The effectiveness of nutritional supplements in improving polycystic ovary syndrome in women: a systematic review and network meta-analysis

Abstract

Background

Nutritional supplements are known to ameliorate polycystic ovary syndrome (PCOS) and have been shown to modulate endocrine and metabolic markers, oxidative stress markers and inflammatory biomarkers in patients with PCOS. A variety of nutritional supplements have been applied in clinics, but a more comprehensive ranking of their efficacy has not yet been investigated.

Objectives

To assess the comparative effectiveness of nutritional supplements in women with PCOS.

Methods

A systematic search was conducted across PubMed, EMBASE, Web of Science, and the Cochrane Library for randomized controlled trials (RCTs) that met the inclusion criteria up to October 12, 2023. We performed a network meta-analysis (NMA) to evaluate the effectiveness of various nutritional supplements on different indicators of PCOS by synthesizing both direct and indirect evidence from the trials.

Results

Seventy-nine RCTs involving 5,501 participants were enrolled in the NMA. It suggested that chromium was notably effective in improving follicle-stimulating hormone levels, total antioxidant capacity, and very low-density lipoprotein levels. Soy isoflavones were more beneficial for enhancing glutathione levels and reducing malondialdehyde levels. Inositol significantly decreased total cholesterol and triglyceride levels, while curcumin was most effective in improving low-density and high-density lipoprotein cholesterol levels. Additionally, Omega-3 was superior in reducing homeostatic model assessment of insulin resistance levels. No significant differences were observed in total testosterone, sex hormone-binding globulin, dehydroepiandrosterone, high-sensitivity C-reactive protein, and plasma nitric oxide levels between the supplements and placebo.

Conclusions

Chromium, inositol, and Omega-3 were found to be beneficial for improving lipid profile. For improving obesity, sex hormone levels, inflammatory factors and oxidative stress indicators of PCOS patients, carnitine, chromium, and soy isoflavones are effective options, respectively. These findings confirm that nutritional supplements can effectively alleviate PCOS symptoms, potentially providing valuable guidance for the clinical selection of appropriate nutritional interventions.

Trial registration

PROSPERO registration number CRD42023483534.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12958-025-01409-9.

Introduction

Polycystic ovary syndrome (PCOS), associated with reproductive and metabolic dysfunction, is a complex, familial, polygenetic metabolic condition with impacts across the lifespan in women [1]. As the most common endocrine disease in women of reproductive age, it was reported to affect 5–18% of women worldwide [2]. However, the pathogenesis of PCOS is complex and numerous factors have been reported to participate in its occurrence and development in recent years, including genetic and epigenetic predisposition, hypothalamic and ovarian dysfunction, excessive androgen exposure, insulin resistance, metabolic abnormalities, obesity and inflammation [3–5]. Furthermore, patients with PCOS are often at an increased risk of developing additional complications, especially those related to metabolic and reproductive malformations, such as type 2 diabetes (T2DM), obesity, endometrial cancer, infertility, mood and eating disorders and cardiovascular disease [6–8], significantly affecting their health and quality of life.

Given the complex pathogenesis of PCOS, its management primarily focuses on symptomatic treatment, aiming to reduce the clinical manifestations of hyperandrogenism, normalize the menstrual cycle, promote ovulation, and regulate metabolic dysregulation [9]. Currently, dietary interventions have been evaluated as a first-line treatment for patients with PCOS, but the optimal diet has not been determined. Following a proper diet and maintaining adequate nutritional status are considered essential for preventing this disease and improving patient outcomes [10]. In recent years, there has been increasing concern about whether supplementation with vitamins, vitamin-like nutrients, or minerals can promote good health outcomes for PCOS [11, 12]. It has been reported that patients with PCOS often lack many of the most common vitamins and minerals [13]. Moreover, a growing number of studies have also shown that supplementation with nutritional supplements (such as minerals, vitamins, melatonin, etc.) may help improve some adverse outcomes, psychological conditions and life status of PCOS patients [14, 15].

However, although several systematic reviews and meta-analyses have reported the benefits of nutritional supplements in patients with PCOS, the number of included studies and the variety of interventions are relatively limited, and direct comparisons among various nutritional supplements are lacking [16, 17]. Furthermore, a network meta-analysis (NMA) [18] analyzed the differences in the effects of some nutritional supplements on improving glycolipid metabolism and endocrine function in PCOS patients, but a relatively detailed analysis of certain indicators was lacked (e.g., free testosterone index, total antioxidant capacity and luteinizing hormone), and some common nutritional supplements were not enrolled in comparisons.

Therefore, we conducted this comprehensive NMA based on all available randomized controlled trials (RCTs) to compare the effectiveness of different nutritional supplements in improving sex hormones, oxidative stress and inflammatory biomarkers, and metabolic outcomes in patients with PCOS women, looking forward to providing further solid evidence for the application of nutritional supplements.

Materials and methods

This NMA was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA-NMA) [19]. The study protocol was pre-registered at the International Prospective Register of Systematic Reviews (PROSPERO), registration number CRD42023483534.

Data sources

Four electronic databases, including PubMed, EMBASE, Web of Science and the Cochrane Library of Trials, were searched from inception to October 12 th, 2023. The medical subject heading terms used in the literature search were as follows: “Polycystic ovary syndrome”, “PCOS”, “ovarian cysts”, “hyperandrogenism”, “hirsutism”, “Magnesium”, “Selenium”, “Chromium”, “Calcium”, “Zinc”, “Probiotic”, “Synbiotics”, “Vitamin”, “Coenzyme Q10”, “Cinnamon”, “Omega-3”, “Carnitine”, “Melatonin”, “Folate acid”, “Alpha lipoic acid”, “Soy isoflavones”, “Dietary soy”, “Nutritional supplements”, “Randomized control trial”, “Clinical trial”, “RCT”. The language of literature was restricted to English, and the references of the articles were also screened for relevant articles. Two authors independently screened abstracts, and then the relevant full-text articles based on inclusion and exclusion criteria. Disagreements were resolved through consensus or third-party arbitration.

Inclusion and exclusion criteria

Studies that met the following inclusion criteria were considered eligible: (1) conducted in women of reproductive age 18–49 years old diagnosed with PCOS according to the Rotterdam criteria (ESHRE/ASRM 2004) [20] or the National Institute of Child Health and Human Development (NICHD) standards [21] or the Androgen Excess Society criteria (AES) [22], without racial restrictions; (2) RCTs and the language limited to English; (3) studies that compared nutritional supplements (alone or in combination) with a control group (placebo for the control group and other adjunctive agents were maintained consistently between the groups), and patients in the intervention group received at least one of the following nutritional supplements (in any form, dose, or duration): Magnesium (Mg), Selenium (Se), Chromium (Cr), Calcium (Ca), Zinc (Zn), Probiotic, Synbiotics, Vitamin (vit), Coenzyme Q10 (CoQ10), Cinnamon, Omega-3, Carnitine, Melatonin, Folate, Soy isoflavones, Dietary soy, or a combination of these supplements, and placebo or one of the above nutritional supplements was used in the control group; (4) studies reported at least one parameter including fasting plasma glucose (FPG), Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), Insulin, triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C), very low-density lipoprotein (VLDL-C), high sensitivity C-reactive protein (hs-CRP), plasma nitric oxide (NO), total antioxidant capacity (TAC), glutathione (GSH), malondialdehyde (MDA), total testosterone (TT), sex hormone-binding globulin (SHBG), and dehydroepiandrosterone (DHEAS), follicle-stimulating hormone (FSH), luteinizing hormone (LH), free testosterone index (FAI).

