In 1993, the World Health Organization (WHO) classified osteoporosis (OP) as a systemic skeletal disease. OP is characterized by low bone density, deterioration of bone tissue microstructure, increased bone fragility, and susceptibility to fracture (Genant et al., 1999). As a severe, chronic, progressive, and asymptomatic metabolic bone disease, OP is a major public health concern worldwide (LeBoff et al., 2022; Moore et al., 2023). The statistical evidence indicates that over 200 million people worldwide suffer from OP. Women over the age of 50 are particularly affected by this condition, with almost one-third being affected. Additionally, about one-fifth of men experience fractures related to osteoporosis during their lifetime (Sozen et al., 2017; Gregson et al., 2022).

In recent years, OP has become a pressing emerging health issue that requires prompt attention. It imposes considerable economic and caregiving burdens on societal healthcare resources and individual households (Mottaghi and Nasri, 2021). The pathophysiology of OP involves a dysregulation in bone metabolism, characterized by an excessive rate of bone resorption over bone formation. This, in turn, results in reduced bone mass and declining bone mineral density (BMD) (Liang et al., 2022). Historically, estrogen deficiency has been identified as a key trigger for OP. However, recent epidemiological and experimental animal studies have highlighted the significant role of oxidative stress in the OP disease process. This underscores the close relationship between bone density and the oxidative state within the organism (Manolagas, 2010; Zhang et al., 2023).

Normally, intracellular reactive oxygen species (ROS) are effectively scavenged by the body’s antioxidant systems to maintain cellular redox homeostasis (de Almeida et al., 2022). ROS act as dual-effect molecules, serving as crucial secondary messengers involved in regulating essential life processes such as apoptosis, differentiation, and proliferation at appropriate concentrations. This has significant implications for the dynamic balance between osteoblasts and osteoclasts (Zhu et al., 2022; Endale et al., 2023). However, studies have shown that aging, inflammatory responses and estrogen deficiency can increase the production of ROS. When the rate of ROS generation exceeds the body’s intrinsic antioxidant clearance capacity, it results in a state of oxidative stress (Mohamad et al., 2020; Zhu et al., 2022; Iantomasi et al., 2023) Excessive accumulation of ROS can suppress osteoblast differentiation and proliferation, while promoting osteoclast differentiation. This can cause structural disruption of bone tissue and subsequent loss of bone mass (Kimball et al., 2021). Additionally, it is noteworthy that the decline in estrogen levels often coincides with elevated levels of pro-inflammatory factors. This further exacerbates ROS production and consequently elevates the risk of OP (Zhou et al., 2021).

Research has shown that supplementing with antioxidants is a key approach to reducing the impact of free radicals, preventing oxidative damage, and countering the negative effects of oxidative stress on bone health (Damani et al., 2022; Davan et al., 2023). Figure 1 illustrates the oxidative stress and antioxidant mechanisms involved in the development of OP. Dietary exogenous antioxidants mainly comprise carotenoids, retinol, vitamin C, and vitamin E. On the other hand, endogenous constituents primarily consist of albumin, bilirubin, and key antioxidant enzymes such as Catalase (CAT), Superoxide Dismutase (SOD), Glutathione Peroxidase (GPX), and Glutathione S-transferase (GST) (Behera and Malik, 1989; Jomová et al., 2023). Research has shown that osteoporotic rats have notably depleted serum glutathione levels (Yalin et al., 2012; Ameen et al., 2020). Furthermore, increasing the GSH/GSSG ratio can help mitigate ROS-induced oxidative injury to osteoblasts by activating the PI3K/Akt-Nrf2 signaling cascade (Casati et al., 2019). However, when examining the relationship between antioxidant biomarkers and OP, conflicting results have been reported. One systematic appraisal found lower SOD activity in postmenopausal women with OP (Zhou et al., 2016), while a case-control investigation showed significantly higher plasma SOD activity in osteoporotic cases compared to controls (Malekian et al., 2023). The effects of dietary-exogenous antioxidants and their metabolites on OP also remain inconsistent. Numerous studies have demonstrated that exogenous antioxidants, including carotenoids, vitamin C, and vitamin E, can reduce the risk of OP (Chavan et al., 2007; Yang et al., 2008; Xu et al., 2016). In studies investigating exogenous antioxidant metabolites, Maggio observed plasma antioxidant levels in patients with osteoporosis, noting consistently lower mean plasma levels of vitamin C, vitamin E, and vitamin A in osteoporosis patients compared to the control group (Maggio et al., 2003). Additionally, Zhang et al. observed a U-shaped correlation between plasma retinol concentration and BMD, suggesting that an appropriate plasma retinol concentration may contribute to improving bone mineral status (Zhang et al., 2022). However, some studies have yielded conflicting conclusions. For instance, Barker et al.’s study failed to find a significant correlation between serum β-carotene levels and the risk of hip fractures in elderly women, indicating a lack of association between β-carotene and reduced OP or fracture risk (Barker et al., 2005). Furthermore, a substudy of the Women’s Health Initiative (WHI) examined the associations between serum levels of α-tocopherol and γ-tocopherol, dietary and total vitamin E intake, and BMD in postmenopausal women. It found no significant correlation between dietary or total vitamin E intake, serum tocopherol concentrations, and BMD (Wolf et al., 2005). Consequently, the scientific community has not yet reached a consensus on the precise role of antioxidants in the prevention and treatment of osteoporosis.

