Selenium supplements and Polycystic Ovary syndrome and its key risk factors: a Mendelian Randomization Study

doi:10.21203/rs.3.rs-2961641/v1

Background

Observational investigations recommend that mineral supplements was associated with a higher risk of Polycystic ovary syndrome (PCOS) and its risk factors(insulin resistance, hyperandrogenism, and obesity), but the relationship with risk of PCOS, hyperandrogenism, obesity and insulin resistance was less clear. The objective of this study was to investigate the potential causal impact of genetically predicted levels of magnesium (Mg), calcium (Ca), selenium (Se), zinc (Zn), iron (Fe), and Omega-3 (ω-3) on polycystic ovary syndrome (PCOS) and its associated risk factors.

Methods

A two-sample Mendelian Randomization (MR) analysis was conducted. The genetic variations obtained from GWAS of individuals with European ancestry were found to be associated with the levels of Ca, Mg, Zn, Se, Fe, or ω-3. The data obtained from the FinnGen Consortium and MAGIC were utilized for the outcome of GWAS. The primary approach employed in the study was the Inverse-Variance Weighted (IVW) MR analysis, which was complemented by additional sensitivity analyses such as the MR-Egger and Weighted Median methods.

Results

The study found that there was a correlation between Se and a reduced risk of insulin resistance, with a decrease of 2.2% according to random-effect IVW (OR 0.978, 95% CI 0.960–0.996, p = 0.015). The association between mineral supplement levels and PCOS was found to be limited, with an odds ratio (OR) ranging from 0.875(95% CI: 0.637–1.202, p-value = 0.411) for Ca. Correspondingly, Limited scientific proof was found for the efficacy of other mineral supplements on hyperandrogenism, obesity, and insulin resistance.

Conclusions

These findings suggested a causal relationship between the decreased Se with the risk of insulin resistance. Nonetheless, there is limited evidence supporting a causal association between various mineral supplements and the risk factors associated with PCOS.

mineral supplements

Mendelian randomization

Polycystic ovary syndrome

hyperandrogenism

obesity

insulin resistance

Polycystic ovary syndrome (PCOS) is a disease that occurs in 6–10% of women worldwide ^[1]. PCOS is a prevalent endocrine disease that affects women of reproductive age. It is characterized by the presence of hyperandrogenism, oligo-ovulation, and polycystic ovarian morphology ^[2]. PCOS is characterized by various clinical manifestations such as hyperandrogenism, infertility, abnormal glucose metabolism, obesity, hirsutism, insulin resistance (IR), and acne vulgarism ^[3]. While the exact etiology and mechanisms of PCOS remains unknown ^[4,5], researchers believe hyperandrogenism, obesity and IR are key risk factors ^[6]. Enhancing comprehension of the etiological pathways underlying PCOS is crucial for preventing its occurrence and associated risk factors, facilitating suitable clinical trials, and identifying targets for efficacious lifestyle and pharmacological interventions. Numerous observational investigations have indicated a strong correlation between minerals and PCOS as well as its principal risk factors. As an illustration, among females diagnosed with PCOS and exhibiting insufficient levels of vitamin D, the administration of a combination of calcium (Ca) and vitamin D resulted in notable enhancements in β-cell function ^[7]. A higher concentration of magnesium (Mg) and zinc (Zn) help reduce IR risk ^[8,9]. Selenium (Se) is able to reduce the BMI and weight of PCOS patients ^[10]. Patients with PCOS have elevated serum ferritin concentrations, suggesting a possible iron overload ^[11]. The utilization of omega-3 (ω-3) fatty acid has the potential to be efficacious in females diagnosed with PCOS who also present with dyslipidemia and/or IR ^[12]. The aforementioned observations present a challenge in terms of explanation, as they are subject to the influence of numerous confounding factors, including but not limited to occupational and environmental exposures, as well as dietary and supplement usage^[13].

Mendelian Randomization (MR) method aims to randomly assign parental alleles to offspring and is therefore less likely to be confounded by confounding factors; it also avoids reverse causality bias since genotypes are not influenced by disease ^[14]. MR methodology is predicated on the utilization of genetic variations that are established at birth to ascertain a causal association between outcome and exposure ^[15]. To the best of our knowledge, the role of mineral supplements in PCOS has not been assessed on a large scale by MR. Therefore, we can use a two-sample MR analysis to obtain evidence of an association between exposure factors (mineral supplements) and outcome factors (PCOS) from different samples ^[16].

The present study verifies the long-held epidemiological hypothesis in the context of the connection between PCOS and mineral supplements in traditional epidemiological investigations. The present study employs the MR framework to examine the causal association between genetically anticipated levels of Ca, Mg, Zn, Se, Fe, or ω-3 and PCOS, as well as its major risk factors, including obesity, hyperandrogenism, and IR.

