Impact of vitamin D supplementation on C-reactive protein; a systematic review and meta-analysis of randomized controlled trials

Background To evaluate the effect of vitamin D supplementation on C-reactive protein (CRP) through a systematic review and meta-analysis of randomized control trials (RCTs). Methods PubMed-Medline, SCOPUS, Google Scholar and Web of Science databases were searched (up until April 2016) to identify RCTs evaluating the impact of vitamin D supplementation on CRP. We used random effects models (using DerSimonian-Laird method) as well as the generic inverse variance methods for quantitative data synthesis. For sensitivity analysis, we applied leave-one-out approach. To examine the heterogeneity we used I2 index. Registration code: CRD42016036932. Results Among 1274 search items, 24 studies met the inclusion criteria in the final evaluation. Pooling the data together indicated a non-significant decrease in CRP level following administration of vitamin D (weighted mean difference [WMD] -0.26(mg/l), (95% CI -0.75 to 0.22, N = 26 arms, heterogeneity p = 0.042; I2 54.2%). The WMDs for IL6 was 0.67 pg/ml, (95% CI 0.29 to 1.06, N = 16 arms, heterogeneity p = 0.234; I2 19.1%), 0.43 pg/ml, (95% CI 0.08 to 1.05, N = 26 arms, heterogeneity p = 0.120; I2 42.1%), for IL10, and −0.11 pg/ml, (95% CI -0.53 to 0.30, N = 12 arms, heterogeneity p = 0.423; I2 9.2%) for TNF-α, 4.03 pg/ml, (95% CI 3.50 to 4.57, N = 3 arms, heterogeneity p = 0.752; I2 8.1%) for adiponectin. Sensitivity analyses confirmed the robustness of the findings. Conclusions This study provided evidence that vitamin D supplementation had no impact on serum CRP, IL10, and TNF-α, while significantly increased serum IL6. We recommend RCTs with longer period of follow-up time (12 months) for future studies to provide explicit results. Electronic supplementary material The online version of this article (10.1186/s40795-017-0207-6) contains supplementary material, which is available to authorized users.


Background
Historically, vitamin D is recognised for its important role for bone health. However, recent studies suggest extraskeletal effects of vitamin D through autocrine and paracrine systems. Low vitamin D concentrations are related with several diseases with inflammatory nature including rheumatoid arthritis, metabolic syndrome, type 2 diabetes, cardiovascular diseases, and some types of cancer [1]. Low vitamin D status is reported to to simulate mild acute phase response in which casue elevated concentrations of C-reactive protein (CRP), several hemostatic factors and different pro-inflammatory cytokines [2][3][4]. Studies suggest vitamin D suppelemnation may reduce circulating CRP levels and some other plasma inflammatory cytokines. However, inconsistent results are reported across completed randomized trials [5][6][7]. Cytokines such as interleukin 6 (IL-6), interleukin 10 (IL-10) and tumor necrosis factor-alpha (TNF-alpha) mediate the inflammatory response in human tehrfore they can serve as potential biomarkers of chronic inflammatory diseases [8][9][10]. The elevated circulating concentrations of pro-inflammatory cytokines, such as IL6, and hepatic acute phase proteins (e.g. CRP) is a common feature of such diseases with chronic inflammation [11,12].
The potential effect of of vitamin D supplementation in decearsing chronic inflammation, if proven, is of public health interest given the disproportionate prevalence of vitamin D deficiency and insufficiency across the globe. A number of recent clinical trials have assessed vitamin D supplementation in different populations for its impact on circulating concentrations of several pro-and anti-inflammatory factors. However, such studies have had limitations such as small sample size, poor research design and subject traits (gender, ethnicity, age, etc.), and underpowered to achieve a comprehensive and reliable conclusion. Therfore there is substantial uncertainty about the net effect of vitamin D supplementation on CRP levels. Asystematic study which has addressed this issue dates back to 2014 inlcuding only a few studies [13]. Therefore, a comprehensive evalaution of evidence is needed to achieve an evidence-based conclusion. Hence, we aimed to address this uncertainty by systematically reviewing the literature, and meta-analysis of all trials, to explore the effects of vitamin D supplementation on CRP levels.

