| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research Communication |
1 From Statistics Collaborative, Inc, Washington, DC.
2 Prepared under contract to the Office of Dietary Supplements, Office of the NIH Director, National Institutes of Health.
3 Address reprint requests to MD Althuis, National Cancer Institute, Division of Cancer Epidemiology and Genetics, Environmental Epidemiology Branch, 6120 Executive Boulevard, Room 7084, EPS MSC 7234, Rockville, MD 20852. E-mail: althuism{at}mail.nih.gov.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: The study aimed to determine the effect of chromium on glucose and insulin responses in healthy subjects and in individuals with glucose intolerance or type 2 diabetes.
Design: The study design was a systematic review and meta-analysis of randomized clinical trials (RCTs).
Results: The authors identified 20 reports of RCTs assessing the effect of chromium on glucose, insulin, or glycated hemoglobin (Hb A1c). This review summarizes data on 618 participants from the 15 trials that reported adequate data: 193 participants had type 2 diabetes and 425 were in good health or had impaired glucose tolerance. The meta-analysis showed no association between chromium and glucose or insulin concentrations among nondiabetic subjects. A study of 155 diabetic subjects in China showed that chromium reduced glucose and insulin concentrations; the combined data from the 38 diabetic subjects in the other studies did not. Three trials reported data on Hb A1c: one study each of persons with type 2 diabetes, persons with impaired glucose tolerance, and healthy subjects. The study of diabetic subjects in China was the only one to report that chromium significantly reduced Hb A1c.
Conclusions: Data from RCTs show no effect of chromium on glucose or insulin concentrations in nondiabetic subjects. The data for persons with diabetes are inconclusive. RCTs in well-characterized, at-risk populations are necessary to determine the effects of chromium on glucose, insulin, and Hb A1c.
Key Words: Chromium • dietary supplements • diabetes • insulin • glucose • hemoglobin • meta-analysis
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
10 million Americans take chromium supplements, sometimes for the prevention or treatment of diabetes (26). Although early case series reported that chromium alleviates severe symptoms of diabetes (27–32), whether randomized placebo-controlled clinical trials have adequately confirmed the findings from uncontrolled studies (33–37) remains controversial. Four recent literature reviews summarized evidence supporting chromium as a nutraceutical that controls glucose and insulin concentrations (9, 12, 15, 16) but provided no quantitative overview. In addition, these reviews did not separate the discussion of randomized controlled trials from the discussion of uncontrolled or observational studies, rendering interpretation of the true effect of chromium supplementation difficult to assess. Although Hellerstein (10) critically reviewed randomized clinical trials of chromium supplementation and diabetes, he presented only a qualitative description of the findings. To help elucidate the effect of dietary chromium, we performed a systematic review of the literature and a meta-analysis of randomized clinical trials assessing the effect of dietary chromium supplements on measures of glycemic control in healthy subjects and in individuals with glucose intolerance or type 2 diabetes.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Identification of trials
We performed a computerized bibliographic search of the Cochrane Controlled Trials Register (April 2000; Cochrane Collaboration, Oxford, United Kingdom) and MEDLINE (1966 to May week 3, 2000; National Library of Medicine, Bethesda, MD) in all languages. For both databases, we applied the optimal MEDLINE search strategy for identifying controlled clinical trials (38) along with the following specific search terms: chromium, diabetes, glucose, insulin, and hemoglobin A1c. We then inspected the bibliographies of the collected articles to identify additional relevant reports. Two persons (NEJ and MDA) independently read the title, abstract, and descriptors of each article to identify potentially relevant trials for full review, and copies of these articles were obtained.
Data collection and statistical analysis
We considered trials to be randomized only if the text stated explicitly that the intervention was allocated randomly. Two of the authors (NEJ and MDA) reviewed and abstracted all potentially relevant articles independently and resolved disagreement by consensus or by third party adjudication (JTW). If necessary, we contacted the investigator for further information on methodology or data. Reviewers were not blinded to author. All contacted investigators were asked whether they had or knew of additional data relevant for this review.
