Translate this page into:
Efficacy & safety of sodium-glucose cotransporter-2 inhibitors in polycystic ovary syndrome: A meta-analysis with trial sequential analysis
For correspondence: Dr Marya Ahsan, Department of Pharmacology, College of Medicine, Imam Mohammad Ibn Saud Islamic University (IMSIU), Riyadh 13317, Saudi Arabia e-mail: marya.ahsan@gmail.com
-
Received: ,
Accepted: ,
Abstract
Background & objectives
Polycystic ovary syndrome (PCOS) is a common endocrinopathy characterised by menstrual irregularities, hirsutism, acne, obesity, infertility, and other features adversely affecting the quality of life of women of childbearing age. Besides lifestyle modifications, limited pharmacological treatments have been used to manage the symptoms of PCOS. This meta-analysis was conducted to evaluate a novel pharmacological approach, sodium glucose cotransporter-2 inhibitors (SGLT2i), in PCOS.
Methods
Electronic databases were searched systematically for literature published before November 2024. Randomised controlled trials (RCTs) evaluating SGLT2i alone or in combination in women diagnosed with PCOS, based on the Rotterdam criteria, were included in the meta-analysis. Preclinical studies, and non-randomised trials, were excluded. Quality of studies was assessed using RoB 2.0. Meta-analysis was performed for change in anthropometry, reproductive hormone levels, glycaemic and cardiometabolic indices. Adverse events (AEs) were also compared between the SGLT2i and control groups, using RevMan 5.4.1. Mean difference using the inverse-variance method and 95% confidence interval was used as a measure of effect size of continuous variables, while odds ratio (OR)using the Mantel-Haenszel method (M-H) with 95% confidence interval was calculated to analyse dichotomous variables. P value less than 0.5 was the cut-off for significance. Trial sequential analysis (TSA) was conducted to test the conventional and required information size (RIS) boundaries. The meta-analysis was registered in PROSPERO (CRD42024608736).
Results
Five RCTs with ‘low’ risk of bias, including 205 patients with PCOS, receiving SGLT2i (empagliflozin, licogliflozin, dapagliflozin, and canagliflozin alone and in combination with metformin) or control (placebo, metformin, or exenatide) were evaluated in the meta-analysis. SGLT2i significantly reduced total testosterone (mean difference=-0.10 [-0.19, -0.01], P=0.03), free androgen index (mean difference = -0.61 [-1.16, -0.05], P=0.03), and homeostasis model assessment-estimated insulin resistance (HOMA-IR) (mean difference = -0.38 [-0.75, -0.02], P=0.04). Total cholesterol (mean difference=0.13 [0.01, 0.26], P=0.04) and low-density cholesterol (MD=0.18 [0.06, 0.31], P=0.003) increased with SGLT2i use. Genitourinary AEs were more common in the SGLT2i group (OR=10.88 [1.33, 89.14], P= 0.03). On performing TSA, the Z-curve did not cross the RIS boundary of 80 per cent power.
Interpretation & conclusions
The findings of this meta-analysis suggest that SGLT2i improves the hormonal and glycaemic indices in patients with PCOS. It can prove to be a safe alternative in patients not responding to or intolerant of standard pharmacological treatments.
Keywords
Insulin resistance
meta-analysis
polycystic ovary syndrome
sodium-glucose transporter-2 inhibitors
trial sequential analysis
Polycystic ovarian syndrome (PCOS), the commonest endocrine disorder affecting women in the reproductive age group, is characterised by hormonal imbalance, metabolic dysfunction, and psychological problems1.
Currently, PCOS management guidelines recommend lifestyle modifications, such as diet and exercise, and preventing excess weight gain as first-line treatment options2. Pharmacological treatments include hormonal contraceptives for menstrual irregularities and hyperandrogenism, while letrozole and clomiphene citrate, often in combination with metformin, are used to manage infertility. Metformin, an insulin sensitiser, is primarily used to address metabolic abnormalities and chronic anovulation. It has been shown to improve IR and some hyperandrogenic symptoms, like hirsutism and menstrual irregularity2.
