Vitamin B12, Folate, Homocysteine, and Bone Health in Adults and Elderly People: A Systematic Review with Meta-Analyses

Elevated homocysteine levels and low vitamin B12 and folate levels have been associated with deteriorated bone health. This systematic literature review with dose-response meta-analyses summarizes the available scientific evidence on associations of vitamin B12, folate, and homocysteine status with fractures and bone mineral density (BMD). Twenty-seven eligible cross-sectional (n = 14) and prospective (n = 13) observational studies and one RCT were identified. Meta-analysis on four prospective studies including 7475 people showed a modest decrease in fracture risk of 4% per 50 pmol/L increase in vitamin B12 levels, which was borderline significant (RR = 0.96, 95% CI = 0.92 to 1.00). Meta-analysis of eight studies including 11511 people showed an increased fracture risk of 4% per μmol/L increase in homocysteine concentration (RR = 1.04, 95% CI = 1.02 to 1.07). We could not draw a conclusion regarding folate levels and fracture risk, as too few studies investigated this association. Meta-analyses regarding vitamin B12, folate and homocysteine levels, and BMD were possible in female populations only and showed no associations. Results from studies regarding BMD that could not be included in the meta-analyses were not univocal.


Introduction
Osteoporosis is a chronic, multifactorial disorder which is characterized by low bone mass and microarchitectural deterioration of bone tissue [1]. Its major consequence is fractures. Especially hip fractures are frequently associated with institutionalization and increased mortality, and thus with an increased social and economic burden. This burden is expected to increase substantially in Europe in the coming decades due to a rise in life expectancy [2].
An elevated plasma homocysteine level (>15 mol/L) is prevalent in 30-50% of people older than 60 years [12][13][14]. The cause is multifactorial; a combination of environmental and genetic factors, nutrition, lifestyle, and hormonal factors [15]. Vitamin B 12 and folate are major determinants of homocysteine metabolism [16,17] and supplementation with vitamin B 12 and folic acid has been shown to be effective in normalizing homocysteine levels [18,19]. Reversing elevated homocysteine levels through folic acid and vitamin B 12 supplementation could theoretically prevent the problem of impaired bone health and osteoporosis. However, at present, no consensus is reached on the magnitude of the association between vitamin B 12 , folate, homocysteine, and bone health nor on the possible effect of vitamin B 12 and folate supplementation on bone health.
Up until now one systematic review including a metaanalysis summarized the evidence on homocysteine and fracture risk, showing that higher homocysteine levels significantly increase the risk of fracture [20]. No meta-analyses are known on the topic of folate and vitamin B 12 in relation to bone health. The purpose of this review is to provide a systematic overview, where possible including pooled estimates of the dose-response association, of the scientific evidence available from randomized controlled trials (RCTs), prospective cohort, and cross-sectional studies addressing vitamin B 12 , folate, and homocysteine levels in association with bone health, that is, fracture risk and BMD, in adults and elderly people.

Methods
This systematic review with dose-response meta-analyses was conducted within the scope of the EURRECA (European Micronutrient Recommendations Aligned) Network of Excellence (http://www.eurreca.org/) [21]. We followed a standardized methodology which is described in short below.

Search Strategy and Selection of Articles.
We conducted systematic literature searches for (1) vitamin B 12 , (2) folate, and (3) homocysteine. The electronic databases MEDLINE, EMBASE, and Cochrane Library Central were searched, using search terms in "MeSH" terms and "title" and "abstract" on study designs in humans, vitamin B 12 , folate, homocysteine, and intake or status. The full Medline search strategy is available online, (see Appendix 1 in Supplementary Material at http://dx.doi.org/10.1155/2013/486186).
To be able to use the same search to identify publications on other health related outcomes both in adults and elderly and in younger population groups, no terms were added to limit the search to health outcome or study population. Moreover, by using a broad search we expected a more complete retrieval of relevant publications. In this review only the results on vitamin B 12 , folate, and homocysteine status (i.e., biomarkers measured in serum or plasma) in relation to bone health indicators (fracture risk and BMD) are presented. In addition to the search, reference lists of 10 review articles were checked to identify potentially relevant references that were not identified with the multidatabase search. The search was not limited by language. This review contains studies up to July 2012.
We selected articles in two steps. The first selection step included screening for title and abstract by three independent investigators (J. P. van Wijngaarden, E. L. Doets, SB). In the second selection step, full texts of the selected abstracts were evaluated on basis of predefined inclusion criteria by four investigators (J. P. van Wijngaarden, E. L. Doets, A. Szczeciń ska, MP).
For the purpose of alignment and quality control 10% of the references in each selection step was screened and selected in duplicate by two investigators independently. Results were compared and discrepancies were resolved by unanimous consensus among all investigators.
Studies were eligible for inclusion if they were conducted in apparently healthy human subjects aged ≥18 y. Furthermore, studies had to report fracture incidence, fracture risk, or bone mineral density (BMD) as a health outcome and had to report baseline data on the outcome measure.
Observational studies were included if they (1) had a prospective cohort, nested case-control, or cross-sectional design, and (2) addressed serum/plasma concentration of markers indicating vitamin B 12 status (serum/plasma vitamin B 12 , methylmalonic acid (MMA), holotranscobalamin (holoTC)), folate status (serum/plasma folate or erythrocyte folate), or homocysteine status (serum/plasma homocysteine). Intervention studies were included if they (1) had a randomized controlled trial design, (2) studied the effects of vitamin B 12 or folic acid supplements, fortified foods or micronutrient intake from natural food sources and included a placebo or untreated comparison group, and (3) had a minimum intervention duration of six months.

