Osteoporosis (OP), recognized as the most prevalent systemic bone disorder, is primarily identified by reduced bone mineral density (BMD) and an increased incidence of brittle fractures resulting from the weakening of bone microarchitecture (Song et al., 2022). This deterioration can escalate the risk of severe disability or mortality, especially among older adults (Borgström et al., 2020). The universally accepted diagnostic approach for OP involves measuring BMD utilizing dual-energy x-ray absorptiometry, a technique that assesses BMD at specific skeletal sites including the femoral neck, lumbar spine, and forearm, which are particularly susceptible to changes in BMD and overall bone health (Kanis, 1997; Kanis, 2002). Additionally, in more recent times, the heel site has emerged as an additional location for estimating OP, further expanding the scope of diagnostic evaluation (Bai et al., 2020). Moreover, measuring total body BMD (TB-BMD) is also considered an equally valid and unbiased approach for evaluating overall BMD in the context of OP diagnosis. Fractures, another significant manifestation of OP, occur most commonly in the leg, arm, heel, and spine, further highlighting the widespread impact of the disease on skeletal integrity (Liu et al., 2019; Black et al., 2020; Leder et al., 2020). According to the most recent findings published by the International Osteoporosis Foundation, it is estimated that globally, one in three women and one in five men who are above the age of 50 are likely to suffer from OP (Cui et al., 2018; Liu et al., 2019; Black et al., 2020; Leder et al., 2020). This statistic not only reflects the extensive prevalence of this disease but also emphasizes the profound effect OP has on the quality of life of patients, as well as the substantial challenge it poses to public health infrastructures and national economies.

Oxidative stress occurs when there is an imbalance between oxidative and antioxidative processes in the body, a condition that significantly contributes to aging and various diseases (Mladenov et al., 2023). This imbalance is largely regulated by reactive oxygen species (ROS) and various antioxidant enzymes. Increasingly, studies are showing that bone health is largely influenced by the regulation of redox balance, with the control of ROS production in bone cells emerging as a critical strategy for preventing bone deterioration (Kaur et al., 2021). The activation of osteoclasts by receptor activator of NF-κB ligand (RANKL) through binding to RANK receptors on the surface of osteoclasts induces the formation of multinucleated giant cells, subsequently facilitating bone resorption and remodeling (Liang et al., 2023). The interaction between RANKL and ROS plays a key role in osteoclast activation and bone remodeling, forming a positive feedback loop to influence bone metabolic homeostasis (Tanaka et al., 2023; Xu et al., 2023). While the superoxide produced by osteoclasts plays a direct role in bone degradation, osteoblasts synthesize antioxidants like glutathione peroxidase (GPX) to combat ROS (Yuan et al., 2022). As a result, in scenarios where there is an imbalance between ROS production and antioxidant defense, it can adversely impact bone metabolism, potentially leading to OP (López-Armada et al., 2022). In addition, recent studies have shown that oxidative stress is involved in the pathogenesis of OP by uncoupling osteoclast and osteoblast functions (Baek et al., 2010; Manolagas, 2018). Using pathways of bone metabolism, such uncoupling can be achieved by promoting osteoclastogenesis, inducing apoptosis of osteoblasts and osteocytes, and inactivating osteoblasts (Kimball et al., 2021). Dietary antioxidants are commonly recognized as the most readily available source of antioxidant defense. Therefore, antioxidants, especially those from dietary sources, have an important role to play in the prevention of OP. There has been some research examining the relationship between certain dietary antioxidants and bone health in humans in the past (Sugiura et al., 2011; Talaulikar et al., 2012). Consuming more fruits, vegetables, and cereals reduces fracture risk, according to Warensjö et al. in a study of 56,736 women (Warensjö Lemming et al., 2017). Similar conclusions have been reached in other studies: natural antioxidant-rich plant foods can improve BMD (Platt et al., 2011; Wallace, 2017). Although diet-based supplements are easily accessible and affordable, it remains critical to identify specific circulating antioxidants that are causally associated with reduced risk of OP.

Randomized controlled trials (RCTs) are widely regarded as the most reliable method for determining causal relationships (Tew et al., 2023). However, RCTs may not always be feasible due to their potential high cost and ethical concerns. Additionally, the dosage and duration of antioxidant supplementations may limit the conclusions drawn from RCTs. As we know, it seems unrealistic to expect a few years of antioxidant treatment to reverse the effects of decades of oxidative stress. To address these limitations, an alternative statistical approach to explore causality in exposure-outcome relationships involves the utilization of instrumental variables. Specifically, Mendelian randomization (MR) is a powerful analytical method that uses genetic variation to study the causal impacts of risk factors on traits (Lawlor et al., 2008; Evans and Davey Smith, 2015; Burgess et al., 2019). Previous research using MR analysis has suggested possible causal links between levels of circulating antioxidants and a range of health conditions (Ni et al., 2023; Tang et al., 2023; Zhu et al., 2023; Chen et al., 2024; Zou et al., 2024). The primary objective of this study was to investigate the causal connection between diet-derived circulating antioxidants and the risk of OP.

