Genetically predicted the causal association between serum mineral elements with immune thrombocytopenia and Henoch-Schonlein purpura: a bidirectional two-sample Mendelian randomization analysis

Abstract

Background

See More

Worldwide, the diagnosis and treatment of immune thrombocytopenia (ITP) and Henoch-Schönlein purpura (HSP) remain a major and ongoing challenge in hematology. Emerging clinical evidences suggest serum mineral elements are associated with ITP or HSP, but the causal relationship between them is still unclear.

Aims

Conducting a two-sample, bidirectional Mendelian randomization (MR) study to evaluate the causal association between serum mineral elements including zinc, copper, magnesium, iron and calcium with ITP and HSP.

Methods

In this two-sample, bidirectional MR study, summary statistics data of genome-wide association studies (GWAS) on exposures including zinc, copper, iron, magnesium and calcium were extracted from the MRC-Integrative Epidemiology Unit (MRC-IEU). The GWAS data on study outcomes, including ITP and HSP, were obtained from the FinnGen consortium. MR-Egger intercept and MR-PRESSO global test were utilized to assess the heterogeneity and horizontal pleiotropic of instrumental variables (IVs) between the exposures and outcomes, respectively. Inverse variance weighted (IVW) test was used as the primary analysis method to evaluate the causal between serum mineral elements with the risk of ITP and HSP, and weighted-median, weighted model, MR steiger, MR-PRESSO and radial MR were used as auxiliary analysis methods, moreover, the odds ratio (OR) and 95% confidence interval (CI) were calculated. Reverse MR analysis was also conducted. Leave-one-out test was further to conduct whether the association between serum mineral elements and the risk of ITP and HSP remain robust.

Results

No significant horizontal pleiotropy and heterogeneity between individuals IVs was found after MR-Egger and MR-PRESSO global test. Genetically predicted that high copper (OR = 0.768, 95%CI: 0.628–0.937) and magnesium (OR = 0.314, 95%CI: 0.112–0.884) concentrations may reduce the risk of ITP and HSP, respectively. High calcium concentration may increase the risk of HSP (OR = 1.823, 95%CI: 1.226–2.712). There was no significant evidence to support a causal association between iron, zinc, magnesium, and calcium with the risk of ITP, or between iron, copper, and zinc and the risk of HSP (all P > 0.005). Moreover, no reverse causal associations between five serum mineral elements with the risk of ITP and HSP were found (all P > 0.05), suggesting the causal associations between serum mineral elements with ITP and HSP were not bidirectional. In addition, consistent results were obtained by multiple sensitivity analyses, indicating the associations of serum mineral elements with the risk of ITP and HSP relatively robust.

Conclusion

In this MR study, we discovered genetically predicted that elevated serum levels of copper and magnesium decreased the risk of ITP and HSP, respectively, and elevated levels of serum calcium increased the risk of HSP. However, no reverse causal association was found between serum mineral elements with the risk of ITP and HSP.

Introduction

Purpura is a set of hemorrhagic disorders caused by abnormal blood vessel or platelet function, accounting for about 1/3 of hemorrhagic disorders. Its clinical manifestations are commonly purple-red ecchymoses or even bleeding on the skin and mucous membranes. Immune thrombocytopenia (ITP) and Henoch-Schönlein purpura (HSP) [also named immunoglobulin A vasculitis (IgAV)] are two typical types of purpuric diseases [1, 2]. ITP is an autoimmune disease in which the immune system mistakenly attacks platelets, resulting in a decrease in platelet count [3]. By contrast, HSP mainly involves immune-mediated inflammation of small blood vessels, which is characterized by the deposition of IgA immune complexes in blood vessels. Up to 75% of adult ITP and HSP patients may develop chronic diseases, which impose a significant burden on patients’ quality of life [4].