Using a specially designed form, the reviewers abstracted data on study design, methodologic quality (39), and glucose, insulin, and glycated hemoglobin (Hb A1c) concentrations after the intervention. The data were then entered into a FILEMAKER PRO (version 4.0; Claris Corporation, Santa Clara, CA) database and exported to SAS (version 6.12; SAS Institute Inc, Cary, NC) for data analysis and to S-PLUS 2000 (Math Soft, Inc, Seattle) for graphics.
The difference between means (and the 95% CIs of the difference) of glucose, insulin, and Hb A1c concentrations was calculated for each trial. For studies with more than one chromium arm or more than one placebo arm, we calculated a weighted average of response rates if the groups were similar by assigning each arm a weight proportional to its sample size. We considered dissimilar groups to be separate treatments. Pooled mean differences and 95% CIs for studies of diabetic and nondiabetic subjects were calculated by using 2 different fixed-effects models: weighted mean differences and standardized mean differences. We planned to report only the results from the weighted mean difference model unless the models yielded dissimilar results. A chi-square test was used to assess heterogeneity among trials. If the P value was <0.05, which indicated heterogeneity among studies, and if only 1 or 2 studies explained the heterogeneity, we removed those studies from the analysis. Otherwise, in the face of evidence of heterogeneity, we planned to use a random-effects model to combine the data.
Formulation, dose, and level of exercise may affect the relation between dietary chromium and glycemic control. For each of these variables, we used a linear regression model to assess whether the relation between dietary chromium and glycemic control differed among levels of the variable.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The main design features of the 15 randomized clinical trials of dietary chromium supplementation and glucose-related factors are summarized in Table 1
. Eleven of the 15 trials (73%) used data from 
90% of the subjects randomly assigned in the outcome analysis (3–5, 7, 11, 13, 14, 17–19, 25). Overall, 15% of the randomly assigned patients were lost to follow-up or were excluded from the analyses.
|
, the mean concentration of glucose for nondiabetic subjects was 5 mmol/L. Only 4 studies reported chromium concentrations at baseline.
|
included diabetic subjects, for a total of 193 diabetic subjects with analyzable data (1, 3, 6, 25); one study included 155 of these subjects (6). Two studies analyzed subjects who were glucose intolerant at baseline; one study enrolled only intolerant subjects (n = 24; 4) and the other enrolled 5 glucose-intolerant and 27 healthy subjects (14). For this review, we combined the data from the healthy subjects and from those with glucose intolerance and refer to these subjects as nondiabetic subjects. Chromium chloride and chromium picolinate were the most frequently used formulations before and after 1990, respectively. Anderson et al (6) reported data on 155 diabetic subjects in China, who were randomly assigned to 3 groups: one-third each to 200 µg chromium picolinate, 1000 µg chromium picolinate, and placebo. We excluded this study from the formal meta-analysis for 2 reasons. First, as the only study conducted in a non-Western country, it may represent the effect of supplementation in a chromium-deficient population. Second, including it in the meta-analysis led to P values for heterogeneity below or very close to 0.05. Therefore, the meta-analysis included only 38 diabetic subjects.
Efficacy of dietary chromium supplementation for reducing glucose concentrations
Fasting glucose
All 14 trials reported data on fasting glucose concentrations (n = 38 diabetic and 425 nondiabetic subjects; Table 3). Combining all 14 trials showed a pooled mean difference in fasting glucose concentrations of 0.027 mmol/L (95% CI: -0.09, 0.15 mmol/L) with no evidence of heterogeneity (P = 0.97) (Figure 1). Similarly, dietary chromium supplementation was not associated with glucose concentrations among nondiabetic subjects (pooled mean difference: 0.028 mmol/L; 95% CI: -0.086, 0.14 mmol/L).
