ABSTRACT
Introduction
A growing body of evidence supports an association between hyperglycemia and zinc metabolism. For example, many studies have provided evidence that low zinc status is associated with impaired insulin secretion, decreased insulin sensitivity, and increased inflammatory biomarkers, including IL-6, TNF-α, and high-sensitivity C-reactive protein (hs-CRP) (5–12). Moreover, previous studies have revealed significant differences between diabetic patients and healthy subjects with respect to blood concentrations of zinc, and patients with low zinc concentrations are more likely to have had diabetes for a longer time, have poorer glucose control, and have reduced pancreatic β cell function (13, 14).
Results from previous meta-analyses suggest that zinc supplementation may exert a favorable effect on several plasma lipid parameters, including significantly reduced concentrations of total cholesterol, LDL cholesterol, and triglycerides (15, 16). Although few meta-analyses have examined the effects of zinc supplementation on glycemic control, some analyses have revealed that fasting glucose (FG) and glycated hemoglobin (HbA1c), either individually or in combination, are modestly—albeit significantly—reduced following zinc supplementation (17, 18). However, due to the limited amount of available data, the effects of zinc supplementation with respect to preventing or treating diabetes are currently inconclusive (19). In recent years, a growing number of well-designed randomized controlled trials have focused on examining the putative effects of zinc supplementation on obesity, metabolic syndrome, prediabetes, and diabetes, thus providing important data regarding the relation between zinc supplementation and diabetes prevention and management. Unfortunately, however, these studies varied with respect to several parameters, including sample size, the subjects’ health status, and the dose and efficacy of the zinc supplementation, thereby leading to inconsistencies among studies (20). To overcome these issues, we performed a comprehensive meta-analysis of all published randomized controlled trials in order to assess the effects of zinc supplementation on glycemic control in subjects who either have diabetes or have a high risk of developing diabetes.
Methods
Our systematic review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) set of items for systematic reviews of interventional studies (21).
Search strategy
We systematically searched PubMed, Embase, and the Cochrane Library for randomized controlled trials published through February 10, 2019. The following keywords were used in the literature search: “zinc” AND (“supplementation” OR “co-supplementation” OR “supplement”) AND (“diabetes” OR “metabolic syndrome” OR “impaired fasting glycemia” OR “impaired glucose tolerance” OR “hyperglycemia” OR “insulin resistance” OR “prediabetic” OR “prediabetes” OR “pre-diabetes” OR “obesity” OR “obese” OR “glycemic” OR “glucose” OR “insulin” OR “Homeostasis Model Assessment” OR “HOMA” OR “HOMA-IR” OR “glycated hemoglobin” OR “HbA1c” OR “C-reactive protein” OR “CRP” OR “hs-CRP”). Our search was restricted to studies conducted in humans, and no restriction was imposed with respect to the language of the publications. The references cited within the retrieved relevant articles were also reviewed in order to identify additional studies. The selection procedure is summarized in Figure 1.
Study selection
Studies were selected for inclusion by 2 independent reviewers (XW and WW), subject to approval by a third reviewer (WZ). Studies that satisfied the following 4 criteria were included in our meta-analysis: 1) the study was a controlled clinical trial in humans; 2) the study population included persons with prediabetes, impaired fasting glucose, or impaired glucose tolerance, who were overweight, or obese, or a combination of these; 3) the study included a zinc-supplement intervention; and 4) the study included ≥1 of the following outcomes: FG, 2-h postprandial glucose (2h-PG), fasting insulin (FI), HOMA-IR, HbA1c, or hs-CRP, or a combination of these. In addition, the following exclusion criteria were applied: 1) studies that were not placebo controlled; 2) studies that were conducted solely in healthy subjects or in patients with type 1 diabetes; 3) the mean change or SD in the outcome measure(s) was not reported; 4) the study had an observational, case-control, cross-sectional, or cohort design; and 5) the publication was a narrative review, comment, opinion piece, methodologic report, editorial, letter, or conference abstract.