2.1. Study design

Figure 1. Framework for the Mendelian randomization analysis of mineral supplements with risk of PCOS and its risk factors. The instrumental variables must meet three key assumptions: (i) The IVs exhibit association with the level of exposure; (ii) The IVs have a direct impact on the outcome solely via their influence on the exposure; and (iii) The IVs are not affected by any confounding factors in relation to the correlation between the exposure and the final outcome.

2.2. Exposure Measures

The overall statistics for all exposures, with the exception of ω-3, were obtained from the NHGRI-EBI GWAS database (https://www.ebi.ac.uk/gwas). These exposures include Mg, Ca, Se, Zn, and Fe. The summarized data pertaining to ω-3 were obtained from the UK Biobank (UKB). The genetic variation connected to calcium was obtained from the results of 17 population-based GWAS with a sample size of 39,400. Furthermore, an additional 21,679 individuals of European origins were utilized to determine the pertinent genetic loci^[17]. The genetic variability of Mg was determined through GWAS involving people of European origins (n = 15,366) from the International CHARGE Alliance. Furthermore, a further 8463 individuals of European ancestry were evaluated to determine the SNPs ^[18]. The Queensland Institute of Medicine's twins and their families (n = 2603) were the sources of genetic variations in erythrocyte Zn levels, which were quantified employing inductively coupled plasma mass spectrometry ^[19]. The genetic variation associated with Fe was identified through GWAS across 11 European populations. This finding was subsequently validated in an additional 8 populations, resulting in a total sample size of 48,972 individuals ^[20]. The genetic variation associated with Se was identified through GWAS of individuals with Australian ancestry (n = 9639). The data were subjected to adjustments for covariates such as age, geography, smoking status, and top eigenvectors ^[21]. The study incorporated a sample size of 114,999 individuals to investigate the genetic variation associated with ω-3. The statistical analysis involved controlling for age, gender, and age squared ^[22].

Multiple quality control measures were implemented to identify SNPs as IVs from the exposure dataset. Initially, it was required that SNPs attain genome-wide significance, which was defined as having a p-value less than 5 × 10^− 8. Next, it was necessary to ensure the independence of the SNPs. To achieve this, clumping was conducted with the specifications of r² = 0.01 and kb = 1,000, in order to eliminate any linkage disequilibrium (LD) among the SNPs. Only the SNP-shaving LDs with the most statistically significant p-values were retained. Subsequently, SNPs that were linked to a phenotype connected to a confounding variable were excluded. The identification of potentially associated phenotypes was conducted through utilization of the "PhenoScanner" online database (www.phenoscanner.medschl.cam.ac), employing a filtering criterion of r² > 0.8 and p < 5 × 10^− 8. Table 1 displays data connected to the minerals.

Table 1

Dataset of the Mineral Information
Exposures	Number of SNPs	Participants	Ethnicity
Ca	7	39,400	European
Mg	6	15,366	European
Zn	11	2603	European
Se	12	9639	Australian
Fe	11	48,972	European
ω-3	42	114,999	European

2.3. Outcome Measures

The summary statistics for GWAS pertaining to PCOS were purchased from the FinnGen Consortium ^[23]. The current investigation was a comprehensive cohort analysis that examined over 16 million SNPs, with adjustments made for age, genders, and genotyping batches. The FinnGen Consortium was also the source of the GWAS summarized data associated with hyperlipidaemia and obesity. Furthermore, the GWAS summarized data pertaining to insulin resistance were acquired from MAGIC. Table 2 illustrates the outcomes data.

Table 2

Dataset of the Outcomes Information
Outcomes	Data sources	Participants	Ethnicity
Polycystic ovarian syndrome	FinnGen Biobank	118,870	European
Hyperlipidaemia	FinnGen Biobank	201,497	European
Obesity	FinnGen Biobank	218,735	European
Insulin resistance	MAGIC	37,037	European