Emerging evidence suggests that sodium-glucose cotransporter-2 inhibitors (SGLT2i) or ‘gliflozins’ may offer additional benefits for women with PCOS3,4. These novel agents are phlorizin analogues, which were historically obtained from apple tree bark5. SGLT2i primarily inhibit the renal reabsorption of glucose in the proximal convoluted tubule, leading to increased urinary loss of glucose6. They have proven anti-hyperglycaemic and cardio-renal protective effects in both diabetic and non-diabetic patients7. The cardioprotective benefits of SGLT2i can be accounted for by multiple mechanisms such as reduction of body weight, HBA1C, IR, endothelial dysfunction, and generalised inflammatory state8. As these pathologies overlap with PCOS, randomised controlled trials (RCTs) have been conducted to explore the efficacy of SGLT2i alone or in combination with other treatments for PCOS9-13. These studies have demonstrated the beneficial metabolic effects of SGLT2i, such as promoting weight loss and improving IR. However, the effect of SGLT2i in improving the reproductive hormone levels has been inconsistent across studies. We conducted this systematic review and meta-analysis to evaluate the efficacy and safety of SGLT2i in the management of PCOS. Through this meta-analysis, we aim to synthesise evidence to determine the impact of SGLT2i on anthropometric indices, reproductive hormone levels, glycaemic and cardiometabolic parameters in patients of PCOS and address the potential AEs reported in the RCTs.
Materials & Methods
The meta-analysis was registered at PROSPERO (CRD42024608736) and drafted in accordance with the PRISMA-2020 checklist14. The study was conducted between December 1, 2024 to January 31, 2025.
Literature search
PubMed, Web of Science, and EMBASE were systematically searched for literature published in the English language before November 2024. Boolean phrases with keywords (‘Sodium-Glucose Cotransport-2 inhibitors’ OR SGLT-2i OR Empagliflozin OR Dapagliflozin OR Canagliflozin OR Ertugliflozin OR Licogliflozin OR Bexagliflozin) and (‘Polycystic Ovary Syndrome’ OR PCOS OR PCOD OR ‘Stein-Leventhal Syndrome’) were used for the search. Filters were applied in the databases to refine the search (Supplementary Table IA). The Clinicaltrials.gov website was also searched for relevant studies. The bibliography of identified reviews was scanned for any additional studies. The search was exported to a web-based software Rayyan (https://www.rayyan.ai/) to remove duplicates and screen the studies15. Titles and abstracts were screened, followed by screening of abstracts to identify eligible studies. Full texts were analysed by two independent reviewers. Any conflicts were addressed by discussion or consultation with a third reviewer.
Eligibility criteria
The PICO strategy was used to identify eligible studies (Supplementary Table IB). RCTs in patients diagnosed with PCOS based on the Rotterdam criteria, and prescribed SGLT-2i alone or in combination, were included for the meta-analysis. Non-randomised trials, preclinical studies, studies among pregnant or lactating mothers, and those trying to conceive were excluded. Studies among those with an alternate diagnosis, such as congenital adrenal hyperplasia (CAH), hyperprolactinemia, adrenal or ovarian tumours, and other malignancies, were also excluded.
Outcome measures
Both efficacy and safety outcomes were included for meta-analysis. Change in anthropometric indices included body weight, body mass index (BMI), and body fat percentage (BF%). Change in reproductive hormone levels, such as total testosterone (TT), sex hormone binding globulin (SHBG), free androgen index (FAI), androstenedione, dehydroepiandrosterone sulphate (DHEAS), luteinizing hormone (LH), and follicle stimulating hormone (FSH), were also included. Change in the fasting blood glucose (FBG), fasting insulin (FINS) levels, and homeostasis model assessment-estimated insulin resistance (HOMA-IR) were the glycaemic indicators, while lipid profile and blood pressure were the cardiometabolic parameters evaluated. Meta-analysis of the proportion of participants experiencing adverse events (AEs) was performed to assess the safety outcome measure.