Observational studies often face complexities in discerning a direct causal relationship between antioxidants and OP due to confounding factors, potential reverse causality, limitations imposed by sample size, and measurement errors (Sekula et al., 2016). To overcome these issues, Mendelian Randomization (MR) research emerges as an ideal methodology due to its unique theoretical framework and practical advantages. This technique uses naturally occurring genetic Instrumental Variables (IVs) to indirectly infer causality between genetically determined biological exposures and health outcomes. In this study, we will use external or exogenous antioxidants and related metabolites as the exposure variable and OP as the outcome variable. Using a unidirectional two-sample MR design, we investigated whether there is a causal link between exogenous antioxidants and OP incidence. Additionally, employing a bidirectional two-sample MR approach, we assessed the relationship between endogenous antioxidants and the risk of OP. These analyses aim to provide insights into the causal relationships between antioxidants and OP, contributing to the understanding of their roles in the pathogenesis and potential prevention strategies for OP.

In the present research, we utilized a dual-sample MR strategy to conduct a comprehensive examination of the causal relationship between both extrinsic and intrinsic antioxidants and OP. Despite the results indicating that neither the consumption of exogenous antioxidants, which include ascorbate, tocopherol, carotenoids, retinol, and lycopene, nor augmented activities within the intrinsic antioxidant defense system, composed of enzymes such as CAT, SOD, GPX, GST, serum albumin, and total bilirubin, showed a statistically discernible causal link with OP risk, our study disclosed a noteworthy observation suggesting that OP itself might act as a causative agent contributing to reduced total albumin concentrations. Additionally, exploratory analyses of data from the FinnGen and MRC-IEU repositories suggested a potential beneficial role of genetically high γ-tocopherol levels and increased serum total bilirubin in mitigating OP risk. Nonetheless, when extending our analysis through a meta-analysis incorporating data from three separate databases, we did not obtain uniform supportive evidence across these larger datasets.

In the course of biological activities, especially within cellular energy metabolism and the biosynthesis of adenosine triphosphate (ATP), the production of ROS as a byproduct of oxygen reduction is unavoidable. An overabundance of ROS can provoke oxidative stress, a condition that is broadly accepted as a key contributor to the advancement of a broad spectrum of age-related and degenerative diseases, OP being one among them (León-Reyes et al., 2023). A multitude of studies highlights the core position occupied by oxidative stress in the senescence process and the pathophysiology of OP. Antioxidants serve a vital protective function by countering ROS, preserving cellular redox equilibrium, and thereby effectively shielding cells from oxidative injury caused by ROS, thereby offering the potential for suppressing the development of OP (Damani et al., 2022; Davan et al., 2023).