Data Extraction and Statistical Analysis.
We extracted data for each of the identified studies on population characteristics, study design, assessment of vitamin B 12 , folate and homocysteine status, and fracture risk or bone mineral density.
Opportunities for meta-analysis were evaluated based on comparability of health outcome and status marker. If less than three comparable studies were available, results were qualitatively described. If three or more comparable studies were available, the results of these individual studies were expressed in a standardized format to allow comparison in the form of a continuous dose-response meta-analysis that pools the regression coefficient ( ) (SE) from multiple adjusted models. We chose to express association measures for serum/plasma vitamin B 12 per 50 pmol/L. When s were not reported in the original article, we transformed Relative Risk (RR), Hazard Ratio (HR), or Odds Ratio (OR) to s, using a standardized method [22]. The transformations to obtain s and SEs and statistical analyses were performed using R statistics version 2.9.2 (http://www.R-project.org/), with statistical significance defined as < 0.05. HR and OR were considered as RR because the outcome was relatively rare. If articles reported insufficient data (missing data, inconsistencies, or any other uncertainties), we contacted corresponding authors for additional information.
We calculated summary estimates of comparable studies using random effects meta-analysis. Applying the methods of DerSimonian and Laird, the between study variance was estimated which was used to modify the weights for calculating the summary estimate [23]. Residual heterogeneity between studies was evaluated using -statistic and 2 -statistic.
In total, from 3 searches we identified 11837 potentially relevant articles, of which 9835 articles were excluded based on title and abstract. Of the remaining 2002 articles, 1961 articles were excluded based on full texts, leaving 41 articles. As the searches were partly overlapping and some articles addressed more than one association this resulted in 20 unique articles, 19 observational and 1 intervention. A search update on July 2nd, 2012 resulted in an additional 8 observational studies, which makes a total of 28 included articles. All addressed the association between vitamin B 12 , folate or homocysteine status, and fracture risk or BMD. The flow diagram of the process of screening and selection is shown in Figure 1.

Fractures
Four longitudinal observational studies [3,[24][25][26], including 7475 elderly people with 3 to 16 years of follow-up and a total of 458 cases addressed the association between serum/plasma vitamin B 12 and fracture (Table 1). Pooled analysis of the association between 50 pmol/L increase in plasma/serum B 12 and change in fracture risk showed an inverse association (RR = 0.96, 95% CI = 0.92 to 1.00) with no heterogeneity between studies ( 2 = 0%, = 0.76) (Figure 2). This indicates that a vitamin B 12 increase of 50 pmol/L tends to decrease the risk of fracture with 4%.

Folate.
Three longitudinal observational studies examined the association between plasma folate and fractures [24][25][26] (Table 2). One study showed that women, but not men, with plasma folate in the lowest quartile had a higher fracture risk (HR 2.40, 95% CI 1.50 to 3.84) compared to the highest (reference) quartile ( for trend <0.001) [24]. Ravaglia et al. (2005) [26] showed a significant association between low folate status and fracture risk when folate was analyzed as a dichotomous variable (lowest quartile of folate status versus other 3 quartiles), but when analyzed as a continuous variable, no significant association was observed [26]. One study did not observe an association [25].
Three studies that were not included in the meta-analysis also showed significant associations between homocysteine levels and fracture risk. These studies were not included because the necessary data could not be retrieved from the articles; either homocysteine levels were log-transformed [4,5] or data was not shown for population homocysteine status [25]. Regardless the type of analysis, women and men in the highest homocysteine quartile had a 1.7 to 3.8 higher RR or HR than those in the lowest or the lowest three quartiles [4,5,25].