In clinical practice, there are no specific diagnostic criteria for ITP and HSP; the diagnosis is primarily exclusionary, requiring the elimination of secondary causes of thrombocytopenia and other confounding factors before a definitive diagnosis can be established. Patients are often diagnosed only at advanced disease stages, which significantly compromises the timeliness of treatment [5]. Therefore, exploring biomarkers associated with the risk of ITP and HSP is helpful to assist in the identification and screening of high-risk groups for ITP and HSP. Serum mineral elements play vital biological functions in the body, such as maintaining fluid balance and osmotic pressure, enzyme activity, and body composition [6]. Although the pathological mechanism of ITP and HSP have not been fully elucidated, there is evidence that oxidative stress plays an important role in the pathogenesis of ITP and HSP [7, 8]. Zinc and copper are vital mineral elements in the human antioxidant system. Zinc can prevent the formation of free radicals and has an antioxidant effect. It prevents the formation of superoxide by inhibiting the pro-oxidant enzyme, nicotinamide adenine dinucleotide phosphate oxidase [9]. Copper also has antioxidant properties, and its deficiency reduces the activity of ceruloplasmin and increase lipid peroxidation [10, 11]. Endothelial damage is another key mechanism in ITP disease progression [12]. Magnesium can increase the synthesis of nitric oxide by endothelial cells by upregulating endothelial nitric oxide synthase, thereby promoting endothelial repair and maintaining endothelial integrity [13, 14]. Calcium is a crucial intracellular second messenger involved in immune cell activation and the release of pro-inflammatory cytokines. Recent studies have revealed that calcium overload plays a significant role in the pathogenesis of ITP and HSP, particularly due to its close association with oxidative stress, inflammatory responses, and vascular endothelial injury [15]. In addition, studies have shown that iron overload can promote lipid peroxidation and reduce the activity of antioxidant enzymes, leading to an imbalance of oxidative stress. Iron deficiency may also cause abnormal glycosylation of IgA1 and promote the deposition of IgA immune complexes [16].

Based on previous related studies, we hypothesized that serum mineral elements may affect the risk of ITP and HSP. However, current evidence based on clinical data is not sufficient to support this conclusion. A case control study reported serum copper and zinc concentration are lower in patients with ITP than in healthy cases. However, the association between the levels of other mineral elements with ITP and HSP is not clear [17]. Moreover, there is also lack of research on the causal relationship between serum mineral elements with the risk of ITP and HSP.

To address this research gap, the present study took a genetic approach and explored the causal association between serum mineral elements with ITP and HSP. Observational studies are susceptible to reverse causation and potential confounding. Mendelian randomization (MR) is similar to random assignment in randomized clinical trials and can overcome the typical reverse causation and confounding problems in non-randomized observational studies [18]. Hence, in present study, a bidirectional two-sample MR analysis was conducted to evaluate the causal association between serum mineral elements with the risk of ITP and HSP, purposing to provide an evidence-based foundation for the screening of high-risk populations for ITP and HSP and potential treatment approaches.

Methods

Principle of MR study

The present study strictly adhered to the three principle of MR analysis: (1) the relevance assumption: instrumental variables (IVs) are strongly associated with the exposures’ (2) the independence assumption: IVs are not associated with the confounding factors between exposures and outcomes; (3) the exclusion restriction assumption: IVs affect the outcomes only through exposures, not directly (Fig. 1).

Fig. 1

The directed acyclic graph between serum mineral elements and ITP and HSP

Sources of data on exposures and outcomes

The summary statistics data of genome-wide association study (GWAS) on exposures including zinc, copper, calcium, magnesium and iron were extracted from the MRC Integrative Epidemiology Unite (MRC-IEU) (https://gwas.mrcieu.ac.uk/). We obtained summary-level data for all single nucleotide polymorphisms (SNPs) associated with serum zinc, copper, calcium, magnesium and iron in GWAS, based on 2,603, 2,603, 315, 153, 64, 979 and 23, 986 participants of European ancestry, respectively.

The GWAS data of outcomes including ITP and HSP were obtained from the FinnGen consortium (https://r10.finngen.fi/). The FinnGen project, launched in 2017, purposes to improve health through genetic research by collecting biospecimens from 500, 000 participants over a six-year period in Finland [19]. For our study, ITP-GWAS and HSP -GWAS included 59, 124 and 431, 365 cases. The detailed information on the GWAS datasets was exhibited in Table S1.

Selection of IVs

The screening process of IVs selection was exhibited in Fig. 2. In brief, the serum mineral elements including zinc, copper, calcium, magnesium and iron were served as exposure factors, ITP and HSP served as outcome factors. To ensure the accuracy and reliability of the conclusion on the causal relationship between the exposure and outcome, multiple quality control methods were performed to select the eligible IVs: (1) above all, in order to satisfy the first principle of MR analysis, we used the common genome-wide statistical significance threshold of P < 5$\:\times\:$10-⁸ to screen the SNPs strongly associated with zinc, copper, calcium, magnesium and iron as IVs. However, only a few SNPs associated with iron, copper, zinc and magnesium were screened out, which may weaken the analytical power of MR to a certain extent. To capture a more comprehensive association between serum mineral elements with ITP and HSP, we considered use a more lenient threshold of P < 5$\:\times\:$10-⁶ to further screen the IVs associated with iron, copper, zinc and magnesium; (2) secondly, SNPs with high linkage disequilibrium (LD) were eliminated with a strict r² cutoff of 0.0001 and a clumping window greater than 10,000 kb [20]; (3) thirdly, pledging that the SNPs effect on exposure correspond to the same allele as the effects on the outcome is the vital point in MR study. Therefore, we deleted palindromic SNPs to avoid distortion of strand orientation or allele coding (such as A/T or G/C alleles).