Albumin is the most abundant protein in plasma and serves as an important endogenous antioxidant reservoir in response to chronic oxidative stress (Atukeren et al., 2010). Its antioxidant properties are mainly derived from the exposed thiol moiety at the Cys34 site. During oxidative stress, Cys34 can form disulfide bridges with free cysteine or glutathione, which alters the tertiary structure of human serum albumin (HSA) and effectively eliminates ROS such as hydrogen peroxide, peroxynitrite, superoxide radicals, and hypochlorous acid (Kawakami et al., 2006; Belinskaia et al., 2021). Previous observational research has presented varying perspectives on the relationship between albumin and OP risk. For instance, in a large study of 21,121 individuals with a mean age of 61 years, Afshinnia et al. found that higher serum albumin levels were negatively correlated with OP prevalence, indicating a decline in OP incidence with rising albumin (Afshinnia et al., 2016). In contrast, Rico et al. discovered in a case-control study that serum albumin levels were significantly lower in elderly women with OP compared to healthy counterparts (Rico et al., 2009). However, some studies have questioned the independent correlation between serum albumin and BMD. For instance, a study conducted in Rancho Bernardo reported a slight positive correlation between serum albumin and BMD, which disappeared after adjusting for age (Lunde et al., 1998). On the other hand, a different study conducted on healthy postmenopausal women did not find any significant correlation between serum albumin and BMD (D’Erasmo et al., 1999). The preliminary analysis of the FinnGen database in the current study suggests a potential positive association between albumin and the onset of OP. However, after removing outliers using MR-PRESSO, the causal relationship between albumin and OP no longer had statistical significance. The reverse MR analysis revealed a new causal pathway, indicating that OP may reduce serum albumin levels. It is hypothesized that this reduction in albumin levels may be linked to oxidative stress, poor nutritional status, and chronic inflammation during the course of OP (Vaughan et al., 2012; Cauley et al., 2016; Tugba et al., 2019; Xu et al., 2022). Extensive research has highlighted the significant role of oxidative stress in the pathogenesis of osteoporosis (Manolagas, 2010; Zhang et al., 2023). Oxidative stress has been identified as a significant contributor to hypoalbuminemia (Chojkier, 2005), with the potential to induce molecular modifications in human serum albumin, including carbonylation and the formation of advanced oxidation protein products and advanced glycation end-products (Michelis et al., 2010). Studies have demonstrated that the cobalt binding capacity of albumin is diminished in individuals diagnosed with osteoporosis, reflecting oxidative damage and a decline in albumin functionality (Tugba et al., 2019). Albumin, an essential protein in the bloodstream, is significantly correlated with muscle mass and strength in elderly individuals with osteoporosis, suggesting a potential nutritional link (Xu et al., 2022). Poor nutritional status is prevalent among the elderly, particularly those diagnosed with osteoporosis, which may contribute to hypoalbuminemia (Xu et al., 2022). Additionally, the synthesis of albumin may be inhibited due to the body’s response to inflammation (Vaughan et al., 2012; Cauley et al., 2016). Overall, our study provides new insights for clinical practice by establishing a causal link between serum albumin and OP. Monitoring changes in serum albumin levels could serve as a biomarker for evaluating disease progression and treatment responsiveness in OP (Fujii et al., 2018). In the future clinical handling of OP, measuring serum albumin levels could provide valuable information for early disease risk detection, tracking disease progression, and devising personalized therapeutic regimens, in addition to standard bone density measurements. However, this proposition requires further substantiation through extensive prospective studies and clinical trials to verify its practicality and effectiveness in OP management.

Bilirubin was previously considered a cytotoxic byproduct. Despite this, bilirubin has been shown to hinder osteoblast function, including growth, differentiation, and mineralization (Turkeli et al., 2008). Recent research, however, has shown that it has antioxidant potential, sometimes even surpassing certain vitamin E analogs in combating lipid peroxidation (Wu et al., 1994; Žiberna et al., 2021). A cross-sectional study of 918 postmenopausal women found a positive correlation between elevated total bilirubin levels and improved BMD as well as reduced risk of osteoporosis (Bian et al., 2013). The impact of α-tocopherol and γ-tocopherol, which are key forms of vitamin E found in the human body and diet and possess antioxidant and anti-inflammatory qualities, on skeletal metabolism remains unclear (Macdonald et al., 2004; Hamidi et al., 2012). MacDonald et al.’s research suggests that excess vitamin E intake may negatively correlate with BMD (Macdonald et al., 2004). However, another study suggests that γ-tocopherol may promote bone formation in postmenopausal women by modulating bone turnover processes (Hamidi et al., 2012). Our study, which utilized MR analyses on FinnGen and MRC-IEU datasets, has preliminarily discovered a potentially significant association between genetically higher levels of γ-tocopherol and serum total bilirubin levels, and a reduced risk of OP. This suggests a positive role for these compounds in bone health, potentially aiding in the prevention of OP due to their potent antioxidant and anti-inflammatory characteristics. Although our FinnGen and MRC-IEU analyses revealed an association, attempts to generalize this finding across multiple independent databases through a meta-analysis did not yield strong and consistent statistical support. This inconclusiveness could result from several factors, including population-specific genetic heterogeneity, sample size constraints, potential confounders, and complex interplays between environmental factors and study outcomes.

