Inositol for Polycystic Ovary Syndrome: A Systematic Review and Meta-analysis to Inform the 2023 Update of the International Evidence-based PCOS Guidelines

Abstract Context Insulin resistance is common in women with polycystic ovary syndrome (PCOS). Inositol may have insulin sensitizing effects; however, its efficacy in the management of PCOS remains indeterminate. Objective To inform the 2023 international evidence-based guidelines in PCOS, this systematic review and meta-analysis evaluated the efficacy of inositol, alone or in combination with other therapies, in the management of PCOS. Data Sources Medline, PsycInfo, EMBASE, All EBM, and CINAHL from inception until August 2022. Study Selection Thirty trials (n = 2230; 1093 intervention, 1137 control), with 19 pooled in meta-analyses were included. Data Extraction Data were extracted for hormonal, metabolic, lipids, psychological, anthropometric, reproductive outcomes, and adverse effects by 1 reviewer, independently verified by a second. Data Synthesis Thirteen comparisons were assessed, with 3 in meta-analyses. Evidence suggests benefits for myo-inositol or D-chiro-inositol (DCI) for some metabolic measures and potential benefits from DCI for ovulation, but inositol may have no effect on other outcomes. Metformin may improve waist-hip ratio and hirsutism compared to inositol, but there is likely no difference for reproductive outcomes, and the evidence is very uncertain for body mass indexI. Myo-inositol likely causes fewer gastrointestinal adverse events compared with metformin; however, these are typically mild and self-limited. Conclusion The evidence supporting the use of inositol in the management of PCOS is limited and inconclusive. Clinicians and their patients should consider the uncertainty of the evidence together with individual values and preferences when engaging in shared decision-making regarding the use of inositol for PCOS.

recommended in the 2018 international evidence-based PCOS guideline (5) for management of anthropometric and metabolic outcomes in PCOS.However, other options to treat insulin resistance may be desired by women with PCOS due to gastrointestinal side effects related to metformin.Hence, there is a need for alternatives to metformin for managing insulin resistance in PCOS.
Inositol was originally isolated from muscle cells in 1850 by Johann Joseph Scherer (6).Inositols are structural components of cell membranes (eg, phosphatidyl-inositol phosphate lipids) and participate in hormone signal transduction (eg, inositol triphosphate).There are 9 stereo-isomers of inositol of which myo-inositol (MI) is the most abundant in the human body (7).In humans, dietary intake and endogenous synthesis of MI occurs.
Mechanistically, MI promotes translocation of the glucose transporter type 4 to the plasma membrane for glucose uptake (7) and also reduces release of free fatty acids from adipose tissue (8).MI is also involved in reproductive functions, including follicle stimulating hormone (FSH)-mediated pathways, which regulate the proliferation and maturation of granulosa cells (9).Insulin stimulates unidirectional conversion of MI to D-chiro-inositol (DCI) (10), which stimulates glycogen production and facilitates additional uptake of glucose through mobilization of glucose transporter type 4 transporters (11).It has been hypothesized that overproduction of insulin in PCOS enhances MI to DCI conversion, which results in an increased DCI and decreased MI concentration in follicular fluid (12).In women without PCOS, the ovarian MI:DCI ratio is 100:1, but in women with PCOS, the ratio is 0.2:1 (13).MI is also postulated to enhance aromatase synthesis in granulosa cells and therefore reduce androgen production (8).It has been suggested that a 40:1 ratio of MI:DCI is physiological, and provision of inositol in this ratio has reverted PCOS phenotypes in mouse models (14).
In view of the potential benefits of inositol, thorough and critical evaluation of its efficacy in the treatment of PCOS is essential.The aim of this systematic review and meta-analysis was to evaluate the effectiveness of inositol alone or in combination with other therapies for the management of hormonal and clinical PCOS features, weight, and reproductive outcomes.

