Vitamin D and bone health in adults: a systematic review and meta-analysis Vitamina

Resumo A baixa saúde óssea está associada à deficiência de vitamina D em indivíduos mais velhos; no entanto, isso não está bem estabelecido em adultos. O estudo objetivou-se analisar a associação entre concentrações séricas de 25-hidroxivitamina D e baixa saúde óssea em adultos por revisão sistemática e metanálise. A pesquisa foi realizada nas bases LILACS, PubMed, Scopus, Web of Science, ScienceDirect de março de 2017 a outubro de 2018 com indivíduos adultos (20-59 anos). A avaliação da saúde óssea foi realizada através da absorciometria dupla de raios X e concentrações séricas de 25(OH)D. O modelo de efeito aleatório foi utilizado para analisar dados do conteúdo mineral ósseo e densidade mineral óssea. Modelos de efeitos aleatórios foram utilizados e a heterogeneidade foi explorada por meio de meta -regressão. Trinta e cinco artigos foram selecionados. Houve correlação positiva entre a vitamina D e a saúde óssea na maioria dos locais avaliados. Observou-se correlação na análise de subgrupos da coluna lombar entre homens. Quando estratificados, os estudos apresentaram alta heterogeneidade, explicada pelo tamanho da amostra, pelos níveis séricos médios da vitamina e pelo risco de viés. A vitamina D está positivamente correlacionada com a saúde óssea em indivíduos adultos. Palavras-chave Densidade mineral, Vitamina D, Adultos Abstract Low bone health is associated with vitamin D deficiency in older individuals; however, this association is not well established in adults. The aim of the study was to analyze the association between serum concentrations of 25-hydroxyvitamin D and bone health in adults by systematic review and meta-analysis. The search was carried out in the LILACS, PubMed, Scopus, Web of Science, ScienceDirect databases from March 2017 to October 2018 with adult individuals (2059 years). Bone health was evaluation performed through dual X-ray absorptiometry and serum concentrations of 25(OH)D. The random effect model was used to analyze data from bone mineral content and bone mineral. Random effects models were used and the sources of heterogeneity were explored by means of meta-regression. Thirty-five articles were selected. There was positive correlation between vitamin D and bone health in most of the evaluated sites. Correlation was observed in the analysis of subgroups for lumbar spine among men. When stratified, the studies presented high heterogeneity, which was explained by the sample size, mean serum vitamin D levels and risk of bias. Vitamin D is positively correlated to bone health in adult individuals.


introduction
Over the last four decades there have been epidemiological and sociodemographic changes, with significant repercussions on living conditions and on the burden of chronic non-communicable diseases which constitute a global health problem 1 . Among the most common, the World Health Organization (WHO) has highlighted those related to complications arising from low bone mass, such as osteoporosis 2 .
Osteoporosis is defined as a progressive, skeletal disease characterized by alterations in microarchitecture and consequent bone fragility 2 . This is an asymptomatic disease, usually identified when the individual presents a fracture, not only bringing damage in relation to the biological aspects, but also to the quality of life, as well as contributing to the increase in mortality and overloading the public health system due to the need for continued care [2][3][4] .
The evolution of osteoporosis and associated fractures are conditioned by some risk factors [5][6][7][8][9][10] , which lead to an osteometabolic imbalance caused by the deficiency of essential nutrients to maintain active bone metabolism, with the main nutrients being calcium and vitamin D 7,11,12 .
Vitamin D plays a determining role in the initial stages of skeletal development 12,13 , constituting a factor for preventing rickets and osteomalacia 14 . However, in adulthood there are still controversies regarding the relationship between this vitamin and low bone mineral density (BMD) and bone mineral content (BMC), which are biophysical parameters used to assess bone health 15,16 . Some studies report a positive association 17,18 , while others suggest that these are not correlated 19,20 .
Considering the controversial results of studies on the influence of serum vitamin D levels on bone metabolism in adults and that vitamin D deficiency has been presented as a global public health problem, it is necessary to summarize the available evidence on the subject. In addition, there are few systematic review studies with a meta-analysis employing this approach in adults 21,22 . Thus, the objective of this systematic review was to analyze the association between serum concentrations of 25-hydroxyvitamin D (25(OH)D) and BMC

Methods
This is a systematic review study with meta-analysis supported by the PRISMA rules (Preferred Reporting Items for Systematic reviews and Meta-Analyses) 23 on the relationship between vitamin D and BMD/BMC in adults. In this perspective, we sought to answer the following question: are serum concentrations of 25(OH)D associated with BMC and BMD in adults?

