Causal relationship between iron status and preeclampsia-eclampsia: a Mendelian randomization analysis

ABSTRACT

Background

Preeclampsia/eclampsia is a severe pregnancy-related disorder associated with hypertension and organ damage. While observational studies have suggested a link between maternal iron status and preeclampsia/eclampsia, the causal relationship remains unclear. The aim of this study was to investigate the genetic causality between iron status and preeclampsia/eclampsia using large-scale genome-wide association study (GWAS) summary data and Mendelian randomization (MR) analysis.

Methods

Summary data for the GWAS on preeclampsia/eclampsia and genetic markers related to iron status were obtained from the FinnGen Consortium and the IEU genetic databases. The “TwoSampleMR” software package in R was employed to test the genetic causality between these markers and preeclampsia/eclampsia. The inverse variance weighted (IVW) method was primarily used for MR analysis. Heterogeneity, horizontal pleiotropy, and potential outliers were evaluated for the MR analysis results.

Results

The random-effects IVW results showed that ferritin (OR = 1.11, 95% CI: .89–1.38, p = .341), serum iron (OR = .90, 95% CI: .75–1.09, p = .275), TIBC (OR = .98, 95% CI: .89–1.07, p = .613), and TSAT (OR = .94, 95% CI: .83–1.07, p = .354) have no genetic causal relationship with preeclampsia/eclampsia. There was no evidence of heterogeneity, horizontal pleiotropy, or possible outliers in our MR analysis (p > .05).

Conclusions

Our study did not detect a genetic causal relationship between iron status and preeclampsia/eclampsia. Nonetheless, this does not rule out a relationship between the two at other mechanistic levels.

KEYWORDS:

• Mendelian randomization
• iron status
• preeclampsia
• causal relationship
• genetic variants

Introduction

Preeclampsia/eclampsia, a complex hypertensive disorder arising during pregnancy, remains a significant cause of maternal and perinatal morbidity and mortality worldwide.¹ Characterized by new-onset hypertension and organ damage, particularly affecting the kidneys and liver, preeclampsia/eclampsia poses substantial challenges in obstetric care.² While extensive research has focused on understanding its underlying mechanisms, the exact etiology of preeclampsia/eclampsia remains multifactorial and elusive.³ One area of growing interest is the role of maternal iron status in the development and progression of preeclampsia/eclampsia. Iron is crucial for a healthy pregnancy, with approximately 1 gram of iron needed for maternal red blood cell expansion and the development of both the placenta and the fetus.⁴ Severe iron deficiency and iron deficiency anemia during pregnancy have been linked to elevated risks of maternal mortality, perinatal mortality, preterm birth, low birth weight, compromised immune function, and enduring cognitive impairments in newborns and infants.⁵ To mitigate these adverse outcomes, the World Health Organization currently recommends daily iron supplementation for all pregnant adolescents and adult women. However, the potential repercussions of elevated maternal iron levels during pregnancy remain uncertain. Excessive iron intake can trigger the production of harmful reactive oxygen species, resulting in the damage of proteins, lipids, and nucleic acids, or potentially increasing susceptibility to conditions such as preeclampsia/eclampsia.^6–8 Nevertheless, the relationship between maternal iron status and preeclampsia/eclampsia is complex and has not been conclusively elucidated.

Traditionally, observational studies have reported conflicting findings regarding the association between iron levels and preeclampsia/eclampsia. Some studies suggest that higher iron levels may be associated with an increased risk of preeclampsia/eclampsia, possibly due to oxidative stress caused by excess iron.^9–13 Margaret P. Rayman and her colleagues discovered that preeclampsia exhibited significantly elevated serum iron concentration, ferritin levels, and transferrin saturation percentages compared to the control group.¹⁴ Conversely, other studies propose that iron deficiency could contribute to the endothelial dysfunction and oxidative stress in preeclampsia.¹⁵ These studies have elucidated the connection between iron status and the pathogenesis of preeclampsia/eclampsia, highlighting the potential significance of investigating their interrelation for a comprehensive understanding of mechanisms and prevention. Nevertheless, the precise causal link between these factors remains unestablished. To address this knowledge gap, we aim to conduct a thorough genetic analysis, delving into the genetic underpinnings of this relationship.