With the significant increase in the incidence of OP over the past few decades, it has become increasingly important to recognize modifiable risk factors, particularly dietary factors, as a promising strategy to curb the occurrence and progression of these diseases. Building on this understanding, in this study, we used genetic variations in the concentration of antioxidants as instrumental variables for exposure. Expanding the scope of our research, our outcome data encompasses not just OP and OPF, but also BMD across different body areas and age groups, along with fractures in various body locations. In terms of specific findings, our results demonstrate a notable positive association between absolute levels of retinol and eBMD. Interestingly, we also found that this significant causal relationship was related to gender. Additionally, we identified weak causal relationships of absolute ascorbate in TB-BMD (age over 60), β-carotene in LS-BMD, α-tocopherol in TB-BMD (age 45–60), relative ascorbate in TB-BMD and TB-BMD (age 0–15), and relative retinol in FA-BMD and SF. However, there was no evidence to support a causal relationship between other circulating antioxidants and the outcomes.

The relationships between antioxidants and the risk of OP have been evaluated in a number of observational studies. Yet, the question of whether antioxidants can mitigate the risk of OP remains debated. While some epidemiological research indicates that high levels of vitamin A (retinol) intake or serum vitamin A are linked to a negative impact on bone mass, this is not a universally accepted finding. Other studies contradict these findings, with some researchers suggesting that vitamin A may actually enhance bone health. For instance, an epidemiological study from Norway demonstrated that elevated serum vitamin A levels did not lead to a higher incidence of hip fractures (Holvik et al., 2015). In contrast, a meta-analysis by Wu et al. (Wu et al., 2014), encompassing eight studies on vitamin A intake, revealed a correlation between high vitamin A consumption and an increased risk of hip fractures. Similarly, a meta-analysis by Zhang et al. (Zhang et al., 2017a), involving 13 studies, showed that increased vitamin A intake was linked to a lower risk of fractures at various sites, but with a notable increase in the risk of hip fractures. However, another piece of research reported an increase in BMD levels with higher vitamin A intake, leading to conflicting conclusions (Barker et al., 2005). In the context of our MR study, absolute retinol levels were significantly positively correlated with eBMD in men. In addition, there were suggestive associations found between metabolic levels of retinol and increased FA-BMD (OR = 1.041, p = 0.042) and decreased risk of SF (OR = 0.999, p = 0.049). Those with OP may benefit from a diet rich in antioxidants, especially vitamin A, and an antioxidant-preserving lifestyle. Vitamin A exerts its effects through the metabolite all-trans-retinoic acid (ATRA), a potent transcriptional regulator that modulates the expression of genes crucial for mediating the indirect antioxidant properties of vitamin A. Consequently, vitamin A can activate the NRF2/KEAP1 or NF-κB signaling pathways to mitigate oxidative stress and decrease the risk of OP (Dinkova-Kostova and Talalay, 2008; Abraham et al., 2019). However, larger clinical trials are needed to increase the credibility of clinical recommendations, especially given the presence of weak causality and sex-specific effects of retinol.

As a key antioxidant, Vitamin C (ascorbate) plays a crucial role in removing ROS, thereby reducing oxidative stress (Kim and Lee, 2016). Vitamin C has been shown to stimulate the formation of osteoclasts and osteoblasts in vitro, although high doses of the vitamin can also exhibit cytotoxic effects on both cell types (Chin and Ima-Nirwana, 2018). New et al. discovered a non-linear relationship between vitamin C consumption and BMD in a cohort of 994 healthy premenopausal women aged 45–49 participating in the Osteoporosis Screening Program in Aberdeen (New et al., 1997). According to a report conducted by Kim and Lee (Kim and Lee, 2016), an elevated intake of vitamin C may decrease the likelihood of OP among Korean adults over 50 who have low levels of physical activity, but surprisingly, this benefit does not extend to those with higher levels of physical activity. Moreover, increased dietary intake of vitamin C has been linked to a decreased risk of hip fractures and OP, as well as to enhanced BMD, notably in the femoral neck and lumbar spine areas (Malmir et al., 2018). Simon and colleagues discovered a significant nonlinear relationship between serum vitamin C levels and hip BMD in men (Simon and Hudes, 2001). Specifically, hip BMD initially increased with rising serum vitamin C levels, but subsequently decreased as serum vitamin C levels continued to rise. Interestingly, while primarily known for its antioxidant effects, vitamin C can display prooxidant characteristics at high doses, as indicated by a previous study where a daily supplementation of 500 mg of vitamin C in individuals aged 17 to 49 for 6 weeks led to oxidative DNA damage (Podmore et al., 1998), a factor potentially relevant to OP (Finck et al., 2014). The current study identified three suggestive causal associations between absolute ascorbate concentrations and reduced TB-BMD (age over 60) and metabolic concentrations of ascorbate with reduced TB-BMD and TB-BMD (age 0–15). However, higher-quality trials are necessary to determine if Vitamin C significantly reduces the risk of OP.

