Vitamin D Supplementation Does not Affect Serum Leptin and Adiponectin Levels in Adults: A Systematic Review and Meta-analysis of Randomised Controlled Trials

Introduction

Obesity is considered as a significant nutritional problem in low, middle and high-income countries [1]. The urbanization and considerable changes in lifestyle are associated with the high prevalence of overweight and obesity, especially in developing countries [2].  It is estimated that more than 57.8% of the world’s adult population will suffer from overweight or obesity by 2030 [2]. Obesity is linked to the increased rate of morbidity and mortality, which may create massive socio-economic and public health burdens for poorer nations [1].  Moreover, overweight and obesity increase the risk of diabetes, cardiovascular disease, cancer and premature death [3].

Obesity is defined as an accumulation of large amounts of fat mass in the body [4, 5]. The fat mass secretes bioactive peptides, named adipokines, which have an essential role in several processes, including food intake, insulin action, lipid, and glucose metabolism and regulation of the energy balance [6, 7]. The dysfunction in adipokines’ pathways is considered as an important reason for diseases caused by obesity [6, 7]. Leptin and adiponectin are well-known adipokines that are involved in the regulation of metabolic homeostasis, especially obesity [8].  It is suggested that serum levels of adiponectin and leptin are negatively and positively associated with body fat, respectively [8]. Studies have shown that the reduction in circulating adiponectin concentrations in obese subjects might increase the risk of obesity, insulin resistance, dia­betes, metabolic syndrome, cardiovascular disease and hypertension [9, 10]. Leptin is another adipokine that inhibits appetite and increases energy expenditure by influencing specific receptors in the hypothalamus when body fat is elevated [11].

Recently, the association between vitamin D deficiency and obesity and its possible mechanisms has been considered by investigators [12]. A meta-analysis of observational studies indicated a significant inverse association between serum 25 (OH) D levels and BMI [13]. However, the meta-analysis of clinical trials did not show the significant effect of vitamin D supplementation on body weight [14, 15], fat mass [14, 15], percentage of fat mass or lean body mass [15], and body mass index (BMI) [14, 16]. Besides, it is proposed that vitamin D might affect adipokines. Observational studies suggest that serum 25 (OH) D levels are positively correlated with adiponectin, and inversely correlated with leptin [17, 18].

In contrast, a number of clinical trials showed that vitamin D supplementation might increase the plasma levels of leptin [19, 20] and decrease serum adiponectin levels [21]. On the other hand, some other studies reported the reduction in serum leptin levels [22] and the increase in serum adiponectin levels [23, 24] after vitamin D supplementation. However, some clinical trials did not show any significant effect [25-27]. Recently, two meta-analyses of clinical trials [28, 29] tried to summarize the effect of vitamin D supplementation on leptin and adiponectin levels, but their results were inconsistent. While no significant effect of vitamin D supplementation on leptin and adiponectin was reported in a study by Dinca et al. [28], the other study concluded that serum level of leptin is significantly increased following vitamin D supplementation [29].  While both meta-analyses have adopted the same search strategies, different studies were included in their analysis, and consequently, the different findings were found. Furthermore, it is claimed that the number of included studies in both meta-analyses were limited and more clinical trials are needed in this regard. The present study attempted to elucidate the effect of vitamin D supplementation on plasma leptin and adiponectin concentrations by conducting an updated systematic review and meta-analysis.

Materials And Methods

The present systematic review is reported based on the preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines. The study protocol was also registered in the prospective international register of systematic reviews (PROSPERO) under the registry number of CRD 42018092110.

Search strategy

This systematic review and meta-analysis evaluated the effect of vitamin D supplementation on serum adiponectin and leptin levels through reviewing controlled clinical trials conducted in humans. The electronic databases including PubMed, Google Scholar, and Scopus were searched until November 2017, using two sets of the following MeSH and non-MeSH keywords: 1) “Vitamin D” OR “Ergocalciferols” OR “Cholecalciferol” OR “Calcitriol” OR “Calcifediol” OR “25-Hydroxyvitamin D 2” OR “25-hydroxyvitamin D” OR “1-25-dihydroxy-23,23-difluorovitamin D3” OR “25(OH)D” OR “25-OH vitamin D” OR “1,25 (OH) (2) D” OR “1,25 (OH) D” OR “1,25-(OH) (2) D (3)” OR “25-hydroxyvitamin D” OR “Vitamin D” OR “25-(OH) D (3)” OR “25-(OH) D (2)” OR “Ergocalciferols” OR “Cholecalciferol” OR “Calcitriol” and 2) “Intervention Studies” OR “intervention” OR “controlled trial” OR “randomized” OR “randomised” OR “random” OR “randomly” OR “placebo” OR “assignment” OR “clinical trial” OR “trial” And “Adipocytokines” OR “Adipokine” OR “Adiponectin” OR “Adipokines” OR “leptin.”  The wild-card term “*” was used to increase the sensitivity of the search strategy. The search was conducted in the title, abstract and keywords without language limitation. Furthermore, the reference lists of retrieved articles were also reviewed for additional studies. All these steps were performed by two researchers individually (SJS and SS), and disagreements were resolved by discussion.