Methods
We conducted his systematic review based on the international referred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Guidelines [13,14]. We registed our study in the International Prospective Register of Systematic Reviews, PROSPERO (registration no: CRD42016036932).

Literature search strategy
In litrature search we considred the effect of vitamin D supplementation on plasma CRP concentration as the primary exposure of interest. The secondary exposure was the effect of vitamin D supplementation on inflammatory and anti-inflammatory markers and cytokines. We considred multiple databases including PUBMED/ Medline, Cochrane Central Register of Controlled Trials (CCTR), Cochrane Database of Systematic Reviews (CDSR), Web of Science and Google Scholar, until April 2016 for litrature search. Weapplied relevant search terms to find a published and unpublished studies for our interested outcome (Additional file 1: Table S1). We conisdred no limitation on language.

Selection criteria
We selected published randomized control trials (RCTs) assessing the impact of vitamin D administration on the infmlamatory parametres. Criteria for selecting studies: (i) clinical trial with single-arm, parallel or cross-over designs; (ii) RCTs of participants received vitamin D supplement compared to control group (either no vitamin D or placebo); and (iii) studies with information on primary outcome at the baseline and the endpoint in each group or the net change values. Exclusion criteria were: (i) non-clinical trials including those with case-control, cross-sectional or cohort designs; and (ii) studies missing to report mean (or median) plasma concentrations of our measures of interest at the baseline and/or at the endpoint. We carried out the selection by removing the of duplicates followed by titles and abstracts screening by two reviewers. The agreement between the reviewers was conisderable (Kappa index: 0.87; p < 0.001). We resolved the disagreements at a meeting between reviewers prior to selected articles being retrieved.

Data extraction and management:
Two reviewers (MM, PR) retrieved the full text of studies the met the inclusion criteria, and screened to determine the eligibility. After assessment of methodological quality, the two reviewers extracted data onto a purpose-designed data extraction form. The same reviewers independently summarised the most significant results of each study f. We compared the summaries and any variations of ideas resolved through a discussion with the third reviewer (HV). Details information from selected RCTs is summarized in Table 1. An independent reviewer confirmed all data entries.

Quality assessment
We used the Cochrane criteria to assess potential bias [14].

Data preparation for meta-analysis:
According to Cochrane Handbook recommendations, the mean change from baseline in the level of variables of interest and standard deviation (SD) for both groups were collected and used to compute the effect size [15]. The following formula was used: SD = SEM × square root (n), where n is the number of subjects. We sued the GetData Graph Digitizer 2.24 [16] to extract the required data when theer were presented in graphs.
We applied random effects model (using the DerSimonian-Laird method) and the generic inverse variance method to take to account the heterogeneity of studies in terms of demographic characteristics of populations [17].A quantitaved assessment of Heterogeneity was conducted using I 2 index, where values of 25%, 50%, and 75% reflect low, medium and high heterogeneity, respectively. We expressed the effect sizes as the weighted mean differences (WMD) and 95% confidence interval (CI). Sensetivity analysis was applied to assess the effct of each RCT on the overall effect size [18][19][20].

Selection RCTs
From searches in different search engines overall 1273 single citations recognized, of these, 126 were duplicates. From 1147 items, 35 left after assessment based on titles and abstracts, of which, 11 were not selected due to fact that: genetic, non-human studies, or molecular studies (n = 4); editorial or review articles (n = 3); incomplete data (2); Fig. 1. Consequently, 23 RCTs were used for pooling the data.