Mendelian randomization (MR), an innovative approach that utilizes genetic variants as instrumental variables, offers an opportunity to assess causality in observational epidemiology.¹⁶ Genetic variants associated with a modifiable exposure, such as iron status, are used as proxies to estimate the causal effect of that exposure on an outcome. MR can overcome some of the limitations of traditional observational studies, including confounding and reverse causation.¹⁷ In this study, we aim to explore the causal relationship between maternal iron status and the risk of preeclampsia/eclampsia using MR analysis. By leveraging genetic data from large-scale genome-wide association studies (GWAS), we seek to provide more robust evidence on whether iron status contributes to the pathogenesis of preeclampsia/eclampsia.

Materials and methods

Study design description

In this study, we utilized single-nucleotide polymorphisms (SNPs) obtained from GWAS datasets as instrumental variables (IVs) to explore the potential causal relationship between iron status and the occurrence of preeclampsia/eclampsia. Figure 1 provides an overview of the study design. Our investigation adhered to the fundamental assumptions integral to the MR approach: (1) all chosen IVs exhibited a high degree of correlation with the exposure (p < 5 × 10⁻⁶, F statistic > 10); (2) none of the selected IVs were influenced by confounding factors connecting the exposure and the outcomes; (3) the impact of all selected IVs on the outcomes was solely through the exposure and not via alternative pathways. It is essential to note that all datasets employed in this study are publicly accessible. As a result, no additional ethical approvals or consent for participation were deemed necessary.

Figure 1. Schematic overview of the study design.

MR study from iron status, including ferritin, serum iron, TSAT, and TIBC to the occurrence of preeclampsia/eclampsia: SNPs for iron status were selected from the IEU Open GWAS project database as exposure variables, whereas data of preeclampsia/eclampsia was retrieved separately from FinnGen. MR analyses were conducted per exposure and outcome. Abbreviations: MR, Mendelian randomization; SNPs, single‐nucleotide polymorphisms; TSAT, transferrin saturation; TIBC, total iron-binding capacity.

Summary statistics of serum iron status and PE or eclampsia

We obtained genetic data pertaining to iron status from the Integrative Epidemiology Unit (IEU) open GWAS project, accessible at https://gwas.mrcieu.ac.uk/. Our search utilized specific codes, namely “ieu-a-1049,” “ieu-a-1050,” “ieu-a-1051,” and “ieu-a-1052,” to retrieve relevant data for serum iron, ferritin, transferrin saturation (TSAT), and total iron-binding capacity (TIBC), respectively. This genetic data was sourced from a comprehensive GWAS involving 48 972 individuals of European descent.¹⁸ Within this GWAS, the discovery phase incorporated summary data, highlighting genome-wide allelic associations between SNP genotypes and various iron markers. This information was derived from 23 986 participants of European ancestry, encompassing 11 cohorts across 9 participating centers. To validate and solidify suggestive and significant associations, replication samples were drawn from an additional 24 986 participants of European ancestry, spanning 8 separate cohorts. In each cohort, distinct genome-wide association tests, genotype imputation, and quality control measures were conducted.¹⁸

Genetic data pertaining to preeclampsia/eclampsia was extracted from the publicly accessible FinnGen database (https://finngen.gitbook.io/documentation). This database represents a collaborative venture between the public and private sectors, amalgamating genotype information sourced from Finnish biobanks with digital health record data from Finnish health registries. The comprehensive GWAS encompassed a substantial cohort, comprising 6,436 cases afflicted with preeclampsia/eclampsia and 176,113 matched controls. Intricate particulars concerning the cohorts engaged, genotype datasets employed, precise endpoint definitions, and the methodologies employed for association testing within the FinnGen consortium are comprehensively documented and readily accessible on the official FinnGen website.