|
|
). Abraham et al (3) reported a nonsignificant increase in fasting glucose concentrations (mean difference: 1.17 mmol/L; 95% CI: -1.04, 3.38 mmol/L). Anderson et al (6), who used 2 doses of chromium supplementation, showed a significant decrease in fasting glucose in the group receiving 1000 µg Cr compared with placebo (mean difference: -1.70 mmol/L; 95% CI: -2.41, -0.99 mmol/L) but not in the group receiving 200 µg Cr (mean difference: -0.10 mmol/L; 95% CI: -0.93, 0.73 mmol/L). The small sample size of the studies by Offenbacher and Pi-Sunyer, Thomas and Gropper, and Abraham et al and the noncomparability of the population used for the study by Anderson et al led to inconclusive results concerning the effect of chromium supplementation on fasting glucose concentrations among diabetic subjects.
|
Two studies reported the relation between chromium and 120-min glucose among diabetic subjects (1, 6). Anderson et al (6) reported a statistically significant reduction in 120-min glucose in the high-dose group compared with the placebo group, but no significant association in the low-dose group. Offenbacher and Pi-Sunyer (1) reported no significant relation between chromium and 120-min glucose (Table 4).
Efficacy of dietary chromium supplementation for lowering insulin concentrations
Fasting insulin
Ten of the 14 trials presented data on fasting insulin concentrations (n = 8 diabetic and 326 nondiabetic subjects; 1, 4, 5, 7, 8, 11, 14, 17, 19, and H Lukaski, unpublished observations, 2000; Table 3). The overall pooled mean difference was 0.28 pmol/L (95% CI: -7.0, 7.5 pmol/L; test of heterogeneity, P = 0.097; Figure 2). Dietary chromium supplementation was not associated with insulin concentrations among nondiabetic subjects (pooled mean difference: 0.25 pmol/L; 95% CI: -6.98, 7.48 pmol/L).
|
Insulin at 120 min
Five of the 14 trials reported data on insulin concentrations 120 min after an oral-glucose-tolerance test (n = 8 diabetic and 133 nondiabetic subjects; 1, 4, 7, 14, 19). The overall pooled mean difference was 11.1 pmol/L (95% CI: -69.0, 91.2 pmol/L; test of heterogeneity, P = 0.15). Dietary chromium supplementation was not associated with an increase in 120-min insulin among nondiabetic subjects (pooled mean difference: 5.5 pmol/L; 95% CI: -74.0, 85.1 pmol/L).
Only Anderson et al (6) and Offenbacher and Pi-Sunyer (1) reported results for 120-min insulin in diabetic subjects. Offenbacher and Pi-Sunyer (1) showed a nonstatistically significant increase in 120-min insulin concentrations after chromium supplementation, with a wide CI (mean difference: 144 pmol/L; 95% CI: -453, 740 pmol/L). Anderson et al (6) showed identical mean reductions in insulin concentrations for both dose groups compared with placebo, with a mean difference of -63 pmol/L (95% CI for 1000 µg: -79.6, -46.4 pmol/L; 95% CI for 200 µg: -78.3, -47.7 pmol/L). More studies of the effect of chromium supplementation in diabetic populations are needed to clarify these relations.
Efficacy of dietary chromium supplementation for lowering Hb A1c
Three very different trials presented data on Hb A1c concentrations: one study each of 33 healthy subjects (7), 24 persons with glucose intolerance (4), and 155 diabetic subjects (6). The study of healthy subjects reported no association between chromium and Hb A1c concentrations (7). Chromium was associated with a slight, nonstatistically significant reduction in Hb A1c concentrations in the study of glucose-intolerant subjects (mean difference: -0.30%; 95% CI: -0.85%, 0.25%) (4). Anderson et al (6) reported a dose-response relation between dietary chromium supplementation and reduction in Hb A1c concentrations among diabetic subjects (mean difference for 1000 µg: -1.90%, 95% CI: -2.34%, -1.46%; mean difference for 200 µg: -1.00%, 95% CI: -1.55%, -0.45%).