Bone Mineral Density.
In the studies included in this review BMD was measured at various sites in the body (e.g., lumbar spine, femoral neck, radius, hip, and total body). As BMD differs per site in the body, we pooled results per biomarker (serum/plasma vitamin B 12 , folate, and homocysteine) and per site for the three sites generally measured (FN, LS, or total hip), thus resulting in 9 meta-analyses. Betas of the individual studies are shown in Tables 1, 2, and 3. The studies included in the meta-analyses took only women into account.
From the studies that could not be compared in a meta-analysis, three studies showed significant associations between folate status and BMD or change in BMD over time [8,10,34]. Five studies did not observe an association between folate status and BMD [7,32,36,38,41].

Intervention Studies.
Up until now, only one RCT ( = 47) which met our inclusion criteria studied the efficacy of B-vitamin supplementation on BMD [42]. This study shows some evidence that BMD may be increased with high doses of B-vitamin supplementation in people with hyperhomocysteinemia (tHcy > 15 mol/L). However, this outcome was only found in a subanalysis of 8 hyperhomocysteinemic subjects [42].

Discussion
Our meta-analyses showed a significant association of homocysteine levels with fracture risk and a weak though significant inverse association of vitamin B 12 levels with fracture risk. We could not draw a conclusion regarding folate levels and fracture risk, as too few studies investigated this association. Meta-analyses regarding vitamin B 12 , folate and homocysteine levels and BMD in women found no associations. Results from studies regarding BMD that could not be included in the meta-analyses are not univocal.
To our knowledge this systematic review with metaanalyses is the most extensive systematic review on the association of vitamin B 12 , folate and homocysteine with bone health until now. Previous non-systematic literature reviews on the association between folate, vitamin B 12 , and homocysteine with bone health reported similar results, that is, conflicting evidence with suggestions towards the association of homocysteine levels with fracture [43][44][45]. These reviews did not report a systematic literature search strategy and did not provide a quantitative cumulative result. In our review the most recent published articles have been taken into account. The search strategy we used was systematic and extensive, and we used well-defined in-and exclusion criteria.
One recent systematic review included a meta-analysis on the association between tHcy and fractures [20]. This metaanalysis is different in design than ours, as it is not a doseresponse meta-analysis. To overcome the variation in cut-off levels for low vitamin B 12 and folate status and high homocysteine status, and to allow comparison and subsequent combination of individual studies in the performed meta-analyses, we expressed results of individual studies in a standardized format. We assumed a linear, continuous dose-response association between markers of vitamin B 12 and folate with fracture rather than a threshold effect. This assumption is generally used in meta-analyses. Furthermore, in some of the key articles addressing the association of homocysteine levels with fractures this association is present [4,5].
A common concern in meta-analyses is heterogeneity between studies. In our meta-analyses we experienced various levels of statistical heterogeneity (no heterogeneity to large heterogeneity). The heterogeneity may be explained by the differences in mean age of the study populations (41-78 years), differences in mean status of vitamin B 12 (190-549 pmol/L), folate (5.2-24.9 nmol/L) and homocysteine (9.3-16.5 mol/L), differences in sex distribution of the study populations, duration of follow-up (3-16 years), and level of adjustment for confounders. Although most included studies adjusted for a wide range of confounders for fracture risk or BMD, residual confounding by other unmeasured or inadequately measured factors cannot be ruled out. For example, low vitamin D status is a risk factor for fracture [46]. From the studies included in our meta-analyses for fracture three out of nine adjusted for vitamin D status [24,25,27]. Outcomes do not seem to differ between studies that corrected for vitamin D status and studies that did not. Homocysteine levels are increased with renal dysfunction, often measured by serum or urine creatinine levels. Five out of eight studies in the meta-analysis regarding homocysteine and fracture risk corrected for creatinine levels [3,24,26,27,29], and outcomes did not seem to differ.
As almost all studies were performed in countries without mandatory folate fortification or were performed before the fortification era in the USA and Australia, we do not consider folate fortification as a source of heterogeneity in our analyses.
The majority of studies included were longitudinal and cross-sectional observational studies. We could only include one intervention study, which had a very small study population ( = 47). One intervention study which found a beneficial effect of vitamin B 12 and folic acid supplementation on fracture risk could not be included in our systematic review, because this study investigated a population of hemiplegic patients following stroke [47]. The generalizability of these findings is confined to a highly selective patient population with a high percentage of vitamin D deficiency and a high fracture risk. As evidence from intervention studies is lacking, currently no causal effect between vitamin B 12 , folate and homocysteine levels and bone health can be established. Consequently, it is yet unknown whether extra vitamin B 12 and folate intake through supplementation could reverse the observed negative effects of vitamin B 12 and folate deficiency and elevated homocysteine levels. Further evidence from an intervention study is expected soon, as a large intervention study on the effect of vitamin B 12 and folic acid supplementation on fracture risk, BMD, and bone turnover markers is currently carried out with results expected in 2013 [48].
