Clinical Review & Education
Review | COMPARATIVE EFFECTIVENESS RESEARCH
Systematic Review of the Benefits and Risks of Metformin in Treating Obesity in Children Aged 18 Years and Younger Marian S. McDonagh, PharmD; Shelley Selph, MD; Alp Ozpinar, BS; Carolyn Foley, BA
IMPORTANCE Childhood obesity is an important public health problem with increasing preva-
Supplemental content at jamapediatrics.com
lence. Because treatment often has limited success, new approaches must be identified. OBJECTIVE To evaluate the effectiveness and safety of metformin for treating obesity in children aged 18 years and younger without a diagnosis of diabetes mellitus. EVIDENCE REVIEW We included randomized clinical trials identified through searches of MEDLINE, the Cochrane Library, and ClinicalTrials.gov. Our primary outcome measure was change in body mass index (BMI, calculated as weight in kilograms divided by height in meters squared). We assessed study quality, pooled data using a random-effects model, and performed subgroup and sensitivity analyses. FINDINGS Fourteen randomized clinical trials were eligible. For BMI, moderate-strength evidence indicated a reduction of −1.38 (95% CI, −1.93 to −0.82) from baseline compared with control at 6 months. A similar, if less dramatic, effect was observed in studies less than 6 months, but the pooled estimate from studies of 1 year of treatment was not statistically significant. Subgroup analyses indicated smaller, but significant, effects for those with baseline BMI below 35, those of Hispanic ethnicity, those with acanthosis nigricans, those who had tried and failed diet and exercise programs, and in studies with more girls or higher mean age (adolescents). Moderate-strength evidence indicated that with metformin, 26% reported a gastrointestinal event compared with 13% in control groups (relative risk, 2.05; 95% CI, 1.193.54), although there was no difference in discontinuations due to adverse events. No serious adverse events were reported. CONCLUSIONS AND RELEVANCE Metformin provides a statistically significant, but very modest reduction in BMI when combined with lifestyle interventions over the short term. A large trial is needed to determine the benefits to subgroups or impacts of confounders. In the context of other options for treating childhood obesity, metformin has not been shown to be clinically superior. JAMA Pediatr. 2014;168(2):178-184. doi:10.1001/jamapediatrics.2013.4200 Published online December 16, 2013.
C
hildhood obesity is one of the most prevalent and challenging health care concerns in the United States, with 16.9% of children found to be obese (>95th percentile of body mass index [BMI, calculated as weight in kilograms divided by height in meters squared] for age1) in 2008.2,3 Nearly 1 in 3 children are considered overweight (BMI ⱖ85th percentile for age).2,3 The incidence of type 2 diabetes mellitus among overweight adolescents has increased dramatically in the past 20 years.4 Obese children are more likely to become obese adults, develop type 2 diabetes mellitus, and have cardiovascular disease risk factors such as elevated blood pressure and serum lipids.4-7 While diet and exercise are the first-line weight-loss methods used, few patients achieve success.5 Since 2003, orlistat (Xenical) has been approved by the US Food and Drug Administration to treat obesity in children ages 12 to 16 years. In severe obesity, bariatric surgery can result in 178
Corresponding Author: Marian S. McDonagh, PharmD, Department of Medical Informatics and Clinical Epidemiology, School of Medicine, Oregon Health & Science University, 3181 SW Sam Jackson Park Road, Mail Stop BICC, Portland, OR 97239 ([email protected]).
weight loss of 32 to 45 kg, but its use is limited to adults, except in extreme circumstances.6-8 While metformin is approved by the Food and Drug Administration for treating type 2 diabetes mellitus in adults and children older than 10 years of age, it has been used off label in recent years to treat childhood obesity.9-14 The most serious adverse effect associated with metformin is lactic acidosis, with kidney disease, heart failure, and alcoholism as known risk factors. Cases in children were not found in the literature. A previous systematic review15 of this topic included studies published through 2008. We aimed to update this review on the comparative benefits and risks of metformin used to treat childhood obesity. We posed the following critical questions: (1) Does metformin use in overweight or obese nondiabetic children reduce BMI, weight, lipids, blood pressure, or the risk for associ-
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
Author Affiliations: Department of Medical Informatics and Clinical Epidemiology, School of Medicine, Oregon Health & Science University, Portland, Oregon (McDonagh, Selph); School of Medicine, Oregon Health & Science University, Portland, Oregon (Ozpinar, Foley).
jamapediatrics.com
Metformin in Treating Obesity
Review Clinical Review & Education
ated comorbidities? (2) What are the adverse effects of metformin use in overweight or obese nondiabetic children? (3) Are results different across subpopulations of children (eg, comorbidities, races, or ethnicities)?