The exclusion criteria were as follows: (1) cohort studies, review articles, descriptive studies and opinion articles; (2) women who were pregnant or intended to be pregnant; (3) studies where data was incomplete or available data cannot be extracted and converted to mean ± SD or cannot be combined.

Data extraction and quality assessment

All studies included were independently screened from the literature according to the preassigned inclusion and exclusion criteria by two investigators. The characteristics of the enrolled studies included basic information (first author’s name, publication date, country), the characteristics of participants (number of patients, age, BMI), interventions, cycles and outcomes.

The bias risk of all eligible RCTs was assessed by two independent investigators using the modified Jadad scale [23] across 4 items (sequence generation, allocation concealment, blinding, withdrawal or loss of visits) [24]. Meanwhile, we used Cochrane’s risk of bias tool was also used to assess bias risk of RCTs, including randomization and sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, completeness of outcome data, selective outcome reporting, and other sources of bias. Any discrepancies were resolved by a consensus discussion between these two investigators or with the third investigator if necessary.

Data synthesis and statistical analysis

The treatment effects of several interventions were compared directly and indirectly by NMA. The summary odds ratio (OR) with 95% confidence interval (CI) was used to estimate the dichotomous outcomes, and the summary Mean Difference (MD) with 95%CI was used to estimate continuous data. If the 95%CI of the OR did not contain 1, or the 0 was not within 95%CI of the MD, the differences between groups were considered statistically significant. The mean and standard deviation (SD) of each outcome indicators at baseline and post-intervention were extracted from each trial for both the intervention and control groups. We calculated SD using standard error (SE) and 95% CI by equations from Sect. 6.5.2.2 of the Cochrane Handbook when was available but SD was not directly reported. [$SD=SE\times \sqrt{N}$ or $SD=\sqrt{N}$ × (upper limit − lower limit)/3.92] [24]. The equation by Follmann et al. was used to estimate change-from-baseline SDs when baseline-to-post-intervention SD changes were missing, assuming the correlation coefficient between baseline and postintervention lipid and lipoprotein values was 0.50, as described in Cochrane Handbook chapter 6.5.2.8

$[\text{SDE change}=\sqrt{{\mathit{SD}}_{E,baseline}^2+{\mathit{SD}}_{E,final}^2-(2\times 0.50\times {\mathit{SD}}_{E,baseline}\times {\mathit{SD}}_{E,final}}$] [25–27]. The correlation coefficient (Corr) of 0.50 was selected based on a similar previous meta-analysis and calculations by Follmann et al. and Koch et al. [26, 27]. Sensitivity analysis for different Corr values (Corr = 0.2 and Corr = 0.8) yielded similar results. The Cochran Q test and I² statistics were used to assess the heterogeneity of the effect of each parameter on treatment effect between trials. An I² value below 25% was low heterogeneity, 25~75% was moderately heterogeneous, and higher than 75% was high heterogeneity. If I² ≤ 50%, it means that statistical heterogeneity among the study results is small, and the fixed effects model is selected for network meta-analysis; if I² > 50%, it means that there is significant statistical heterogeneity among the study results, and the randomized effects model is used [28]. The frequentist method was used to conduct NMAs of randomized RCTs. The node splitting method was used to assess the local inconsistencies of the model by dividing the evidence for specific comparisons into direct and indirect evidence. There was no significant inconsistency in the node splitting analysis (p> 0.1). Furthermore, the variance between the studies were evaluated through τ², and subgroup analyses were also conducted based on treatment period to explore potential sources of heterogeneity [29]. Moreover, the sensitivity analyses were performed by excluding highly biased RCTs. Each outcome was ranked using the surface and average ranking under the cumulative rank curve (SUCRA). Publication bias was assessed using Egger’s tests [30] and funnel plot.

The STATA 15.0 software was used for statistical analysis of NMA, assessment of global inconsistency and local inconsistency, and Egger’s test, and funnel plot.

Results

Characteristics of included studies

A total of 1844 potential studies were identified through the initial and systematic searches. After removing duplicates, 1366 articles were screened by title and abstract. We then considered 264 potentially eligible studies for inclusion and retrieved full-text articles. Finally, 79 RCTs enrolling 5501 participants were included for the NMA. The flow of screening process was shown in Fig. 1.

Fig. 1.

PRISMA flow diagram of search

Among the included 79 RCTs, 67 were two-arm, 4 were three-arm and 8 were four-arm studies. The basic characteristics of included studies are presented in Table 1. Thirty treatments were analyzed, including vit D (the vit D (a) dose was 50000 IU, vit D (b) was 1000 IU, and vit D (c) was 4000 IU), omega-3, Se, curcumin, carnitine, Cr, synbiotics, Mg, Zn, CoQ10, melatonin, Soy Isoflavones, inositol, Probiotic, Ca, folate acid, vit E, prebiotic, vit D + omega-3, Ca + vit D, vit D + Probiotic, omega-3+vit E, CoQ10+vit E, Mg + vit E, melatonin + Mg, Mg + zinc, Cr + carnitine, Probiotic + Se, Ca + vit D + vit K, Mg + Zn + Ca + vit D.

Table 1.