search strategy
Two independent reviewers (KJ Segheto and M Pereira) conducted study searches in the LI-LACS, PubMed, Scopus, Web of Science and Sci-enceDirect electronic databases from March 2017 to October 2018, with the following descriptors/ Mesh terms: "vitamin D", "bone density", "BMD", "BMC", "adult" and "observational study" and their respective corresponding terms in Portuguese and Spanish. The search strategy included truncating the terms to exclude texts which did not fit the objectives of this review, such that they were adjusted according to the search form of each database (Chart 1).
The search results were managed in the Men-deley® program to remove duplicates and apply the inclusion criteria. The manuscript titles were initially read and then the abstracts of those publications which fulfilled the inclusion criteria. Once the articles were selected, the reading was completed in full. The last selection stage was an analysis of the references of the original articles and the identified revisions, thus guaranteeing refinement in searching for relevant works for this review.
The whole selection and evaluation process of the articles was done in pairs. At the end of the review, disagreements on eligibility were resolved by consensus with a third reviewer (CJ Carvalho).

eligibility criteria
The articles selected for this review had to meet the following eligibility criteria: original studies whose objective was to analyze the association between 25(OH)D and BMC and/or BMD, performed with adult individuals aged 20 to 59 years of age, having no association to diseases, with bone health evaluation performed through dual X-ray absorptiometry (DXA) and serum concentrations of 25(OH)D.
In this study, it was decided to evaluate the serum concentrations of 25(OH)D because they are the best indicator of this vitamin 24 . Thus, studies that included individuals with vitamin D supplementation were excluded. In addition, serum concentrations of 25(OH)D were converted when necessary using the following criteria: 1ng/ ml=2.496 nmol/l (24). In this way, it was possible to guarantee standardization and comparison of the presented results.

Data extraction
Eligible articles were read in full and information on the year of study publication, sample size, type of study, vitamin D results, correlation coefficient and/or linear regression and fit variables used in the modeling were registered in specific form.

Risk of bias
The risk of bias of the studies was individually assessed through the Research Triangle Institute Item Bank (RTI-Item Bank) scale 25 . This scale is composed of 29 questions, among which 7 items were selected to assess the risk of bias of the articles due to the methodological diversity of the designs of observational studies. After the analysis, the risk of bias was classified as: high risk of bias -a study with one or more negative answers to the items; moderate risk of bias -when one or more items were considered "partially" or "cannot be determined"; and low risk of biasall items on the scale had a positive response 25 (Chart 2).

statistical analysis
Correlation coefficients and sample size were used to calculate the standard error to evaluate the correlation between 25(OH)D and bone health. The Z-test was used to analyze the data for the following bone sites: BMC, lumbar spine BMD (LS-BMD), hip (H-BMD), femoral neck (FN-BMD) and trochanter (T-BMD). The Cochran Q statistical test and the inconsistency test (I²) 25 were used to evaluate the heterogeneity and consistency of the studies. When heterogene-Chart 2. Assessment of bias risk using RTI Item Bank for the studies included in the meta-analysis. ity (I²>25%) was identified, the random effects model was used 26 . The publication bias was evaluated through the funnel plot symmetry 26 . The statistical evaluation of the effect of small studies was performed by the Egger test 27 . The criterion for the application of these tests was the minimum number of eight studies. Gender-based subgroup analysis was also performed to identify possible sources of heterogeneity. The overall effect was derived from the DerSimonian and Laird method 26 , using the random effects model, which takes into account the variation between the studies.

Authors/Year
In addition to gender, meta-regressions were performed considering the following variables: age, group evaluated (1-men, 2-women, 3-men and women), sample size, mean concentration serum levels of vitamin D, latitude and longitude of the study site, and risk of bias score of each study. The results obtained from the correlation between vitamin D and bone health in adults are presented using a Forest Plot chart. A p-value less than 0.05 was considered significant in all analyzes. The STATA 14 program (Stata Corp, College Station) was used for data analysis.
The main aspects that contributed to the high risk of bias were: uniform inclusion and exclusion criteria; valid outcome evaluation, appropri-

Meta-analysis results
In the meta-analysis, only articles using Pearson's linear correlation were included for the following bone sites: BMC -7/9 18 Table 2). The evaluated studies  (Figures 3 and 4). It was possible to observe that there is no evidence of publication bias through the funnel graph ( Figure 5). Considering the Egger's test, the effect of small studies was only observed for the femoral neck (LS-BMD: p=0.083; H-BMD: p=0.088; FN-BMD: p=0.024; and T-BMD: p=0.074).
Meta-regressions were performed to investigate possible sources of heterogeneity identified in the meta-analysis. We observed statistically significant sources of heterogeneity for: sample size for FN-BMD (p<0.01) and H-BMD (p=0.01); mean serum 25(OH)D concentrations of participants for H-BMD (p=0.03); and risk of bias for BMC (p=0.03). On the other hand, the mean age of the participants, the analyzed group (1-men, 2-women, 3-men and women), as well as the latitude and longitude did not significantly contribute to the occurrence of heterogeneity in the meta-analyzes (Table 3).