Quality assessment
The quality of individual studies was determined using the 2019 version of the Cochrane Collaboration’s RoB 2 tool16. The RoB 2 tool addresses bias under five domains: randomisation process, deviations from intended intervention, missing outcome data, measurement of outcome data, and selection of the reported result. Risk of bias grades: ‘Low risk,’ ‘Some concern,’ and ‘High Risk’ are assigned to each domain. The studies were assessed independently by two authors, and any discrepancy was resolved by discussion. Robvis was used to generate the visual representation of the risk of bias assessment17.
Data extraction
A standardised form was developed in Excel to extract data. Characteristics of individual studies and baseline patient characteristics in each group, such as age, BMI, FBG, HOMA-IR, TT, and FAI levels, were tabulated. Mean changes in anthropometric measures, reproductive hormone levels, glycaemic indices, cardiometabolic parameters, and AE details were recorded. Data were extracted from both the published research and results posted in the clinical trials registry records.
Statistical analysis
Rev Man 5.4.1 was used for performing the meta-analysis18. We calculated the pooled mean difference (MD) using the inverse variance (IV) method with a 95% confidence interval (95% CI) as an effect size measure for continuous data to give more weight to precise studies. For dichotomous data, the odds ratio (OR) using the Mantel-Haenszel method (M-H) with 95% CI was estimated as the sample size in the included studies was small. The fixed effect model or random effect model was used if Higgin’s I2 was less than or more than 50 per cent, respectively. P value less than 0.5 was considered significant. Forest plots were plotted to visualise the effect size of change. The mean change from baseline was used to impute the MD. However, some studies reported the mean change while others reported only the baseline and post-treatment values. The mean change with standard deviations (SD), if not available, were calculated assuming a moderate correlation coefficient, r=0.5, for the paired samples19. Julius.ai powered by Python was employed to carry out the calculations. All measurements were converted to the same unit for performing the meta-analysis20. Trial sequential analysis (TSA) was performed to assess the reliability of the results21. Z-score cumulative line was plotted against the conventional boundary and the required information size (RIS) boundary, assuming O’Brien-Fleming’s alpha=5% and power=80%.
Results
Eligible studies
Literature search of the databases and clinical trials registry yielded 89 results. After removing duplicates and screening, eight reports were sought. Two studies that were completed had not posted their results, and one report included outcome measures other than the eligibility criteria. Remaining five RCTs were finally included for systematic review and meta-analysis (Fig. 1).
- PRISMA flow chart of eligible studies.
Study characteristics
Five RCTs conducted in various locations (United Kingdom, the USA, Germany, and China) were included in our study9-13. Out of these, three RCTs were open-labelled, one was single-blinded, and one was a double-blinded trial. SGLT2i employed in the trials were empagliflozin (25 mg/day), licogliflozin (50 mg) TID, dapagliflozin (10 mg/day) alone, or in combination with metformin or exenatide, and canagliflozin 100mg/day in combination with metformin. These were compared with metformin, exenatide, phentermine, or placebo. The duration of treatment commonly ranged from 12-24 wk. Tan et al10 administered licogliflozin for only two wk. The studies assessed a range of outcomes such as anthropometry, hormone levels, glycaemic indices, endothelial function, cardiometabolic and inflammatory markers, and clinical response to treatment. AEs were also reported by all studies (Table I)9-13.