Study selection

The original studies were included if they met the following criteria: 1) being conducted in human adults, 2) being controlled clinical trial in design (either parallel or cross-over), 3) vitamin D supplementation was considered as an intervention, 5) measured and reported leptin and adiponectin concentrations were reported as an outcome at baseline and the end of follow-up in each group or provided the net change values.

Data extraction

Eligible studies were reviewed, and the following information was extracted: lead author’s last name, year of publication, study location, study design, participant’s gender, the sample size of vitamin D and control groups, type and dose of vitamin D, participants’ characteristic, duration of the treatment, mean values and SD of the baseline. Ultimately, final concentrations and net changes of leptin and adiponectin were extracted independently by two reviewers (SJ, SS) in the intervention and control groups.

Quality assessment

Quality of the included articles was evaluated using the Cochrane Collaboration’s tool [30] based on the following domains: sequence generation (selection bias), allocation sequence concealment (selection bias), blinding of participants and outcome assessors (performance and detection bias, respectively), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias) and other potential sources of bias. The risk of bias was measured as ‘low risk,’ ‘high risk or “unclear risk.” The included studies were classified as fair if they were at low risk for two domains and good if they were low risk in more than two domains.

Statistical analysis

The difference in mean change (MD) for serum leptin, adiponectin levels and their corresponding standarddeviation (SD) were calculated for each study. In case of not reporting the mean and SD for mean change values, they were estimated using correlation coefficient for studies which reported the baseline, after follow-up andchange values (r = 0.57 for leptin [20, 31-33] and r =  0.59 for adiponectin [20, 23, 24, 31, 34, 35]. To check the sensitivity of the meta-analyses to calculated correlation coefficients, the meta-analyses were replicated by computing the effect sizes based on r = 0.2 and r = 0.8. The Hedges’ g and its corresponding standard error (SE) were calculated as the effect size to perform the meta-analysis because the values were reported in different scales, and the conversion was not possible. The meta-analyses were done using a random-effects model, which takes the between-study heterogeneity into account. Statistical heterogeneity between studies was examined using the Cochran’s Q test and I-squared [36]. Subgroup analysis was conducted to evaluate the sources of between studies heterogeneity. In order to explore the extent to which inferences might depend on a particular study or group of studies, a sensitivity analysis was performed by recalculating the pooled effects after 1) removing the highest-weighted study from a given analysis (the “leave-one-out” analysis) [37]; and 2) testing alternatives by the 0.5 correlation between baseline and post-treatment values (0.2 and 0.8). Publication bias was examined by visual inspection of funnel plots [38]. In this funnel, effect sizes were depicted against their corresponding SE. Statistical assessment of funnel plot asymmetry was tested using two formal tests, the Begg’s adjusted rank correlation tests and the Egger’s regression asymmetry test [39]. Statistical analyses were conducted using STATA version 11.0 (STATA Corp. College Station, Texas). P-values less than 0.05 was considered to be a significant difference.

Results

After the initial search, 4,929 articles were identified, and 227 duplicates were removed. The screening of the title/abstracts led to 416 articles which their full-text were assessed for eligibility; 393 items were eliminated because they did not have the inclusion criteria and did not measure adipokines as the outcome variable. Finally, 23 studies were entered into the systematic review and 22 studies in the meta-analysis (Figure 1). The plasma adiponectin concentrations were evaluated in 19 RCTs [20-27, 31, 33-35, 40-46] and 16 RCTs [19, 20, 22, 25-27, 31-33, 40-42, 44, 46-48] reported data on plasma leptin concentrations. The included studies have been published between 2009 to 2017. Among the twenty-three studies, twelve were conducted in Asia [19, 22-24, 31-35, 40, 43, 48], three in America [25, 41, 47], six in Europe [20, 26, 27, 42, 44, 45] and two in Australia [21, 46]. In the included studies, 858 participants were assigned to vitamin D supplementation, and 846 were assigned as controls. Two studies [27, 41] were conducted in only females, and others were done in both sexes. The study duration ranged from 1 to 48 weeks.