As the quality of included studies determines the quality of the review and meta-analysis, we assessed the overall risk of bias of each individual study using standardized procedures largely based on guidance from the Cochrane Collaboration [49], resulting in one of the following judgments: low, moderate, or high risk of bias. Twenty out of the 28 included studies were evaluated as having moderate ( = 15) or high risk ( = 5) of bias. These studies did take one or more of the predefined confounders into account, that is, age, sex, smoking, physical activity, and body weight, or the study was funded or cofunded by a commercial organization. Due to the limited number of studies included in the meta-analyses, we were not able to study the effect of the overall risk of bias, nor of its single components on the pooled effect measures. There seems to be no difference in the outcomes of studies with low risk of bias compared to studies with moderate or high risk of bias, and we therefore assume that the quality of the included studies had no effect on the outcome of this review.
The intake of folate and vitamin B 12 are a determinant of folate, vitamin B 12 , and homocysteine status. To deal with potential malabsorption of vitamin B 12 [50] and reduced bioavailability of folate [51], the use of biomarkers for vitamin B 12 and folate status is preferred over measures of intake when studying associations with bone health in elderly people.
In studies addressing folate status, serum or plasma folate was measured, which is considered as an appropriate marker for folate status in epidemiological studies [52]. Homocysteine is a nonspecific marker for both folate and vitamin B 12 status [53], which makes it a relevant biomarker in this review. Regarding the metabolic interactions between vitamin B 12 , folate, and homocysteine combined with the variety in data presented in the studies, we were not able to investigate the possibility that a low vitamin B 12 or folate status in combination with a high homocysteine level might result in a higher fracture risk in comparison to a low vitamin B 12 or folate status or homocysteine level alone. In most studies regarding vitamin B 12 status, status was assessed with serum or plasma vitamin B 12 . Other, more sensitive markers for vitamin B 12 deficiency, like MMA and HoloTC [54], were addressed only in a few studies. We could therefore not draw conclusions about the association between these biomarkers and outcomes on bone health.
There are several suggested mechanisms for the association between vitamin B 12 , folate, homocysteine, and bone health. Homocysteine may interfere with collagen crosslinking. Cross-links are important for the stability and strength of the collagen network. Interference in cross-link formation would cause an altered bone matrix, resulting in more fragile bones [55]. As collagen cross-links do not alter BMD, this may explain why a more convincing result is found regarding fractures than BMD, as suggested for example by Van Meurs et al. [5]. Vitamin B 12 deficiency has been associated with impaired functional maturation of osteoblasts [56]. Some in vitro studies support the hypothesis of a possible favorable effect of vitamin B 12 supplementation, although results are equivocal. Vitamin B 12 has been shown to stimulate osteoblast proliferation and alkaline phosphatase activity [57] but Herrmann et al. were not able to show any significant and consistent effect of vitamin B 12 or folic acid on osteoblast activity [58]. Recent publications show evidence of osteoclast stimulation in the presence of high homocysteine and low vitamin B 12 concentrations [59][60][61]. Vitamin B 12 and folate are not the only B-vitamins involved in the homocysteine metabolism. Various micronutrients, such as vitamin B 2 (riboflavin), vitamin B 6 (pyridoxine), and choline also affect homocysteine levels [16,17,62], and may consequently affect bone health. Given that vitamin B 12 and folate are the main factors influencing homocysteine levels, and therefore the primary focus in a homocysteine lowering intervention [63], our review focused on vitamin B 12 , folate, and homocysteine.

Considerations for Future Research and Conclusions.
The mechanisms involved in the association between biomarkers of B-vitamins and bone health are still unclear and therefore more fundamental research is required to establish the potential mechanisms. Subsequently, both observational and intervention studies should preferably not focus on just one biomarker in relation to the homocysteine metabolism, but take a biomarker profile into account, including serum/plasma vitamin B 12 , MMA, HoloTC, folate, and homocysteine levels. Evidence is needed from welldesigned, large intervention studies to establish a causal relationship between markers of B-vitamins and bone health.
This systematic review with meta-analyses shows that elevated homocysteine levels are associated with increased fracture risk. Vitamin B 12 status may be associated with fracture risk and evidence for an association between folate status and fracture risk is scarce. Vitamin B 12 , folate, and homocysteine levels are probably not associated with BMD, but results are not univocal.