Discussion
We observed a significant positive correlation between serum 25      studies selected for this review presented high heterogeneity and high coverage in relation to gender, ethnicity, study population, and countries, among other aspects. An effect of small studies were not observed except for the FN-BMD site (Chart 4). The relationship between 25(OH)D and bone health in adults is biologically plausible, since deficiency of this vitamin has been considered an important determinant of several diseases, especially those related to bone health [60][61][62][63] . Vitamin D is important throughout development, as it allows a greater absorption of calcium in the intestine, which also positively contributes to bone health 13,24 . Deficiency of vitamin D in the early stages of development may lead to developing weak, narrow and soft bones, providing a greater probability of fractures 13,24 . In a systematic review, it was identified that low 25(OH)D in childhood increases the risk of fractures in this age group, requiring supplementation in these cases 64 . Although all the factors associated with the occurrence of osteoporosis have not yet been well established, it is known that peak bone mass during childhood and adolescence (the period for accumulating 50% of total bone mass) and the rate of bone loss during aging are determinants 65 . Thus, adequate concentrations of vitamin D during these development stages contribute to optimizing mineral gain, and consequently to better bone health in adulthood 65 . This may ease the process of loss at more advanced ages.
A meta-analysis conducted with randomized clinical trials aimed at assessing the relationship between calcium and vitamin D supplementation and fracture prevention in middle-aged and older adult individuals found that those who underwent vitamin D supplementation had a reduction in the risk of general fractures by 15% and of hip fractures by 30% 63 . Another meta-analysis identified similar results, noting an association between vitamin D insufficiency and hip fracture risk, with a 40% increase in risk of occurrence 62 . Another study with predominantly white adults evaluating 2,294 individuals submitted to vitamin D supplementation found a positive association in six of the ten studies evaluated, four with beneficial effect on only one bone site and two on the hip, de-                           66 . These authors emphasized that, although maintaining adequate serum levels of vitamin D is important, supplementation should be performed only in individuals identified with inadequate levels 66 .
In our review, the most evaluated bone sites were LS-BMD, H-BMD and FN-BMD. These results are consistent with the recommendations for clinical practice, since they are anatomical regions where the initial process of bone loss (osteopenia)    it continues considered in future work, particularly through forward-looking approaches.
The present study has some limitations. The first one refers to the fact that all the selected studies have a cross-sectional design, and therefore by character these studies do not enable an establishment of a temporal relation between the studied variables. Another issue is the non-standardization of the evaluated bone sites, which makes it difficult to summarize the results. Finally, the comparison of vitamin D status may be hampered by a high variation of serum 25(OH)D measurement between different analytical methods.
As a positive, we highlight that this review includes an evaluation of the association between serum vitamin D concentrations in different bone sites in adults and the analysis of data exploring subgroups and heterogeneity sources. Moreover, the performance of all the review stages by inde-pendent authors is also worth mentioning, reducing the chance of selection bias of the studies.
In conclusion, we showed a positive association between serum concentrations of 25(OH) D and bone health from the results of this systematic review and meta-analysis. It should be noted that clinically healthy individuals without osteometabolic diseases were evaluated in this study. This is an important issue because given the positive association identified, there is need to maintain adequate vitamin D levels even in adults due to its biological importance.
It is therefore recommended that bone health evaluation be incorporated into clinical practice aimed at adults with vitamin D insufficiency or deficiency. It is known that osteoporosis is a silent disease and therefore requires preventive measures and early diagnosis in order to avoid critical illness or possible fractures.

Collaborations
KJ Segheto: responsible for writing the article, collecting data, analyzing and interpreting the results. M Pereira: responsible for writing the article, analyzing and interpreting the results. DCG Silva: responsible for writing the article and critical review. CJ Carvalho: responsible for writing the article, responsible for data collection and analysis and interpretation of results. FR Massardi: responsible for data collection and analysis and interpretation of results. AM Kakehasi: responsible for writing the article and critical review. LL Juvanhol: responsible for writing the article and critical review. GZ Longo: responsible for writing the article and critical review.