Analysis of confounders
Chromium supplementation was not associated with fasting insulin or fasting glucose within strata of formulation, dose, or exercise levels.
Safety
None of the 15 trials reported any adverse events associated with dietary chromium supplementation.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The studies included too few patients to make conclusions about the effect of chromium in those with glucose intolerance. Uusitupa et al (4), the only investigators to report findings solely on subjects with impaired glucose tolerance, found no association between dietary chromium supplementation and glucose or insulin concentrations in a study of 24 subjects.
Too few trials in diabetic subjects have been conducted to allow conclusive findings for subjects with type 2 diabetes (1, 3, 6). Anderson et al (6) were the only investigators to report a dose-response relation between chromium and glucose and insulin concentrations in diabetic subjects. Because this study was conducted in China, however, the applicability of the results to the Western hemisphere is uncertain; subjects enrolled had low body mass indexes of 22–23, which may indicate poor nutritional status of the population at baseline. In addition to the study by Anderson et al, only one small randomized controlled trial of chromium supplementation of women with gestational diabetes (n = 20), not included in this review, reported significant reductions in glucose and insulin concentrations in the chromium-supplemented group (44). Three small randomized clinical trials of persons with type 2 diabetes (which included a total of 38 subjects), with data on insulin available for only 8, found no association between chromium supplementation and glycemic control. Two studies not included in the meta-analysis because of insufficient data also reported no association (21, 23).
Glucose and insulin concentrations fluctuate with changes in diet, exercise, and use of some medications. Therefore, measurement of glycated proteins, such as Hb A1c, is a more reliable method of assessing long-term glycemic control (45). Only 3 studies assessed the relation between dietary chromium supplementation and reductions in Hb A1c, one each of healthy subjects (7), subjects with glucose intolerance (4), and diabetic subjects (6). The reduction in Hb A1c concentrations was larger for randomized clinical trials of subjects with more severe disease. Specifically, among diabetic subjects, Anderson et al (6) reported a dose-response relation between dietary chromium supplementation and a decrease in Hb A1c concentrations in the control group compared with the treatment groups (200 and 1000 µg chromium picolinate/d). Reductions in Hb A1c concentrations were evident after 2 mo of chromium supplementation, with more marked reductions after 4 mo. Similarly, Uusitupa et al (4) reported an association of chromium with a reduction in Hb A1c concentrations (0.3%) in subjects with insulin intolerance, but the reduction, which was not statistically significant, was smaller than that reported by Anderson et al among diabetic subjects; however, this association may be attributable to differential weight loss in the treatment groups. In a study not included in this analysis, Jovanovic et al (44) reported that chromium reduced Hb A1c concentrations by 0.4% in women with gestational diabetes receiving 4 µg chromium picolinate/kg body wt for 8 wk compared with a control group (44). The randomized clinical trial of healthy subjects included in this review (7) showed no association of chromium and Hb A1c. These studies assessed the magnitude of change in Hb A1c concentrations. Future studies should also report the number of subjects who shift from abnormal to within-normal ranges for all endpoints.
The 21 randomized clinical trials either summarized in this review or assessed in this meta-analysis, in which both healthy subjects and persons with type 2 diabetes received between 10.8 and 1000 µg Cr/d for 28 d to 16 mo, reported no evidence of any toxic effect. However, the studies included in this review contribute only an estimated 220 person-years of data (both treated and control groups) and <35 person-years in subjects who received high doses of chromium (800 µg). Several case series in the literature have reported chromium toxicity (15). Thus, future studies of chromium supplementation should establish the long-term safety of chromium therapy, particularly at high doses.