Figure 1. Results of Literature Search 54 Records identified from database searches after removal of duplicates
Methods
21 Additional records identified through other sources (ClinicalTrials.gov and reference lists)
75 Records screened
53 Records excluded at abstract level
22 Full-text articles assessed for eligibility
8 Full-text articles excluded 1 With ineligible outcome 4 With ineligible population 3 With ineligible study design
Inclusion Criteria Studies eligible for this review included nondiabetic children or adolescents (aged ⱕ18 years) who were overweight or obese (BMI >25 or had BMI for age ⱖ85th percentile). We included randomized clinical trials of metformin compared with any intervention. Benefit outcomes included change in BMI or BMI-for-age percentile; weight loss (percentage and absolute); change in weight category (obese, overweight, and normal); change in serum lipids, blood pressure, or quality of life; exercise tolerance; cardiovascular events (eg, stroke); and onset of type 2 diabetes mellitus. Adverse event outcomes included discontinuation from study owing to adverse events and the type and incidence of adverse events including lactic acidosis and alterations in height growth. This systematic review did not require institutional review board approval nor patient consent.
14 Randomized clinical trials included
ing of drug intervention, ITT analysis, and loss to follow-up. According to these methods, good studies meet all criteria, poor studies fail to meet combinations of criteria thought to constitute a fatal flaw, and the rest are fair quality. Disagreements were resolved through consensus.
Analysis Literature Search We conducted literature searches in MEDLINE (1996 to September 2012 including in-process and nonindexed citations), the Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov. We used the search term metformin combined with overweight or obese combined with child. Searches were limited to human and English language (owing to resource limitations). Manual searches of reference lists of included studies were conducted.
Study Selection Two reviewers independently screened titles and abstracts and then full-text publications by applying the inclusion criteria just described. Disagreements were resolved through consensus. We excluded studies of adolescent girls with polycystic ovary syndrome because it has been associated with weight gain and obesity and because many studies did not stratify results of teens from the women included.
Data Abstraction The following data were abstracted from each included study: study design, eligibility criteria (eg, BMI), baseline characteristics (eg, failure of prior weight-loss interventions and insulin resistance), intervention (eg, drug, dosage, and length of treatment), control intervention, other permitted concurrent interventions, methods of outcome assessment, population demographics, sample size, loss to follow-up, and results. We recorded intentionto-treat (ITT) results when available. A second reviewer checked abstracted data.
Quality Assessment Internal validity of included trials was independently graded as good, fair, or poor by 2 reviewers based on the methods of the Drug Effectiveness Review Project.16 Elements of internal validity included methods of randomization and allocation concealment, maskjamapediatrics.com
We analyzed results qualitatively and quantitatively. Our primary outcome measure was mean change in BMI. We also analyzed mean change in other continuous variables (eg, change in BMI and blood pressure) and relative risk for events (eg, rates of adverse events). Where we had at least 2 trials reporting the same outcome, we conducted meta-analyses using a Mantel Haenszel random-effects model using Stata (StataCorp). For continuous variables, we analyzed databases on those who had both baseline and follow-up values (ie, not ITT), but for categorical variables, we analyzed based on number randomized (ITT). Statistical heterogeneity was assessed using the Q and the I2 statistics. We analyzed publication bias visually using a funnel plot. We conducted sensitivity analyses to explore the effect of study quality, duration, dose, percentage male, country of study, and presence of acanthosis nigricans.
Strength of Evidence and Applicability We graded strength of the body of evidence for change in BMI, change in lipid and blood pressure parameters, and adverse events based on the GRADE method.17 This approach assesses 4 key domains: risk for bias, consistency, directness, and precision of the evidence. We separately reported the applicability of the body of evidence based on the descriptions of the populations, interventions, comparators, duration of study, and settings of the studies included.