Risk of bias assessment
Results of assessment of bias revealed that some of the selected items recognized by the absence of information about the random sequence generation, blinding of outcome assessment and blinding of participants and study personnel, and allocation concealment. Though, nearly all of the assessed RCTs had a low risk of bias according to selective outcome reporting, with the exemption of two, which did not have sufficient material [25]. Details of the quality of bias assessment are presented in Additional file 1: Table S2.

Characteristics of the studies
A summary of the characteristics of the studies is presented in  [27]. Twoenty one studies used cholecalciferol in a dosage range from 0 IU/d [28] to 200,000 IU/d [29]. In two of the studies, particciopants were summplemned with ergocalciferol at a dose of 50,000 IU at baseline for 26 [30] or 12 [31] weeks. In three studies, calcium supplements had also been administered in doses of 200 [32], 1000 [33], 1200 [34] mg/d, respectively.

Pooled estimate of the effect of vitamin D supplementation on IL-6
The pooled estimate (weighted mean difference) of the impact of vitamin D supplementation on IL-6 was 0.67 pg/ml, (95% CI 0.29 to 1.06, n = 16 arms, heterogeneity p = 0.234; I 2 19.1%) across all studies (Fig. 3).

Pooled estimate of the effect of vitamin D supplementation on IL-10
The pooled estimate (weighted mean difference) of the impact of vitamin D supplementation on IL-10 was 0.43 pg/ml, (95% CI -0.56 to 1.44, N = 9 arms, heterogeneity p = 0.120; I 2 42.1%) across all studies. Pooled estimate of the effect of vitamin D supplementation on IL-12 The pooled estimate (weighted mean difference) of theimpact of vitamin D supplementation on IL-12 was 0.045 pg/ml, (95% CI -0.14 to 0.23, N = 5 arms, heterogeneity p = 0.358; I 2 15.1%) across all studies.

Sensitivity analysis
Tthe pooled effect estimates remained similar across all studies in leave-one-out sensitivity analyses (Table 2).

Publication bias
The visual inspection of funnel plot asymmetry declared no potential publication bias for the comparison of CRP levels between vitamin D supplementation and placebo groups (Fig. 4). Moreover, the presence of publication bias was not suggested by Egger's linear regression (intercept = 2.12, standard error = 2.68; 95% CI = −3.41, 7.66, t = 0.79, df = 24.00, two-tailed P = 0.435) and Begg's rank correlation test (Kendall's Tau with continuity correction =0.04, z = 0.28, two-tailed P value =0.774). After adjustment of effect size for potential publication bias using the 'trim and fill' correction, no potentially missing studies were imputed in funnel plots. Hence no difference in effect size than the initial estimate (WMD −0.26(mg/l), 95% CI -0.75 to 0.22) (Fig. 5). The 'fail-safe N' test showed that 271 studies would be needed to bring the WMD down to a nonsignificant (P > 0.05) value.

