Can Dietary Intake of Vitamin C-Oriented Foods Reduce the Risk of Osteoporosis, Fracture, and BMD Loss? Systematic Review With Meta-Analyses of Recent Studies

Background: Several epidemiological studies have been performed to evaluate the association of dietary intake of vitamin C-oriented foods (DIVCF) with risk of fracture and bone mineral density (BMD) loss, but the results remain controversial. Therefore, we conducted a systematic meta-analysis to assess this correlation. Methods: We searched EmBase, PubMed, Web of Science, and the Chinese database CNKI for relevant articles published up to August 2019. Pooled relative risks (RRs) with 95% confidence intervals (CIs) were calculated using the random- or fixed-effects model. Discrepancies were resolved by consultation with a third expert. Results: A total of 13 eligible articles (including 17 studies) with 19,484 subjects were identified for the present meta-analysis. The pooled RR of hip fracture for the highest vs. lowest category was 0.66 (95% CI, 0.47–0.94) for DIVCF, i.e., people with a greater frequency of Vitamin C uptake had a 34% (95% CI, 6%−53%) lower prevalence of hip fracture. In subgroup analyses stratified by study design, gender, and age, the negative associations were statistically significant. Furthermore, the statistical analysis of the association between DIVCF and risk of osteoporosis (RR, 0.66; 95% CI, 0.48–0.92), BMD at the lumbar spine (pooled r, 0.15; 95% CI, 0.09–0.23), and BMD at the femoral neck (pooled r, 0.20; 95% CI, 0.11–0.34) showed beneficial effects of DIVCF. Conclusion: Our meta-analysis indicates that DIVCF is negatively associated with the risk of hip fracture, osteoporosis, and BMD loss, suggesting that DIVCF decreases the risk of hip fracture, osteoporosis, and BMD loss.


BACKGROUND
Osteoporosis is a systemic skeletal disease, which is common in postmenopausal women and is characterized by an increased risk of bone fragility and a decrease in bone mass (1). Bone homeostasis requires a balance between bone-forming osteoblasts and bone-resorbing osteoclasts (2). When this balance is impaired, normal bone remodeling cannot keep bone mass stable, leading to osteopenia and osteoporosis (3). Traditionally, dual X-ray absorptiometry (DEXA), which measures bone mineral density (BMD), is routinely used to assess the risk of fracture in osteoporosis patients. According to the World Health Organization, osteoporosis is one of the most common disorders, and 30-50% of all women in the world suffer from fractures due to osteoporosis throughout their lives (4).
Increasing evidence demonstrates that osteoporosis is affected by genetic factors, low body mass (weight), and lifestyle factors, such as coffee intake, alcohol consumption, soft drink intake, and dietary intake of vitamin C-oriented foods (DIVCF) (5). Diet and physical activity, the two key determinants of body weight, might influence osteoporosis directly. Vitamin Coriented foods comprise one of the most important components of the diet and have been reported to exert many potential health benefits, because they are rich in vitamins, fiber, phytochemicals, and minerals (6). Several studies have found that DIVCF significantly lowers the risk of breast cancer, hypertension, metabolic syndrome, type 2 diabetes mellitus, depression, inflammatory bowel disease, and all-cause mortality (7,8). Antioxidants and anti-inflammatory components from such diets are hypothesized to play a vital role in the protective effects against osteoporosis (9).

METHODS
The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and others were followed in the current analysis (23,24).

Literature Search Strategy
A systemic search was performed for potential articles in four databases, i.e., EmBase, PubMed, Web of Science, and the Chinese database CNKI, published in or before August 2019. We used the search terms "Vitamin C, " "ascorbic acid, " "ascorbate, " and "acid ascorbic" in combination with "osteoporosis" or "fracture." The reference lists from the articles included were manually screened for undetected relevant studies.