Materials and Methods
This systematic review and meta-analysis was conducted to inform the 2023 update of the International Evidence-based Guideline on the Assessment and Management of PCOS (15) and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (16).The protocol was prospectively registered with PROSPERO (CRD 42022356057).The review informed the guideline recommendation for the question "In adolescents and adults with PCOS, is inositol alone or in combination with other therapies effective for management of hormonal and clinical PCOS features, weight, and reproductive outcomes?"The PROSPERO registration includes the reproductive outcomes of interest; the other outcomes were inadvertently left off the registration.However, all the outcomes including hormonal, clinical, anthropometric, and reproductive were determined a priori by the guideline evidence synthesis team.

Information Sources
We searched the following databases from inception until August 5, 2022: Medline (OVID), PsycInfo (EBSCO), EMBASE (OVID), All EBM (OVID), and CINAHL (EBSCO).A handsearch of reference lists of relevant recent systematic reviews was conducted to identify any trials that were not captured in our search.We asked experts in the field to identify any trials that were not identified in our search.

Search Strategy
The search strategy and selection criteria were developed by the guideline evidence and key contact team, comprising an international team of evidence synthesis experts, reproductive endocrinologists, endocrinologists, pediatricians, and a general practitioner (family physician).The search strategy included terms for PCOS, inositol, and randomized controlled trials and is included in the supplemental data (17).
Only randomized controlled trials (RCTs) were included with an a priori defined Participant-Intervention-Comparison-Outcome framework, using the following criteria: (1) participants: females diagnosed with PCOS by Rotterdam, National Institutes of Health, or Androgen Excess and PCOS Society (AEPCOS) criteria of any age, ethnicity and weight; (2) intervention: inositol (MI or DCI alone or in combination), alone or combined with usual care (as defined by study authors), lifestyle, or any other interventions of any dose or duration; (3) comparator: placebo, usual care alone, lifestyle alone, or any other interventions (listed in intervention) or combinations of those listed in intervention.Where comparisons or co-interventions were used in 2 or more study groups, these should be the same across all groups, with inositol being the main difference between groups in order to isolate the effects of inositol.The only exception was that comparisons of inositol + folic acid (FA) vs another comparator (eg, placebo) was deemed to be eligible as the effect of FA on outcomes was thought to be minimal; (4) outcomes: hormonal, metabolic, lipids, psychological, anthropometric or reproductive outcomes, and adverse events [see supplementary data (17) for a full list of eligible outcomes].

Selection Process and Data Extraction
Citations were imported into Covidence (18), where duplicates were removed in Covidence.Titles and abstracts and then full-text manuscripts were screened independently in duplicate by 3 reviewers (V.F., C.E., S.M.) independently using Covidence, and disagreements were resolved by discussion, with a third reviewer to adjudicate if needed.Data were extracted by 1 reviewer (A.B., C.E., D.N., J.L., L.L., S.G., V.F., V.R., V.V.) on study characteristics, participant characteristics at baseline, intervention, and outcomes using a data extraction template created by the guideline evidence team.Extracted data were then verified by a second reviewer (A.B., C.E., G.Y., L.L., M.A., S.M., V.F., V.R.).

Integrity Assessment
Trial integrity was assessed by an integrity committee following the Research Integrity in Guideline Development (RIGID) framework developed by Mousa et al (2023;unpublished), as detailed in Section 6.7 of the guideline technical report (19).Here, studies were assessed using the Trustworthiness in Randomised Controlled Trials checklist (20), an integrity assessment tool similar to the Cochrane Research Integrity Assessment tool (21), which assesses studies on multiple domains related to integrity.Following the steps of the RIGID framework, studies were classified as low, moderate, or high risk for integrity concerns.Low-risk studies were included, while authors for moderate-and high-risk studies were contacted to clarify integrity concerns.Where a satisfactory response was received, those studies were subsequently "included."Studies with no response were "not included," while studies requiring additional time to provide the necessary information (eg, raw data, ethics protocols, etc.) are "awaiting classification" and have not been included in the review or analysis at this stage.

Quality Appraisal
We assessed risk of bias for each trial using the Cochrane Risk of Bias 1.0 tool (22).Two reviewers independently assessed random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other biases.Any disagreements were resolved by discussion.
The certainty of evidence was assessed by M.A. and V.F. using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach, as outlined in the GRADE handbook (23).The GRADE approach uses 5 considerations (risk of bias, inconsistency of effect, indirectness, imprecision, and other bias including publication bias).The evidence can be downgraded from "high quality" by 1 level for serious (or by 2 levels for very serious) limitations, depending on assessments of these domains.