2.4. Two-sample MR

In order to establish a dependable basis for MR analysis, it is necessary to ensure that three fundamental assumptions have been met ^[24]: (i) The IVs exhibit a relationship with the level of exposure; (ii) The IVs have an impact on the final result solely via their influence on the level of exposure; and (iii) The IVs are not influenced by any confounding factors in the relationship between the outcome and exposure. The current investigation employed the inverse variance weighting (IVW) approach utilizing the random-effects model as the primary analytical approach ^[25]. Various analytical techniques have been employed for repeated analysis, such as the MR-Egger ^[26], weighted median ^[27], and simple mode ^[28]. Various hypothetical models are utilized by each method to assess the causal impact, and these models are subsequently employed to verify the resilience of the outcomes. The MR-Egger method yields estimates that have been adjusted for pleiotropy ^[26]. The utilization of a weighted median approach facilitated the estimation of causal effects considering the presence of 50% invalid SNPs ^[27]. The utilization of the weighted median approach enables the computation of causal effects in situations where 50% of the SNPs are deemed reliable ^[27]. The MR-PRESSO technique is capable of identifying and rectifying outliers, thereby yielding MR estimates that exhibit robustness with respect to heterogeneity subsequent to the elimination of the detected outliers ^[29]. The assessment of heterogeneity was conducted through the implementation of Cochran's Q test on the IVW and MR-Egger estimates. Whenever the Cochran's Q statistic in the MR-IVW fixed impacts analyses yielded a significant result, the MR-IVW random-effects model was utilized. The study employed MR-Egger intercept investigations to determine the presence of horizontal pleiotropy. Additionally, single-SNP and leave-one-out analyses were conducted to detect outlier SNPs that may be influencing the observed relationships ^[30].

The study presents the connection between exposures and outcomes in the form of ORs accompanied by 95% CIs. The statistical analysis was conducted utilizing version 3.6.0 of the R program (https://www.r-project.org/). The software packages employed for conducting MR analyses were "MendelianRandomization" and "MR-PRESSO" ^[29].

3.1. Causality between Minerals and PCOS

Seven distinct SNPs exhibiting a statistical significance of p < 5 × 10^− 8 and r² < 0.01 were found to be correlated with Ca levels. Similarly, five independent SNPs were observed to be linked to Mg, eleven SNPs with Fe, eleven SNPs with Zn, twelve SNPs with Se, and forty-two SNPs with ω-3 (Supplementary Table S1). The results of the study indicate that there is not a significant relationship between genetically predicted Ca and PCOS. This was determined through the use of various models, including random-effect IVW, MR-Egger, and Weighted median, which yielded OR of 0.549, 0.300, and 0.496, respectively, and the corresponding p-values were 0.502, 0.932, and 0.531, respectively, indicating no statistically significant association between the two variables. Furthermore, the elements Zn, Mg, Fe, Se, and ω-3 exhibited no significant correlation with the likelihood of developing PCOS (Table 3).

Table 3

MR estimates of the causality between six minerals and PCOS.
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Heterogeneity		Pleiotropy
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Cochran Q test	P value		Egger-intercept	P value
Ca	7	IVM	-0.600	0.549(0.088,3.410)	0.520
		MR-Egger	-0.600	0.300(0.011,7.890)	0.932	4.071	0.667	0.020		0.680
		MR-RAPS	-0.493	0.611(0.097,3.863)	0.600
		WM	-0.701	0.496(0.055,4.442)	0.531
Mg	5	IVM	0.523	1.688(0.331,8.613)	0.529
		MR-Egger	-0.01	0.990(0.166,59.212)	0.997	10.594	0.032	0.026		0.733
		MR-RAPS	NA	NA	NA
		WM	0.354	1.424(0.254,7.981)	0.688
Zn	11	IVM	0.016	1.016(0.806,1.282)	0.891
		MR-Egger	-0.111	0.895(0.445,1.797)	0.762	8.769	0.554	0.029		0.713
		MR-RAPS	0.007	1.007(0.785,1.292)	0.953
		WM	-0.120	0.887(0.646,1.217)	0.457
Se	12	IVM	0.186	1.204(0.955,1.518)	0.116
		MR-Egger	0.156	1.169(0.659,2.073)	0.605	12.341	0.339	0.006		0.912
		MR-RAPS	0.183	1.201(0.934,1.543)	0.153
		WM	0.216	1.241(0.890,1.730)	0.203
Fe	11	IVM	-0.093	0.911(0.621,1.337)	0.635
		MR-Egger	0.034	1.034(0.562,1.902)	0.916	5.459	0.858	-0.016		0.613
		MR-RAPS	-0.088	0.916(0.616,1.361)	0.664
		WM	-0.010	0.990(0.614,1.598)	0.968
ω-3	42	IVM	0.002	1.002(0.959,1.046)	0.938
		MR-Egger	0.001	1.001(0.900,1.113)	0.990	46.303	0.263	0.001		0.983
		MR-RAPS	0.018	1.018(0.975,1.063)	0.424
		WM	0.041	1.042(0.978,1.110)	0.207

IVW indicates inverse variance-weighted; CI indicates confidence interval; WM indicates weighted median; NA, not applicable.