Characteristic of included studies

Author, year	Country	Sample size (Experiments/Control)	Mean age (year)		Mean BMI (kg/m2)		Experimental group	Control group	Duration (weeks)	Outcomes
			Treatment	Control	Treatment	Control
Bahramian 2023[31]	Iran	18/18	23.60 ± 3.42	22.35 ± 3.12	25.47 ± 8.08	25.49 ± 6.35	Vit D 50000 IU/qw	Placebo (paraffin oil)	8	a, b, c, d, e, f, g, i, j, s
Maktabi 2017a [32]	Iran	35/35	/	/	/	/	Vit D 50000 IU/biw	placebo	12	a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r
Jamilian 2017a [33]	Iran	30/30	26 ± 5	25 ± 5	33 ± 5	30 ± 6	Vit D 1000 IU/qd	placebo	12	h, i, j, l. m, n, o, p, q, r, s
Jamilian 2017a [33]	Iran	30/30	28 ± 5	25 ± 5	31 ± 6	30 ± 6	Vit D 4000 IU/qd	placebo	12	i, j, l. m, n, o, p, q, r, s
Foroozanfard 2015a [34]	Iran	26/26	/	/	/	/	Vit D 50000 IU/qw	placebo	8	l, m, n, o, q
Garg 2015 [35]	India	15/17	22.0 ± 4.61	22.8 ± 4.56	26.8 ± 4.56	26.7 ± 6.11	Vit D 4000 IU/qd	placebo	26	b, c, d, g, p, r
Dastorani 2018 [36]	Iran	20/20	29.9 ± 4.4	30.1 ± 3.4	27.7 ± 3.9	28.4 ± 2.6	Vit D 50000 IU/qw	Placebo (paraffin)	8	a, b, c, d, e, f, g, i, k
Abootorabi 2018 [37]	Iran	19/17	26.21 ± 4.62	22.76 ± 4.40	/	/	Vit D 50000 IU/qw	placebo	8	g, i, k
Foroozanfard 2017 [38]	Iran	30/30	/	/	/	/	Vit D 1000 IU/qd	placebo	12	c
Foroozanfard 2017 [38]	Iran	30/30	/	/	/	/	Vit D 4000 IU/qd	placebo	12	a, b, c, d, e, f, g, i, k
Sfidvajani 2018 [39]		26/28	28.43 ± 6.27	27.83 ± 5.71	31.13 ± 4.99	31.61 ± 4.91	Vit D 50000 IU/qw	Placebo (paraffin oil)	12	h, j, p, r, s
Ardabili 2012 [40]	Iran	24/26	26.8 ± 4.7	27.0 ± 3.7	29.10 ± 4.62	28.28 ± 3.51	Vit D 50000 IU/tiw	placebo	8	f, i, k, s
Asemi 2015 [41]	Iran	26/26	25.6 ± 4.4	24.3 ± 5.2	29.3 ± 3.9	27.5 ± 5.2	Vit D 50000 IU/qw	placebo	8	a, b, c, d, e, f, g, i, k, s
Al-Bayyari 2021[42]	Jordan	29/29	23.6 ± 4.3	23.9 ± 6.0	27.3 ± 1.9	26.9 ± 1.6	Vit D 50000 IU/qw	placebo	12	h, i, j, r
Bahramian 2023[31]	Iran	20/18	22.29 ± 3.63	22.35 ± 3.12	25.36 ± 4.97	25.49 ± 6.35	omega-3 1200 mg/qd	Placebo (paraffin oil)	8	a, b, c, d, e, f, g, i, j, s
Mirmasoumi 2018 [43]	Iran	30/30	28.4 ± 6.4	27.0 ± 3.2	26.9 ± 5.1	26.7 ± 5.3	omega-3 1000 mg/qd	Placebo (liquid paraffin)	12	a, b, c, d, e, f, g, h, i, j, k, n, p, q, r, s
Khani 2017 [44]	Iran	43/44	31.04 ± 5.04	29.23 ± 6.73	31.8 ± 3.61	31.79 ± 3.6	omega-3 2000 mg/qd	Placebo (olive oil)	24	a, b, c, d, g
Nadjarzadeh 2015 [45]	Iran	39/39	/	/	31.46 ± 5.74	31.88 ± 3.86	omega-3 900 mg/qd	Placebo (paraffin)	8	s, t, u
Jamilian 2015a [46]	Iran	35/35	25.4 ± 5.1	25.7 ± 4.8	25.0 ± 3.7	25.2 ± 4.1	Se 200 µg/qd	Placebo (cellulose)	8	a, b, c, d, e, f, g, i, k, s
Razavi 2016[47]	Iran	32/32	25.4 ± 4.9	25.1 ± 4.5	25.3 ± 4.3	24.7 ± 3.5	Se 200 µg/qd	Placebo (cellulose)	8	l, m, n, o, p, q, t, u
Hosseinzadeh 2016 [48]	Iran	26/27	29.23 ± 4.90	28.90 ± 6.08	27.4 ± 4.49	28.39 ± 3.74	Se 200 µg/qd	Placebo (starch)	12	f, g, h, i, j, r
Modarres 2022 [49]	Iran	20/20	32.6 ± 4.6	32.8 ± 4.1	25.2 ± 4.1	25.6 ± 2.5	Se 200 µg/qd	Placebo (starch)	8	a, b, c, d, e, g, i, k, l, m, o
Rashidi 2020 [50]	Iran	34/32	29.4 ± 5.3	28.6 ± 5.5	28.3 ± 5.2	29.5 ± 5.4	Se 200 µg/qd	Placebo (rice flour)	12	a, b, c, d, j
Heshmati 2021 [51]	Iran	34/33	30.97 ± 5.20	30.75 ± 7.97	28.73 ± 4.92	27.28 ± 4.82	Curcumin 500 mg/tid	Placebo (maltodextrin)	12	g, i, k, p, s, t, u
Sohaei 2019 [52]	Iran	27/24	29.40 ± 5.33	29.58 ± 5	29.67 ± 3.72	31.32 ± 4.6	Curcumin 500 mg/bid	placebo	6	a, b, c, d, f, g, i, k, q
Sohrevardi 2021 [53]	Iran	48/50	29 ± 2	28.8 ± 2.46	27.2 ± 2.2	27.01 ± 1.65	curcumin 80 mg/qd + metformin500 mg/qd	MI 500 mg/tid	12	a, b, c, d, g, i, k, p, r, t, u
Jamilian 2020 [54]	Iran	24/26	28.6 ± 4.7	27.2 ± 3.4	27.4 ± 3.9	26.4 ± 3.8	curcumin 500 mg/tid	Placebo (starch)	12	a, b, c, d, e, f, g, i, k, s
Samimi 2016a [55]	Iran	30/30	24.8 ± 5.5	25.5 ± 5.7	29.1 ± 3.4	28.9 ± 3.9	carnitine 250 mg	Placebo (cellulose)	12	a, b, c, d, e, f, g, i, k, p, s
Talari 2019 [56]	Iran	30/30	23.6 ± 4.6	25.0 ± 5.4	29.4 ± 4.1	28.2 ± 6.5	carnitine 250 mg	placebo	12	n, q, s
Sangouni 2022 [57]	Iran	28/28	30.7 ± 6.7	30.8 ± 6.6	31.0 ± 4.7	30.8 ± 3.6	carnitine 1000 mg	placebo(starch)	12	a, d, j, k, s
Nasri 2018[58]	Iran	30/30	25.7 ± 5.5	25.9 ± 5.2	27.4 ± 4.0	27.2 ± 5.3	Synbiotic 2 × 109 CFU/g	placebo	12	h, j, l, m, n, o, p, q, r, s