Fig. 2

Flow chart about the analytical methods and how the MR analysis was performed step-by-step

Ulteriorly, MR-Egger regression and MR pleiotropy residual sum and outlier (MR-PRESSO) test were performed to test the horizontal pleiotropy and heterogeneity. We set F > 10 to exclude weak IVs when quantifying using the F-statistic. The F-value was calculated using the following formula [21]:

$${R^2} = \sum\limits_{i = 1}^k {{{2\beta _i^2MA{F_{SNPi}}(1 - MA{F_{SNPi}})} \over {S{D^2}}}} $$

$$F = {{{R^2}(n - 1 - k)} \over {(1 - {R^2})k}}$$

Note: MAF, minor allele frequency; β, the effect size of SNP on exposure; SD: standard deviation; k, number of IVs; n, sample size of exposure.

Statistical analysis

Several analyses based on the two-sample, bidirectional MR were carried out to evaluate the causal association between serum mineral elements with ITP and HSP, including inverse variance weighted (IVW), weighted median analysis, weighted mode, MR-steiger, MR-PRESSO, and radial MR. The IVW method was utilized as the primary analysis for assessing the causal effects [20]. There was no heterogeneity among IVs, and the causa association between serum mineral elements with ITP and HSP should be interpreted by IVW fixed effect model. The results of bidirectional MR analysis were reported with the evaluation index of odds ratio (ORs) and 95% confidence intervals (CIs).

The MR-steiger test was conducted to assess the directional causality of the serum mineral elements on ITP and HSP, moreover, the reverse MR analysis was also conducted. We believed that there is a significant causal association existing between serum mineral elements and ITP and HSP if the following four conditions are met: (1) significant statistical difference was presented after IVS analysis (P < 0.05); (2) IVW, weighted median and MR-Egger methods have consistent estimates of the results; (3) no significant results were yielded from both the MR-Egger intercept test and the MR-PRESSO global test (P > 0.005); (4) The MR-steiger tests indicated TURE (P < 0.05). In addition, sensitivity analyses based on leave-one-out, MR-PRESSO, racial MR were performed to assess whether the causal association between serum mineral elements with ITP and HSP remain robust. Statistical analyses were conducted using R version 4.3.3 (2023-03-15 ucrt) with the R package “TwoSampleMR”.

Results

The selection of SNPs

Based on a battery of quality control methods, 6 to 189 SNPs associated with ITP or HSP were screened out, as elucidated in Table 1a and b. The F-value of each SNP was greater than 10, suggesting a low possibility of weak IVs bias. The degree of explanation (R²) of each IVs to exposures are ranged from 0.67 to 10.14%. Furtherly, MR-Egger test and MR-PRESSO global test were conducted to evaluate the horizontal pleiotropy and heterogeneity. The results indicated that no horizontal pleiotropy and heterogeneity were existed between each IVs (all P > 0.05) (Table 2).

Table 1a The screening procedure of SNPs in forward MR analysis (P < 5$\:\times\:$10⁻⁸)

Table 1b The screening procedure of SNPs in forward MR analysis (P < 5$\:\times\:$10⁻⁶)

Table 2 The pleiotropic and heterogeneity test of exposures instrument variables in forward MR analysis

Forward MR analysis

Firstly, we conducted a forward MR analysis to evaluate the causal association between serum mineral elements with ITP and HSP, and the results were shown in Table 3. Based on IVW fixed effect models, suggestive evidence was found between genetically increased serum copper (OR = 0.768, 95%CI: 0.628–0.937, P = 0.009) and magnesium (OR = 0.314, 95%CI: 0.112–0.884, P = 0.028) and a reduced risk of ITP and HSP, respectively. We also found low serum calcium concentration (OR = 1.823, 95%CI: 1.226–2.712, P = 0.003) may increase the risk of HSP. However, we did not find a significant causal association between serum iron, zinc, magnesium and calcium and the risk of ITP, nor did we find a significant causal association between serum iron, copper and zinc with the risk of HSP (all P > 0.05). The results of weighted median and weighted mode were attached in Supplementary Table 1. The forest plot of serum mineral elements and ITP and HSP suggest the consistent causal associations (Fig. 3).