In recent MR investigations, researchers have not been able to establish a significant causal link between circulating bilirubin concentrations and either extraskeletal bone mineral density (eBMD) or the incidence of fractures. This suggests that bilirubin may not be a direct, pivotal factor affecting OP susceptibility (Zhao et al., 2021). Li’s study suggests that antioxidants β-carotene and γ-tocopherol may potentially mitigate OP risk. However, the robustness of these claims requires more rigorous examination. Li’s study used a meta-analytic approach to enhance the persuasiveness of the conclusions. However, the credibility and broad applicability of these outcomes were limited by reliance on just two discrete datasets. Of particular relevance, the extensive osteoporosis-centric research endeavor, the GEFOS database, did not reveal any statistically discernible disparities in OP risk between β-carotene and γ-tocopherol levels (Li et al., 2023).

In this research, we leveraged the extensive capabilities of large-scale MR techniques, utilizing aggregated data from GWAS involving 1,178,298 cases and controls, to pioneer a systematic inquiry into the potential causal links between exogenous antioxidants—those obtained through dietary supplements or pharmaceutical ingestion—and endogenous antioxidants with the risk of OP. A prime virtue of our methodology is the employment of genetic proxies instead of conventional interventions, thereby mitigating potential risks associated with direct antioxidant administration and mitigating issues of reverse causality between exposures and outcomes, thus enhancing the precision of causal reasoning. Additionally, the GWAS subjects in this investigation were predominantly of European descent, and we applied meticulous genomic control procedures uniformly across all studies. This greatly diminishes biases resulting from population stratification and genomic inflation effects, thereby strengthening the internal validity of our findings. In pursuit of further validation and consistency, we conducted a meta-analysis of data from three major databases. Our statistical heterogeneity assessments demonstrated that the merged findings across these three databases displayed a relatively low degree of heterogeneity, which partly attests to the robustness of our study’s conclusions.

In this study, we employ MR to explore the causal relationship between antioxidants and the risk of OP. However, it is crucial to acknowledge several limitations. Firstly, one significant limitation not addressed directly in our analysis is the intake of exogenous antioxidants, which was not measured through dietary consumption frequencies or 24-h dietary recalls. This omission might have influenced the observed associations, as dietary antioxidant intake could modulate the effects of genetically predicted antioxidant levels. Future studies should incorporate detailed dietary assessments to better control for this potential confounder. Secondly, the selection of single nucleotide polymorphisms (SNPs) in our study involved different criteria, including various p-value thresholds (e.g., p < 5 × 1 0 − 8 , p < 5 × 1 0 − 6 , p < 1 × 1 0 − 5 ) and different LD thresholds (e.g., LD < 0.02 and LD < 0.001). The variation in the selection of SNPs was due to the limited availability of reliable genetic instruments for the antioxidants under investigation. As larger sample GWAS for exposure and outcomes emerge, this limitation may diminish. Thirdly, the current study exclusively utilized GWAS data for OP and did not include BMD measurements from the forearm, hip, or whole body. This approach may restrict the breadth of our osteoporosis risk assessment. Given that BMD is a pivotal diagnostic marker for osteoporosis, and measurements from different anatomical sites can offer varied insights into bone health, the omission of comprehensive BMD data could impact the accuracy of our findings. Future research should incorporate BMD measurements from multiple sites to enhance the thoroughness of risk assessment and more precisely determine which anatomical site’s BMD most effectively predicts OP risk. Fourthly, the relationship between osteoporosis and reduced serum albumin concentrations remains complex and not fully elucidated in the current literature. While some studies suggest a potential link between the two, others have not found direct evidence supporting osteoporosis as a causative factor in lowering serum albumin levels. Consequently, further investigations are warranted to clarify the causal relationship and clinical significance between osteoporosis and reduced serum albumin levels. Finally, the generalizability of our findings to other racial and ethnic groups may be constrained, as the GWAS data primarily comprised individuals of European descent.

This MR study used a large dataset and rigorous genetic analysis to provide substantial evidence for exploring the causal relationship between antioxidants and OP risk in Europeans. Although the study could not definitively confirm a statistically significant causal link between exogenous and endogenous antioxidants and OP risk, it reveals that osteoporosis may contribute to reduced serum albumin levels. This suggests that a decrease in serum albumin levels may be strongly associated with the progression of OP. Furthermore, this study indicates that monitoring dynamic fluctuations in serum albumin levels can serve as an effective biomarker for evaluating treatment efficacy and tracking the progression of OP in clinical practice.