Karimi 2018 [59]	Iran	44/44	28.1 ± 5.5	29.00 ± 5.1	32.89 ± 6.11	32 ± 4.23	Synbiotic 500 mg	Placebo (starch and maltodextrin)	12	g, i, k
Karimi 2020[60]	Iran	44/44	28.1 ± 5.5	29.00 ± 5.1	32.89 ± 6.11	32 ± 4.23	Synbiotic 500 mg	Placebo (starch and maltodextrins)	12	a, b, c, d, s
Strugała 2021 [61]	Poland	20/19	30.8 ± 4.02	29.1 ± 5.67	33.4 ± 4.47	34.4 ± 5.23	Synbiotic 4粒	placebo	12	a, b, c, d, j, p, r, s, t, u
Samimi 2019[62]	Iran	30/30	27.0 ± 5.6	27.3 ± 6.1	27.3 ± 3.8	27.5 ± 5.3	Synbiotic 2 × 109 CFU/g	Placebo (starch)	12	a, b, c, d, e, f, g, i, k, s
Darvishi 2021 [63]	Iran	34/34	30.4 ± 5.82	28.6 ± 4.82	29.43 ± 5.69	28.47 ± 3.55	synbiotic 500 mg	placebo	8	a, b, c, d, g, f, i
Ashoush 2016 [64]	Egypt	44/41	24.7 ± 3.7	24.6 ± 4	30 ± 3.3	29.7 ± 3.1	Cr 1000ug	placebo	26	l, m, n, o, p, q
Jamilian 2015c [65]	Iran	30/30	/	/	/	/	Cr 200 µg/qd	cellulose	8	a, b, c, d, e, f, g, i, k, s, t, u
Jamilian 2016b [66]	Iran	32/32	24.9 ± 5.0	24.4 ± 4.4	25.8 ± 5.1	25.4 ± 4.0	Cr 200 µg/qd	cellulose	8	n, q, s
Jamilian 2018c [67]	Iran	20/20	30.3 ± 4.6	32.3 ± 3.0	27.4 ± 3.4	26.6 ± 5.1	Cr 200 µg/qd	placebo	8	a, b, c, d, e, f, g, i, k, l, m, o, s
Siavashani 2018 [68]	Iran	20/20	33.3 ± 2.7	33.8 ± 1.9	27.7 ± 2.5	27.0 ± 3.9	Cr 200 µg/qd	placebo	8	g
Mousavi 2022 [69]	Iran	21/20	25.57 ± 4.88	26.200 ± 5.72	27.99 ± 3.22	26.94 ± 3.83	Mg 250 mg/qd	placebo	12	l, m, s
Shahmoradi 2023 [70]	Iran	20/20	27.45 ± 5.04	29.00 ± 4.24	29.73 ± 6.44	30.61 ± 5.03	Mg 250 mg/qd	placebo	8	a, b, c, d, f
Alizadeh 2021 [71]	Iran	21/20	25.57 ± 4.88	26.200 ± 5.72	27.99 ± 3.22	26.94 ± 3.83	Mg 250 mg/qd	placebo	8	b, c, d, g
Mahsa 2022 [72]	Iran,UK	32/32	31.69 ± 5.41	32.44 ± 6.42	26.89 ± 4.68	26.63 ± 4.06	Mg 250 mg/qd	Placebo (starch)	10	p
Jamilian 2015b [73]	Iran	24/24	/	/	/	/	Zn 50 mg/qd	Placebo (starch)	8	l, m, n, o, p, q, t, u
Foroozanfard 2015b [74]	Iran	26/26	24.7 ± 3.7	25.7 ± 5.2	26.3 ± 5.7	25.2 ± 3.9	Zn 50 mg/qd	Placebo (starch)	8	a, b, c, d, e, f, g, i, k
Izadi 2018 [75]	Iran	22/21	27.64 ± 5.2	26.0 ± 4.53	28.97 ± 2.95	28.73 ± 3.39	CoQ10 200 mg/qd	placebo	8	f, h, i, r
Samimi 2016b [76]	Iran	30/30	24.5 ± 4.3	25.3 ± 5.7	27.1 ± 4.3	27.9 ± 6.0	CoQ10 100 mg/qd	Placebo (cellulose)	12	a, b, c, d, e, f, g, i, k, s
Karamali 2022[77]	Iran	28/27	27.1 ± 6.5	29.0 ± 6.9	27.2 ± 3.9	28.2 ± 4.8	CoQ10 100 mg/qd	Placebo (starch)	12	j, l, m, o, p, q, r
Izadi 2019 [78]	Iran	22/21	27.64 ± 5.2	26.0 ± 4.53	28.97 ± 2.95	28.73 ± 3.39	CoQ10 200 mg/qd	placebo	8	a, b, c, d
Taghizadeh 2021 [79]	Iran	22/21	27.64 ± 5.19	26.00 ± 4.53	28.97 ± 2.95	28.73 ± 3.16	CoQ10 200 mg/qd	placebo	8	q
Mousavi 2022 [69]	Iran	21/20	25.57 ± 4.99	26.200 ± 5.72	28.40 ± 3.86	26.94 ± 3.83	MI 6 mg	placebo	12	l, m, s
Shabani 2019 [80]	Iran	29/29	26.5 ± 3.5	26.0 ± 3.3	27.1 ± 4.6	27.8 ± 4.7	MI 10 mg/d	placebo	12	a, b, c, d, e, f, i, k, s
Alizadeh 2021 [71]	Iran	21/20	25.57 ± 4.99	26.200 ± 5.72	28.40 ± 3.86	26.94 ± 3.83	MI 3 mg	placebo	8	b, c, d, g
Jamilian 2019c [81]	Iran	28/28	28.7 ± 2.1	28.3 ± 2.3	29.1 ± 4.6	29.2 ± 3.5	MI 5 mg/tid	placebo	12	h, l, m, n, o, q, r, s
Jamilian 2016a [82]	Iran	35/35	27.5 ± 6.4	25.9 ± 4.8	24.9 ± 5.6	26.7 ± 4.7	Soy Isoflavones 50 mg/d	placebo	12	a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s
Karamali 2018a [83]	Iran	30/30	25.0 ± 5.7	25.9 ± 5.0	28.7 ± 6.1	27.8 ± 3.9	Soy 30 g	placebo	8	a, b, c, d, e, g, l, m, n, o, q, r, t, u
Janati 2022 [84]	China	40/40	34.21 ± 4.9	35.37 ± 5.4	28.43 ± 2.3	27.82 ± 3.4	inositol + olic acid 200 μg bid)	metformin + folic acid (400 μg/day)	8	l, m
Jamilian 2017b [85]	Iran	30/30	27.7 ± 5.2	25.9 ± 4.8	25.8 ± 3.8	27.1 ± 6.4	inositol + folic acid	metformin	12	j, n, p, q, r, s
Shokrpour 2019 [86]	Iran	26/27	28.3 ± 4.9	27.7 ± 3.2	28.1 ± 3.1	27.3 ± 3.3	inositol + 200 mg folic acid	500 mg metformin tid	12	a, b, c, d, e, f, g, i, k, s
Costantino 2009 [87]	Italy	23/19	28.8 ± 1.5	27.1 ± 1.4	22.8 ± 0.3	22.5 ± 0.3	inositol + folic acid	Placebo (folic acid 400 mcg)	6	a, b, j, p, r
Artini 2013 [88]	Italy	25/25	34.9 ± 2.1	36.2 ± 2.3	26.5 ± 6.1	26.3 ± 6.8	Inositol + folic acid 200 mg	folic acid	12	f, t, u
Kaur 2022 [89]	India	52/52	23.6 ± 3.9	24.4 ± 4.8	26.5 ± 5.3	27.8 ± 4.8	Probiotic 2 × 109 CFU/g	Placebo (maltodextrin)	26	j, l, m, n, o, p, q, r