| Author/yr | Study/study design/ Place of study | Number of participants (N)/ Ethnicity | Interventions (n) | Duration of treatment | Outcome measures |
|---|---|---|---|---|---|
| Javed et al9, 2019 | NCT03008551 Randomized open-label comparative study, United Kingdom | 39 Ethnicity not reported | EMPA 25 mg (n=19) MET SR 1500 mg (n=20) | 12 wk | Endothelial function as measured by reactive hyperaemia index; inflammatory markers; HOMA-IR; fasting glucose & insulin, body weight; blood pressure; lipid profile; hormonal parameters; quality of life questionnaire |
| Tan et al10, 2021 | NCT03152591 Double-blinded randomized placebo-controlled study multicentric (Germany & USA) | 29 Whites | LICO 50 mg TID (n=15) Placebo (n=14) | 14 days | Change from baseline in reproductive hormones; adverse events |
| Cai et al11, 2021 | NCT04700839 Randomized open-label study, Shanghai | 68 Chinese | CANA 100 mg (n=33) MET 1500-2000 mg (n=35) | 12 wk | Anthropometry; HOMA-IR; FBG & PPG &FINS; liver enzymes; uric acid; lipid profile; reproductive hormone assay; Ferriman-Gallway score; acne index; male pattern index; adverse events |
| Elkind-Hirch et al12, 2021 | NCT02635386 Randomized single-blinded comparative study, USA | 119 White (71.4%) Blacks (28.6%) | EQW 2 mg weekly* (n=20) EQW/DAPA 2 mg weekly/10 mg daily (n=20) DAPA 10 mg daily* (n=17) DAPA/MET XR (10 mg/2000 mg) (n=19) PHEN/TPM (7.5 mg/ 46 mg ER daily) (n=16) | 24 wk | Oral disposition (insulin sensitivity-insulin secretion) index, anthropometric measures, FPG, HOMA-IR, lipid profile, reproductive hormones, BP |
| Zhang et al13, 2022 | NCT04973891 Randomized open-label study, China | 41 Chinese | CANA100 mg once daily plus MET 1000 mg twice daily (n=21) MET 1000 mg twice daily (n=20) | 12 wk | Changes in anthropometric indices; reproductive hormones; glucose tolerance and insulin sensitivity; lipid profile; assessment of menstruation; adverse events |
N=total no. of participants enrolled in the study, n=no. of participants in each group. *Treatment arms included in the meta-analysis. BP, blood pressure; CANA, canagliflozin; DAPA, dapagliflozin; DAPA/MET, dapagliflozin-metformin combination; EMPA, empagliflozin; EQW, exenatide; FBG, fasting blood glucose; FINS, fasting insulin; HOMA-IR, homeostasis model assessment-estimated insulin resistance; LICO, licogliflozin; MET, metformin; MET SR, metformin sustained release; PHEN/TPM, phentermine-topiramate combination; PPG, post-prandial glucose
Baseline patient characteristics
In the five RCTs included, 205 patients completed the trial. The mean age of participants ranged from 25 to 31 yr and involved overweight or obese patients with a BMI ranging from 27 to 39 kg/m2. The TT and FAI levels were also significantly different across studies. All patients had significant IR (Supplementary Table II).
Risk of bias assessment
All the studies were of robust quality and had a low risk of bias for most domains (Fig. 2). Though three RCTs were open-label, this did not influence the outcomes assessed. In the study by Zhang et al13, some concern was there as participants adhered to different doses of metformin.
- Risk of bias assessment of the included studies. Domains are D1: bias arising from the randomization process. D2: Bias due to deviations from intended intervention. D3: Bias due to missing outcome data. D4: Bias in measurement of the outcome. D5: Bias in selection of the reported result.
Change in anthropometric indices
Our meta-analysis did not demonstrate any significant impact of SGLT2i inhibitors on body weight MD= 0.34 kg (-0.82, 1.50)], BMI [MD= 0.19 kg/m2(-0.31, 0.69)] or BF% [MD=0.20 (-0.43, 0.84)] in comparison to the control (Table II and Supplementary Fig. 1).