Table 3 The forward MR analysis of serum mineral elements with ITP and HSP

Fig. 3

The forest plot of causal association between serum mineral elements with ITP and HSP. Note, SNP, single nucleotide polymorphisms; analyses were conducted using the IVW methods

Reverse MR analysis

Then, the reverse MR analysis, considering the ITP and HSP as the exposures and five serum mineral elements as the outcomes, were conducted to assess the bidirectional causal association between them. As shown in Table 4, we found no horizontal pleiotropic and heterogeneity were existed between each IVs (all P > 0.005). Table 5 exhibited the results the reverse MR analysis between ITP and HSP with the five serum mineral elements. Based on IVW analysis, the results shown no reverse causal association between the risk of ITP and HSP with the serum mineral elements (all P > 0.005). In conclusion, the results of our study do not support a bidirectional causal association between serum mineral elements and the risk of ITP and HSP.

Table 4 The pleiotropic and heterogeneity test of exposures instrument variables in reverse MR analysis

Table 5 The reverse MR analysis of serum mineral elements with ITP and HSP

Sensitivity analyses

To verify the causal association between serum mineral elements with the risk of ITP and HSP, we conducted sensitivity analyses based on leave-one-out. The scatter plots illustrated the consistent direction of several methods (Fig. 4a, b and c). The results of leave-one-out analysis reported no significant difference in causal evaluations of serum mineral elements and ITP and HSP, suggesting that none of the identified causal associations were driven by any single IVs (Fig. 5a, b, c). Above results supported the potential causal association between copper, magnesium and calcium with the risk of ITP and HSP.

Fig. 4a

The scatter plot for copper with ITP

Fig. 4b

The scatter plot for magnesium and HSP

Fig. 4c

The scatter plot for calcium and HSP. Note, ITP, immune thrombocytopenia; HSP, Henoch-Schönlein purpura; the four trajectories depict effect estimates derived from four distinct MR analyses; individual SNPs are represented by plotted points, with the abscissa (x-axis) quantifying exposure effects and the ordinate (y-axis) measuring outcome effects

Fig. 5a

Leave-one-out sensitivity analysis of copper and ITP

Fig. 5b

Leave-one-out sensitivity analysis of magnesium and HSP

Fig. 5c

Leave-one-out sensitivity analysis of calcium and HSP

Discussion

At the genetic prediction level, we conducted two-sample, bidirectional MR analysis to evaluate the causal association between serum mineral elements and the risk of ITP and HSP. Based on IVW analysis, we found elevated serum copper and magnesium concentrations may decrease the risk of ITP and HSP, respectively, and elevated serum calcium may increase the risk of HSP. In consideration of the potential bias in IVW analysis techniques, auxiliary MR methods may help to determine causality more precisely. Our sensitivity analyses, using various MR methods, presented the consistent results to the primary results. Additionally, reverse MR analysis did not report significant inverse causal associations between ITP and HSP with any serum mineral elements studied in present research. To our best knowledge, this was the first attempt to explore the causal association between serum mineral elements with the risk of ITP and HSP from the genetic epidemiology perspective.

Although ITP and HSP have different etiology and clinical manifestations, they are both immune-mediated bleeding disorders [1, 2]. A properly functioning immune system supports physical and mental health, and the level of minerals in the body greatly influence how the immune system functions [22]. In recent years, epidemiological studies have been conducted on the association between serum mineral elements with ITP and HSP, but clinical research evidence is insufficient and there is lack of causal research. A case-control study from Turkey examined 154 patients reported that compared with healthy participants, serum copper and zinc concentrations were lower in ITP patients [16]. In present study, we discovered serum copper is the protective factor for ITP, which was consistent with above case-control study. Copper is an essential trace element in the human body that is related to systemic immune activity and is also an important cofactor for oxidative balance. It is mainly found in the liver, bones and muscles [23, 24]. Previous animal experiments demonstrated copper-deficient mice have a reduced thymus size, an enlarged spleen, and a decrease in neutrophils and T cells [25]. The copper concentration in vivo should be balanced from the insight of the immune system, as low concentration is enough to reach the optimal immune function, while too high concentration can have harmful functions [26]. However, the association of optimal serum copper concentration with ITP risk needs further elucidation.