Karamali 2018b [90]	Iran	30/30	27.2 ± 4.6	27.7 ± 4.7	23.7 ± 3.6	23.6 ± 3.5	Probiotic 2 × 109 CFU/g	Placebo (starch)	12	g, h, p, r
Shoaei 2015 [91]	Iran	32/33	26.5 ± 0.1	25.72 ± 0.1	26.06 ± 0.1	25.8 ± 0.1	Probiotic 500 mg	Placebo (starch and maltodextrins)	8	f, g, i, k
Ghanei 2018 [92]	Iran	30/30	30.06 ± 1.06	28.96 ± 0.98	26.84 ± 0.53	25.97 ± 0.42	Probiotic 2 × 109 CFU/g	Placebo (maltodextrin)	12	q, s
Foroozanfard 2015a [34]	Iran	26/26	/	/	/	/	Calcium 1000 mg	placebo		l, m, n, o, q
Asemi 2015 [41]	Iran	26/26	25.0 ± 6.7	24.3 ± 5.2	28.3 ± 4.7	27.5 ± 5.2	Calcium	placebo	8	a, b, c, d, e, f, g, i, k, s
Bahmani 2014 [93]	Iran	23/23	24.1 ± 5.4	25.1 ± 4.9	26.1 ± 6.2	29.0 ± 5.9	folate 1 mg/qd	placebo	8	l, m, n, o, q
Bahmani 2014 [93]	Iran	23/23	24.9 ± 5.9	25.1 ± 4.9	27.6 ± 5.7	29.0 ± 5.9	folate 5 mg/qd	placebo	8	l, m, n, o, q
Asemi 2014 [94]	Iran	27/27	24.3 ± 5.0	24.7 ± 5.0	27.2 ± 5.0	27.9 ± 4.7	folate 1 mg/qd	placebo	8	a, b, c, d, e, f, g, i, k, s
Asemi 2014 [94]	Iran	27/27	25.1 ± 4.5	24.7 ± 5.0	29.3 ± 4.6	27.9 ± 4.7	folate 5 mg/qd	placebo	8	a, b, c, d, e, f, g, i, k, s
Izadi 2018 [75]	Iran	22/21	27.18 ± 5.77	26.0 ± 4.53	29.28 ± 4.24	28.73 ± 3.39	VitE 400 IU	placebo	8	f, h, i, r
Izadi 2019 [78]	Iran	22/21	27.18 ± 5.77	26.0 ± 4.53	29.28 ± 4.24	28.73 ± 3.39	VitE 400 IU	placebo	8	a, b, c, d
Bahramian 2023 [31]		20/18	24.62 ± 3.11	22.35 ± 3.12	24.85 ± 2.62	25.49 ± 6.35	Vit D + omega-3	placebo	8	a, b, c, d, f, g, i, j, s
Jamilian 2018c [95]	Iran	30/30	26.8 ± 4.4	25.1 ± 3.7	27.4 ± 3.9	27.1 ± 7.0	Vit D 50000 IU/biw + omega-3 1000 mg/d	placebo	12	j, l. m. n. o, q, r, s
Asemi 2015 [41]	Iran	26/26	24.9 ± 5.1	24.3 ± 5.2	27.3 ± 5.3	27.5 ± 5.2	Ca 1000 mg + Vit D 50000 IU/qw	placebo	8	a, b, c, d, e, f, g, i, k
Foroozanfard 2015a [34]	Iran	26/26	/	/	/	/	Ca 1000 mg + Vit D 50000 IU/qw	calcium placebo + vitamin D placebo	8	l, m, n, o, q
Ostadmohammadi 2019 [96]	Iran	30/30	24.4 ± 4.7	25.4 ± 5.1	24.3 ± 4.2	25.1 ± 4.9	Vit D 50000 IU/biw + Probiotic 8*109 CFU	Placebo (corn oil and starch)	12	j, l. m, n, o, q, r, s
Sadeghi 2019 [97]	Iran	32/30	26.67 ± 3.35	26.98 ± 3.78	27.67 ± 5.09	27.89 ± 5.40	omega-3 + Vit E	Placebo (oral paraffin)	8	l, m
Ebrahimi 2017 [98]	Iran	34/34	23.8 ± 4.6	25.2 ± 5.2	28.0 ± 4.3	28.5 ± 6.6	omega-3 1000 mg + Vit E 400 IU	placebo	12	f, g, h, i, j, k, p, r, s, t, u
Rahmani 2017 [99]	Iran	34/34	24.9 ± 5.5	26.6 ± 5.6	28.4 ± 4.4	29.0 ± 6.5	omega-3 + Vit E	placebo	12	a, b, c, d, e, l, m, o, s
Izadi 2018 [75]	Iran	21/21	28.33 ± 5.52	26.0 ± 4.53	29.28 ± 3.23	28.73 ± 3.39	CoQ10 + Vit E	placebo	8	f, h, i, r
Izadi 2019 [78]	Iran	21/21	28.33 ± 5.52	26.0 ± 4.53	29.28 ± 3.23	28.73 ± 3.39	CoQ10 + Vit E	placebo	8	a, b, c, d
Shokrpour 2019 [100]	Iran	30/30	27.2 ± 7.1	26.0 ± 3.7	27.1 ± 4.2	27.9 ± 4.2	Mg 250 mg/qd + Vit E 400 mg/qd	placebo	12	j, l, m, n, o, q, r, s
Jamilian 2018b [101]	Iran	30/30	29.2 ± 7.2	28.3 ± 3.8	25.5 ± 3.5	26.0 ± 4.7	Mg 250 mg/qd + Vit E 400 mg/qd	placebo	12	a, b, c, d, e, f, g, i, k, s
Mousavi 2022 [69]	Iran	22/20	28.22 ± 6.38	26.200 ± 5.72	29.64 ± 3.71	26.94 ± 3.83	MI 6 mg/qd + Mg 250 mg/qd	placebo	12	l, m, s
Alizadeh 2021 [71]	Iran	22/20	28.22 ± 6.38	26.200 ± 5.72	29.64 ± 3.71	26.94 ± 3.83	MI + Mg	placebo	8	b, c, d, g
Ebrahimi 2018 [102]	Iran	30/30	/	/	/	/	Mg 250 mg/qd + Zn 50 mg/bid	placebo	12	l, m, n, o, q
Jamilian 2019a [103]	Iran	27/27	29.6 ± 4.3	27.4 ± 5.3	29.1 ± 2.8	28.0 ± 2.3	Cr 200ug/qd + carnitine 1000 mg/qd	Placebo (starch)	12	a, b, c, d, e, f, g, i, k, s
Jamilian 2019b [104]	Iran	26/27	26.3 ± 5.0	27.5 ± 3.5	/	/	Cr 200ug/qd + carnitine 1000 mg/qd	placebo	12	j, l, m, n, o, q, r
Jamilian 2018a [105]	Iran	30/30	26.0 ± 5.3	25.6 ± 3.8	24.6 ± 3.3	24.0 ± 3.0	Probiotic 8*109 + Se 200ug/qd	placebo	12	j, l, m, n, o, q, r, s
Razavi 2016 [106]	Iran	27/27	/	/	/	/	Ca + Vit D + Vit K	placebo	8	l, m, n, o, p, q, t, u
Karamali 2017 [107]	Iran	28/27	23.5 ± 4.2	23.3 ± 3.4	24.2 ± 4.8	24.3 ± 3.9	Ca 200ug/d + Vit D 200 IU/tid + Vit K 90 µg/tid	Placebo (cellulose)	8	a, b, c, d, e, f, g, i, k, s
Jamilian 2017c [108]	Iran	30/30	/	/	/	/	Mg 100 mg + Zn 4 mg + Ca 400 mg + Vit D 200 IU bid	placebo	12	a, b, c, d, e, f, g, i, k
Shamasbi 2018 [109]	Iran	31/31	28 ± 6.3	26.0 ± 6.2	25.0 ± 4.4	27.0 ± 5.6	prebiotic 20 g	Placebo (maltodextrin)	12	a, b, c, d, g, p, q