This review summarized primarily small, underpowered randomized controlled trials of diverse populations. We reviewed or analytically combined studies of subjects of different nationalities (6, 23), one of which was non-Western (6). Some subjects were elderly (1, 4, 17, 18, 21), were overweight (7, 8, 14, 19), were athletes (5, 11), were participating in exercise (5, 7, 11, 13, 14) or dietary interventions (13 and H Lukaski, unpublished observations, 2000), had mild forms of diabetes (1), or had atherosclerotic disease (3). Only one of the studies of nondiabetic subjects (13) and no trial of diabetic subjects focused on a large subgroup of older African American women, a segment of the population with a high prevalence of diabetes (46).
It is difficult to recommend whether to launch a large randomized clinical trial in the face of so many unanswered questions. Who is likely to benefit? Currently no biomarker of chromium status exists. What is an efficacious but safe dose? Very few person-years of data are available. Which formulation is best? No single trial has addressed this question. If chromium were a new drug on the pathway to Food and Drug Administration approval, we would recommend a phase II clinical trial, that is, a smaller study with more than one dose or formulation arm. Because the study by Anderson et al (6) has been interpreted as showing some efficacy in diabetic subjects, manufacturers are extensively advertising the benefits of chromium and the general public uses chromium widely. Thus, it is important to either confirm or refute these findings soon with a well-designed study conducted in North America.
Dietary chromium supplements are inexpensive (26), and the limited safety data suggest that chromium is safe even at high doses (6). Thus, if dietary chromium supplementation is efficacious, it would be an attractive option for management of diabetes and for control of insulin and glucose concentrations of persons at high risk of type 2 diabetes. To date, data from randomized clinical trials are sparse and inconclusive. Placebo-controlled randomized clinical trials in well-characterized, at-risk populations are necessary to determine the effects of chromium on concentrations of glucose, insulin, and Hb A1c.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. E. Woods, V. Ghodsi, A. Engel, J. Miller, and S. James Serum Chromium and Gestational Diabetes J Am Board Fam Med, March 1, 2008; 21(2): 153 - 157. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 American Diabetes Association Nutrition Recommendations and Interventions for Diabetes: A position statement of the American Diabetes Association Diabetes Care, January 1, 2008; 31(Supplement_1): S61 - S78. [Full Text] [PDF]
|
|
|
|
|
|
N. Kleefstra, S. T. Houweling, and H. J.G. Bilo Effect of Chromium Supplementation on Glucose Metabolism and Lipids: A Systematic Review of Randomized Controlled Trials: Response to Balk et al. Diabetes Care, September 1, 2007; 30(9): e102 - e102. [Full Text] [PDF]
|
|
|
|
|
|
N. Kleefstra, S. T. Houweling, S. J.L. Bakker, S. Verhoeven, R. O.B. Gans, B. Meyboom-de Jong, and H. J.G. Bilo Chromium Treatment Has No Effect in Patients With Type 2 Diabetes in a Western Population: A randomized, double-blind, placebo-controlled trial Diabetes Care, May 1, 2007; 30(5): 1092 - 1096. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
American Diabetes Association Nutrition Recommendations and Interventions for Diabetes: A position statement of the American Diabetes Association Diabetes Care, January 1, 2007; 30(suppl_1): S48 - S65. [Full Text] [PDF]
|
|
|
|
 |
|