Inclusion and Exclusion Criteria
Articles were included in this review if they met the following criteria: (1) observational study published as an original study; (2) exposure of interest was defined as DIVCF, i.e., dietary intake of foods that are rich in Vitamin C; (3) outcome of interest was defined as osteoporosis, fracture, and/or BMD loss in the subjects; and (4) study included estimates of relative risks (RRs), comparing the highest DIVCF category with the lowest, with corresponding 95% confidence intervals (CIs) for the associations or other information that could be used for inference. Furthermore, the following exclusion criteria applied: (1) studies describing animal experiments, review papers, and mechanistic studies; (2) studies lacking specified data of osteoporosis in relation to low bone mass; and (3) articles that only contained an abstract. All identified articles were reviewed independently by two researchers, and discrepancies were discussed with and resolved by a third expert.

Data Extraction and Quality Assessment
Two researchers extracted data independently, following the standardized norms for literature collection. Any discrepancies were resolved by consulting with a third expert. The following information was extracted from the articles: name of the first author, publication year, country where the study was conducted, baseline age, gender, study design, sample size and number of cases, assessment approach of DIVCF, assessment approach of osteoporosis, the measurement location of osteoporosis, and adjusted RR with 95% CI of osteoporosis for DIVCF consumption. The Newcastle-Ottawa Scale (NOS) was used to assess the quality of case-control and cohort studies (25). The Agency for Healthcare Research and Quality guidelines were applied to assess the quality of cross-sectional studies (26).

Statistical Analysis
The risk of osteoporosis, fracture, and BMD loss was estimated, comparing the highest DIVCF category to the lowest. Pooled RRs with 95% CIs were calculated using the random-effects model (REM) if moderate or high heterogeneity (I 2 ≥ 50%) was observed; otherwise, the fixed-effects model (FEM) was used. The I-squared (I 2 ) statistic and chi-square test were adopted to explore the possible heterogeneity of the studies. Potential publication bias was inspected by funnel plots. Subgroup analyses were performed by study type, gender, and age to assess potential influencing factors. All statistical analyses were performed using the statistical software package STATA SE, version 14.1 (Stata Corp, College Station, TX, USA).

Patient and Public Involvement Statement
The study is a systemic literature review that did not involve experimental participants or human subjects, and we did not collect any private data or sensitive information during the review process.

Results of the Literature Search
As shown in Figure 1, a total of 1,968 articles were identified in the database search; 1,349 articles were excluded because of the following reasons: 711 articles were duplications, and 638 articles were excluded due to their nature (e.g., reviews, animal studies, or laboratory studies). Of the remaining 619 articles, 189 did not assess the association between DIVCF and the risk of osteoporosis, fracture, or BMD loss; 157 studies did not report the RR (with 95% CI) for the relationship between DIVCF and osteoporosis risk; 92 articles were excluded for other reasons; and 168 studies were based on a hospital-based sample (not on a population-based sample). As a result, 13 published articles (with 17 studies, i.e., 4 cohort, 11 case-control, and 2 cross-sectional studies) were identified as eligible for the present meta-analysis (10-22) (Figure 1).

Study Characteristics and Quality Assessment
The basic characteristics for DIVCF with risk of osteoporosis are shown in Table 1. In our study quality assessment, the studies received a quality score of ≥5, indicating that the methodological quality of the studies was generally good.

Meta-Analysis of DIVCF and the Risk of Hip Fracture
Eleven studies from seven articles (10)(11)(12)(13)(14)(15)(16) reported the events of hip fracture, and our random-effects meta-analysis showed that there is a significant difference between subjects in the highest category of DIVCF and the lowest. The RR for hip fracture events between the two groups was 0.66 (95% CI, 0.47-0.94) (Figure 2).
These results indicate that the people with a higher consumption of Vitamin C-oriented foods had a 34% (95% CI, 6%−53%) lower risk of hip fracture. Because of the high heterogeneity among the abovementioned studies (I 2 = 79.5%; P = 0.000), we conducted subgroup analyses based on study type (i.e., cohort and casecontrol studies), gender, and age (middle-aged subjects, older subjects) (Figures 3-5).