3.2. Causality between Minerals and Hyperlipidaemia

Seven distinct SNPs exhibiting a statistical significance of p < 5 × 10^− 8 and r² < 0.01 were found to be correlated with Ca levels. Similarly, five independent SNPs were observed to be connected to Mg levels, whereas eleven SNPs were identified as being correlated with Zn levels, eleven SNPs were connected with Fe levels, twelve SNPs were related to Se levels, and forty-two SNPs were linked with ω-3 levels (Supplementary Table S2). The results of the study indicate that there is not a significant connection between genetically predicted Ca and hyperlipidaemia, as determined by the implementation of random-effect IVW, MR-Egger, and Weighted median models. The OR for random-effect IVW was 0.959 with a 95% CI of 0.350–2.626 and a p-value of 0.935. The OR for MR-Egger was 0.877 with a 95% CI of 0.122–6.285 and a p-value of 0.901. Lastly, the OR for the weighted median was 0.780 with a 95% CI of 0.344–1.856 and a p-value of 0.602. In addition, Mg, Zn, Se, Fe, and ω-3 had no association with the risk of hyperlipidaemia (Table 4).

Table 4

MR estimates of the causality between six minerals and Hyperlipidaemia.
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Heterogeneity			Pleiotropy
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Cochran Q test	P value		Egger-intercept	P value
Ca	7	IVM	-0.042	0.959(0.473,1.941)	0.907
		MR-Egger	-0.132	0.877(0.122,6.285)	0.901	12.234		0.057	0.003		0.919
		MR-RAPS	-0.025	0.975(0.402,2.365)	0.955
		WM	-0.224	0.780(0.344,1.856)	0.602
Mg	5	IVM	0.488	1.629(0.868,3.058)	0.129
		MR-Egger	0.475	1.609(0.680,3.804)	0.358	1.320		0.858	0.001		0.969
		MR-RAPS	NA	NA	NA
		WM	0.501	1.651(0.883,3.087)	0.117
Zn	11	IVM	-0.054	0.947(0.866,1.036)	0.236
		MR-Egger	-0.075	0.928(0.710,1.213)	0.597	3.449		0.969	0.005		0.876
		MR-RAPS	-0.053	0.948(0.862,1.044)	0.280
		WM	-0.011	0.989(0.878,1.114)	0.858
Se	12	IVM	-0.044	0.957(0.875,1.047)	0.334
		MR-Egger	-0.089	0.915(0.749,1.118)	0.406	9.095		0.613	0.009		0.638
		MR-RAPS	-0.053	0.948(0.860,1.046)	0.287
		WM	-0.080	0.923(0.811,1.050)	0.222
Fe	11	IVM	-0.163	0.850(0.653,1.106)	0.226
		MR-Egger	0.008	1.008(0.664,1.530)	0.971	31.336		0.001	-0.021		0.330
		MR-RAPS	-0.110	0.896(0.729,1.101)	0.291
		WM	-0.158	0.854(0.702,1.039)	0.114
ω-3	42	IVM	-0.001	0.999(0.965,1.035)	0.966
		MR-Egger	0.025	1.025(0.947,1.111)	0.540	176.097		0.000	-0.015		0.483
		MR-RAPS	-0.017	0.983(0.895,1.080)	0.727
		WM	0.005	1.005(0.980,1.031)	0.706

IVW indicates inverse variance-weighted; CI indicates confidence interval; WM indicates weighted median; NA, not applicable.

3.3. Causality between Minerals and Obesity

Seven distinct SNPs exhibiting a statistical significance of p < 5 × 10^− 8 and r² < 0.01 were found to be correlated with Ca levels. Similarly, five independent SNPs were observed to be associated with Mg levels, while eleven SNPs were linked to Zn levels, eleven SNPs were correlated with Fe levels, twelve SNPs had a connection to Se levels, and forty-two SNPs were linked with ω-3 levels (Supplementary Table S3). Genetically anticipated Ca revealed no relationship with hyperlipidaemia depending on models including random-effect IVW (OR 0.946, 95% CI 0.556–1.607, p = 0.836), MR-Egger (OR 0.605, 95% CI 0.235–1.560, p = 0.346) and Weighted median (OR 0.814, 95% CI 0.438–1.513, p = 0.514). In addition, Mg, Zn, Se, Fe and ω-3 had no association with the risk of hyperlipidaemia (Table 5).

