Serum iron levels as a risk factor for Osteoporosis-Related pathological fractures

Abstract

Background

Osteoporosis is a prevalent bone disease characterized by decreased bone mass and deterioration of bone microstructure, leading to an increased risk of fractures. Pathological fractures, particularly in postmenopausal women, are a severe complication of osteoporosis. This study investigates the potential molecular role of micronutrients, particularly iron, in osteoporosis and its association with pathological fractures.

Methods

We performed a Mendelian randomization (MR) analysis using genome-wide association study (GWAS) summary data to explore the relationship between 15 trace elements (including iron, calcium, copper, and vitamins) and osteoporosis-related outcomes. Outcome data were obtained from the FinnGen database. The inverse-variance weighted (IVW) method was used to evaluate the causal effects of micronutrients on osteoporosis and osteoporosis-related fractures, including those occurring in postmenopausal women. Sensitivity analyses for pleiotropy and heterogeneity, including MR-Egger and leave-one-out analysis, were also performed.

Results

Among the 15 micronutrients analyzed, serum iron levels were significantly associated with an increased risk of osteoporosis with pathological fractures. The IVW analysis revealed a strong association between iron levels and the occurrence of osteoporosis with pathological fractures (OR = 2.630, 95% CI: 1.161–5.957, P = 0.020), as well as postmenopausal osteoporosis with fractures (OR = 2.714, 95% CI: 1.156–6.368, P = 0.022). No significant genetic association was observed between iron and osteoporosis alone (OR = 1.238, 95% CI: 0.856–1.791, P = 0.256). Other micronutrients did not show significant effects on osteoporosis or fractures. Sensitivity analyses indicated no evidence of heterogeneity or pleiotropy.

Conclusion

This study identifies serum iron levels as a molecular risk factor for osteoporosis-related pathological fractures, particularly in postmenopausal individuals, but not for osteoporosis itself. These findings highlight the potential role of iron in the pathogenesis of osteoporotic fractures and suggest further investigation into its molecular mechanisms in bone health.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40520-025-03225-y.

Introduction

Osteoporosis is a widespread skeletal disorder, with a particularly high incidence among older women. It is characterized by decreased bone density and progressive degradation of bone tissue, leading to increased bone fragility and a substantially higher risk of fractures [1]. This risk is particularly elevated in postmenopausal women, where the marked decline in estrogen levels accelerates bone mass loss, further raising the likelihood of pathological fractures [2]. Pathological fractures, the most severe complication of osteoporosis, not only greatly impair patients’ quality of life but also contribute to increased healthcare costs and societal burden. Consequently, developing a comprehensive insight into the determinants of osteoporosis and its associated fractures, particularly modifiable ones, is crucial for enhancing patient outcomes.

Trace elements are essential for maintaining bone health. Numerous studies have demonstrated that trace elements, including zinc, selenium, and copper, influence bone health through various mechanisms, such as contributing to bone matrix synthesis [3], counteracting oxidative stress [4], regulating osteoblast and osteoclast activity, and preserving bone density [5]. However, the cause-and-effect link between these trace elements and osteoporosis, particularly in postmenopausal women, remains a topic of debate, especially concerning their involvement in the development of complications. Iron, a key trace element, is primarily found in hemoglobin, myoglobin, and various metabolic enzymes. Although essential for maintaining normal physiological functions, its role in bone metabolism has only recently attracted widespread attention. Research has indicated that iron may affect bone health through various mechanisms, such as promoting oxidative stress and triggering inflammatory responses, which modify osteoblast and osteoclast activity, ultimately disrupting the equilibrium between bone formation and resorption [6]. Specifically, both excessive and deficient iron levels have been associated with heightened oxidative stress and inflammatory responses, which contribute to reduced bone density and a greater likelihood of developing osteoporosis along with pathological fractures [7]. For instance, excessive iron has been shown to promote oxidative damage to proteins, lipids, and DNA within bone tissue, impair osteoblast function, and accelerate bone resorption [8]. However, the precise mechanisms by which abnormal iron metabolism contributes to osteoporosis and related pathological fractures remain insufficiently understood, warranting further investigation. These findings suggest that regulating serum iron levels could be a promising strategy for reducing the risk of osteoporosis and managing its associated complications, particularly in postmenopausal women. Nevertheless, additional evidence is required to elucidate its mechanisms of action and clinical significance.