| Parameter | No. of studies | Participants | MD (inverse variance, fixed, 95% CI) | Heterogeneity (I2, P value) | P value for effect |
|---|---|---|---|---|---|
| Change in anthropometric indices | |||||
| Body weight (kg) | 4 | 87/89 | 0.34 (-0.82, 1.50) | 41%, 0.16 | 0.57 |
| BMI (kg/m2) | 4 | 87/89 | 0.19 (-0.31, 0.69) | 18%, 0.3 | 0.45 |
| BF% | 3 | 65/69 | 0.20 (-0.43, 0.84) | 0%, 0.9 | 0.53 |
| Change in reproductive hormone levels | |||||
| TT (nmol/L) | 5 | 102/103 | -0.10 (-0.19, -0.01) | 0%, 0.7 | 0.03* |
| SHBG (nmol/L) | 4 | 85/83 | -0.56 (-3.68, 2.57) | 0%, 0.54 | 0.73 |
| FAI | 4 | 72/74 | -0.61 (-1.16, -0.05) | 3%, 0.38 | 0.03* |
| Androstenedione (nmol/L) | 4 | 85/83 | -0.96 (-2.04, 0.11) | 0%, 0.71 | 0.08 |
| DHEAS (µmol/L) | 4 | 81/83 | -0.94 (-1.94, 0.06) | 54%, 0.09 | 0.07 |
| LH (U/L) | 3 | 65/64 | 0.27 (-0.19, 0.72) | 0%, 0.4 | 0.25 |
| FSH (U/L | 3 | 66/63 | 0.78 (-0.82, 2.38) | 83%, 0.003 | 0.34 |
| Change in glycaemic indicators | |||||
| FBG (mmol/L) | 5 | 101/103 | -0.03 (-0.24, 0.17) | 53%, 0.07 | 0.75 |
| FINS (mIU/L) | 4 | 85/83 | -2.49 (-5.41, 0.43) | 0%, 0.46 | 0.10 |
| HOMA-IR | 5 | 101/103 | -0.38 (-0.75, -0.02) | 0%, 0.56 | 0.04* |
| Change in cardiometabolic indicators | |||||
| TC (mmol/L) | 4 | 87/89 | 0.13 (0.01, 0.26) | 0%, 0.92 | 0.04* |
| LDL-C (mmol/L) | 4 | 87/89 | 0.18 (0.06, 0.31) | 0%, 0.58 | 0.003** |
| HDL-C (mmol/L) | 3 | 66/69 | 0.01 (-0.03, 0.05) | 0%, 0.58 | 0.73 |
| TG (mmol/L) | 4 | 87/89 | -0.06 (-0.16, 0.05) | 0%, 0.55 | 0.29 |
| SBP (mmHg) | 2 | 36/40 | 2.91 (-3.44, 9.25) | 56%, 0.13 | 0.37 |
| DBP (mmHg) | 2 | 36/40 | 1.68 (0.62, 2.74) | 22%, 0.26 | 0.002** |
P*<0.05, **<0.01. BMI, body mass index; BF%, body fat percentage; DBP, diastolic blood pressure; DHEAS, dehydroepiandrosterone sulphate; FAI, free androgen index; FBG, fasting blood glucose; FINS, fasting insulin; FSH, follicle stimulating hormone; HDL-C, high density cholesterol; HOMA-IR, homeostasis model assessment-estimated insulin resistance; LDL-C, low density cholesterol, LH, homeostasis model assessment-estimated insulin resistance SBP, systolic blood pressure; SHBG, sex hormone binding globulin; TC, total cholesterol; TG, triglyceride; TT, total testosterone
Change in reproductive hormone levels
Significant decrease in TT [MD = -0.10 nmol/L (-0.19, -0.01), P=0.03] and FAI [MD=-0.61 (-1.16, -0.05), P=0.03] was seen in the SGLT2i group (Fig. 3). However, changes in SHBG, androstenedione, DHEAS, LH, and FSH were not significant from the control (Table II).
- Forest plot of impact of SGLT2i vs. control on reproductive hormone levels: (A) total testosterone (nmol/L), (B) Sex hormone binding globulin (nmol/L), (C) Free androgen index (%), and (D) Dehydroepiandrosterone sulphate (µmol/L).
Change in glycaemic indices
There was a significant decrease in HOMA-IR in the SGLT2i group (MD = -0.38 [-0.75, -0.02], P=0.04) (Fig. 4). However, changes in FBG and FINS were not significantly different from the control (Table II).
- Forest plot of impact of SGLT2i vs. control on glycaemic indices: (A) Fasting blood glucose (mmol/L), (B) Fasting insulin (mIU/L), and (C) HOMA-IR.