Data Analysis
The key contacts (clinical leads) for the guideline judged the following outcomes as critical (for GRADE purposes): free testosterone (FT), homeostatic model assessment of insulin resistance (HOMA-IR), 2-hour glucose, BMI.The remaining outcomes were judged as important but not critical.Outcome data were extracted from original intention-to-treat results wherever possible or from per-protocol results if these were the only outcomes available.Heterogeneity was assessed using the I 2 statistic.Outcomes from individual studies were pooled using random effects models.For trials that used the same assessment method and provided continuous data, we reported mean difference (MD), converting units of measurement to standard units where required.Categorical outcomes were reported as odds ratios (OR).All pooled analyses were reported with 95% confidence intervals (CIs).We used RevMan (24) for statistical analysis.Publication bias was assessed by visual inspection of funnel plot asymmetry, where applicable.Subgroup analyses for menopausal stage, adolescents/adults, and BMI category were planned but were not conducted due to the small number of trials.

Changes Since Protocol Was Registered
Anthropometric, hormonal, psychological, and metabolic outcomes were collected as determined a priori by the research team but inadvertently not included in the PROSPERO registration.We conducted a sensitivity analysis according to risk of bias, where only the studies that were assessed as low risk of selection bias and not at high risk of performance bias were included.We conducted supplementary analyses for critical outcomes only where we also included data from studies that were not included due to moderate-high risk of integrity concerns.We conducted supplementary analyses using standardized mean differences (SMD) for endocrinological outcomes where outcomes were collected at different laboratories and therefore potentially using different assays (see Supplementary Data S6).

Results
See Fig. 1 for the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart of study selection.A total of 1535 citations were identified, of which 1534 were identified through database searches and 1 through expert identification, with 734 remaining after duplicates were removed.After title and abstract screening, an additional 592 were excluded, and 142 full-text manuscripts were reviewed for eligibility.Ninety-nine were excluded based on full text review, which left 43 trials then assessed for integrity.Thirty were assessed as low risk and included.Thirteen were assessed to be of moderate risk and were not included.(see supplementary data and Supplementary Table S1 for characteristics of moderate risk studies) (17).Table 1 provides details of study characteristics of the included studies.In total, there were 31 articles representing 30 unique trials and 2230 participants.There was 1 study including only adolescents (n = 106) (37).Fifteen trials were conducted in Italy (27, 29, 31-34, 38, 39, 44, 45, 47, 48, 50, 54, 55), 6 in India (26,28,30,36,41,42,51), 2 in Iran (25,49), 2 in Venezuela (35,43), and 1 each in Pakistan (37), Turkey (46), Spain (40), Denmark (52), and Bosnia and Herzegovina (53).Sample sizes for arms relevant for this study ranged from 8 to 195 with a mean sample size of 36 participants.Nineteen studies contributed to meta-analyses.

Participants
All participants were individuals with PCOS diagnosed by Rotterdam, National Institutes of Health, or AEPCOS criteria.Mean age ranged from 15 to 36.5, and mean baseline BMI ranged from 20.96 to 32.4 kg/m 2 .Three studies enrolled women with PCOS with an overweight or obese BMI classification (29,44,54).Ten studies specifically enrolled women with PCOS who also had infertility (25, 27, 40, 46-51, 55, 56).These studies used co-interventions of fertility treatment, either ovulation induction with letrozole (49), gonadotropin (50), or gonadotropin and intrauterine insemination (46) or in vitro fertilization (IVF) with or without intracytoplasmic sperm injection (ICSI) (25,27,40,47,48,51,55).The fertility cointerventions were either concurrent with the intervention or following a period of pretreatment depending on study design.
One trial evaluated MI + DCI + metformin (28) and 1 MI + metformin (30).One had 3 arms comparing MI + monacolin K to MI alone and also to metformin alone (39).We included data only from the MI and metformin arms as the comparison with MI + monacolin K was not considered relevant.One trial evaluated MI + DCI compared to a combined hormonal contraceptive (CHC) (36).
Most trials provided interventions for at least 6 weeks, up to a maximum of 6 months.One study did not clearly define treatment duration (48), and 1 study only provided inositol during IVF stimulation (47).Follow-up ranged from 8 weeks to 6 months and was not reported by 1 study (48).