T: treatment group, C: control group, VitD: vitamin D, Mg: Magnesium, Se: Selenium, Cr: Chromium, Ca: Calcium, Zn: Zinc, Soy: Soy isoflavones, MI: melatonin, CoQ10: Coenzyme Q10, VitE: vitamin E, a: total cholesterol (TC), b: triglyceride (TG), c: high-density lipoprotein cholesterol (HDL-C); d: low-density lipoprotein cholesterol (LDL-C); e: very low-density lipoprotein cholesterol(VLDL-C); f: insulin, g: fasting plasma glucose (FPG), h: free testosterone index (FAI), i: Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), j: sex hormone-binding globulin (SHBG), k: QUICKI, l: malondialdehyde (MDA), m: total antioxidant capacity (TAC), n: plasma nitric oxide (NO), o: glutathione (GSH), p: dehydroepiandrosterone (DHEAS), q: high sensitivity C-reactive protein (hs-CRP), r: total testosterone (TT), s: BMI, t: luteinizing hormone (LH), u: follicle-stimulating hormone (FSH)

Methodological quality of the studies

According to the Jadad Quality Assessment Scale score, 76 studies were considered to be high-quality (Jadad score >3), 3 studies were low-quality (Jadad score ≤3), and the details are shown in Appendix 3. A bias risk rating for each study was provided in section Appendix 8. Two studies did not mention random allocation methods, six studies did not mention allocation concealment methods, and one study that did not mention blinding were rated at high risk of bias. The remaining studies were of moderate to high quality (Fig. 2).

Fig. 2.

Bias risk map

Network structure diagrams

Multiple different nutritional interventions were covered in these included studies. The network structure diagram of the direct association between the different nutritional interventions and placebo is shown in the Appendix Figure S1. In order to minimize the bias of the results, interventions with study numbers greater than two (except for FAI, LH, and FSH) were selected to perform direct and indirect comparisons of nutritional interventions in this study, and the corresponding graph was also provided in Fig. 3. In terms of the geometry of the network map, head-to-head trials between nutritional supplement interventions were relatively few, and most outcomes were mainly dependent on placebo-controlled, resulting in star-shaped network plots. In addition, the thickness of the line is proportional to the amount of literature or the number of comparisons, and the diameter of the circle is proportional to the sample size included in the network meta-analysis.

Fig. 3.

Network structure diagrams (Except for FAI, LH, and FSH, comparison between nutritional supplements with more than two studies)

Results of the network meta-analysis

Hormonal outcomes

The Cr (MD −2.80, 95% CI −4.75 to −0.85) was more effective in increasing FSH compared to the placebo, and Cr was more effective in increasing FSH (MD −1.60, 95% CI −3.11 to −0.09) than the synbiotics. Ca + vit D(a) + vit K (MD −11.97, 95% CI −23.26 to −0.68) were significantly reduced LH levels compared to Zn (Fig. 4). NMA results showed that none of the supplements significantly improved the levels of TT, SHBG, FAI, and DHEAS, and there were no significant differences in the relative effects among the various supplements (Appendix Table S3-S5).

Fig. 4.

Results of network meta-analysis for LH/FSH. Notes: FSH: estimates are presented as MD and 95% CI in parentheses. Comparisons should be read from left to right and the estimate is in the cell in common between the column-defining treatment and the row-defining treatment. MD of less than 0 indicates that the outcome is more likely with treatment (column) than reference (row). LH: estimates are presented as MD and 95% CI in parentheses. Comparisons should be read from right to left and the estimate is in the cell in common between the column-defining treatment and the row-defining treatment. MD of more than 0 indicates that the outcome is more likely with treatment (column) than reference (row). Significant results are presented in bold

Appendix 5 shows the mean values of SUCRA for establishing the hierarchical ranking of different treatments on our outcomes. For SHBG (SUCRA = 87,4%) and DHEAS (SUCRA = 81.6%), inositol may be the best nutritional supplement. Synbiotics may be more effective and FSH (SUCRA = 96.0%). Probiotic may be more effective for TT (SUCRA = 84.6%). Furthermore Ca + vit D +vit K was the best nutritional supplement to reduce the level of LH (SUCRA = 90.5%).

Inflammatory factors and oxidative stress indicators

Six nutritional supplements across 12 RCTs were analyzed using NMA to evaluate the effects of nutritional supplements on MDA in PCOS patients (Fig. 3o). The NMA showed that soy isoflavones (MD −1.44; 95% CI −2.19 to −0.69) and vit D(a) (MD −0.86; 95% CI −1.67 to −0.06) significantly reduced MDA in PCOS patients compared with placebo. In indirect comparisons of other nutritional supplements, soy isoflavones (MD −1.55; 95% CI −2.54 to −0.56) improved MDA better compared to MI, with statistically significant differences (Fig. 5). In terms of SUCRA ranking, soy isoflavones were the best nutritional supplement to reduce the MDA (SUCRA = 94.7%) (Appendix Table S9).

Fig. 5.

Results of network meta-analysis for MDA. Notes: estimates are presented as MD and 95% CI in parentheses. Comparisons should be read from left to right and the estimate is in the cell in common between the column-defining treatment and the row-defining treatment. MD of less than 0 indicates that the outcome is more likely with treatment (column) than reference (row). Significant results are presented in bold

The inflammatory factors and indicators of oxidative stress were evaluated, including TAC, NO, hs-CRP, and GSH levels. The network plot was shown in Fig. 3. In total, the relative effects of 6 nutritional supplements on TAC levels were evaluated from 12 RCTs, the relative effects of four nutritional supplements on the levels of NO were evaluated from 6 RCTs, ten RCTs evaluated the effect of five nutritional supplements on hs-CRP, and the relative effects of four nutritional supplements on GSH levels were analyzed from eight RCTs. The NMA showed that most nutritional supplements increase the decline in TAC and NO levels during patient treatment relative to placebo, but the differences were not statistically significant. In addition, soy isoflavones significantly increased GSH in PCOS patients compared to placebo and other nutritional supplements (Appendix Table S10-S12). The SUCRA ranking showed that Cr may be the best choice to maintain the TAC (SUCRA = 96.6%), soy isoflavones were the best nutritional supplement to increase NO (SUCRA = 81.4%) and GSH (SUCRA = 98.7%) (Appendix 5).

Metabolic markers

Twenty-four RCTs investigated the relative impacts of nine nutritional supplements on improving the HOMA-IR. It was observed that, compared with placebo, omega-3 (MD −1.08; 95% CI −2.14 to −0.02) and Cr (MD −0.93; 95% CI −1.75 to −0.11) better improved HOMA-IR in PCOS patients (Fig. 6). According to our present results of SUCRA, omega-3 was the best nutritional supplement for improving HOMA-IR (SUCRA = 78.6%) (Appendix Table S21).