Results Our searches identified 75 studies, from which we included 14 randomized clinical trials (Figure 1). 1,18-30 These small trials enrolled a total of 946 children and adolescents, ranging in age (mean) from 10 to 16 years, with baseline BMIs ranging from 26 to 41 (eTable in the Supplement). Only 3 studies used age-adjusted JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
179
Clinical Review & Education Review
Metformin in Treating Obesity
Figure 2. Change in Body Mass Index With Metformin Compared With Control by Length of Follow-up Effect Size (95% CI)
Source 6 mo follow-up Lavine et al,23 2011
–0.60 (–1.66 to 0.46)
Wilson et al,29 2010
–1.10 (–2.49 to 0.29)
Subtotal (I 2 = 0.0%, P = .58)
–0.79 (–1.63 to 0.06)
Overall (I 2 = 60.0%, P = .003)
–1.16 (–1.60 to –0.73) –5
–1
0
1
5
Mean Change in Body Mass Index
BMI of greater than the 95th percentile for inclusion.25,26,29 Of these, 1 study25 reported that at baseline, the participants were in the 98th percentile and had a mean BMI of 33. Ten studies documented insulin resistance at baseline (4 did not report on this characteristic). Two studies required that patients be normoglycemic or nondiabetic at baseline to enroll,18,29 and 8 reported participants were normoglycemic at baseline (6 did not report fasting glucose levels at baseline). Only 2 studies required that patients have tried and failed prior weight-loss efforts (eg, diet and exercise program) prior to enrolling in the study.26,28 Race and ethnicity distribution was not reported in a consistent manner across the studies, with 3 not reporting these data at all: 1 in Iran, 1 in Turkey, and 1 in Mexico.1,20,26 Three studies reported enrolling more than 90% white children,19,22,28 while the remainder reported a more mixed population including a study from Australia, where 64% were ethnically Indian subcontinent or Pacific Islanders. The balance of males and females ranged from 32% to 81% male across the studies, with a mean of 45% male. Three trials were determined to be good quality,23,29,30 2 were poor quality,19,25 and the remainder were fair quality. The studies that were rated poor quality were open label (everyone involved was unblinded including those making assessments of outcomes), were unclear on the methods of randomization and allocation concealment, and either did not use an ITT analysis or had potentially important differences between groups at baseline. The other 12 studies used a placebo control. The study durations varied, with 4 being of short duration (6
−0.79 (−1.63 to 0.06)
0
1000
−1.09 (−2.50 to 0.32)
64
>1000 to 35
−1.23 (−1.66 to −0.79)
0
≤12
−1.56 (−2.29 to −0.84)
58
13-14
−0.61 (−0.71 to −0.50)
0
>14
−0.96 (−1.59 to −0.32)
0
–6
–4
–2
0
2
4
Effect
Baseline BMI
Age, y
Intensity of control intervention Low
−1.06 (−1.60 to −0.51)
0
Moderate
−1.33 (−1.99 to −0.67)
72
High
−0.85 (−2.31 to 0.61)
18
Country of study United States
−1.01 (−1.44 to −0.58)
0
Outside United States
−1.35 (−2.26 to −0.45)
80
Failed diet/exercise Yes
−0.60 (−0.66 to −0.54)
0
No
−1.34 (−1.79 to −0.90)
33
Acanthosis nigricans Yes
−0.99 (−1.52 to −0.46)
0
No
−1.34 (−1.96 to −0.72)
73
Abbreviation: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared).
blinded follow-up for another 48 weeks, and both groups regressed toward baseline.29 Subgroup analyses (Table) showed greater differences in BMI reduction in patients whose baseline BMI was 35 or more (−1.23; 95% CI, −1.66 to −0.79) than for those with BMI less than 35 (−1.05; 95% CI, −1.81 to −0.29), although both were statistically significant. Studies with greater proportions of girls had smaller decreases in BMI relative to studies with more boys, but both subgroup analyses were statistically significant. While the study reports did not allow an analysis based on pubescent / prepubescent status, analysis based on mean age indicated that younger children (aged ⱕ12 years; −1.56; 95% CI, −2.29 to −0.84) had greater reduction in BMI compared with adolescents (aged >14 years; − 0.96; 95% CI, −1.59 to −0.32). Analysis of metformin dose showed only a small difference in effect with 2000 mg per day compared with lower doses. Analysis of studies including children with insulin resistance or insulin sensitivity or those who jamapediatrics.com
were known to be normoglycemic did not result in any clear differences in effect on BMI. This was most likely because the comparison groups were studies that did not report these characteristics rather than studies of children without them. Children with acanthosis nigricans had smaller, but significant, decreases in BMI. Similarly, analyzing 2 studies of children who failed diet and exercise prior to enrollment resulted in a smaller decrease in BMI compared with 10 studies that did not have this criterion. Studies conducted only in the United States found a smaller mean change in BMI than studies conducted outside the United States, although this analysis resulted in a large amount of statistical heterogeneity. Sensitivity analysis removing the 2 studies rated poor quality changed the overall estimate of effect from −1.16 to −1.09, but it remained statistically significant (95% CI, −1.54 to −0.63). We conducted an exploratory metaregression analysis of subgroup characteristics potentially associated with change in BMI. While there were too few studies to provide statistically significant results, our findings suggested that higher proportions of patients with Hispanic ethnicity or a history of failing lifestyle intervention in the past should be examined more closely in future studies as potential predictors of smaller change in BMI. The funnel plot of our primary outcome measure (difference in change in BMI) indicates a potential for publication bias, given the gap in studies on the side of no effect/favoring placebo at the top of the plot (Figure 3).