Data extraction
Data were extracted with the use of a standardized data-collection form. Two investigators (XW and WW) independently extracted detailed information from each included article. This information included the first author, publication year, study location, study design, health status, gender, age, sample size, formulation, dose and duration of the zinc supplementation, and outcome measures (baseline and postintervention concentrations). Any necessary calculations that used the data were performed by 2 investigators (XW and WW) and were checked by a third investigator (WZ), and any discrepancies were resolved through group discussion with 2 additional investigators (JM and FW).
Quality assessment
The Jadad scale was used to assess the methodologic quality of the trials included in the systematic review (22). Each study was given a quality score ranging from 0 (“poor”) to 5 (“good”) based on the following set of criteria (with 1 or 0 points assigned for an answer of “yes” or “no”, respectively): 1) Was the study described as randomized? 2) Was the study described as double blind? 3) Are there descriptions of withdrawals and dropouts? 4) Are the methods of randomization both described in the paper and appropriate? 5) Are the methods of blinding both described and appropriate?
Statistical analysis and data synthesis
We used the Cochrane Handbook to calculate effect size. For both the control and intervention groups, we calculated the mean (and SD) changes in the outcome measures between the baseline value and the value at the end of the intervention (23). We based our meta-analysis on calculating the net change between baseline and the endpoint when the mean and SD values of the changes were reported, as follows: [(measure at endpoint in the treatment group – measure at baseline in the treatment group) – (measure at endpoint in the control group – measure at baseline in the control group)]. Effect sizes are expressed as the between-group weighted mean differences with 95% CIs and were assessed with a random-effects model.
Read more Are Compression Socks Good For Diabetics?
Plasma glucose concentrations (FG and 2h-PG) were collated in mg/dL; where necessary, a multiplication factor of 18.016 was used to convert glucose concentrations from mmol/L to mg/dL (24–28). FI concentrations were collated in mU/L; where necessary, a multiplication factor of 0.1436 was used to convert FI from pmol/L to mU/L (26, 29). hs-CRP values presented in nmol/L were multiplied by 0.105 to obtain values in mg/dL (30). Each dose of zinc gluconate and zinc sulfate was multiplied by 0.143 (31) or 0.227 (32, 33), respectively, to obtain the appropriate dose of elemental zinc.
Heterogeneity among the studies was estimated through the use of the I2 statistic, with values of 0–25%, 26–75%, or 76–100% representing a low, moderate, or high degree of heterogeneity, respectively (34, 35). We performed meta-regression analyses to compare subgroup differences (36). Publication bias was evaluated through the use of funnel plots and Egger’s linear regression test, with significance set to P < 0.10 (37, 38). A sensitivity analysis was performed in order to examine the effect of individual studies by omitting 1 study at a time (39). All statistical analyses were performed with Stata version 12. Differences with a P value <0.05 were considered significant.
Results
Study characteristics
We performed a literature search as described in the Methods; the search strategy is summarized in Figure 1. Our initial search identified 734, 386, and 377 articles in PubMed, Embase, and the Cochrane Library, respectively. After removing duplicate publications and screening the articles by reading the title, abstract, and—where appropriate—the full-text article, a total of 36 publications were eligible for inclusion in the meta-analysis, including 22 intervention studies in 26 publications that used zinc as a single factor (24–30, 32, 33, 40–56), and 10 intervention studies in 10 publications in which zinc was administered together with a co-supplement (57–66). The publications included in the meta-analysis are described in Table 1.
Of the 22 studies in which zinc was a single factor, 20 intervention studies (24–27, 29, 30, 32, 33, 40–56) investigated the effects of zinc supplementation alone compared with placebo; the other 2 intervention studies measured the effects of zinc supplementation for zinc together with a co-supplement compared with the co-supplement alone (28, 29, 41). The remaining 10 interventions (57–66) compared the effects of zinc with a co-supplement group with a group that received only placebo. The mean dose of elemental zinc included in these interventions was 35 mg/d (range: 4–240 mg/d; median: 30 mg/d), and the duration of zinc supplementation ranged from 1 to 12 mo (24–30, 32, 33, 40–45, 47–66). A variety of anions were used in the zinc preparations, including sulfate (24, 25, 28, 29, 32, 40–46, 48, 49, 52, 58, 59), gluconate (26, 27, 30, 33, 47, 51, 53, 54, 56, 65), amino chelate (50, 66), oxide (62), and acetate (60); in some cases, the anion was not specified (55, 57, 61, 63, 64).