Outcomes
The outcomes reported were heterogeneous.Follow-up duration ranged from completion of an IVF cycle to 6 months after treatment.Up to 23 studies reported on metabolic outcomes including fasting glucose, fasting insulin, HOMA-IR, and quantitative insulin-sensitivity check index, area under the curve (AUC) insulin, and AUC glucose.Twelve studies reported total cholesterol, 10 reported low-density lipoprotein (LDL), highdensity lipoprotein (HDL), and triglyceride levels.Twenty-nine studies reported on BMI, 16 reported waist-hip ratio (WHR), 12 reported weight, and 8 reported waist circumference.
Androgenicity outcomes were reported by up to 14 studies and included hirsutism, which was reported as the Ferriman-Gallwey (FG) or modified Ferriman-Gallwey, total and FT levels, sex hormone binding globulin (SHBG), dehydroepiandrosteronesulphate (DHEAS), androstenedione and free androgen index.Reproductive outcomes were reported by up to 13 studies, specifically menstrual regularity, ovulation rate, pregnancy rates, and live birth rate.Gastrointestinal adverse events (GI AEs) were reported by 6 trials, and other adverse events were reported by 2 trials.One study comparing MI and metformin reported on psychological outcomes (depression and quality of life) (52).

Effects of Interventions
We report on a total of 13 comparisons.Meta-analyses were conducted on the following 3 comparisons: (1) DCI vs placebo (2 RCTs), (2) MI + FA vs FA (8 RCTs), and (3) MI vs metformin (10 RCTs).For the remaining comparisons, a metaanalysis was not possible on any outcome either due to the comparison only having 1 representative RCT or RCTs reporting nonparametric data (median and interquartile range) or change scores without any information on standard deviation or standard error.For those, we have provided a narrative synthesis.

Meta-analyses
D-Chiro inositol DCI (600-1200 mg/day) v placebo.Two trials reported on this comparison (35,43).Sample sizes were 20 and 44.The dose of DCI provided ranged from 600 mg (35) to 1.2 g (43) daily.One trial enrolled women with BMI >28 kg/m 2 (43).There were no co-interventions in either trial.There was very serious imprecision due to very small sample size in both studies; hence all outcomes were of low certainty (see Table 2 for a summary of GRADE assessment for this comparison).Figure 3A-3E shows the forest plots for comparisons regarding androgenicity outcomes.There was low certainty evidence that DCI was superior to placebo for SHBG, DHEAS, and FT (MD = 1.79 mcg/dL, −139.43 mcg/dL and −0.46 ng/dL, respectively; Fig. 3A, 3B, 3D) with no differences in total testosterone or androstenedione (Fig. 3C and  3E).Similarly, DCI was superior to placebo for the metabolic outcomes of AUC insulin (MD −3.65 µU/mL/min, Fig. 3I) and AUC glucose (MD −20.85 mg/dL/min, Fig. 3H), as well as triglycerides (MD −31.95 mg/dL, Fig. 3K), whereas there were no differences for fasting glucose (Fig. 3F), fasting insulin (Fig. 3G), total cholesterol (Fig. 3J), or from single study results for LDL and HDL 43 .Meta-analysis also showed that placebo was superior to DCI for the outcome of BMI (MD 0.67, Fig. 3L) wIth no difference in WHR (Fig. 3m).Ovulation rate was improved with DCI compared with placebo in metaanalysis of 2 trials (OR 11.5; Fig. 3m), both of which counted an ovulatory event if serum progesterone level was >8 ng/mL.Supplemental analysis using SMD for endocrine outcomes were similar to findings from analyses using MD except for FT and AUC glucose, which were no longer statistically significant (17).Six of the 8 studies enrolled women with PCOS who were presenting for infertility treatment (25,27,(46)(47)(48)(49). MI was provided prior to ovulation induction with letrozole for women with letrozole resistance (40), gonadotropin for ovulation induction and intrauterine insemination (46), IVF (25,27), and IVF with ICSI (48).One trial randomized women to MI + FA or FA alone during the IVF cycle until 14 days after embryo transfer (47).The remaining 2 studies were not limited to women undergoing infertility treatment (31,34).One recruited women who were overweight (34) and randomized them to 2 g MI + 200 mcg FA daily or FA 200 mcg daily for 12 weeks.See Table 3 for a summary of GRADE assessments for this comparison.
Supplemental analysis using SMD for endocrine outcomes was unchanged except for total testosterone and androstenedione where MI + FA was superior to FA (17).