Selection of instrumental variables for MR analyses

IVs for our MR analyses were selected from two distinct GWAS summary datasets. Initially, we identified SNPs that achieved a genome-wide significance threshold of p < 5 × 10⁻⁶. Subsequently, we pruned these SNPs for linkage disequilibrium, thereby retaining only independent IVs devoid of any appreciable linkage disequilibrium (r² = .001, kb = 10,000).¹⁹ Further refinement ensued as we excluded SNPs demonstrating any discernible association with the outcome variable (p < 5 × 10⁻⁶).¹⁹ During the crucial harmonization phase, designed to ensure congruence between the exposure and outcome data, we eliminated palindromic SNPs characterized by intermediate allele frequencies.²⁰ To gauge the instrumental strength, we computed the F-statistics. An F-statistic exceeding the threshold of 10 serves as a robust indicator, suggesting that the presence of weak instrument bias is improbable.²¹ This selection process was undertaken to furnish our MR analysis with robust and reliable instrumental variables.

Statistic

In this investigation, we employed the random-effects inverse-variance weighted (IVW) method as the primary statistical approach to evaluate potential causal associations between iron status and the occurrence of preeclampsia/eclampsia. The IVW method operates under the fundamental assumption that all core prerequisites of MR are met.²² Nonetheless, recognizing that IVs may potentially exert effects on the outcome via alternate pathways, indicating the presence of potential horizontal pleiotropic effects and consequent bias in the causal estimate derived from IVW, we conducted a series of sensitivity analyses. These included the MR-Egger, Weighted median, Simple mode, and Weighted mode methods, each of which furnishes precise estimates of the causal relationship, even in the presence of potentially invalid SNPs. To scrutinize the core assumptions of MR, we employed several diagnostic tools. In assessing the relevance assumption, we computed R², offering insight into the proportion of variance in the exposure variable that can be attributed to genetic variants. Additionally, we evaluated the instrument strength by estimating F-statistics, with values below 10 indicative of weak instrument strength.²³ For an in-depth exploration of the exclusion restriction assumption, we employed the MR-Egger regression intercept and its corresponding 95% confidence intervals (CIs) to gauge the extent of bias in the causal estimates due to directional pleiotropy.²⁴ Furthermore, we assessed horizontal pleiotropy using the Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) global test and excluded any outlying SNPs identified by the MR-PRESSO outlier test.²⁵ Subsequently, we investigated potential variations in results following the removal of outlying instrumental variables. Heterogeneity in outcomes, as derived from both the IVW and MR-Egger methods, was tested employing Cochran’s Q statistics and subsequently visualized through funnel plots.²⁶ Moreover, we conducted an array of sensitivity analyses, encompassing leave-one-out analysis and single SNP analysis, with the aim of elucidating the influence of individual SNPs on the primary causal relationship.²⁷ For binary outcomes, we presented odds ratios (ORs) accompanied by their corresponding 95% CIs to quantify the strength of the causal relationship. In scenarios involving both binary and continuous outcomes, we reported the causal estimate, p-value, β coefficient, and standard error. All analytical procedures were executed utilizing the MR packages within the R environment (version 4.3.1).

Results

Quality control for genetic instruments

In this comprehensive analysis, we incorporated 10, 8, 17, and 11 autonomous genetic variants into our MR investigations, specifically tailored to ferritin, serum iron, TIBC, and TSAT, respectively. It’s noteworthy that our meticulous scrutiny, encompassing the MR-PRESSO global test, did not yield any discernible evidence of horizontal pleiotropy for each of the aforementioned iron status biomarkers (all p > .05). Furthermore, in our evaluation of the data, we detected no indications of between-SNP heterogeneity or directional pleiotropy. This was corroborated by both Cochran’s Q test and MR-Egger regression, conducted across the spectrum of ferritin, serum iron, TIBC, and TSAT (Table 1). These findings substantiate the robustness and reliability of our MR analyses in elucidating potential causal relationships concerning iron status biomarkers.

Table 1. Pleiotropy and heterogeneity analyses.

		MR‐Egger regression		Heterogeneity analyses
Exposure	Outcome	Intercept	p intercept	Method	Q	Q pval
Ferritin	PE or eclampsia	−0.043	0.080	IVW	7.885	0.546
				MR‐Egger	3.883	0.867
Serum iron	PE or eclampsia	−0.011	0.499	IVW	9.425	0.224
				MR‐Egger	8.678	0.193
TSAT	PE or eclampsia	−0.009	0.488	IVW	13.460	0.176
				MR‐Egger	12.721	0.176
TIBC	PE or eclampsia	0.011	0.224	IVW	24.547	0.078
				MR‐Egger	22.166	0.104

Abbreviations: IVW, inverse variance weighted; MR, mendelian randomization; PE, preeclampsia; TSAT, transferrin saturation; TIBC, total ironbinding capacity.