In total, 1700 participants were assigned to either a zinc intervention or a control group (24–30, 32, 33, 40–45, 47–66). The participants were all adult subjects aged ≥18 y, with the exception of 1 study involving children 6–10 y of age (44, 48). Of the 32 intervention studies, 4 involved pregnant women; 3 interventions were for GD (33, 47, 61, 65), and 1 intervention was for pregnant women who had impaired glucose tolerance (IGT) but not GD (56). Of the remaining trials, 19 involved patients with T2D (25, 27–29, 32, 40, 41, 43, 45, 46, 49, 51, 52, 57–60, 64, 66), 6 involved obese individuals (26, 30, 42, 44, 48, 50, 53, 54, 63), 2 involved subjects with prediabetes (24, 55), and 1 involved subjects with metabolic syndrome (62). The mean Jadad score for all 32 intervention studies included in our meta-analysis was 3.7; 11 studies had the maximum score of 5, and none of the studies had a score of 0 points (Supplemental Table 1).
Effects of zinc supplementation on glycemic control
Effect of zinc supplementation on FG
To determine whether zinc supplementation affects hyperglycemia, FG data were extracted from the eligible studies. The effect of zinc on FG was reported in 29 interventions, including 20 studies that used zinc as a single factor and 9 studies that used zinc with a co-supplement (Figure 2). The following analysis revealed a significant reduction in FG when zinc supplementation was a single factor, with a weighted mean difference (WMD) of −13.26 mg/dL (95% CI: −17.69, −8.83 mg/dL; I2 = 94.5%) between the zinc and control groups. In addition, we found a significant reduction in FG in the zinc with co-supplements group (WMD: −16.42 mg/dL; 95% CI: −27.54, −5.31 mg/dL; I2 = 95.6%) and in the overall, ungrouped analysis (WMD: −14.15 mg/dL; 95% CI: −17.36, −10.93 mg/dL; I2 = 94.7%).
Effect of zinc supplementation on 2h-PG
Next, we examined whether zinc supplementation affects the glycemic response by extracting 2h-PG data from the eligible studies. The effect of zinc on 2h-PG was reported for 7 interventions, including 6 studies that used zinc as a single factor and 1 study that used zinc with a co-supplement (Figure 3). Similar to our results obtained with FG, zinc supplementation significantly reduced 2h-PG when zinc was a single factor (WMD: −39.87 mg/dL; 95% CI: −68.16, −11.57 mg/dL; I2 = 98.7% ), when zinc was used with a co-supplement (WMD: −19.70 mg/dL; 95% CI: −27.67, −11.73 mg/dL), and in the overall, ungrouped analysis (WMD: −36.85 mg/dL; 95% CI: −62.05, −11.65 mg/dL; I2 = 98.6%).
Effect of zinc supplementation on FI
Next, we examined whether zinc supplementation affects circulating insulin by extracting and analyzing FI concentrations. The effect of zinc on FI was reported in 20 interventions, including 13 studies that used zinc as a single factor and 7 studies that used zinc with a co-supplement (Figure 4). Our analysis showed no significant reduction in FI when zinc was either a single factor (WMD: −1.96 mU/L; 95% CI: −4.27, 0.35 mU/L; I2 = 98.1%) or used with a co-supplement (WMD: −1.74 mU/L; 95% CI: −3.65, 0.16 mU/L; I2 = 96.9%). However, we found a significant reduction in FI based on an overall, ungrouped analysis (WMD: −1.82 mU/L; 95% CI: −3.10, −0.54 mU/L; I2 = 97.7%).
Effect of zinc supplementation on HOMA-IR
We also examined whether zinc supplementation affects insulin sensitivity by extracting HOMA-IR data from the eligible studies. The effect of zinc on HOMA-IR was reported in 16 interventions, including 11 studies that used zinc as a single factor and 5 studies that used zinc with a co-supplement (Figure 5). We found a significant reduction in HOMA-IR when zinc was a single factor (WMD: −0.87; 95% CI: −1.55, −0.19; I2 = 98.6%), but not when zinc was used with a co-supplement (WMD: −0.44; 95% CI: −1.43, 0.55; I2 = 97.7%). Nevertheless, the overall, ungrouped analysis showed that zinc caused a significant reduction in HOMA-IR (WMD: −0.73; 95% CI: −1.22, −0.24; I2 = 98.8%).