Magnesium has numerous important tasks in regulatory cellular functions, especially in regulations of immunological functions [27]. Magnesium has effects on the acute phase response and function of macrophages, moreover, it also has a major influence on the development, differentiation, and proliferation of lymphocytes [28]. Like copper-deficient, magnesium-deficient mice also exhibit early regression of the thymus, which negatively affects the T-cells [29]. In addition, magnesium stimulates vascular endothelial function by affecting the release if nitric oxide, endothelin-1, and prostacyclin [30], and is also a key mechanism in HSP [31].

We also found serum calcium concentration is the potential risk factor on the risk of HSP. Purpura is characterized by severe ADAMTS13 deficiency due to anti-ADAMTS13 autoantibodies, followed by accumulation of ultra-large von Willebrand factor (VWF) multimerizes in the circulation [32]. Serum calcium maintains vascular endothelial integrity by regulating cadherins and tight junction proteins. Hypocalcemia induces endothelial cell contraction, widening intercellular gaps and promoting IgA immune complex deposition. As an essential cofactor for clotting Factor IV (FXIII activation), calcium deficiency may exacerbate microthrombosis in HSP. Concurrently, calcium signaling activates phospholipase A2 (PLA2), triggering prostaglandin and leukotriene release, which amplifies the vasculitic response. Moreover, elevated reactive oxygen species (ROS) in HSP open mitochondrial permeability transition pores (mPTP), causing calcium efflux that further increases ROS production– establishing a vicious cycle of oxidative-endothelial injury [33, 34].

Our study has several standout characteristics. First, this study emphasized the importance of genetic-based analysis methods in revealing the causal association between serum mineral elements with ITP and HSP by using two-sample, bidirectional MR method. This method helps to overcome the limitations of traditional observational studies, such as the influence of confounding factors and reverse causality, and thus provide more reliable causal inferences. Second, based on the findings of this study, we found that serum copper and magnesium are protective factors for ITP and HSP, respectively, while serum calcium concentration is a risk factor for HSP. These results enrich clinicians’ diagnosis and prevention of ITP and HSP, and provide new ideas for the genetic research of ITP and HSP. Third, a rigorous process was utilized to screen for significant SNPs and for the first time performing a comprehensive causal analysis of the association between serum mineral elements with the risk of ITP and HSP.

Nevertheless, our study still has several limitations, and the results need to be interpreted with caution. First, the present MR analysis relied on existing genetic data and may not be able to take into account other non-genetic factors, such as demographic distribution characteristics, gender, and lifestyle, which may affect the incidence of ITP and HSP. Future studies can integrate the effects of genetic and non-genetic factors by incorporating more epidemiological data and multicenter collaboration, evaluate the universality and cross-ethnic applicability of these findings, and further accurately evaluate the causal association between serum mineral elements with ITP and HSP. Second, the current study only considered five serum mineral elements, zinc, iron, copper, magnesium and calcium, and could not rule out the association effects between other serum mineral elements that were not included with ITP and HSP, Further studies should expand the coverage of serum mineral elements and determine a more comprehensive set of serum mineral factors through systematic literature review to further reveal the potential causal association between serum mineral elements with ITP and HSP. Third, MR studies reflect a lifetime cumulative effect and cannot capture the time window of exposure or the dose-response relationship. Fourth, the genetic variants used in MR are screened from GWAS and reflect population averages rather than individual patient-specific variants. Even if an SNP is significantly associated with disease risk, the effect size (e.g., OR or β value) is a population statistic and cannot be directly mapped to the pathological mechanism of a particular patient. Fifth, in order to capture more IVs related to iron, copper, zinc and magnesium, we used a looser threshold to structure IVs, which may increase the possibility of violating the first assumption of the MR principle to a certain extent. However, the F-statistics of each SNP were greater than 10, indicating that the possibility of weak instrumental variables was low. Last but not least, the number of SNPs related to ITP and HSP included in current MR study was finite, and the diseases mechanisms cannot be fully elucidated. Further studies should consider techniques such as molecular biology, and increase sample size and multi-center collaboration to obtain more comprehensive insights into the diseases.

Conclusion

In conclusion, this two-sample bidirectional MR analysis indicates a potential causal association between serum copper, magnesium and calcium with the risk of ITP and HSP, whereas, no reverse causal association was existed in ITP and HSP with serum mineral elements. This study provides new ideas for genetic research on ITP and HSP. In the future, prospective, multicenter, large-sample, long-term follow-up randomized controlled trials are needed to clarify this causal relationship.