Subgroup Analyses by Study Design Concerning DIVCF and the Risk of Hip Fracture
Regarding the studies assessing the correlation between DIVCF and risk of hip fracture, two articles (including four case-control studies) (11,16) reported that higher DIVCF could lower the risk of hip fracture, while other case-control studies (13)(14)(15) failed to reflect such a relationship. Also, two articles (including three cohort studies) (10,12) failed to identify such a positive association. The RR for DIVCF and risk of hip fracture was much lower in the case-control studies (RR, 0.58; 95% CI, 0.38-0.89), and no significant results were found in the cohort studies (RR, 0.92; 95% CI, 0.59-1.44) (Figure 3).

Subgroup Analyses by Gender Concerning DIVCF and the Risk of Hip Fracture
Seven articles (with 11 studies) (10-16) reported the effects of gender in the association between DIVCF and risk of hip fracture. A reduced risk of hip fracture was found for the pooled results of three articles (with five studies) ( (Figure 4).

Publication Bias Analysis
Funnel plots were used to evaluate the potential publication bias, with RRs generated from the articles concerning DIVCF and the risk of hip fracture. In the absence of publication bias, most of the formulated points are symmetrically placed around the vertical line given by the pooled RRs. No evidence of publication bias was found in the evaluation of DIVCF and the risk of hip fracture (Supplementary Figure 1).

Meta-Analysis of DIVCF and BMD at the Lumbar Spine and Femoral Neck
The risk of BMD loss at the lumbar spine was considered in four studies (17,18,21,22). In our meta-analysis, we found that higher DIVCF was negatively associated with the risk of BMD loss at the lumbar spine (pooled r, 0.15; 95% CI, 0.09-0.23). Furthermore, four studies (17,(20)(21)(22) reported on DIVCF and the risk of BMD loss at the femoral neck. We found a significant negative association between them (pooled r, 0.20; 95% CI, 0.11-0.34) (Figure 7).

Summary of Evidence
Our meta-analysis included 13

Comparison of the Findings With Other Results in the Literature
In this paper, we carried out a meta-analysis of the possible protective effect of DIVCF against osteoporosis, fracture, and BMD loss. A similar article (27) was published in April 2018. The authors found no significant association between Vitamin C intake and the risk of hip fracture (RR, 0.74; 95% CI, 0.51-1.08). In our analysis, three more articles (with four studies) (16,19,20) were included, providing the chance to access documents that were published in Chinese. In this paper, we also construct a sound basis for further clinical research in this field (28). In addition to assessing the potential correlation between DIVCF and the risk of osteoporosis, we included meta-analyses on the effect of DIVCF on the risk of hip fracture and BMD loss. Finally, our pooled-effects analysis revealed a negative correlation between DIVCF and risk of hip fracture (RR, 0.66; 95% CI, 0.47-0.94); i.e., subjects with a higher intake of Vitamin Coriented foods had a 34% (95% CI, 6%−53%) lower risk of hip fracture (Figure 2). Our results differ from those of Dr. Malmir, published in April 2018 (27). Due to the major heterogeneity among the studies, we performed subgroup analyses based on study type (i.e., cohort and case-control studies), gender, and age (middle-aged subjects, older subjects). Between DIVCF and risk of osteoporosis, a negative correlation was found (RR, 0.66; 95% CI, 0.48-0.92) (Figure 6). Also, we found that DIVCF was negatively associated with the risk of BMD loss at the lumbar spine (pooled r, 0.15; 95% CI, 0.09-0.23) and at the femoral neck (pooled r, 0.20; 95% CI, 0.11-0.34) (Figure 7).
Although a series of studies on Vitamin C and the risk of osteoporosis have been published, including controlled trials, case series, and case reports, to the best of our knowledge, no systematic meta-analyses have been published focusing on the correlation between DIVCF and osteoporosis, hip fracture, and BMD loss.
Several biological mechanisms may underlie the impact of DIVCF on the risk of osteoporosis. First, DIVCF includes not only Vitamin C but also Vitamin A, Vitamin B1, Vitamin B2, Vitamin B3, iron, and potassium intake. The positive effects of these vitamins and minerals on bone health have been well-established (29). Osteoporosis is a systemic skeletal disease, which is common in postmenopausal women and is characterized by an increased risk of bone fragility and a decrease in bone mass. Osteoporosis is often associated with metabolic syndrome (MBS), which involves oxidative stress and insulin resistance (30). Vitamin C-oriented foods are rich in several antioxidants, including Vitamin C, Vitamin K, and other phytochemicals. A high antioxidant consumption could contribute to lowering reactive oxygen species levels and enhancing the antioxidant status in animal models and human subjects (31,32); this may slow down the progression of systemic oxidative damage resulting from overproduction of reactive oxygen species (33)(34)(35). Moreover, many phytochemicals can increase insulin production (36), and therefore potentially aid in the prevention of insulin resistance and hence MBS. Furthermore, DIVCF could protect against MBS by regulating the levels/activity of C-reactive protein and other inflammatory markers (37,38). Higher DIVCF is correlated with reduced plasma concentrations of C-reactive protein (39,40), and the latter might protect against MBS. Finally, people who eat many Vitamin C-oriented foods may consume much dietary fiber and little fat, decreasing the risk of MDS (41,42). All the above mentioned factors are positively associated with BMD and negatively correlated with bone loss, thus reducing the risk of osteoporosis and fractures.