Fig. 6.

Results of network meta-analysis for HOMA-IR. Notes: estimates are presented as MD and 95% CI in parentheses. Comparisons should be read from left to right and the estimate is in the cell in common between the column-defining treatment and the row-defining treatment. MD of less than 0 indicates that the outcome is more likely with treatment (column) than reference (row). Significant results are presented in bold

Se (MD 0.02, 95%CI 0.01 to 0.04) was more beneficial than placebo in improving QUICKI (Appendix Figure S8). The outcomes were analyzed in 17 studies involving eight interventions. According to the SUCRA and NMA results, Se (SUCRA = 85.8%), Synbiotics (SUCRA = 75.8%), and carnitine (SUCRA = 65.5%) were likely to be more effective in improving QUICKI (Appendix Table S22).

As shown in Appendix 4, 27 studies (1272 patients) and 26 studies (1543 patients), evaluated the effects of different nutritional supplements on levels of TG and TC, respectively. When compared with the placebo, inositol [(MD_TG −52.09, 95% CI −76.67 to −27.51) and MD_TC −31.91, 95% CI −48.12 to −15.70)] significantly reduced the TC and TG in patients with PCOS. Moreover, inositol could better improve TG and TC levels in PCOS patients compared to other nutritional supplements, as detailed in Appendix Figures S9-S10. Furthermore, the analysis results of LDL-C were based on 28 trials with 1640 participants, HDL-C was based on 1524 participants from 26 studies, and VLDL-C was based on 506 PCOS patients from 9 studies. When compared with the placebo, curcumin (MD – 12.03; 95% CI −20.55 to −3.51]) and CoQ10 (MD − 9.81; 95% CI −19.20 to −0.43) significantly reduced LDL-C Levels, and curcumin (MD 5.31; 95% CI 1.99 to 8.64) increased HDL-C levels of patients during treatment. Compared with placebo, the inclusion of nutritional supplements mostly improved VLDL-C levels and the difference was statistically significant. When comparing between nutritional supplements, chromium had advantages over selenium and vitamin D(a).

Other metabolic markers (Appendix 4) results of network meta-analysis for were evaluated, including FPG and insulin levels. Vit D(a) (MD −3.81; 95% CI −6.64 to −1.16) and curcumin (MD −2.73; 95% CI −5.27 to −0.18) significantly reduced FPG compared with placebo, and the remaining supplements relatively improved FPG, but none were statistically significant. The relative effects between different nutritional supplements were also not statistically significant (Appendix Figure S14). synbiotics (MD −4.82; 95% CI −8.63 to −1.02) and Cr (MD −4.03; 95% CI −7.79 to −0.26) significantly reduced insulin levels in PCOS patients then the control groups (Appendix Figure S15).

Based on the SUCRA, omega-3 (SUCRA = 78.6%) was most likely to rank best for improving HOMA-IR. The Se (SUCRA = 85.8%) was shown to be the best choice in raise QUICKI. Inositol was found to be the most effective in reducing levels of TG (SUCRA = 98.7%) and TC (SUCRA = 97.5%). Curcumin was the best nutritional supplement in reducing LDL-C (SUCRA = 90.3%) levels and raising HDL-C (SUCRA = 95.2%) levels. Moreover, vit D (SUCRA = 90.8%) was likely to be the most choice for lowering FPG, and omega-3 (SUCRA =77.3%) was the best nutritional supplement in reducing insulin. All sorts are presented in Appendix 5.

Anthropometric units of measurement

It was revealed that carnitine (MD −0.88; 95% CI −1.05 to −0.71) and curcumin (MD −0.20; 95% CI −0.40 to −0.01) significantly reduced the BMI level compared to placebo. Meanwhile, carnitine was more effective for weight loss when compared to other nutritional supplements. Furthermore, curcumin also had a similar advantage (MD −0.21; 95% CI −0.42 to 0.00) in reducing the BMI compared to Mg + vit E (Appendix Figure S16). According to SUCRA, carnitine ranks best for weight loss (Appendix Table S23).

Subgroup and sensitivity analyses

We performed subgroup analyses based on the treatment period (8 and 12 weeks), and the results were summarized in Appendix 10. The results indicated that a 12-week treatment period with nutritional supplements, compared to an 8-week period, can more effectively improve the lipid metabolic profile of PCOS patients.

For the sensitivity analyses, we found that the results of TT, QUICKI, TG, and LDL-C were generally similar to the overall findings of the network analyses. However, with the exception of synbiotics showing a notable correlation with HDL-C levels, no significant relationship was observed between nutritional supplements and HDL-C levels, which may be related to the limited number of studies and samples (Appendix 11).

Publication bias assessment and inconsistency test

Publication bias was evaluated regarding all indicators using funnel plots and Egger’s test. As shown in Fig. 7, the funnel plots of TT, FPG, TAC, MDA, hs-CRP exhibited skewed distribution, suggesting potential publication bias, which was consistent with the results of the Egger’s test (P<0.05). Furthermore, Egger’s test showed potential publication bias of TG, but no significant bias was detected for TT. The indexes of network maps in ring structures were tested for inconsistency and there was no evidence of global inconsistency in all networks except for HOMA-IR. A little heterogeneity was found among included studies (Appendix 6–7). As shown in Appendix 7, no significant evidence of inconsistency for any indicators was observed by node-splitting method.

Fig. 7.

funnel plot

Discussion

As far as we all know, this is the most comprehensive systematic review and NMA to evaluate the efficacy of nutritional supplements as adjunctive therapy to improve irregular menses, hyperandrogenism, abnormal glucose and lipid metabolism and oxidative stress and inflammatory biomarkers in women with PCOS. According to our results, it was demonstrated that Cr may be the most effective intervention to improve FSH, TAC, and VLDL-C levels. Synbiotics were the most effective on insulin and LH. Soy isoflavones were more effective at GSH and MDA compared to placebo. In addition, inositol was more significant in reducing TC and TG levels and curcumin was the most significant in improving LDL-C and HDL-C compared to placebo and other supplements. Omega-3 played a significant role in reducing HOMA-IR levels. No statistically significant effects on TT, SHBG, DHEAS, hs-CRP, and NO were observed between nutritional supplements and placebo. These findings, based on 79 RCTs and 5501 participants may provide some reference for clinicians to make initial choices about nutritional supplements in the adjuvant treatment of PCOS.

Nutritional supplements may play a potential adjunctive role in the clinical treatment of PCOS by regulating hormonal imbalances. It is known that women with PCOS may experience hormonal imbalances, including androgen levels, the luteinizing/follicle-stimulating hormone (LH/FSH) ratio, and growth hormone levels, which can affect their psychological status and bone metabolism. Furthermore, guidelines also recommended that the FAI can be used to assess biochemical hyperandrogenemia [110], and FAI has been used to assess androgen activity in vivo [111]. Currently, the regulatory effects of different nutritional supplements on FAI levels have not been evaluated. Previous published two systematic reviews [112–114] indicated that no any roles were found in TT and SHBG levels regarding the effects of nutritional supplements on reproductive hormone profiles, which was accordance with our findings. Furthermore, our results further suggested a potential influence of nutritional supplements on FSH and LH levels, providing additional information over these two publications above mentioned. These contradicting findings may be attributed to the characteristics of included participants, sample size, intervention protocols and other known or/and unknown potential confounding variables, which are required to be identified in future RCT with high-quality and large samples.