Read more 6 Health Benefits of Drinking Pickle Juice
Effect of zinc supplementation on HbA1c
Next, we examined whether zinc supplementation affects long-term glycemic regulation by extracting and analyzing HbA1c data. The effect of zinc on HbA1c was reported in 14 interventions, including 8 studies that used zinc as a single factor and 6 studies that used zinc with a co-supplement (Figure 6). Our analysis revealed a significant reduction in HbA1c when zinc was a single factor (WMD: −0.35%; 95% CI: −0.67%, −0.02%; I2 = 97.4%), when zinc was used with a co-supplement (WMD: −0.79%; 95% CI: −1.50%, −0.07%; I2 = 97.4%), and in the overall, ungrouped analysis (WMD: −0.55%; 95% CI: −0.84%, −0.27%; I2 = 97.4%).
Effect of zinc supplementation on hs-CRP
Lastly, given that previous studies suggest that chronic inflammation is closely associated with the risk of diabetes (67, 68), we evaluated whether zinc supplementation affects the concentration of circulating hs-CRP. The effect of zinc supplementation on hs-CRP was reported in 11 interventions, including 9 studies that used zinc as a single factor and 2 studies that used zinc with a co-supplement (Figure 7). Our analysis revealed a significant reduction in hs-CRP when zinc was used as a single factor (WMD: −1.63 mg/L; 95% CI: −2.68, −0.58 mg/L; I2 = 98.6%), when zinc was used with a co-supplement (WMD: −0.35 mg/L; 95% CI: −0.65, −0.06 mg/L; I2 = 64.8%), and in the overall, ungrouped analysis (WMD: −1.31 mg/L; 95% CI: −2.05, −0.56 mg/L; I2 = 98.3%).
Subgroup analyses of zinc supplementation on glycemic control
We also performed subgroup analyses on studies that used zinc as a single factor for the 6 glycemic indicators stratified according to 6 specific factors (Table 2).
Subgroup analyses based on health status
First, we examined whether health status influenced the effects of zinc supplementation (Table 2). Notably, the effect of zinc supplementation on FG was stronger among diabetic subjects than among subjects at high risk of diabetes. A meta-regression analysis revealed a significant difference in FG concentrations based on diabetic status (P = 0.026).
Subgroup analyses based on the dose of zinc supplementation
We also performed a subgroup analysis to determine whether the dose of zinc supplementation affected outcome by comparing studies with low (<30 mgd) and high (≥30 dailyzinc supplementation (Table 2). Our meta-regression analysis revealed no significant difference in glycemic indicators with respect to the dose of zinc supplementation.
Subgroup analyses based on the duration of zinc intervention
Subgroup analyses based on the formulation of the zinc supplement
Next, we performed a subgroup analysis to determine whether the formulation of the zinc supplement influenced the effects of the intervention (Table 2). Specifically, we compared the effects of inorganic zinc supplements with organic zinc supplements. Our analysis revealed that the effect of zinc supplementation on FG was stronger for inorganic supplements than for organic supplements. A meta-regression analysis revealed a significant difference in FG between inorganic and organic formulations (P = 0.008).
Subgroup analyses based on geographic area
We also examined whether geographic area influenced the effects of zinc supplementation (Table 2). We compared the effects of zinc in studies in which the intervention was conducted in Eastern (Asian) and Western (American, European, and Australasian) countries. A subsequent meta-regression analysis revealed a slight—albeit not significant—difference in HbA1c (P = 0.082) between the 2 geographic subgroups.
Subgroup analyses based on methodologic quality
Next, we examined whether the methodologic quality of the included studies, as measured by the Jadad score (22), differed between low-quality studies (defined as a total score of <3) and high-quality studies (defined as a total Jadad score ≥3) (Table 2). Notably, a meta-regression analysis was performed based on the quality assessment subgroups and revealed no significant differences.