Conventional observational research often encounters difficulties in determining causal links between trace elements and bone health because of its vulnerability to confounding factors and reverse causality [9]. The Mendelian randomization (MR) approach utilizes genetic variants as proxies to replicate randomized controlled trials, allowing for a more accurate assessment of the cause-and-effect relationship between exposure factors, such as serum iron levels, and disease outcomes [10]. A key advantage of the MR method is its ability to effectively control for confounding variables and reduce the impact of reverse causality, thereby providing more compelling support for determining causality.

The aim of this study is to utilize large-scale genetic data to determine the causal relationship of 15 trace elements with susceptibility to osteoporosis and its associated pathological fractures using the MR method. Specifically, the study aims to establish whether a significant link exists between iron metabolism and the likelihood of developing postmenopausal osteoporosis along with related pathological fractures. This study represents the inaugural systematic use of the MR method to investigate the causal link between multiple trace elements, particularly iron, and the risk of osteoporosis and its complications. The findings from this study offer a novel perspective on the role of trace elements in bone health, while providing a scientific foundation for future clinical practice and public health strategies.

Methods

Data on genetic prediction of Circulating micronutrient levels

This study utilized data sourced from various genome-wide association study (GWAS) databases comprising European ancestry populations, focusing on 15 trace elements potentially linked to osteoporosis and its associated risk of pathological fractures. The elements considered in this study include trace minerals such as copper, calcium, carotene, folate, iron, magnesium, potassium, and selenium, along with vitamins A, B12, B6, C, D, and E, and zinc. These micronutrients were selected based on evidence from prior studies indicating their close association with various health conditions, including osteoporosis and fracture risk. Specifically, the data for serum iron encompassed 64,979 samples and 9,851,867 SNPs, while the data for copper included 2,603 samples and 2,543,646 SNPs. Detailed data for the remaining elements are provided in the Table S1.

Selection and screening criteria for instrumental variables

To ensure the robustness of causal inference, stringent criteria were applied in the selection of instrumental variables (IVs). Single nucleotide polymorphisms (SNPs) linked to each micronutrient were selected as IVs according to the following criteria:

1.Association threshold: SNPs having a P value below 5 × 10⁻⁸ were chosen to ensure a robust correlation with the target micronutrient. 2.Linkage disequilibrium (LD) pruning: To reduce bias caused by LD, SNPs were pruned across a 10 Mb window, applying an R² cut-off of < 0.01 to maintain the independence of the chosen SNPs. 3.F-statistic test: To confirm the validity of SNPs as IVs, it was ensured that all selected SNPs had an F-statistic greater than 10. This criterion enhanced confidence in causal inference and reduced the risk of bias from weak instruments.

Data on genetic prediction of osteoporosis and related pathological fracture risk

Data on osteoporosis and related pathological fractures were obtained from the FinnGen consortium, an initiative involving collaboration between public and private sectors, which combines information from the Finnish Biobank and electronic healthcare records from various national registries. The analysis utilized the FinnGen version 10 dataset, which provides comprehensive information on osteoporosis and its associated pathological fractures. Specifically, the dataset for osteoporosis includes 8,017 cases of osteoporosis and 391,037 controls. For postmenopausal osteoporosis with pathological fractures, 1,486 cases and 228,601 controls are included. Additionally, the data for osteoporosis with pathological fractures comprise 1,822 cases and 311,210 controls. Detailed information regarding participant recruitment, genotyping, data processing, and quality control protocols is available on the FinnGen official website (https://www.finngen.fi/en).

