SGTL2 inhibitors have been shown to cause reduction in body weight and fat mass by enhancing lipolysis, fatty acid oxidation and adipose tissue browning.[14,24] These findings coincide with the alterations seen in substrate utilization from carbohydrates to lipids and potentially, ketone bodies.[9,25] Reductions in other measures of adiposity including waist and hip circumference, visceral and subcutaneous fat depots or indices, which may better reflect risk for metabolic disturbances and cardiovascular disease, have also been demonstrated in patients with type 2 diabetes following treatment with SGLT2 inhibitors,[22,23] with these findings being in agreement with the improvements in waist and hip circumference seen in our women with PCOS assigned to receive empagliflozin.

In the present study, the metformin group experienced modest increases in body weight. Studies on the effect of metformin on body weight in women with PCOS have yielded mixed results.[8,26,27] While some studies have suggested that metformin therapy may result in weight reduction, some randomized controlled trials have failed to confirm this. For example, a large, randomized, double-blind, placebo-controlled trial evaluated the combined effects of lifestyle modification and metformin (850 mg twice daily), by studying 143 anovulatory women in the UK with a mean BMI of 38 kg/m2 and showed that it is no different than placebo in terms of weight reduction.[26] However, women in this study were not required to have clinical and biochemical evidence of hyperandrogenemia—an essential component of the diagnosis of PCOS. Conversely, a Finnish multicentre randomized study[27] compared metformin with placebo in 320 women with PCOS and the authors reported significantly higher live birth rates in the metformin group (41.9% vs 28.8%; P = 0.014) and maximal effect was seen in obese women with PCOS. A recent Cochrane review explored the effect of metformin on PCOS (40 studies, total n = 3848 women) failed to provide any conclusive evidence against or for metformin in women with PCOS.[8] These mixed results in the literature with regards to metformin treatment indicate that there are subtypes of PCOS which might respond beneficially to PCOS. Our study is not powered to assess the phenotypic heterogeneity in PCOS with regards to response to metformin—and highlights the need to study this with large scale studies looking at the effect of metformin on PCOS subtypes.

Women with PCOS experience higher prevalence of insulin resistance, type 2 diabetes mellitus, dyslipidaemia, endothelial dysfunction and atherosclerosis compared to age-matched women without PCOS.[2] Such metabolic disturbances are characterized by chronic low-grade inflammation and vascular impairments which increase cardiovascular risk.[2] The effects of SGLT-2 inhibitors on glycaemic control have been evaluated as the primary outcome in the majority of the studies investigating this new class of glucose lowering agents in type 2 diabetes. A meta-analysis of 13 randomized trials on the efficacy of SGLT-2 inhibitors compared to placebo demonstrated improvements in glycaemic control in type 2 diabetes patients, as evidenced by reductions in HbA1c (−0.49% and −0.50% after one and 2 years of treatment) and fasting plasma glucose levels (−0.81 and −0.76 mmol/L after 1 and 2 years of treatment).[28] Similar results were shown in a meta-analysis (10 studies, total n = 6203 participants) on the efficacy and safety of empagliflozin only.[29] In addition to glycaemic control, use of SGLT-2 inhibitors results in a reduction in TG levels and increases in HDL-cholesterol levels, but also LDL-cholesterol levels, possibly due to the shifted metabolism favouring lipid utilization.[30,31] In contrast to these beneficial effects on glycaemic control and less pronounced lipids effects of SGLT-2 inhibitors reported in type 2 diabetes, we did not observe significant changes in fasting glucose, insulin fasting lipids or hs-CRP at 3 months after empagliflozin treatment compared to baseline or any differences between our treatment groups. These results may be related to the short duration of the study or to the baseline characteristics of our participants with PCOS, who were young and did not have diabetes.

Further evidence from studies in patients with type 2 diabetes suggest that empagliflozin, and other SGLT2 inhibitors such as canagliflozin and dapagliflozin cause reductions in blood pressure, as a result of their natriuretic effects or due to the intensification of anti-hypertensive therapy.[32,33] No such blood pressure changes were demonstrated in our women with PCOS, though these subjects were normotensive and changes may not have been expected. Similarly, measures of endothelial function (RHI) or arterial stiffness (AI) were not altered compared to baseline in either treatment groups. Empagliflozin has been shown to improve endothelial dysfunction in preclinical studies in diabetic rat models,[34,35] but human data are scarce. A recent 16-week study demonstrated that dapagliflozin add-on therapy to metformin improved endothelial function, as evaluated by flow-mediated dilation, in patients with inadequately controlled early-stage type 2 diabetes mellitus.[36] Although there are no comparative data from studies that have investigated the effects of SGLT-2 inhibitors in women with PCOS, the results of our 12-week intervention contrast those of a longer study which demonstrated that metformin treatment for 6 months improved or even normalized abnormal flow-mediated dilation on the brachial artery and improved plasma endothelin-1 levels in women with PCOS.[37] The discrepancies in these results may be at least partially explained by differences in study duration and the use of different endothelial function measures.

There were significant increases in the SHBG and oestradiol levels in the empagliflozin group, but no significant reductions were seen in FAI and serum total testosterone levels. The % changes from baseline in hormonal levels did not differ to metformin. Metformin use in women with PCOS has been associated with improvements in hormonal levels. A recent meta-analysis demonstrated that metformin treatment resulted in small improvements in serum testosterone, but no changes in free testosterone, FAI, SHBG, DHEAS LH, FSH, LH/FSH ratio, oestradiol or progesterone compared to placebo in women with PCOS.[38] Metformin may also have some beneficial effects on ovulation and menstrual frequency.[8] Given the short follow-up of the present study, we did not assess these parameters, which is a limitation of the present study.