MI vs metformin.
Ten studies compared MI to metformin and were included in meta-analyses (26,30,33,41,42,(49)(50)(51)(52)(53)(54).Sample sizes in each arm ranged from 12 to 60 per study, with a total of 353 participants in the intervention arms and 356 in the control arm across the 10 studies.The dose of MI ranged from 1 to 4 g, with most studies using 4 g.Doses of metformin ranged from 1000 mg (26,30) to 2000 mg daily (52).Duration of intervention ranged from 3 to 6 months.Three studies enrolled women with PCOS who were attempting to conceive and utilized co-interventions of letrozole (49), gonadotropin (50), or IVF (51).See Table 4 for the GRADE summary for this comparison.
In a meta-analysis of 6 trials, GI AEs were less common in the MI group compared with metformin (OR 0.09, 0.02 to 0.37, I 2 = 69%, 6 trials, Fig. 5Q) (26,33,(51)(52)(53)(54).One study comparing MI and metformin reported on depression and quality of life measures (40).There was no significant difference between the interventions for any of the outcomes measured.
Supplemental analyses using SMD for endocrine outcomes were unchanged except for SHBG where MI was no longer superior to metformin (see Supplementary Data S6).Supplemental analyses adding studies at moderate risk of compromised integrity for critical outcomes of HOMA-IR and BMI still showed no differences between groups for these outcomes (see Supplementary Data S7, Figs. 2 and 3) (17).
Sensitivity analysis.Sensitivity analysis including only the 2 trials judged to be at low risk of bias (26,51) was conducted for the outcomes of clinical pregnancy rate, total testosterone, fasting insulin, fasting glucose, HOMA-IR, BMI, waist circumference, regular menstruation, and GI AEs.See Supplementary Data for more details (17).Metformin was superior to MI for fasting insulin (MD 1.73.95% CI 0.76-2.71,I 2 = 0%, 2 trials) while MI was superior to metformin for regular menstruation (OR 2.84, 95% CI 1.26-6.37,I 2 = 42%), but there was no difference between groups for GI AEs.

MI + DCI vs placebo.
Another study (n = 106) (37) compared 500 mg MI in combination with 13.8 mg DCI twice daily for 6 months to FA among a population of teenage girls (age 13-19 years) with PCOS.This study had high risk of bias and reported no significant difference in fasting glucose between intervention and placebo group.There was a significant reduction in number of participants with elevated BMI in the intervention group; however, no BMI values or cut offs used for categorization were reported.

Inositol vs CHC
MI+ DCI vs CHC.One trial (n = 70) (36) assessed outcomes after 6 months of treatment and again 3 months after stopping treatment with MI + DCI compared to CHC. Results suggested no significant difference in weight, BMI, WHR, HOMA-IR, fasting insulin, fasting blood glucose, or testosterone after 6 months of treatment.There was improvement in HOMA-IR and fasting insulin level in the MI + DCI group when assessed 3 months after stopping treatment.Menstrual cycle length was shorter in the CHC group.Inositol + metformin v metformin alone MI + metformin vs metformin.One study (n = 50) (30) showed no difference between groups for BMI, weight, and fasting insulin.Women received either 1 g MI + 1 g metformin daily or 1 g metformin only daily for 4 months.
Supplemental analyses adding studies at moderate risk of compromised integrity for the critical outcome of BMI still showed no differences between groups (see Supplementary Data S7, Fig. 1) (17).