Mendelian randomization analysis

The inverse-variance weighted (IVW) method was primarily employed to evaluate the causal effects of micronutrients on osteoporosis and pathological fractures. The IVW method estimates causal effects by applying a weighted linear model to the relationship between genetic instruments (such as SNPs) and exposure outcomes [11]. As one of the most widely applied methods in MR, the IVW approach functions on the premise that all instrumental variables are reliable and there is an absence of horizontal pleiotropy. When these assumptions are satisfied, the IVW method provides unbiased estimates of causal effects. To confirm the robustness and validity of the findings, additional evaluations were performed using the weighted median method and MR-Egger regression. The weighted median approach offers dependable causal inferences as long as at least half of the instrumental variables are valid by assigning weights to the effect sizes estimated by each SNP and taking the median value [12]. MR-Egger regression, on the other hand, removes the constraint on the intercept, allowing it to be non-zero, which enables the detection and correction of potential horizontal pleiotropy [13]. Additionally, both the simple model and weighted model methods were applied. The simple model assumes that most instrumental variable effects represent true causal effects and selects the most common effect value as the estimate to address potential biases [14]. The weighted model approach, which assigns weights to the estimated effects of instrumental variables based on their precision, is capable of delivering a reliable effect estimation even when some instrumental variables are invalid [15]. To confirm the validity of the genetic instruments, SNPs linked to each micronutrient were extracted from the GWAS database. Those exhibiting P-values below 5 × 10⁻⁸ were selected as IVs. The robustness of these IVs was evaluated through the calculation of the F-statistic. The F-statistic was calculated through a formula, in which R² represents the proportion of exposure variance explained by the genetic instrument, and N refers to the GWAS sample population. Additionally, R² was calculated using the formula , where β is the regression coefficient for the allele’s effect on the exposure, and MAF denotes the frequency of the minor allele. An F-statistic exceeding 10 was considered indicative of strong instrumental variables.

Sensitivity analysis

Multiple sensitivity analyses were performed to assess potential horizontal pleiotropy and other factors that might impact the findings, ensuring the reliability of the MR results. First, Cochrane’s Q statistic was used to assess heterogeneity among SNP estimates, providing a quantitative measure of the consistency of genetic instrumental variable effects across different SNPs. A high Q value indicating significant heterogeneity may point to horizontal pleiotropy, which can undermine the precision of causal estimates [16]. Second, MR-Egger regression was used to identify and correct for pleiotropic effects. By relaxing the intercept constraint inherent in the IVW method, MR-Egger allows for pleiotropic effects to be accounted for. Special consideration was given to the intercept in MR-Egger analysis to evaluate the degree of pleiotropic effects, as this is essential for confirming the validity of the causal estimates. Additionally, the MR-PRESSO (Mendelian Randomization Pleiotropy Residual Sum and Outlier) method was applied to further investigate pleiotropy and identify outliers. MR-PRESSO detects and excludes outliers caused by pleiotropic effects, thereby enhancing the precision of causal estimates [17]. This comprehensive approach provided greater insight into the potential influence of pleiotropy on the analysis, ensuring the robustness of the findings.

Statistical analysis

Statistical analyses were conducted in R software (version 4.4.1) using several specialized R packages for data processing and interpretation of results. Beyond the basic packages employed across all analyses, VariantAnnotation was used for genetic variant annotation [18], while Gwasglue facilitated the integration and manipulation of GWAS data [19]. For visualization, the circlize [20] and ComplexHeatmap packages [21] were utilized to create complex circular diagrams and heat maps, respectively, and forestploter was employed to generate forest plots [22]. The combined application of these tools allowed for comprehensive and precise sensitivity analyses, thereby ensuring the rigor and credibility of the Mendelian randomization results.

Results

Genetically predicted causal relationships between trace elements and osteoporosis associated with pathological fractures

The causal relationship between 15 micronutrients and osteoporosis with pathological fractures was explored, with a significant analysis conducted for iron using the IVW method. The results indicated a notable causal relationship linking iron concentration to osteoporosis with pathological fractures. Specifically, the IVW analysis identified a strong correlation between iron and osteoporosis accompanied by pathological fractures (OR = 2.630, 95% CI: 1.161–5.957, P = 0.020), as shown in Fig. 1A. This finding indicates that iron could act as a key contributor to osteoporosis with pathological fractures. In contrast, no meaningful causal links were identified between other micronutrients and osteoporosis with pathological fractures (Fig. 2A). Notably, the results for vitamin A approached significance, with a P value of 0.069, as shown in Table S2. Although this result fell short of the conventional significance threshold, the P value close to 0.05 and the wide confidence interval suggest a potential correlation between vitamin A and osteoporosis with pathological fractures, warranting further investigation. None of the analyses indicated significant heterogeneity or pleiotropic effects (Table 1). The leave-one-out analysis further supported the stability and reliability of the results, as no abnormal SNPs were identified (see Supplementary Figure). In summary, this study underscores the significance of iron in the onset of osteoporosis accompanied by pathological fractures. While the results for vitamin A did not reach statistical significance, they suggest a potential correlation that warrants further investigation. These findings enhance our understanding of how micronutrients influence osteoporosis and its complications.