Table 5

MR estimates of the causality between six minerals and Obesity.
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Heterogeneity			Pleiotropy
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Cochran Q test	P value		Egger-intercept	P value
Ca	7	IVM	-0.056	0.946(0.556,1.607)	0.836
		MR-Egger	-0.502	0.605(0.235,1.560)	0.346	2.352		0.885	0.015	0.316
		MR-RAPS	-0.042	0.959(0.560,1.641)	0.879
		WM	-0.206	0.814(0.438,1.513)	0.514
Mg	5	IVM	0.15	1.162(0.722,1.869)	0.536
		MR-Egger	0.228	1.256(0.432,3.654)	0.703	8.229		0.084	-0.084	0.847
		MR-RAPS	NA	NA	NA
		WM	0.158	1.171(0.714,1.922)	0.531
Zn	11	IVM	0.017	1.017(0.950,1.088)	0.626
		MR-Egger	-0.041	0.960(0.783,1.176)	0.701	4.960		0.894	0.013	0.567
		MR-RAPS	0.017	1.017(0.946,1.094)	0.644
		WM	0.016	1.016(0.929,1.110)	0.731
Se	12	IVM	0.006	1.007(0.941,1.077)	0.851
		MR-Egger	-0.016	0.984(0.847,1.143)	0.839	8.709		0.649	0.004	0.750
		MR-RAPS	0.010	1.010(0.939,1.086)	0.799
		WM	0.022	1.022(0.930,1.122)	0.652
Fe	11	IVM	-0.037	0.964(0.862,1.078)	0.520
		MR-Egger	0.030	1.030(0.856,1.240)	0.761	10.685		0.383	-0.008	0.390
		MR-RAPS	-0.036	0.965(0.854,1.089)	0.560
		WM	-0.042	0.959(0.827,1.112)	0.582
ω-3	42	IVM	0.005	1.005(0.992,1.018)	0.435
		MR-Egger	0.034	1.034(1.002,1.067)	0.043	52.081		0.115	-0.016	0.054
		MR-RAPS	0.004	1.004(0.992,1.016)	0.573
		WM	0.008	1.008(0.991,1.026)	0.358

IVW indicates inverse variance-weighted; CI indicates confidence interval; WM indicates weighted median; NA, not applicable.

3.4. Causality between Minerals and Insulin resistance

Six independent SNPs (p < 5 × 10^− 8, r² < 0.01) were related toCa, eleven SNPs were linked with Zn, five independent SNPs were connected with Mg, ten SNPs were associated with Fe, twelve SNPs were associated with Se, and twenty-five SNPs were associated with ω-3 (Supplementary Table S4). Genetically predicted Se showed the significant associated with insulin resistance. Se was associated with the 2.2% decreased risk of insulin resistance based on random-effect IVW(OR 0.978, 95% CI 0.960–0.996, p = 0.015). However, Ca, Mg, Zn, Fe and ω-3 had no association with the risk of insulin resistance (Table 6).

Table 6

MR estimates of the causality between six minerals and Insulin resistance.
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Heterogeneity		Pleiotropy
Mineral	SNPs	MR method	Beta	OR (95%CI)	P value	Cochran Q test	P value	Egger-intercept	P value
Ca	6	IVM	-0.133	0.875(0.637,1.202)	0.411
		MR-Egger	0.154	1.167(0.706,1.929)	0.579	30.850	0.000	-0.010		0.240
		MR-RAPS	NA	NA	NA
		WM	-0.025	0.976(0.839,1.135)	0.749
Mg	5	IVM	-0.063	0.939(0.836,1.055)	0.287
		MR-Egger	-0.144	0.866(0.738,1.016)	0.176	2.539	0.638	0.004		0.246
		MR-RAPS	NA	NA	NA
		WM	-0.076	0.927(0.817,1.050)	0.232
Zn	11	IVM	-0.009	0.991(0.973,1.010)	0.365
		MR-Egger	-0.010	0.999(0.942,1.060)	0.975	4.637	0.914	-0.002		0.796
		MR-RAPS	-0.010	0.990(0.971,1.010)	0.375
		WM	-0.008	0.992(0.970,1.015)	0.496
Se	12	IVM	-0.023	0.978(0.960,0.996)	0.015
		MR-Egger	-0.042	0.959(0.909,1.012)	0.157	5.412	0.910	0.003		0.472
		MR-RAPS	-0.023	0.977(0.958,0.997)	0.027
		WM	-0.020	0.980(0.957,1.004)	0.106
Fe	10	IVM	0.000	1.000(0.974,1.026)	0.977
		MR-Egger	0.025	1.026(0.986,1.068)	0.247	8.785	0.457	-0.004		0.132
		MR-RAPS	-0.002	0.998(0.971,1.026)	0.907
		WM	0.004	1.004(0.970,1.039)	0.828
ω-3	25	IVM	0.000	1.000(0.993,1.007)	0.945
		MR-Egger	0.002	1.002(0.986,1.017)	0.842	76.411	0.000	-0.001		0.795
		MR-RAPS	-0.003	0.997(0.993,1.001)	0.156
		WM	0.000	1.000(0.995,1.006)	0.888

IVW indicates inverse variance-weighted; CI indicates confidence interval; WM indicates weighted median; NA, not applicable.