Comparisons between MI and DCI MI vs DCI.
One study in euglycemic women undergoing ovulation induction for ICSI (n = 84) (55) with unclear risk of bias suggests that 4 g MI daily for 8 weeks before FSH administration improves total pregnancy rate [22 (51%) vs 10 (24%), P < .05]vs 1.2 g DCI daily, but there is no difference for clinical pregnancy, biochemical pregnancy, and miscarriage rates.Evidence is of low certainty.

GRADE assessments
The majority of outcomes were assessed as being low to very low certainty, as outlined in Tables 2 to 4. Within the comparison of MI vs metformin, the outcomes of total testosterone, clinical pregnancy rate, GI AEs, menstruation, fasting glucose, fasting insulin, and waist circumference were assessed as moderate certainty.The main reasons for downgrading included risk of bias (eg, most studies had a high or moderate risk of bias), inconsistency due to wide CIs or CIs not overlapping, and imprecision (small numbers of studies or sample sizes).

Discussion
In this systematic review and meta-analysis conducted to inform the 2023 International Evidence-Based PCOS Guideline (15,(57)(58)(59)(60), we report on findings from 30 RCTs comparing inositol with various comparators for PCOS.At this time, available evidence regarding the benefits of inositol (in all forms) was inadequate to make evidence-based recommendations regarding efficacy for clinical outcomes.Yet, limited and inconsistent data support some metabolic and hormonal benefits although certainty of the evidence was low to very low.No benefit was found for anthropometric outcomes.While MI resulted in fewer GI AEs compared to metformin, we note that the majority of trials (23/29) did not report on adverse events.Our findings are consistent with a recent meta-analysis by Jethaliya and colleagues (61) who examined the efficacy of MI against any comparator for anthropometric, metabolic, and endocrine outcomes in PCOS and included 17 RCTs.Similar to our findings, they did not find any improvements in anthropometric (BMI, WHR), metabolic (fasting insulin, fasting glucose, HOMA-IR), or hormonal (luteinizing hormone, FSH, oestradiol, SHBG, DHEAS, and total testosterone levels) outcomes with the exception of androstenedione levels, which favored inositol (61).Our review is more recent than Jethaliya al's, who searched until January 2022, and includes DCI as well as MI.Additionally, we conducted GRADE assessments, which were not performed in Jethaliya et al's review.
Greff and colleagues examined the efficacy of MI or DCI against placebo or metformin in PCOS and reported on 26 RCTs (62).In contrast to our findings, Greff et al (62).reported a higher rate of cycle normalization and greater reduction of BMI and weight in the inositol group compared with placebo.A potential reason for this discrepancy is that Greff and colleagues (62) included some studies that we had excluded from analyses due to uncertainty about research integrity; hence, their results may have been influenced by data that we considered to be questionable in the context of informing guidelines for clinical practice.Moreover, follow-up values were used here, whereas Greff et al used mean change values (delta) between baseline and post-treatment, which tend to produce larger effect sizes and thus more significant results in meta-analyses (63).Ideally, using both estimates in sensitivity analysis would have been the optimal method to ensure robustness of results; however, this was not possible in the present review due to the small number of studies.
Metformin was more efficacious than MI for clinically important outcomes such as central adiposity (WHR) and hirsutism (Ferriman-Gallwey score).Although moderate certainty evidence also suggests that metformin results in a substantially greater number of GI AEs than MI, we note that these adverse events are generally mild and self-limiting (5) and that a sensitivity analysis on studies judged to be at low risk of bias did not detect a difference between groups for GI AEs.
Metformin is recommended in the PCOS guidelines for management of anthropometric and metabolic outcomes in women with BMI ≥25 kg/m 2 and should be used as a first-line treatment.However, it may be reasonable to consider the use of MI as an alternative to metformin in women who cannot tolerate metformin and who require management of anthropometric and metabolic outcomes.It is unclear whether there is any benefit from using a combination of metformin and MI.Data on the prevalence of use of complementary therapies in PCOS has not been updated for almost 10 years.A 2014 survey of Australian women reported that 70% of respondents used complementary medicines, which were mostly nutritional and herbal supplements (64).In the general population, individuals choose complementary medicines for a number of reasons including feeling unsatisfied with conventional therapy results, alignment with the ideology of holistic health, and for preventive health and well-being, an increased sense of control over one's health, and a perception of safety (65).Similarly, women with PCOS use complementary therapies because they wish to concurrently manage comorbidities such as sleep problems and to improve general well-being (64).
Nevertheless, 70% of women with PCOS cited disadvantages of using complementary medicine such as financial cost and lack of supporting scientific evidence (64).It is important for clinicians to be provided with evidence-based guidance and to understand the motivations for use of complementary therapies and the potential risks, including financial cost, when engaging with patients in informed discussions about use of inositol for PCOS.While complementary medicines are often perceived to be safer than pharmaceuticals, it is important to note that regulation of nutrient supplements varies globally.Most nutritional supplements are not regulated by the European Medicines Agency or US Food and Drug Administration.Studies showing safety and efficacy are generally lacking.In addition, manufacturing standards may vary resulting in inconsistent quality.The cost of inositol is also greater than the cost of metformin.Clinicians should incorporate these considerations together with scientific evidence and individual patient values and preferences when engaging in shared decision-making with their patients about the benefits and risks of using inositol for managing PCOS.
The study has several strengths including the rigorous design and methodology, endorsed by 39 organizations globally as part of the guideline.This topic was highly prioritized by clinicians, patients, and experts who participated in an extensive global prioritization process, and the review was conducted by a multidisciplinary team with substantial expertise in PCOS.Additionally, the validity and trustworthiness of our results is strengthened by the integrity checks incorporated using the RIGID framework, which is now being implemented in 2 further international guidelines.We included GRADE assessments to provide an overview of the certainty of the evidence at the outcome level, in addition to traditional study-level quality appraisal using risk of bias.Limitations of the study are that we only included studies published in English and did not search grey literature.The risk of bias of included studies was generally high, and sample sizes and study numbers were small, which precluded meta-analyses in some instances and decreased the certainty of the evidence overall.Adverse events were not reported by the majority of studies, and there was only 1 study examining adolescents, limiting comparisons in this population subgroup.