 C. Rink, S. Roy, S. Khanna, T. Rink, D. Bagchi, and C. K. Sen Transcriptome of the subcutaneous adipose tissue in response to oral supplementation of type 2 Leprdb obese diabetic mice with niacin-bound chromium Physiol Genomics, November 21, 2006; 27(3): 370 - 379. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
American Diabetes Association Nutrition Recommendations and Interventions for Diabetes-2006: A position statement of the American Diabetes Association. Diabetes Care, September 1, 2006; 29(9): 2140 - 2157. [Full Text] [PDF]
|
|
|
|
|
|
N. Kleefstra, S. T. Houweling, F. G.A. Jansman, K. H. Groenier, R. O.B. Gans, B. Meyboom-de Jong, S. J.L. Bakker, and H. J.G. Bilo Chromium Treatment Has No Effect in Patients With Poorly Controlled, Insulin-Treated Type 2 Diabetes in an Obese Western Population: A randomized, double-blind, placebo-controlled trial. Diabetes Care, March 1, 2006; 29(3): 521 - 525. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
American Diabetes Association Standards of Medical Care in Diabetes-2006 Diabetes Care, January 1, 2006; 29(suppl_1): S4 - S42. [Full Text] [PDF]
|
|
|
|
|
|
K. Yamaoka and T. Tango Efficacy of Lifestyle Education to Prevent Type 2 Diabetes: A meta-analysis of randomized controlled trials Diabetes Care, November 1, 2005; 28(11): 2780 - 2786. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Guallar, F. J. Jimenez, P. van 't Veer, P. Bode, R. A. Riemersma, J. Gomez-Aracena, J. D. Kark, L. Arab, F. J. Kok, J. M. Martin-Moreno, et al. Low Toenail Chromium Concentration and Increased Risk of Nonfatal Myocardial Infarction Am. J. Epidemiol., July 15, 2005; 162(2): 157 - 164. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Gunton, N. W. Cheung, R. Hitchman, G. Hams, C. Sullivan, K. Foster-Powell, and A. McElduff Chromium Supplementation Does Not Improve Glucose Tolerance, Insulin Sensitivity, or Lipid Profile: A Randomized, Placebo-Controlled, Double-Blind Trial of Supplementation in Subjects With Impaired Glucose Tolerance: Response to Komorowski and Juturu Diabetes Care, July 1, 2005; 28(7): 1842 - 1843. [Full Text] [PDF]
|
|
|
|
|
|
W. T. Cefalu and F. B. Hu Role of Chromium in Human Health and in Diabetes Diabetes Care, November 1, 2004; 27(11): 2741 - 2751. [Full Text] [PDF]
|
|
|
|
|
|
S. Rajpathak, E. B. Rimm, T. Li, J. S. Morris, M. J. Stampfer, W. C. Willett, and F. B. Hu Lower Toenail Chromium in Men With Diabetes and Cardiovascular Disease Compared With Healthy Men Diabetes Care, September 1, 2004; 27(9): 2211 - 2216. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Juturu and J. R Komorowski Chromium supplements, glucose, and insulin responses Am. J. Clinical Nutrition, July 1, 2003; 78(1): 190 - 190. [Full Text] [PDF]
|
|
|
|
|
|
M. F McCarty Chromium meta-analysis Am. J. Clinical Nutrition, July 1, 2003; 78(1): 191 - 192. [Full Text] [PDF]
|
|
|
|
|
|
D. S Kalman Chromium picolinate and type 2 diabetes Am. J. Clinical Nutrition, July 1, 2003; 78(1): 192 - 192. [Full Text] [PDF]
|
|
|
|
|
|
M. D Althuis and J. T Wittes Reply to DS Kalman, MF McCarty, and V Juturu and JR Komorowski Am. J. Clinical Nutrition, July 1, 2003; 78(1): 192 - 193. [Full Text] [PDF]
|
|
|
|
|
|
G. Y. Yeh, D. M. Eisenberg, T. J. Kaptchuk, and R. S. Phillips Systematic Review of Herbs and Dietary Supplements for Glycemic Control in Diabetes Diabetes Care, April 1, 2003; 26(4): 1277 - 1294. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. D. D. Hepburn, J. Xiao, S. Bindom, J. B. Vincent, and J. O'Donnell Bioinorganic Chemistry Special Feature: Nutritional supplement chromium picolinate causes sterility and lethal mutations in Drosophilamelanogaster PNAS, April 1, 2003; 100(7): 3766 - 3771. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
, pooled mean difference. Error bars indicate 95% CIs.