 Twelve studies administered daily doses [19, 21, 24-26, 31, 32, 40, 41, 44, 45, 47], four studies managed weekly doses [22, 33, 34, 43], two studies did monthly doses [46, 48], one study prescribed vitamin D for every two weeks [35],  and four studies used a single bolus dose [20, 23, 27, 42]. Fourteen studies were conducted on participants with vitamin D deficiency [21-23, 25, 27, 31, 33, 34, 40-42, 45, 47, 48]. The doses of vitamin D supplementation varied from 400 IU to 10000 IU/day in multiple daily doses prescription and 300000 IU to 600000 IU when single bolus doses were prescribed. All randomized clinical trials have a parallel design. So, if a study used different doses of vitamin D [46], it is included the highest dose in the meta-analysis. Besides, one study investigated different types of vitamin D, including ergocalciferol and cholecalciferol for intervention [44]. Therefore, the data for vitamin D₃ supplementation was included.

Furthermore, the data of both male and female sex was combined in a study conducted by Sharifi et al. [35], and then they were analyzed. The researchers of this study contacted the corresponding author of Al-Sofiani et al. study [40] for data regarding the serum level of leptin and adiponectin. However,  a requested data were not provided. So, the study was included in the systematic review but not in the meta-analysis. The characteristics of the included studies are shown in Table 1. One study did not report the final measurement of adipokines [40]; thus this study was also excluded from the meta-analysis.

Figure  1: Study selection process

Risk of bias in individual studies

The result of evaluating the bias assessment in included studies is presented in Table 2. A high risk of bias was assessed according to random sequence generation [48], allocation concealment [48], blinding of participants and outcome assessors [32, 42, 48], and incomplete outcome data [44] in a number of studies. Some trials were classified as unclear risk of bias regarding allocation concealment [20-25, 31, 33, 41, 42, 44-46], blinding of participants and outcome assessors [45]; but all studies [19-27, 31-35, 41-48] were low-risk in terms of other sources of bias. The overall quality of all the included studies was assessed to be good.

The effect of vitamin D supplementation on serum leptin levels

As shown in Figure 2, the meta-analysis of 15 studies [19, 20, 22, 25-27, 31-33, 41, 42, 44, 46-48] did not show any significant effect of vitamin D supplementation on serum leptin levels (Hedges’ g = 0.042, 95 % CI: -0.294 to 0.378, p = 0.807). There was a high-level of heterogeneity between included studies (Cochrane Q test, Q statistic = 93.79, P < 0.001, I² = 85.1%). Subgroup analyses were performed based on baseline vitamin D status, disease status (with/without diabetes), kidney disease status (yes/no) of participants, vitamin D type used for supplementation, vitamin D supplementation method (single dose/daily/weekly/two weeks/ monthly), participants’ sex, study duration, and vitamin D fortification (yes/no) to evaluate whether the effect is different in a specific group of studies or not. Vitamin D supplementation significantly decreased serum leptin level in patients with end-stage renal disease (ESRD) (Hedges’ g = -0.634, 95 % CI: -1.221 to -0.047, P = 0.034). However, the serum leptin increased significantly when a single dose of vitamin D supplementation was prescribed (Hedges’g = 0.941, 95 % CI: 0.368 to 1.514, p = 0.001). Besides, subgroup analysis revealed that study duration was another source of heterogeneity.  The results of this study were not sensitive to the selected correlation coefficient (Table 3)

Table 1. Characteristics of randomized controlled trials that evaluated the effect of vitamin D supplementation on serum leptin and adiponectin that were eligible to be included in the systematic review

Table 2. Study quality and risk of bias assessment using the Cochrane collaboration tool

L: Low; H:High

Table 3. Meta-analysis showing the effect of vitamin D supplementation on serum leptin based on several subgroups (all analyses were conducted using a random-effects model).

¹ ESRD: End-stage renal disease

Table 4. Meta-analysis indicated the effect of vitamin D supplementation on serum adiponectin based on several subgroups (all analyses were conducted using a random-effects model).