Conclusion
Inositol is extensively used and promoted as effective for management of PCOS and was prioritized by patients internationally in the 2023 international evidence-based PCOS guideline.Yet, this high-quality systematic review and meta-analysis conducted to inform the 2023 international evidence-based guideline on management of PCOS has found that the evidence supporting the use of inositol in the management of PCOS is limited and inconclusive.Some evidence suggests potential benefits of MI or DCI for certain metabolic measures and DCI for ovulation, but inositol may have no effect on other outcomes.The available evidence shows that metformin is superior to MI in improving clinical outcomes such as WHR and FG score.While MI likely causes fewer GI AEs compared with metformin, these GI AEs are typically mild and self-limited.This meta-analysis, and the guideline, should be used to guide clinical practice and to limit potential harm (including financial harm) to women from using therapies without evidence for effectiveness.Clinicians and patients should consider the uncertainty of the evidence together with individual values and preferences when engaging in shared decisionmaking regarding the use of inositol for PCOS.Given the consumer interest in using inositol, further research and clinical trials are needed to establish a more robust evidence base and to better understand the potential role of inositol in PCOS management.

Funding
The

Disclosures
S.F.W. declares that she has received royalties from UpToDate for authorship of an article and an honorarium from the Canadian Society of Endocrinology and Metabolism for a talk and that she is a member of the board of directors of the Pediatric Endocrine Society.C.T. declares that she received support for registration for the 2023 AEPCOS and ASPIRE conferences and is xhair of the Early Career Researcher Network, CRE WHIRL (unpaid).C.E. declares that she has received honoraria for speaking at complementary medicine seminars and funding from nutraceutical companies (not relating to PCOS or inositol) and that she is program lead (academic, no financial benefit) of a universitybased integrative medicine center.As a medical research institute, NICM Health Research Institute receives research grants and donations from foundations, universities, government agencies, and industry.Sponsors and donors provide untied and tied funding for work to advance the vision and mission of the institute.All other authors declare that they have no conflicts of interest.