Our analysis was an attempt to elucidate the potential of Ca, Mg, Zn, Se, Fe and ω-3 in PCOS prevention and management and its key risk factors using MR methods. Our MR methods found Se was associated with the 2.2% decreased risk of insulin resistance. Nonetheless, we dicovered little evidence for a underlying relationship between genetically predicted concentrations of other mineral supplements and PCOS.

The current study's outcomes were corroborated by biological and epidemiological investigations that demonstrate correlations between Se and insulin resistance, which could potentially mediate the risks of PCOS ^[31–33]. Se is an essential micronutrient, which plays an vital function in redox reactions including glutathione peroxidase and thioredoxin reductase ^[34]. More recently, in vitro and in vivo investigations have exhibited that Se possesses insulin-like actions ^[35]. In an investigation by Jamilian et al ^[32]revealed that the supplementation of 200 µg/day Se for eight weeks caused a significant reduction in insulin resistance levels among females with PCOS. In accordance with the aforementioned findings, our study revealed a correlation between genetic upregulation of SE and the development of insulin resistance.

Furthermore, even though no epidemiological investigation has been conducted to explore the correlation between the six minerals and PCOS along with its associated risk factors such as hyperlipidemia and obesity, prior research indicates that women with PCOS may exhibit a deficiency in Zn and Mg, which could potentially contribute to the insulin-resistant state observed in certain PCOS phenotypes. This finding holds significant implications for future research efforts in the field of PCOS. The study's conclusion indicates the absence of a causal relationship between minerals such as Ca, Mg, Zn, Se, Fe, and ω-3 and PCOS, hyperlipidemia, or obesity. This suggests that the prior observations may be the result of confounding factors for these diseases rather than a true causal relationship.

MR studies have validated outcomes for numerous established risk factors, including estimated bone mineral density of the heel and type 2 diabetes ^[36]. MR has also exhibited no causal impact of micronutrient concentration on COVID-19 outcome ^[21], which could have saved a lot of manpower and medical drug costs if reported earlier. MR evidence can provide significant perspective on the prospects for treatment. Here, present MR analyses do not promote a causal association between Ca, Mg, Zn, Fe and ω-3 and outcomes(PCOS and its risk factors).

A vital benefit of this investigation is that reverse causality or residual confounding can be mitigated by using genetic variation as an indicator of mineral levels ^[37]. The MR methodology operates under the assumption that SNPs connected to exposure are not influenced by confounding variables, and that genetic variability solely impacts outcomes through exposure-related factors ^[38]. Another advantage is that the genetic instruments in this study were selected from a great number of participants from UK Biobank and NHGRI-EBI GWAS.

This study also has some limitations. First, the restricted European ethnic background of our study sample restricts the generalizability of the results to other populations of varying racial backgrounds. The impact of minerals on human health may vary based on race and environmental factors, but there is no evidence to suggest that they operate through pathways specific to certain populations. Secondly, none of the three strong MR hypotheses were empirical. The utilization of SNPs as instrumental variables in GWAS is deemed acceptable; however, it may elevate the probability of false positive outcomes owing to restrictions in sample size. The existence of weaker IVs may potentially introduce bias into the outcomes ^[39]. Finally, it should be noted that the employment of a limited number of SNPs as IVs is only capable of elucidating a restricted set of causal associations ^[40]. There is potential for future improvement in this area, as more advanced GWAS may become available for both exposures and outcomes.

In conclusions, we discovered that Se was connected with the 2.2% reduced risk of insulin resistance. Nevertheless, we detected little evidence for a causal relationship between genetically predicted concentrations of other mineral supplements and PCOS with its risk factors. Further a well-designed epidemiological combination and MR studies are needed to confirm these results and determine the causal mechanism.

Data Availability

Data are available from the website (https://gwas.mrcieu.ac.uk/ ).

Funding sources

It was supported by Huzhou Science and Technology Plan (2022GY27).

Competing interests

The authors declare that there is no competing interests.

Acknowledgements

We thank Home for Researchers editorial team (www.home-for-researchers.com) for language editing service.