Publication bias and sensitivity analysis
Next, we examined publication bias by generating funnel plots (Supplemental Figure 1) and performing Egger’s test; for this analysis, we used Egger’s test rather than Begg’s rank correlation test, as the data are quantitative. Our analysis revealed no evidence of publication bias for FG (P = 0.093), 2h-PG (P = 0.643), FI (P = 0.221), HOMA-IR (P = 0.167), HbA1c (P = 0.964), or hs-CRP (P = 0.203). Finally, a sensitivity analysis revealed that no single study likely affected the pooled results or total effect size (Supplemental Figure 2).
Discussion
In this meta-analysis of placebo-controlled trials, we analyzed data from 32 interventions involving a total of 1700 participants spanning 14 countries. In this respect, this meta-analysis provides the most up-to-date evidence supporting the putative favorable effects of zinc supplementation in preventing and managing diabetes. The results of our meta-analysis show that zinc supplementation can modulate glycemic control both in diabetic patients and in patients with a high risk of developing diabetes. Specifically, we found that zinc supplementation alone is associated with reduced blood glucose concentrations, increased insulin sensitivity, decreased nonenzymatic glycosylation, and reduced inflammation in these subjects.
In mammals, zinc plays an important role in glycemic control. Under physiologic conditions, zinc is abundant in pancreatic islets (69, 70), where it plays a role in the crystallization and secretion of insulin (71). In addition, evidence suggests that zinc regulates the glucose transporter GLUT4 translocation and the glucose utilization (72). Importantly, zinc deficiency is associated with increased chronic inflammation (73). Another meta-analysis also reported zinc supplementation has a beneficial effect on circulating hs-CRP concentrations (74).
Compared to the standard CRP (C-reactive protein) test, which measures CRP concentrations from 10 to 1000 mg/L, the high-sensitivity CRP test can accurately detect concentrations of CRP protein as low as 0.5–10 mg/L; furthermore, Pradhan et al. (68) used hs-CRP to predict the risk of developing T2D. In addition, Dehghan et al. (67) reported that genetic variations are also associated with differences in hs-CRP concentrations and the risk of developing diabetes. Based on the recommendation of the US Institute of Medicine, the recommended dietary allowance of zinc is 11 mg/d for men, 8 mg/d for women, 11 mg/d for pregnant women, and 5–8 mg/d for children aged 4–13 y (75). The Institute of Medicine–suggested tolerable upper limit of zinc intake is 40 mg/d for healthy adults (regardless of gender and pregnancy status) and 12–23 mg/d for children aged 4–13 y (75). And based on the recommendation of the European Food Safety Authority, the population reference intake of zinc is 9.4–16.3 mg/d for men, 7.5–12.7 mg/d for women, 9.1–14.3 mg/d for pregnant women, and 5.5–7.4 mg/d for children aged 4–10 y (76). The European Food Safety Authority–suggested upper limit of zinc is 25 mg/d for adults and 7–10 mg/d for children aged 4–13 y (77).
Read more Become a Diabetes Nurse in 3 Steps
A recent meta-analysis showed that zinc supplementation can significantly reduce total cholesterol, LDL cholesterol, and triglyceride concentrations (16). However, because health status, the dose, duration, and formulation of the zinc intervention, and the genetic background of the individuals vary among studies, the overall role of zinc on preventing and treating diabetes has remained inconclusive (78). Previous systematic reviews and meta-analyses found that FG (17, 18), HbA1c (17), and CRP (74) concentrations can be reduced by zinc intervention. However, the effects of zinc supplementation on other glycemic indicators—particularly insulin sensitivity-related factors such as 2h-PG, FI, and HOMA-IR—are inconclusive due to a limited number of clinical studies. In recent years, interest in the effects of zinc on glycemic control has increased considerably, allowing us to perform a more comprehensive meta-analysis in order to investigate the effects of zinc supplementation with respect to preventing and managing diabetes.
Notably, our up-to-date meta-analysis included 32 studies (in 36 publications) that evaluated the effect of zinc supplementation in 1700 subjects with T2D and subjects at high risk of developing diabetes. By comparison, 2 previous meta-analyses conducted by Jayawardena et al. (17) and Capdor et al. (18) included 12 publications with 580 T2D subjects and 14 publications with 584 T2D and high-risk subjects, respectively.