Figure
Figure2Aand 2B describe the risk of bias across studies.Around half to one-third of included trials were judged at

Figure 1 .
Figure 1.PRISMA flowchart of study selection.Abbreviations: PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.
MI ± FA vs FA.There were 8 studies(25,27,31,34,(46)(47)(48)(49) that compared MI + FA to FA alone.Sample sizes in each arm ranged from 10 to 195 per study with a total of 432 participants in the intervention arms and 457 in the control arm across the 8 studies.The dose of MI used was 4 g daily in 6 of the studies(25,31,(46)(47)(48)(49) with 200 to 400 mcg of FA.The other 2 studies evaluated 2 g of MI daily with 200 mcg(34) or 400 mcg(27) of FA.Duration of intervention ranged from 4 to 12 weeks.

Figure 2 .
Figure 2. Risk of bias summary and graph.
MI + DCI + FA vs MI + FA.One study (n = 22) (38), with very low certainty, evaluated MI 1.1 g + DCI 27.6 mg and FA 400μg taken daily to MI 4 g + FA 400μg daily after 3 and 6 months.Both arms also included dietary interventions.There was no difference for the outcomes of hirsutism, weight, BMI, WHR, or waist circumference at 3 months or 6 months.

Table 1 . Continued Author, year, country, study design Population/Setting Sample size per group, age, BMI Intervention details Comparison/control details Co-interventions Outcomes, follow-up duration
The Journal of ClinicalEndocrinology & Metabolism, 2024, Vol.109, No. 6

Table 1 . Continued Author, year, country, study design Population/Setting Sample size per group, age, BMI Intervention details
The Journal of ClinicalEndocrinology & Metabolism, 2024, Vol.109, No. 6

Table 2 . GRADE table for comparison of DCI vs placebo Comparison: DCI vs Placebo Quality assessment No. participants No. studies Design Risk of bias Inconsistency Indirectness Imprecision Other DCI Placebo Effect, random [95% CI]
The Journal of ClinicalEndocrinology & Metabolism, 2024, Vol.109, No. 6

Table 3 . Continued Comparison: MI + FA vs FA Quality assessment No. participants No. studies Design Risk of bias
Downgraded 2 levels due to very serious risk of bias; 3 studies contribute to this outcome: 1 of the 3 included studies is at unclear risk of bias for most domains, and 1 of the studies is at unclear risk for randomization.Downgraded 2 levels due to very serious risk of bias; 5 studies contribute to this outcome: 4 have unclear risk of bias in most domains, and 1 has high risk of bias for randomization.Downgraded 1 level for serious risk of bias; 3 studies contribute to this outcome: 1 study has mostly unclear risk of bias across domains, and the other 2 studies have an unclear risk of bias in a blinding domain.
a b Downgraded 1 level for serious imprecision: small sample size (112 participants).c Downgraded 1 level due to serious risk of bias: 1 of the 2 included studies is at unclear risk of bias for most domains.d Downgraded 2 levels due to very serious imprecision: very small sample size (70 participants).e f g Downgraded 1 level for serious imprecision: small sample size for relatively rare events (160 participants).

Table 4 . GRADE table for comparison for MI vs metformin Comparison: MI vs Metformin Quality assessment No. participants No. studies Design Risk of bias Inconsistency Indirectness Imprecision Other MI Metformin Effect, random [95% CI] Favors Certainty Importance
The Journal of ClinicalEndocrinology & Metabolism, 2024, Vol.109, No. 6

Table 4 . Continued Comparison: MI vs Metformin Quality assessment No. participants No. studies Design Risk of bias Inconsistency Indirectness Imprecision Other MI Metformin Effect, random [95% CI] Favors Certainty Importance
guidelines are funded by a National Health and Medical Research Council (Australia) (NHMRC) Centre of Research Excellence scheme grant (Centre of Research Excellence in Women's Health in Reproductive Life/CRE WHIRL/ 1171592).H.T. and A.M. are supported by NHMRC fellowships.C.T. is supported by the NHMRC CRE WHIRL.No other funding was obtained for this study.