The casual effect of iron status on PE or eclampsia

The effect sizes of the SNPs on both the exposures (ferritin, serum iron, TIBC, and TSAT) and the outcome (preeclampsia/eclampsia) have been presented through scatter plots (Table 2 and Figure 2). According to the recognized IVW method, our estimations indicate the absence of a causal link between ferritin (OR = 1.11, 95% CI: 0.89–1.38, p = .341), serum iron (OR = 0.90, 95% CI: .75–1.09, p = .275), TIBC (OR = 0.98, 95% CI: 0.89–1.07, p = .613), TSAT (OR = 0.94, 95% CI: 0.83–1.07, p = .354), and the risk of developing preeclampsia/eclampsia. In addition to the IVW approach, we scrutinized various statistical models; however, none of them yielded a statistically significant correlation between genetically predicted ferritin, serum iron, TIBC, and TSAT and the risk of preeclampsia/eclampsia (Figures 3a, 4 a, 5 a, and 6a, respectively). The funnel plots consistently underscore the absence of heterogeneity within the identified association (Figures 3b, 4 b, 5 b, and 6b, respectively). Furthermore, the leave-one-out sensitivity analysis demonstrates that no individual SNP exerts disproportionate influence on these results (Figures 3c, 4 c, 5 c, and 6c, respectively). Lastly, the cumulative estimations derived from both MR Egger and IVW methodologies collectively fail to establish a statistically significant association between iron status and the risk of developing preeclampsia/eclampsia (Figures 3d, 4 d, 5 d, and 6d, respectively).

Figure 2. Causal associations between iron status and preeclampsia or eclampsia risk.

MR analysis results of the four exposures (ferritin, serum iron, TIBC, and TSAT) and outcomes (preeclampsia/eclampsia). Five methods: random-effects IVW, MR Egger, weighted median, simple mode, and weighted mode. Abbreviations: MR, Mendelian randomization; TIBC, total iron-binding capacity; TSAT, transferrin saturation; IVW, inverse-variance weighted.

Figure 3. Casual effect of ferritin on preeclampsia or eclampsia risk.

(a) Scatter plot illustrating the relationship between ferritin and preeclampsia/eclampsia. The five methods utilized in this study are depicted, with lines in light blue, dark blue, light green, dark green, and red representing IVW, MR-Egger, Simple model, Weighted median, and Weight mode methods. (b) Funnel plot employed to evaluate the presence of potential heterogeneity in the observed association. (c) Leave-one-out analyses conducted to ascertain whether any individual instrumental variable was exerting disproportionate influence on the estimated causal effect. (d) Forest plot utilized to visually present the MR estimate and corresponding 95% confidence interval values for each SNP. Furthermore, the IVW and all-MR Egger results are presented at the bottom of the plot. Abbreviations: MR, Mendelian randomization; SNP, single-nucleotide polymorphism; CI, confidence interval; IVW, inverse variance weighted.

Figure 4. Casual effect of serum iron on preeclampsia or eclampsia risk.

(a) Scatter plot illustrating the relationship between serum iron and preeclampsia/eclampsia. The five methods utilized in this study are depicted, with lines in light blue, dark blue, light green, dark green, and red representing IVW, MR-Egger, Simple model, Weighted median, and Weight mode methods. (b) Funnel plot employed to evaluate the presence of potential heterogeneity in the observed association. (c) Leave-one-out analyses conducted to ascertain whether any individual instrumental variable was exerting disproportionate influence on the estimated causal effect. (d) Forest plot utilized to visually present the MR estimate and corresponding 95% confidence interval values for each SNP. Furthermore, the IVW and all-MR Egger results are presented at the bottom of the plot. Abbreviations: MR, Mendelian randomization; SNP, single-nucleotide polymorphism; CI, confidence interval; IVW, inverse variance weighted.

Figure 5. Casual effect of TIBC on preeclampsia or eclampsia risk.