Discussion
In the current meta-analysis of randomized trials, we investigated the impact of high-dose vitamin D supplementation on circulating inflammatory and anti-inflammatory indexes. We detected no effect of vitamin D supplementation on circulating CRP. However our analysis revealed that vitamin D supplementation significantly increased IL-6 level by 0.67(pg/dl), while no significant effect was found on serum IL10 and TNF-α. From a theoretical point of view, there are several possible mechanisms that may explain vitamin D may affect serum CRP and IL-6. The physiological impact of vitamin D is not limited to the homeostasis of calcium and phosphate. For instance, vitamin D receptors (VDR) play role in the decreased activation of the pro-inflammatory transcription factor nuclear factor kappa B (NF-κB). This suggests that VDR has an intrinsic inhibitory role in inflammation [35,36]. One important target of vitamin D is NF-κB, which is inhibited by vitamin D, and via NF-κβ downstream release of the pro-inflammatory cytokines. NF-κB activation participates in the endogenous induction of CRP. Accordingly, the activated NF-κB may increase the effects of an activator of transcription-3 (STAT3) [37]. Studies have shown the active form of vitamin D (1,25dihydroxyvitamin D3 [1,25(OH)2D] inhibits NF-κB activation. This inhibatory effect is done by upregulating the inhibitor of NF-κB (IκB-α) and reducing IkB-α phosphorylation in lipopolysaccharide-stimulated murine macrophage cells as well as submissively sensitized human airway smooth muscle cells [38,39]. Thus, it may be hypothesised that vitamin D supplementation may suppress CRP via NF-κB and STAT3 signaling. Decreased parathyroid hormone (PTH) production with vitamin D supplementation may also explain the effects of vitamin D on hs-CRP. Low PTH may lead to decreased production of inflammatory factors [40].
According to our results, existing studies report mixed results regarding the impact of vitamin D supplementation on CRP. The Framingham Offspring Study cohort reported no significant association was found between vitamin D and CRP (n = 1381) [41]. While, in a metaanalysis of 10 randomized controlled trials (Chen et al. 2014), investigating the effect of vitamin D supplimentation on CRP [13], vitamin D supplementation significantly decreased the circulating CRP level by 1.08 mg/L [13]. In addition,, a recent meta-analysis of randomized controlled trials indicated a favourable impact on markers inflammation with vitamin D treatment [40]. It has been stated that heterogeneity across the findings of the studies may be due to supplemental dose of vitamin D, intervention durtaion and baseline hs-CRP level.
In the study by Forman et al.(2008) in a 1484 young women (aged 32 to 52 years)found that although vitamin D supplementation did not lower CRP specifically, it did lead to improvements in other inflammatory markers, for example IL-10 and TNF-α [42]. In addition, Ngo et al. studied 253 adults (aged 51 to 77 years) with mean CRP level of 3.6 ± 4.0 mg/mL and reported serum vitamin D have significant converse association with CRP level [43]. This association was seen in 147 morbidly obese subjects with CRP levels ranged from 1.88 to 4.01 mg/L [44]. In one study impact of vitamin D supplementation on CRP and IL-6 was different [45]. The one-year vitamin D supplementation in overweight and obese participants resulted in reduced serum IL-6 concentrations, while serum CRP concentrations were significantly increased. The contradictory findings in these studies may be attributed to the length of the study, seasonal change or geographical location [45]. In a randomized control trial in patients with acute myocardial infarction, a short duration of treatment with vitamin D has significant impact on weakening the rise of CRP and IL-6 (but not TNF-α) [46]. IL-6 is a multi-potential inflammatory cytokine that has a fundamental role in host defence including the immune responses, acute phase reactions and haematopoiesis [47]. Our analysis presented a positive association between vitamin D supplementation and circulating IL-6 levels. Our results may be influenced by seasonal differences in vitamin D level that cause changes in this increased levels of IL-6, IL-6-related signalling pathways, chronic diseases, congenital diseases, baseline IL-6 level, age, sex of subjects and a supplemental dose of vitamin D. Hence, this finding needs to be reexamined in larger randomized trials specifically designed to investigate the relationship between inflammatory indexes and vitamin D.
Our study has some potential limitations. Internal validity of our results relies on the quality of individual studies as it is seen in all meta-analsyes. Several limitations can be named in this regard. Firstly, most studies in this meta-analysis had medium sample sizes. This may lead to overestimation of vitamin D supplementaion effects. Smaller trials might be methodologically less robust and more prone to report larger effect sizes [48,49]. The number of available studies on this topic was rather small. Only four of the studies inlcuded in current meta-analyses were with the duration of 12 months. Among them, only one had a relatively large sample. Heterogeneity exist in doses of vitamin and health status of target population at the baseline. Further, most of the studies were conducted in clinical population rather than general healthy population. This may likely affect the baseline levels of vitamin D and the inflammatory markers. Across all studies 0.027 (pg/ml), (95% CI -0.72 to 0.77)

IL-13
Across all studies