Change in cardiometabolic parameters
SGLT2i significantly altered the lipid profile and blood pressure (Supplementary Fig. 2). There was a significant increase in the TC [MD=0.13 mmol/L (0.01, 0.26), P=0.04] and LDL-C [MD=0.18 mmol/L (0.06, 0.31), P=0.003], with no significant difference in change in HDL-Cor TG levels. Decrease in diastolic blood pressure (DBP) was slightly greater in the control group compared to SGLT2i [MD=1.68 (0.62, 2.74), P=0.002] (Table II).
Safety outcome
In the five RCTs, 40 per cent of patients on SGLT2i reported at least one AE in comparison to 48.7 per cent in the control group [OR=0.88 (0.24, 3.22), P=0.84]. Gastrointestinal symptoms such as nausea, diarrhoea, and flatulence were the most common AEs in both the SGLT2i (27.8%) and control group (43.3%) [OR=0.50 (0.03, 8.83), P=0.63]. Genito-urinary infections were significantly higher in the SGLT2i group [OR=9.33 (1.14, 76.26), P=0.04] (Supplementary Fig. 3). Other AEs reported in the SGLT2i group included headache, dizziness, rash, and nasopharyngitis (Supplementary Table III).
TSA
The Z-curve for TT levels crossed the conventional boundaries of significance at 5 per cent, but did not reach the RIS boundary for 80 per cent power (required n=346). Similar results were obtained for other efficacy and safety parameters (Fig. 5).
- Trial sequential analysis for impact of SGLT2i vs. control on reducing total testosterone levels. Z-curve crossing the conventional boundary but failing to cross the RIS 80 per cent boundary.
Discussion
This meta-analysis evaluated SGLT2i as a novel therapeutic approach in PCOS, drawing on data from five RCTs. Our analysis of 205 patients found that SGLT2i resulted in a statistically significant reduction of TT [MD = -0.10 (-0.19, -0.01), P=0.03], FAI [-0.61 (-1.16, -0.05), P=0.03] and HOMA-IR [MD = -0.38 (-0.75, -0.02), P=0.04] in comparison to control. Improvement in other reproductive hormones, FBG, and FINS was not different from the control. However, four of the RCTs involved active control (MET or EQW), and only the study by Tan et al10 compared LICO to a placebo. On eliminating the study by Tan et al10, the advantage of SGLT2i over the active controls persisted with respect to TT and FAI. This advantage was lost with respect to HOMA-IR [MD=-0.35 (-0.72, 0.02), P =0.06]. This is due to the active control’s proven ability to decrease IR. In a meta-analysis of 23 RCTs comparing metformin to placebo among women suffering from PCOS, metformin, a known insulin sensitizer, was found to be superior to placebo in decreasing HOMA-IR22. Though the drop in TT and FAI was statistically significant compared to the active control, the difference was too small to reflect any clinically significant improvement. However, the decline in androgen levels could reflect a trend if accompanied by improvements in symptoms of hyperandrogenism such as acne, hirsutism, and menstrual irregularities. While Cai et al11 and Zhang et al13 reported improvement in menstrual irregularities, and Cai et al11 also reported improvement in Ferriman-Gallway score, it was not significantly different from the metformin group13,15. Other studies did not assess clinical symptoms.
Despite the fact that all of the SGLT2i (EMPA, CANA, DAPA, and CANA/MET) improved the anthropometric indices in the included RCTs, our meta-analysis revealed no statistically significant difference between SGLT2i and the active controls. This is because both MET and GLP-1 receptor antagonists are also known to help obese patients lose weight22,23. When compared to placebo, the reduction in BMI is greater with metformin in a meta-analysis of RCTs among PCOS patients, especially in those with a BMI >25 kg/m.