In addition to the clinical trials that used zinc as a single factor, several other studies investigated the effects of zinc supplementation in combination with other micronutrients. Most of these studies reported the beneficial effects of combined supplementation (60, 61, 65, 66), whereas other studies found less pronounced effects (58, 62, 63)—or even an opposite effect (64)—on glycemic control. Differences in the physiologic functions of the supplements and interactions, including differences in ingredients, may have contributed to these inconsistent findings. Therefore, we included studies that used zinc as single factor as well as studies that combined zinc with other supplements in our meta-analysis; moreover, we used studies that used zinc as a single factor in our subgroup analyses in order to examine in further detail the effects of zinc supplementation on glycemic control.
Despite its advantages, our study has several limitations that warrant discussion. First, relatively high heterogeneity was found among the intervention studies included in our analysis. Because zinc homeostasis is tightly regulated by SLC39 (79) and SLC30 family members (80), gene-environment interactions might contribute to high heterogeneity among different individuals. Indeed, a previous meta-analysis found a nominally significant interaction between total zinc intake and the SLC30A8 rs11558471 variant on FG concentrations (81). Finally, a recent prospective study found that genetic variations in SLC30A8 were associated with dietary and supplemental zinc intake and the risk of T2D (82).
Second, although we observed stronger effects of inorganic zinc supplements on FG, the dose of zinc supplements and health status of subjects varied among studies. The average doses of zinc supplements were higher in the inorganic zinc subgroup (52 mg/d) than in the organic zinc subgroup (40 mg/d). Notably, the inorganic zinc subgroup consisted of more interventions with diabetes (11 trials) than high-risk nondiabetes (3 trials), and stronger beneficial effects on FG were indeed observed in diabetic subjects than in high-risk nondiabetic subjects. Future studies should be designed in order to examine the effects of different formulations of zinc supplements in diabetic subjects.
Third, some of the studies had relatively small groups (i.e., <30 participants/intervention group); nevertheless, although this may have reduced the reliability of these individual studies, taken together their results support the results of the larger trials.
Lastly, the majority of the original clinical trials were performed in Asian countries, with relatively few studies performed in Europe, the United States, and Latin America. Thus, well-designed placebo-controlled trials with larger sample sizes in a variety of geographic regions should be conducted in order to overcome these limitations.
In conclusion, the results of our meta-analysis indicate that zinc supplementation has beneficial effects with respect to glycemic control in diabetics and in individuals with a high risk of developing diabetes. Importantly, we provide the first meta-analysis‒based evidence that 2h-PG, FI, and HOMA-IR concentrations are significantly reduced by zinc supplementation. In addition, our subgroup analyses revealed stronger beneficial effects on FG in subjects with diabetes and in subjects who received an inorganic zinc supplement. We therefore believe that our findings provide practical information with clinical implications.
Taken together, these results indicate that zinc supplementation may be used as an adjunct therapy for preventing or managing diabetes. In the future, large multinational prospective randomized controlled trials should help determine the ideal dose, duration, and formulation of the zinc intervention specific to each individual patient’s health status.
Acknowledgments
The authors’ responsibilities were as follows—XW, WW, JM, and FW: designed the study; XW, WW, and WZ: conducted the research; XW, WW, WZ, XF, and LC: analyzed the data; XW, WW, JM, FW, and LR: wrote and revised the manuscript; and XW, WW, WZ, XF, LC, LR, JM, and FW: read and approved the final version. None of the authors reported a conflict of interest related to the study.
Notes
This study was supported by research grants from the National Natural Science Foundation of China (31600953 to XW, 31530034 to FW, 31570791 and 91542205 to JM), the National Key R&D Program of China (2018YFA0507801 to JM and 2018YFA0507802 to FW), and the Zhejiang Provincial Natural Science Foundation of China (LQ15C110002 to XW).
Supplemental Table 1 and Supplemental Figures 1 and 2 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: CRP, C-reactive protein; FG, fasting glucose; FI, fasting insulin; GD, gestational diabetes; HbA1c, glycated hemoglobin; hs-CRP, high-sensitivity C-reactive protein; T2D, type 2 diabetes; WMD, weighted mean difference; 2h-PG, 2-hour postprandial glucose.