Vitamin E, known for its antioxidant properties, is divided into two primary subgroups, namely, tocopherols and tocotrienols. Each of these subgroups contains four unique analogs: alpha, beta, gamma, and delta (Zhang et al., 2017b). Significantly, α-tocopherol is one of the homologs within the vitamin E category. Research suggests that vitamin E may benefit bone health due to its anti-inflammatory properties, yet this effect has not been conclusively demonstrated in human subjects (Reid and New, 1997). To further explore this, Shi et al. (Shi et al., 2016) investigated the correlation between vitamin E levels and BMD in middle-aged and elderly adults (40–75 years) in Guangzhou, China. Their findings indicated a positive association between higher intake and serum levels of vitamin E and increased BMD in Chinese women, but this association was not mirrored in Chinese men. These results are similar to our findings, as our MR analysis revealed a suggestive association between the metabolic level of α-tocopherol and increased TB-BMD in individuals aged 45–60. In contrast, a negative correlation has been observed between serum α-tocopherol levels and femoral neck bone mineral density among the elderly population in the US (both men and women over 50), indicating a potentially detrimental impact of α-tocopherol on bone health (Zhang et al., 2017b). γ-tocopherol, the second most prevalent isoform of vitamin E found in dietary sources, accounts for 20%–25% of total vitamin E intake and exhibits the ability to scavenge both reactive nitrogen species (RNS) and ROS, distinguishing it from α-tocopherol (Abraham et al., 2019). While a clinical trial indicated that supplementation with γ-tocopherol reduces systemic oxidative stress, findings from cross-sectional studies have been inconclusive. Yang et al. (Yang et al., 2016) did not observe statistically significant associations between γ-tocopherol and BMD in a study involving over 5,000 older women. In contrast, Ilesanmi-Oyelere et al. (Ilesanmi-Oyelere et al., 2019) reported that increased vitamin E intake was linked to lower BMD levels. The current body of evidence regarding the impact of γ-tocopherol on OP remains limited, highlighting the need for more rigorous trials to elucidate its effects.

Lycopene is an acyclic isomer derived from β-carotene (Agarwal and Rao, 2000). A long-term study over 17 years found that increased lycopene consumption was associated with a lower incidence of hip and nonvertebral fractures, highlighting its potential role in bone health (Sahni et al., 2009). This elevation in lycopene potentially lowers oxidative stress and bone resorption markers, thereby suggesting a possible reduction in the risk of OP. Nevertheless, a cross-sectional study involving 4,820 individuals in the United States revealed no significant association between lycopene intake and OP (Kan et al., 2021). There is weak evidence that β-carotene levels are associated with increased LS-BMD in the current study. However, the results of a previous MR performed by li et al. (Li et al., 2023) showed that serum β-carotene levels significantly reduced OP risk. For the instrumental variables of β-carotene, we used more strict screening criteria (p < 5 × 10⁻⁸ and r² < 0.001), which helped us to make more reliable conclusions. Furthermore, in our study, we also analyzed subgroups of TB-BMD according to age. In conclusion, the weak causal relationship between β-carotene levels and increased LS-BMD provides some insight into the reduction of OP risk by dietary antioxidants.

The study we conducted possesses a range of strengths. Primarily, the MR approach, utilizing two separate samples, reduces the potential risks that participants may encounter in clinical trials. Additionally, the application of distinct sets of instrumental variables for both absolute circulating antioxidants and their metabolites increases the informative value of the MR study outcomes. However, our study does have its limitations. One of the primary limitations is the constrained number of SNPs for antioxidant instrumental variables that were sourced from published GWAS data. In other MR analysis, absolute retinol and β-carotene were also unable to validate the robustness of the results using complementary MR analysis methods due to limitations in the number of SNPs (Mondul et al., 2011; Hendrickson et al., 2012; Zhao et al., 2023). It should be noted, however, that these SNPs are situated within essential genes pertinent to antioxidant metabolism, according to the PhenoScanner database (Kamat et al., 2019). In future studies, identifying more relevant loci through broader GWAS will be critical to strengthen the effectiveness of the instrumental variables. Another point is that antioxidant concentrations vary greatly with different detection methods, underscoring the need for standardizing these techniques in future studies. Finally, postmenopausal status was not included in the subgroup analyses due to data limitations, which may have weakened the reliability of our conclusions. Broader GWAS data are needed for more detailed subgroup analyses in the future.

In conclusion, there is a positive causal relationship between absolute retinol levels and eBMD. Our study, however, was unable to establish evidence of a causal relationship between retinol and BMD at other body sites or fractures. The implications of these results should be taken into account in future studies and in the creation of public health policies and OP prevention tactics.