The included RCTs varied with respect to the impact of SGLT2i on lipid profiles. However, on evaluating, we found that SGLT2i significantly increased TC [MD=0.13 mmol/L (0.01, 0.26), P=0.04] and LDL-C [MD=0.18 mmol/L (0.06, 0.31), P=0.003]. Though TG levels decreased with SGLT2i, it was not significant [MD = -0.06 mmol/L (-0.16, 0.05), P=0.29]. A similar effect of SGLT2i on lipid profile has been established in earlier studies24. It has been postulated that delayed turnover of LDL-C and increased lipoprotein lipase activity are responsible for the changes in lipid profile25. Despite increasing LDL-C levels, SGLT2i are proven to exert cardio-renal protection. This is explained by the differential effect of SGLT2i on the LDL-C subclasses. While the SGLT2i increases large buoyant (lb) LDL, they decrease small dense (sd) LDL particles. The sd LDL particles are highly susceptible to oxidation and are responsible for metabolic disorders, obesity, and coronary artery disease25. In addition, both SBP and DBP decreased with SGLT2i across the studies. However, the decrease in DBP with SGLT2i was significantly less than that seen in the control group [MD=1.68 mmHg (0.62, 2.74), P=0.002].
Our results differ from a similar meta-analysis conducted by Sinha et al26, who reported a significant reduction in body weight and DHEAS. This could be because Sinha et al26 included only two studies for the comparison of anthropometric indices9,12. For the analysis of the effect on reproductive hormones, only three studies were included by Sinha et al26. Moreover, we involved studies conducted among different ethnicities, such as Whites and Chinese, while Sinha et al26 only involved studies conducted among mostly Whites. This could have impacted the variable results. However, with respect to HOMA-IR, our results comply with those reported by Sinha et al26.
From the pooled analysis of self-reported AEs, we inferred that SGLT2i were well tolerated by patients with PCOS. However, there was high heterogeneity, which could be due to differences in AE reporting. Though genitourinary infections were statistically significant in the SGLT2i group, it was reported by only two studies. Most of these studies were short-term studies, which could have impacted the manifestation of AEs. Further long-term studies are needed to confirm the safety of SGLT2i in PCOS.
We employed five high-quality RCTs, but each RCT was constrained by the small sample size. There was also inconsistency in reporting across the studies. We did not evaluate the clinical impact of SGLT2i on PCOS symptoms, such as menstrual regularity, hirsutism, psychological symptoms, or quality of life, due to a lack of data. Nonetheless, we are the first to evaluate cardiometabolic and safety data. On performing the TSA, we also found that the pooled data yielding statistically significant improvement of hormonal and glycaemic indices with SGLT2i was insufficient. Though five RCTs have been conducted evaluating the efficacy and safety of SGLT2i in PCOS, the results of the meta-analysis are inconclusive as they did not cross the RIS boundary of 80 per cent power. This highlights the need for more thorough large scale RCTs, especially evaluating the long-term impact of SGLT2i on clinical, laboratory, and safety outcomes. Furthermore, RCTs can be designed to explore the role of SGLT2i on diverse outcomes like ovulation rates, mental health outcomes such as anxiety and depression, and the cost-effectiveness of SGLT2i in PCOS.
We conclude that SGLT2i are associated with improvements in hormonal and glycaemic profiles in PCOS in comparison to active controls. Though SGLT2i proved to be well tolerated in clinical trials, safety concerns exist regarding genitourinary infections. Further large-scale studies are required to increase the power of the result.
Financial support & sponsorship
This work was supported and funded by the Deanship of Scientific Research at Imam Mohammad Ibn Saud Islamic University (IMSIU) (grant number IMSIU-DDRSP2501).
Conflicts of Interest
None.
Use of Artificial Intelligence (AI)-Assisted Technology for manuscript preparation
The authors confirm that there was no use of AI-assisted technology for assisting in the writing of the manuscript and no images were manipulated using AI.
References
- Clinical presentation and diagnosis of polycystic ovarian syndrome. Clin Obstet Gynecol. 2021;64:3-11.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- Recommendations from the 2023 international evidence-based guideline for the assessment and management of polycystic ovary syndrome. J Clin Endocrinol Metab. 2023;108:2447-69.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- SGLT‐2 inhibitors enhance the effect of metformin to ameliorate hormonal changes and inflammatory markers in a rat PCOS model. Physiol Rep. 2023;11:e15858.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- Impact of dapagliflozin on metabolic phenotype, hormone levels, and fertility in female mice after prolonged high-fat diet. Front Endocrinol (Lausanne). 2025;15:1457268.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- SGLT2 inhibitors: Beyond glycemic control. J Clin Transl Endocrinol. 2024;35:100335.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- Basic and clinical pharmaco-therapeutics of SGLT2 Inhibitors: a contemporary update. Diabetes Ther. 2020;11:813-3.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- SGLT2 inhibitors: Mechanisms of cardiovascular benefit beyond glycaemic control. Nat Rev Cardiol. 2020;17:761-72.