(a) Scatter plot illustrating the relationship between TIBC and preeclampsia/eclampsia. The five methods utilized in this study are depicted, with lines in light blue, dark blue, light green, dark green, and red representing IVW, MR-Egger, Simple model, Weighted median, and Weight mode methods. (b) Funnel plot employed to evaluate the presence of potential heterogeneity in the observed association. (c) Leave-one-out analyses conducted to ascertain whether any individual instrumental variable was exerting disproportionate influence on the estimated causal effect. (d) Forest plot utilized to visually present the MR estimate and corresponding 95% confidence interval values for each SNP. Furthermore, the IVW and all-MR Egger results are presented at the bottom of the plot. Abbreviations: MR, Mendelian randomization; SNP, single-nucleotide polymorphism; CI, confidence interval; IVW, inverse variance weighted; TIBC, total iron-binding capacity.

Figure 6. Casual effect of TSAT on preeclampsia or eclampsia risk.

(a) Scatter plot illustrating the relationship between TIBC and preeclampsia/eclampsia. The five methods utilized in this study are depicted, with lines in light blue, dark blue, light green, dark green, and red representing IVW, MR-Egger, Simple model, Weighted median, and Weight mode methods. (b) Funnel plot employed to evaluate the presence of potential heterogeneity in the observed association. (c) Leave-one-out analyses conducted to ascertain whether any individual instrumental variable was exerting disproportionate influence on the estimated causal effect. (d) Forest plot utilized to visually present the MR estimate and corresponding 95% confidence interval values for each SNP. Furthermore, the IVW and all-MR Egger results are presented at the bottom of the plot. Abbreviations: MR, Mendelian randomization; SNP, single-nucleotide polymorphism; CI, confidence interval; IVW, inverse variance weighted; TSAT, total iron-binding capacity.

Table 2. Iron status and its association with preeclampsia/eclampsia in the MR analyses.

Exposure	Outcome	No. of SNPs	Methods	OR (95% CI)	β (SE)	p
Ferritin	PE or eclampsia	10
			IVW	1.11(0.89–1.38)	0.11 (0.11)	0.341
			MR‐Egger	2.09(1.08–4.03)	0.74 (0.33)	0.059
			Weighted median	1.22(0.92–1.63)	0.20 (0.15)	0.165
			Simple mode	1.16(0.76–1.79)	0.15 (0.22)	0.508
			Weighted mode	1.20(0.85–1.71)	0.18 (0.18)	0.327
Serum iron	PE or eclampsia	8
			IVW	0.90(0.75–1.09)	−0.10 (0.09)	0.275
			MR‐Egger	1.01(0.70–1.44)	0.01 (0.18)	0.968
			Weighted median	0.96(0.80–1.16)	−0.04 (0.10)	0.669
			Simple mode	0.66(0.41–1.06)	−0.41 (0.24)	0.132
			Weighted mode	0.95(0.78–1.15)	−0.05 (0.10)	0.622
TSAT	PE or eclampsia	20
			IVW	0.94(0.83–1.07)	−0.06 (0.06)	0.354
			MR‐Egger	0.99(0.82–1.20)	−0.01 (0.10)	0.943
			Weighted median	0.95(0.84–1.09)	−0.05 (0.07)	0.48
			Simple mode	1.01(0.80–1.29)	0.02 (0.12)	0.888
			Weighted mode	0.96(0.85–1.10)	−0.04 (0.07)	0.595
TIBC	PE or eclampsia	11
			IVW	0.98(0.89–1.07)	−0.02(0.04)	0.613
			MR‐Egger	0.93(0.83–1.04)	−0.07(0.06)	0.238
			Weighted median	0.94(0.87–1.03)	−0.06 (0.04)	0.186
			Simple mode	0.96(0.75–1.23)	−0.04 (0.12)	0.766
			Weighted mode	0.95(0.88–1.04)	−0.05 (0.04)	0.276

Abbreviations: CI, confidence interval; IVW, inverse variance weighted; MR, Mendelian randomization; PE, preeclampsia; SNPs, single‐nucleotide polymorphisms; OR, odds ratio; TSAT, transferrin saturation; TIBC, total iron binding capacity.