Fig. 1.

MR analysis results of serum iron levels in osteoporosis with pathological fractures (A) and postmenopausal osteoporosis with pathological fractures (B)

Fig. 2.

Results of five MR methods for the relationship between 15 trace elements and three outcomes. (A) Osteoporosis with pathological fractures. (B) Postmenopausal osteoporosis with pathological fractures. (C) Osteoporosis

Table 1.

Results of heterogeneity and Pleiotropy tests for the exposure factor of serum iron levels and the three outcomes: osteoporosis with pathological fractures, postmenopausal osteoporosis with pathological fractures, and osteoporosis

Outcome	Exposure	Pleiotropy test	Heterogeneity test
		p_value	MR Egger	IVW
Osteoporosis with pathological fracture	Iron	0.61	0.29	0.35
Postmenopausal osteoporosis with pathological fracture	Iron	0.90	0.35	0.43
Osteoporosis	Iron	0.71	0.99	0.99

Causal relationship between genetically predicted trace elements and postmenopausal osteoporosis with pathological fractures

The MR analysis identified a noteworthy causal link between serum iron concentrations and postmenopausal osteoporosis with pathological fractures. Specifically, the IVW approach revealed a strong link between increased iron levels and a heightened likelihood of developing postmenopausal osteoporosis with pathological fractures (OR = 2.714, 95% CI: 1.156–6.368, P = 0.022), as shown in Fig. 1B. This finding indicates that higher iron levels may significantly increase the likelihood of developing osteoporosis with pathological fractures among postmenopausal women. Additionally, the analysis of 14 other micronutrients showed no significant genetic causality for postmenopausal osteoporosis with pathological fractures, apart from iron (Fig. 2B). There was no indication of heterogeneity or pleiotropy detected in any of the MR analyses concerning other micronutrients and postmenopausal osteoporosis with pathological fractures (see Table 1). Furthermore, the leave-one-out analysis identified no abnormal SNPs (see Supplementary Figure), validating the strength of the observed causal link between iron and postmenopausal osteoporosis with pathological fractures.

Causal relationship between genetically predicted trace elements and osteoporosis

As shown in Fig. 2C, no causal relationship was identified between trace elements and general osteoporosis. Although the analysis did not demonstrate a notable correlation between iron and osteoporosis (OR = 1.238, 95% CI:0.856–1.791, P = 0.256), iron emerged as a substantial risk factor in cases where osteoporosis is accompanied by pathological fractures. This indicates that while iron could be an important contributor to the progression of osteoporosis with pathological fractures, it might not be a key factor in osteoporosis by itself. Furthermore, no significant correlations were observed for other micronutrients (see Fig. 2C).

Discussion

The present study systematically examined the causal links among 15 trace elements and their influence on the likelihood of osteoporosis and pathological fractures using the MR method. The findings revealed a significant association between increased iron levels and the likelihood of developing postmenopausal osteoporosis with pathological fractures, as well as osteoporosis with accompanying pathological fractures. Nonetheless, the analysis did not identify any significant causal link between iron levels and the likelihood of osteoporosis alone. These results highlight the potential impact of iron on osteoporosis-related complications, indicating that iron could be a critical factor in fracture susceptibility, beyond its effect on bone density.

The metabolism of iron and its impact on bone health are both complex and multifaceted. Studies have shown that iron significantly influences bone metabolism through oxidative stress, inflammatory responses, and the regulation of bone cell function [23]. In postmenopausal women, the decline in estrogen levels is often accompanied by iron overload, which may further amplify iron’s negative effects on bone metabolism [24]. Estrogen regulates both iron metabolism and bone cell activity, as well as the overall process of bone remodeling [25]. Abnormal iron metabolism, particularly iron overload, may be a key factor contributing to the heightened likelihood of pathological fractures among postmenopausal women with osteoporosis [26]. As a result, elevated iron levels due to estrogen deficiency may accelerate fracture risk by promoting bone resorption and inhibiting bone formation. These mechanisms could account for the strong link identified between iron levels and the risk of pathological fractures in this study, despite the lack of correlation with osteoporosis itself. Based on these findings, closer monitoring and regulation of serum iron levels may be required in the clinical management of postmenopausal osteoporosis, particularly when assessing fracture risk, as serum iron may serve as a key biomarker.