- [CrossRef] [PubMed] [Google Scholar]
- Cardiovascular protection by sodium glucose cotransporter 2 inhibitors: Potential mechanisms. Am J Med. 2017;130:S30-9.
- [CrossRef] [PubMed] [Google Scholar]
- Effects of empagliflozin on metabolic parameters in polycystic ovary syndrome: a randomized controlled study. Clin Endocrinol (Oxf). 2019;90:805-13.
- [CrossRef] [PubMed] [Google Scholar]
- Licogliflozin versus placebo in women with polycystic ovary syndrome: a randomized, double‐blind, phase 2 trial. Diabetes Obesity Metabolism. 2021;23:2595-9.
- [CrossRef] [PubMed] [Google Scholar]
- Efficacy of canagliflozin versus metformin in women with polycystic ovary syndrome: a randomized, open‐label, noninferiority trial. Diabetes Obesity Metabolism. 2022;24:312-20.
- [CrossRef] [PubMed] [Google Scholar]
- Exenatide, dapagliflozin, or phentermine/topiramate differentially affect metabolic profiles in polycystic ovary syndrome. J Clin Endocrinol Metab. 2021;106:3019-33.
- [CrossRef] [PubMed] [Google Scholar]
- Canagliflozin combined with metformin versus metformin monotherapy for endocrine and metabolic profiles in overweight and obese women with polycystic ovary syndrome: a single-center, open-labeled prospective randomized controlled trial. Front Endocrinol (Lausanne). 2022;13:1003238.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- Rayyan—a web and mobile app for systematic reviews. Syst Rev. 2016;5:210.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- RoB 2: A revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898.
- [CrossRef] [PubMed] [Google Scholar]
- GitHub. Robvis: An R package and web application for visualising risk-of-bias assessments. Available from: https://github.com/mcguinlu/robvis, accessed on February 15, 2025.
- The Cochrane Collaboration. Review Manager (RevMan). Version 5.4. Available from: https://training.cochrane.org/system/files/uploads/protected_file/RevMan5.4_user_guide.pdf, accessed on February 13, 2025.
- Chapter 6: Choosing effect measures and computing estimates of effect. In: Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, eds. Cochrane Handbook for Systematic Reviews of Interventions version 6.5 (updated August 2024). Cochrane; 2024.
- [Google Scholar]
- Julius AI. (2025). Julius AI: Advanced data science and machine learning tools. Available from https://julius.ai, accessed on February 13, 2025.
- The Copenhagen Trial Unit, Centre for Clinical Intervention Research. Trial Sequential Analysis. Available from: https://ctu.dk/tsa/, accessed on February 13, 2025.
- The impact of metformin with or without lifestyle modification versus placebo on polycystic ovary syndrome: a systematic review and meta-analysis of randomized controlled trials. Eur J Endocrinol. 2023;189:S37-63.
- [CrossRef] [PubMed] [Google Scholar]
- The efficacy and safety of GLP-1 agonists in PCOS women living with obesity in promoting weight loss and hormonal regulation: A meta-analysis of randomized controlled trials. J Diabetes Complications. 2024;38:108834.
- [CrossRef] [PubMed] [Google Scholar]
- Effect of sodium-glucose co-transporter 2 inhibitors on lipid profile: a systematic review and meta-analysis of 48 randomized controlled trials. Pharmacol Res. 2020;160:105068.
- [CrossRef] [PubMed] [Google Scholar]
- The effects of SGLT2 inhibitors on lipid metabolism. Metabolites. 2021;11:87.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- A meta-analysis of the effect of sodium glucose cotransporter-2 inhibitors on metabolic parameters in patients with polycystic ovary syndrome. Front Endocrinol (Lausanne). 2022;13:830401.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]