Bozdag, G., Mumusoglu, S., Zengin, D., Karabulut, E. & Yildiz, B.O. The prevalence and phenotypic features of polycystic ovary syndrome: a systematic review and meta-analysis[J]. Hum Reprod, 2016, 31, (12): 2841-2855.
Guan, C., Zahid, S., Minhas, A.S. , et al. Polycystic ovary syndrome: a "risk-enhancing" factor for cardiovascular disease[J]. Fertil Steril, 2022, 117, (5): 924-935.
Shojaeian, Z., Sadeghi, R. & Latifnejad Roudsari, R. Calcium and vitamin D supplementation effects on metabolic factors, menstrual cycles and follicular responses in women with polycystic ocvary syndrome: A systematic review and meta-analysis[J]. Caspian J Intern Med, 2019, 10, (4): 359-369.
Bednarska, S. & Siejka, A. The pathogenesis and treatment of polycystic ovary syndrome: What's new?[J]. Adv Clin Exp Med, 2017, 26, (2): 359-367.
Ibáñez, L., Oberfield, S.E., Witchel, S. , et al. An International Consortium Update: Pathophysiology, Diagnosis, and Treatment of Polycystic Ovarian Syndrome in Adolescence[J]. Horm Res Paediatr, 2017, 88, (6): 371-395.
Alesi, S., Ee, C., Moran, L.J., Rao, V. & Mousa, A. Nutritional Supplements and Complementary Therapies in Polycystic Ovary Syndrome[J]. Adv Nutr, 2022, 13, (4): 1243-1266.
Foroozanfard, F., Jamilian, M., Bahmani, F. , et al. Calcium plus vitamin D supplementation influences biomarkers of inflammation and oxidative stress in overweight and vitamin D-deficient women with polycystic ovary syndrome: a randomized double-blind placebo-controlled clinical trial[J]. Clin Endocrinol (Oxf), 2015, 83, (6): 888-894.
Hamilton, K.P., Zelig, R., Parker, A.R. & Haggag, A. Insulin Resistance and Serum Magnesium Concentrations among Women with Polycystic Ovary Syndrome[J]. Curr Dev Nutr, 2019, 3, (11): nzz108.
Jamilian, M., Foroozanfard, F., Bahmani, F. , et al. Effects of Zinc Supplementation on Endocrine Outcomes in Women with Polycystic Ovary Syndrome: a Randomized, Double-Blind, Placebo-Controlled Trial[J]. Biol Trace Elem Res, 2016, 170, (2): 271-278.
Shabani, A., MaryamJamilian, MehriChamani, MaryamMohammadi, SotoudehAsemi, Zatollah. The effects of a novel combination of selenium and probiotic on weight loss, glycemic control and markers of cardio-metabolic risk in women with polycystic ovary syndrome[J]. 2018, 46.
Martínez-García, M.A., Luque-Ramírez, M., San-Millán, J.L. & Escobar-Morreale, H.F. Body iron stores and glucose intolerance in premenopausal women: role of hyperandrogenism, insulin resistance, and genomic variants related to inflammation, oxidative stress, and iron metabolism[J]. Diabetes Care, 2009, 32, (8): 1525-1530.
Yang, K., Zeng, L., Bao, T. & Ge, J. Effectiveness of Omega-3 fatty acid for polycystic ovary syndrome: a systematic review and meta-analysis[J]. Reprod Biol Endocrinol, 2018, 16, (1): 27.
Smith, G.D. & Ebrahim, S. Data dredging, bias, or confounding[J]. Bmj, 2002, 325, (7378): 1437-1438.
Zheng, J., Baird, D., Borges, M.C. , et al. Recent Developments in Mendelian Randomization Studies[J]. Curr Epidemiol Rep, 2017, 4, (4): 330-345.
Deshmukh, H., Shah, N., Papageorgiou, M. , et al. Genetic risk for the polycystic ovary syndrome, bone mineral density and fractures in women and men: A UK Biobank Mendelian randomisation study[J]. Bone, 2022, 155: 116285.
Lawlor, D.A. Commentary: Two-sample Mendelian randomization: opportunities and challenges[J]. Int J Epidemiol, 2016, 45, (3): 908-915.
O'Seaghdha, C.M., Wu, H., Yang, Q. , et al. Meta-analysis of genome-wide association studies identifies six new Loci for serum calcium concentrations[J]. PLoS Genet, 2013, 9, (9): e1003796.
Meyer, T.E., Verwoert, G.C., Hwang, S.J. , et al. Genome-wide association studies of serum magnesium, potassium, and sodium concentrations identify six Loci influencing serum magnesium levels[J]. PLoS Genet, 2010, 6, (8).
Evans, D.M., Zhu, G., Dy, V. , et al. Genome-wide association study identifies loci affecting blood copper, selenium and zinc[J]. Hum Mol Genet, 2013, 22, (19): 3998-4006.
Benyamin, B., Esko, T., Ried, J.S. , et al. Novel loci affecting iron homeostasis and their effects in individuals at risk for hemochromatosis[J]. Nat Commun, 2014, 5: 4926.