OS, an imbalance between oxidants and antioxidants in cells, is one of many factors that play an important role in the pathogenesis of PCOS and may lead to increased risk of metabolic syndrome, IR, hyperandrogenemia, cardiovascular disease, reproductive failure, and cancer [115]. Relevant studies have shown that nutritional supplements can effectively reduce OS and inflammatory marker levels in PCOS patients, improve the body’s antioxidant capacity and anti-inflammation, improve OS imbalance, and reduce the increased risk of PCOS-related diseases. Moreover, there is also a link between mitochondrial dysfunction and PCOS [116], and it has been suggested that reduced mitochondrial O2 consumption and increased glutathione and reactive oxygen species (ROS) might lead to mitochondrial dysfunction in patients with PCOS [117]. Some nutritional supplements regulate ROS metabolic homeostasis by activating oxidative enzymes, thereby modulating OS. Bioflavonoids are thought to have metabolic and anti-inflammatory effects, inhibiting NF-κB and enhancing glucose uptake through glucose transporter protein-4 (GLUT4) induction and activation of AMP-activated protein kinase (AMPK) [118]. Some studies have found that n-3 fatty acids significantly reduce hs-CRP, thus reducing the inflammatory status of PCOS women [119]. Other nutritional supplements also showed improvements in various indices. Our results showed that soy isoflavones have a better ameliorating effect on MDA, NO and GSH. Both Cr and vit D ranked high in important indicators for improved OS and the Cr obviously reduced protein glycosylation and lipid peroxidation. It was hypothesized experimentally that Cr might reduce OS by regulating the ROS metabolic balance in muscle through activation of glutathione peroxidase (GPx) or other antioxidant enzymes [120]. The SUCRA curve showed that soy isoflavones may be the best intervention among all anti-oxidative stress methods. In addition, omega-3 may have the best ameliorating effect in anti-inflammation. Therefore, we found that most nutritional supplements may have beneficial effects on PCOS patients by improving some endocrine indicators.

Nutritional supplements may be an important means of modulating endocrine metabolism in patients with PCOS in the context of clinical drug applications. Various reports indicate that endocrine disorders are the most common physiological conditions of PCOS and have a major health impact. Meanwhile, they can also easily to induce various endocrine diseases. According to our results, inositol and curcumin played the most beneficial roles in improving endocrine metabolism. Furthermore, inositol is beneficial in all aspects of PCOS and is not inferior to metformin in all parameters. A significant decrease in fasting blood glucose and AUC insulin levels was found in the inositol-treated group. The inositol provided the greatest benefit among the isoforms analyzed and exhibited fewer side effects compared to metformin [121]. Inositol also significantly decreased the TC, TG, HDL-C, and LDL-C levels in PCOS patients, which is conducive to regulating the endocrine status of the patients. Curcumin has an obvious protective effect on ovarian tissue. In fact, this compound may be involved in inhibiting the expression of vascular endothelial growth factor (VEGF), a proangiogenic factor closely related to the formation of PCOS, thereby inhibiting ovarian angiogenesis, preventing ovarian fibrosis, and promoting matrix degradation [122]. Nanocurcumin can significantly improve oxidative markers, glucose index and tumor necrosis factor α (TNF-α) level, restore phosphoinositide 3-kinase (PI3k)/serine-threonine kinase (Akt)/mammalian target of rapamycin (mTOR) signaling, and then reduce insulin resistance and maintain the integrity of outlet function [123]. In addition, selenium is a trace mineral essential for a variety of bio-metabolic functions and may be the best intervention in improving QUICKI. However, according to the latest findings, the role of selenium supplementation in ameliorating the metabolic pathology associated with PCOS is inconclusive [124], and caution is needed when treating PCOS with high doses of selenium due to potential toxicity. Finally, omega-3 is best for improving HOMA-IR levels. Nutritional supplements may be more beneficial in the adjuvant treatment of PCOS if they have a clear target and mechanism of action.

PCOS is strongly associated with obesity, which exacerbates and worsens the endocrine metabolic and reproductive consequences of patients with PCOS. Previous studies found that carnitine was also one of the best interventions to reduce body weight in patients with polycystic ovary syndrome and significantly reduced their BMI [125], which was consistent with our findings.

Limitation

Although this NMA conducted a relatively comprehensive analysis of the efficacy of nutritional supplementation in patients with PCOS across various indicators, some limitations should be considered. First, other unpublished literature on relevant websites was not searched and only trials in English were included, and this may lead to potential language bias and selection bias. Moreover, the variability in regiment, dosage and treatment duration may also affect our results. The efficacy of certain nutritional supplements with a favorable dose–response will generally increase with higher dosage. However, as the available information and original data of included articles were relatively limited, only the subgroup analyses based on 8- and 12- week treatment period were performed, while other subgroup analyses and even the dose–response analysis were not performed. Secondly, we did not compare the interventions involving the combined use of nutritional supplements and did not evaluate the use of the drugs alone and in combination to derive an improved effect of the combination. Third, the literature included was almost exclusively from a single team in Iran, with geographical limitations. Fourth, almost all of the included studies were placebo controls, and there were insufficient comparative studies to adequately assess these interventions. Fifth, the combined analysis regarding the safety of these nutritional supplements was not performed due to the limited information and available data of included original article. Therefore, the application of our results should take into account any limitations of the analysis and the specific clinical situation. Meanwhile, we will continue to focus on these issues in subsequent studies and accordingly update our results when applicable in the future.

Conclusions

In conclusion, our results indicated that chromium, inositol, and omega-3 are beneficial for improving the lipid profile. For improving obesity, sex hormone levels, inflammatory factors and oxidative stress indicators in PCOS patients, carnitine, chromium, and soy isoflavones are effective options, respectively. Nutritional supplements may be effective in improving PCOS. Our study did an indirect comparison between nutritional supplements for different indicators of patients with PCOS, which may provide some reference for the selection of clinical nutritional. Nevertheless, due to certain limitations in our study, large-scale and high-quality randomized controlled trials are needed to further confirm the beneficial effects of these supplements on PCOS and to explore the effectiveness and safety of various combinations of nutritional supplements.

Authors’ contributions

“GHZ, YXF, RXL are responsible for the design, data analysis and preparation of manuscript. YH and WJL are responsible for the literature research. YHZ and MDZ are responsible for the data collection and statistics analysis. GHZ are responsible for the first manuscript review. JCH is responsible for the manuscript review and checking. And the final version of this manuscript has been read and agreed by all authors”.

Funding

The study was supported by the Construction of research platform for early clinical trials of innovative drugs from Yunnan Province Science and Technology Department (No. 202302 AA310007).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

All outcomes and analyses were based on previous ethically approved studies. Therefore, no further ethical approval and patient consent were required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.