Our MR analysis also yielded null results for vitamin D, which may appear unexpected given the well-recognized role of vitamin D deficiency in osteoporosis. However, it is important to note that MR estimates reflect the average lifelong effect of genetically predicted vitamin D levels across the general population, rather than targeted supplementation in deficient individuals. If the skeletal benefits of vitamin D are concentrated in deficient subgroups or rely on calcium co-supplementation, then population-wide MR estimates may attenuate toward the null. This helps reconcile our findings with randomized clinical trials, which often show little benefit in replete community-dwelling adults, but stronger protective effects in deficient or institutionalized populations, particularly when calcium is co-administered [27–29]. Thus, our null MR estimate does not contradict current clinical practice of correcting vitamin D deficiency, but rather indicates limited evidence for a broad, population-wide causal effect.

Although earlier research has explored the effects of various trace elements, including zinc, selenium, and copper, on bone health [30–33], the majority of these investigations were observational in nature, which makes establishing causality challenging. Applying the MR method, this research is the first to definitively establish a causal link between iron and osteoporosis with pathological fractures, offering more robust evidence. The discrepancies with previous studies may stem from differences in methodology. Conventional observational research is frequently affected by confounding variables, whereas this study utilized genetic variations as IVs, thereby mitigating this limitation. Furthermore, the absence of significant associations for other elements like zinc, selenium, and copper could be attributed to variations in study design, participant demographics, or statistical methods, suggesting that their effects might be more complex or limited to specific subpopulations.

It is also important to clarify how iron was modeled in this study. Our MR approach estimated the average linear effect of higher genetically proxied iron levels across the population distribution. Because the analysis was based on summary-level data, it is not possible to divide participants into “iron deficiency” and “iron overload” groups and estimate separate causal effects. Therefore, our findings should be interpreted as a population-level effect, without excluding potential risks associated with very low or very high iron levels.

Although this study offers strong causal inferences through the MR method, there are certain limitations to consider. First, the data primarily originated from individuals of European descent, and additional research is required to assess whether these findings are applicable to other ethnicities or populations. Second, although several sensitivity analyses were performed to minimize the effects of pleiotropy and population stratification, factors such as IV selection, genotyping errors, and phenotypic misclassification could still affect the accuracy of the findings. Third, as noted above, this MR design cannot stratify participants into clinical categories such as iron deficiency versus iron overload, and thus cannot estimate subgroup-specific effects. Additionally, while no significant causal link between iron and osteoporosis was found, this does not rule out iron’s possible involvement in osteoporosis progression, which warrants further mechanistic investigation.

Future studies should further investigate the specific mechanisms linking abnormal iron metabolism to osteoporosis and its complications, with a particular focus on how interactions between iron and other trace elements influence bone health. Moreover, prospective clinical trials are necessary to evaluate whether decreasing iron levels can effectively mitigate the likelihood of pathological fractures among postmenopausal women with osteoporosis. Such research will deepen our understanding of iron’s contribution to osteoporosis management and serve as a basis for developing new strategies for prevention and therapy.

Conclusion

In summary, this study underscores the critical role of iron in osteoporosis and pathological fracture risk, emphasizing the importance of iron metabolism in managing bone health. These findings offer new perspectives on the role of iron in osteoporosis and its complications, while also suggesting directions for future research and clinical practice. Effective monitoring and regulation of iron levels may enhance bone health in postmenopausal women and help reduce the incidence of pathological fractures.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Shengwu Yang: Writing – review & editing, Visualization, Supervision, Methodology, Formal analysis. Zhengxiang Huang: Writing – review & editing, Visualization, Project administration, Data curation, Validation, Software. Tao Cui: Methodology, Validation, Software, Data curation, Investigation. Yongqian Fan: Visualization, Validation, Resources, Formal analysis, Funding acquisition. Xingcai Deng: Writing – original draft, Visualization, Validation, Software, Resources. Shangjin Lin: Writing – original draft, Methodology, Visualization, Validation, Software, Investigation.

Funding

This study was supported by the Clinical Research Center for Geriatric Fractures of Huadong Hospital (No. LCZX2208).

Data availability

The datasets used in this study were sourced from publicly available data through the MR-Base platform and the Finnish Biobank.

Declarations

Conflict of interest

The authors confirm the absence of any financial conflicts or personal affiliations that might have affected the work presented in this study.

Ethics statement

Not applicable.