Discussion

Preeclampsia/eclampsia is a significant pregnancy-related disorder characterized by hypertension and damage to multiple organs, primarily affecting the placenta.²⁸ Despite extensive research, the exact etiology of preeclampsia/eclampsia remains elusive. One area of investigation has focused on maternal iron status, specifically serum iron, ferritin, TIBC, and TSAT, as potential risk factors.^11,13,29 In this study, we conducted a comprehensive MR analysis to investigate the causal relationship between maternal iron status and the risk of preeclampsia/eclampsia. Our results indicate that there is no causal relationship between these iron status biomarkers and preeclampsia/eclampsia. Our findings provide novel insights into the association of serum iron status with the occurrence of preeclampsia/eclampsia.

During pregnancy, the demand for iron increases by approximately 30% to support the processes of hematopoiesis in both the mother and the fetus.³⁰ In addition to its role in hemoglobin synthesis and red blood cell production, iron plays a crucial role in various biological processes within all cells, including energy production and DNA replication/repair. Excessive iron can display toxic effects, as free iron can induce oxidative stress, inflammation, and in severe cases, cell death through iron-dependent lipid peroxidation (ferroptosis).³¹ Recent studies have revealed various physiological adaptations in pregnant women to meet the increased iron demand. Notably, circulating hepcidin levels decrease significantly and become nearly undetectable in the latter stages of pregnancy.³² This decrease in hepcidin levels promotes the absorption of dietary iron and the release of iron from tissue stores. These adaptive changes, combined with the reduction in iron loss due to the absence of menstruation during pregnancy, might potentially restrict the effectiveness of feedback mechanisms in responding to environmental iron exposure.³³ Paradoxically, these adaptations may increase pregnant women’s susceptibility to clinical disorders associated with excessive iron accumulation. Previous epidemiological studies have demonstrated that the serum iron levels in patients with preeclampsia/eclampsia are higher compared to healthy pregnant women, and this association is also significant in case-control studies. Additionally, in both Asian and European populations, pregnant women with preeclampsia/eclampsia generally have higher serum iron levels than the healthy control group.³⁴ The results of previous observational studies did not align with our findings, and several potential explanations for this inconsistency are plausible. First, it’s worth noting that previous investigations often involved small study populations and limited sample sizes, which can result in wide confidence intervals for association estimates. Furthermore, estimates derived from observational studies are susceptible to inherent issues like residual confounding and reverse causality. While experimental studies have explored the connection between iron status and preeclampsia/eclampsia, their findings have not provided conclusive evidence of a direct effect. Instead, they have uncovered indirect pathways linking iron and preeclampsia/eclampsia, such as the promotion of oxidative stress, cell death, and compensatory proliferation. In this context, it’s possible that abnormal iron status could be a consequence of preeclampsia/eclampsia rather than a causal factor. Thus, the debate over the causal relationship between iron status and preeclampsia/eclampsia development remains ongoing and warrants further investigation.

Despite the strengths of our MR analysis, it is essential to acknowledge its limitations. MR assumes that genetic instruments are not associated with confounding factors other than the exposure of interest. While we conducted extensive quality control and sensitivity analyses to address potential biases, there might still be unaccounted factors or interactions that could influence our results. Additionally, our study focused on genetic determinants of iron status and did not consider other aspects of iron metabolism or potential interactions with other risk factors for preeclampsia/eclampsia.

Conclusions

In conclusion, our comprehensive MR analysis suggest that there is no causal relationship between maternal iron status, as determined by serum iron, ferritin, TIBC, and TSAT, and the risk of developing preeclampsia/eclampsia. These findings challenge previous observational studies and suggest that factors other than maternal iron status may play a more substantial role in the pathogenesis of preeclampsia/eclampsia. Future research should continue to explore the multifactorial nature of this complex pregnancy-related disorder to improve our understanding and develop effective prevention and management strategies.

Abbreviations

GWAS	=	genome-wide association study
MR	=	mendelian randomization
TIBC	=	total iron binding capacity
TSAT	=	transferrin saturation
IVW	=	inverse variance weighted
SNPs	=	single-nucleotide polymorphisms
Ivs	=	instrumental variables.

Acknowledgments

Thanks to all the contributions of authors.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by the China Postdoctoral Science Foundation (No: 2023M740431).