Sobczyk, M.K. & Gaunt, T.R. The Effect of Circulating Zinc, Selenium, Copper and Vitamin K(1) on COVID-19 Outcomes: A Mendelian Randomization Study[J]. Nutrients, 2022, 14, (2).
Sun, Q., Gao, N. & Xia, W. Association between omega-3/6 fatty acids and cholelithiasis: A mendelian randomization study[J]. Front Nutr, 2022, 9: 964805.
The FinnGen consortium. The R4 release of FinnGen study. Accessed January 15, 2021.https://finngen.gitbook.io/documentation/.
VanderWeele, T.J., Tchetgen Tchetgen, E.J., Cornelis, M. & Kraft, P. Methodological challenges in mendelian randomization[J]. Epidemiology, 2014, 25, (3): 427-435.
Lawlor, D.A., Harbord, R.M., Sterne, J.A., Timpson, N. & Davey Smith, G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology[J]. Stat Med, 2008, 27, (8): 1133-1163.
Bowden, J., Davey Smith, G. & Burgess, S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression[J]. Int J Epidemiol, 2015, 44, (2): 512-525.
Bowden, J., Davey Smith, G., Haycock, P.C. & Burgess, S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator[J]. Genet Epidemiol, 2016, 40, (4): 304-314.
Hartwig, F.P., Davey Smith, G. & Bowden, J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption[J]. Int J Epidemiol, 2017, 46, (6): 1985-1998.
Verbanck, M., Chen, C.Y., Neale, B. & Do, R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases[J]. Nat Genet, 2018, 50, (5): 693-698.
Hemani, G., Zheng, J., Elsworth, B. , et al. The MR-Base platform supports systematic causal inference across the human phenome[J]. Elife, 2018, 7.
Kanafchian M, Mahjoub S, Esmaeilzadeh S et al. Status of serum selenium and zinc in patients with the polycystic ovary syndrome with and without insulin resistance[J]. Middle East Fertil Soc J, 2018; 23: 241–245.
Jamilian M, Razavi M, Fakhrie Kashan Z et al. Metabolic response to selenium supplementation in women with polycystic ovary syndrome: a randomized, double-blind, placebo-controlled trial[J]. Clin Endocrinol, 2015; 82: 885–891.
Hajizadeh-Sharafabad F, Moludi J, Tutunchi H, Taheri E, Izadi A, Maleki V. Selenium and polycystic ovary syndrome; current knowledge and future directions: a systematic review[J]. Horm Metab Res, 2019; 51(05):279–87.
Lu J, Holmgren A. Selenoproteins[J]. J Biol Chem. 2009; 284(2):723-7.
Stapleton SR. Selenium: an insulin-mimetic[J]. Cell Mol Life Sci. 2000; 57(13-14):1874-9.
Gan W, Clarke RJ, Mahajan A, Kulohoma B, Kitajima H, Robertson NR, Rayner NW, Walters RG, Holmes MV, Chen Z, McCarthy MI. Bone mineral density and risk of type 2 diabetes and coronary heart disease: A Mendelian randomization study[J]. Wellcome Open Res. 2017; 22(2):68.
Gan W, Neeland, I.J.; Kozlitina, J. Mendelian Randomization: Using Natural Genetic Variation to Assess the Causal Role of Modifiable Risk Factors in Observational Studies[J]. Circulation 2017; 135: 755–758.
Gan W, VanderWeele, T.J.; Tchetgen Tchetgen, E.J.; Cornelis, M.; Kraft, P. Methodological challenges in mendelian randomization[J]. Epidemiology 2014； 25: 427–435.
Burgess, S.; Thompson, S.G.; Collaboration, C.C.G. Avoiding bias from weak instruments in Mendelian randomization studies[J]. Int. J. Epidemiol. 2011; 40: 755–764.
Swerdlow, D.I.; Kuchenbaecker, K.B.; Shah, S.; Sofat, R.; Holmes, M.V.; White, J.; Mindell, J.S.; Kivimaki, M.; Brunner, E.J.; Whittaker, J.C., et al. Selecting instruments for Mendelian randomization in the wake of genome-wide association studies[J]. Int. J. Epidemiol. 2016; 45: 1600–1616.

No competing interests reported.

Selenium supplements and Polycystic Ovary syndrome and its key risk factors: a Mendelian Randomization Study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

1.Introduction

2. Materials and Methods

2.1. Study design

2.2. Exposure Measures

2.3. Outcome Measures

2.4. Two-sample MR

3. Results

3.1. Causality between Minerals and PCOS

3.2. Causality between Minerals and Hyperlipidaemia

3.3. Causality between Minerals and Obesity

3.4. Causality between Minerals and Insulin resistance

4. Discussion

5. Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1