Meaning of the Study for Possible Clinical Use
A positive link was found between DIVCF and a reduced risk of osteoporosis, fracture, and BMD loss. Several observations support the use of this information in clinical settings.  First, our meta-analysis included a large number of subjects, reducing the sampling error to a large extent. Second, large efforts have been made to control the interference of possible confounding factors (such as age, BMI, physical activity level, and smoking status), resulting from the adjusted RR being fully extracted. Third, similar correlations were observed after performing subgroup analyses stratified by study type, gender, and age, confirming that the results are credible and robust.

Limitation of the Review
Nevertheless, our study has several limitations. One of the key points is that only DIVCF was taken into account (and no dietary intake of other foods, such as fruit and vegetable), while many people with high DIVCF may in general eat a healthy diet, so an analysis of other healthy diets may give similar results. Therefore, it is necessary to verify that DIVCF is the active component of the diet. Also, it would be helpful to include intake of other dietary components (such as fruit and vegetable) for further study. Furthermore, Vitamin D plays a large role in the maintenance of BMD; its influence on the development of secondary hyperparathyroidism and osteomalacia, states where BMD is reduced reversibly, has been reported. Only a few studies in our meta-analysis adjusted for Vitamin D intake, though it is of great importance in the majority of nutritional studies (10,12,(18)(19)(20). All the above mentioned factors could result in bias or inaccurate interpretations.
Another important point is that most of the articles we identified were cross-sectional or case-control studies; larger prospective cohort studies are warranted to verify the results. Although we have adjusted for most of the confounders the articles included, other confounders may still have existed. Moreover, the confounders that were adjusted for were different in each research; this may certainly have affected the results. Moreover, the approach of osteoporosis assessment differed between articles, which could influence the observed correlation. Because of the limited number of included articles, the effects of an inconsistent assessment approach on the strength of the correlation could not be further explored. Finally, due to the limited available data, we could not assess the dose-response association between DIVCF and the risk of osteoporosis.
Between-study heterogeneity is identified in most metaanalyses, and it is of importance to explore the sources of heterogeneity between studies. In our meta-analysis, there was high heterogeneity in the correlation between DIVCF and risk of osteoporosis. The factors contributing to such heterogeneity are complex. First, differences in the definition, classification, and assessment of DIVCF were found, which could result in heterogeneity. Second, the variables adjusted for (including age, BMI, energy intake, parathyroid hormone levels, 25-hydroxyvitamin D levels, smoking, alcohol intake, physical activity, supplement use, and oral contraceptive use) were not consistent between studies. In addition, because of the limited sample size, the evaluation of publication bias calls for further research.

CONCLUSION
Our meta-analysis indicates that DIVCF is negatively associated with the risk of osteoporosis, hip fracture, and BMD loss and suggests that people should consume more DIVCF to reduce these risks.

AUTHOR CONTRIBUTIONS
JL and Z-PL conceived and designed the study. The literature searches, study selection, critical appraisal, data extraction, and contacting authors of included studies for additional