Weight Moderate-strength evidence, based on 8 studies, showed a statistically significant greater weight loss with metformin compared with control (−3.26; 95% CI, −4.23 to −2.30; I2, 0%).1,18,20,22,23,25,27,30 The largest difference in weight loss was seen at 6 months (−3.77; 95% CI, −5.03 to −2.51), while a year-long study found no significant difference between groups (0.70; 95% CI, −3.62 to 5.02; P = .75).23 The difference between groups was inversely associated with the intensity of the control intervention (low, medium, and high) but statistically significant for all 3 subgroups (data not shown). Similarly, greater differences in weight loss were found for patients whose BMI was 35 or more (−3.45; 95% CI, −4.23 to −2.30) than for those whose baseline BMI was less than 35 (−2.98; 95% CI, −4.61 to −1.36), although both were statistically significant. Stratifying by the percentJAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
181
Clinical Review & Education Review
Metformin in Treating Obesity
Figure 4. Change in Total Cholesterol After Metformin Treatment Effect Size (95% CI)
Study Favors Metformin Favors Control Atabek and Pirgon,1 2008
–3.10 (–12.70 to 6.50)
Bugert et al, 2008
–8.60 (–18.03 to 0.83)
Freemark and Bursey,21 2001
–0.30 (–23.00 to 22.40)
18
Kay et al, 2001 22
–18.00 (–34.86 to –1.14)
Lavine et al,23 2011
0.80 (–10.34 to 11.94)
Wiegand et al, 2010 28
0.00 (–15.18 to 15.18)
Yanovski et al,30 2011
–4.54 (–12.54 to 3.46)
Overall (I2 = 0.0%, P = .60)
–4.65 (–8.90 to –0.41) –35
–5
0
5
25
Mean Change in Total Cholesterol, mg/dL
age male or by metformin dose did not indicate important differences across groups.
Lipids and Blood Pressure Pooled analysis of 7 trials showed slightly greater decrease in total cholesterol with metformin compared with control (−4.65 mg/dL; 95% CI, −8.90 to −0.41; I2, 0%; Figure 4).1,18,21,23,25,28,30 A small but significantly greater change in triglycerides was found with metformin (−17.42 mg/dL; 95% CI, −33.68 to −1.16); however, owing to statistical heterogeneity (I2, 81%), we rated this as low strength of evidence.1,18,19,21-23,25,28-30 Other lipid outcomes were not different between groups. Only 4 studies reported changes in blood pressure, but these changes were inconsistent across the studies, and pooled analyses did not result in statistically significant differences.1,18,25,28
Adverse Events and Discontinuations In these studies, metformin was relatively well tolerated, with discontinuations owing to adverse events occurring in 1.6% of metformin patients overall and in 1.7% of control group patients. Evidence on the specific risks of metformin relative to placebo treatment was poorly reported in these studies. Ten of 14 reported limited details on specific adverse events stated during the trials, but none adequately reported how these events were ascertained. In these studies, there were no serious adverse events related to treatment and no cases of lactic acidosis reported. Likewise, alterations in height growth were not reported. The most commonly reported adverse events were gastrointestinal in nature. Moderate-strength evidence indicated that in metformin groups, 26% of patients reported some type of gastrointestinal event compared with 13% of those in the control groups (relative risk, 2.05; 95% CI, 1.19 to 3.54; I 2 , 35%). The types, frequency, and severity of events reported varied, and it was not clear that adverse events were ascertained equally in control group patients in all studies. Several studies noted that gastrointestinal adverse events generally resolved with time or dose reduction (often followed by slower dose increase). Other adverse events, such as increases in liver function tests, were reported inconsistently, but among those studies that did report values, differences between groups were not seen. 182
Discussion During the past decade, there has been interest in the potential benefits of metformin in obese children and adolescents, as evidenced by 14 trials examining various metabolic outcomes. Our focus was on the potential benefit in a primary health outcome for these children: reduction in BMI. Because the studies were generally small, ranging from 24 to 173 participants, a pooled analysis is ideal for determining the effectiveness with adequate statistical power. Our analysis indicated that metformin combined with lifestyle interventions is efficacious in helping obese children aged 10 to 16 years reduce their BMI and weight as compared with lifestyle interventions alone in the short term. However, the magnitude of change was small relative to known reductions needed to impart long-term health benefits. Our pooled analysis indicated a maximum reduction in BMI of 1.4 relative to lifestyle interventions alone in studies with durations of 6 to 12 months. Given that the mean BMI at study outset was 33, this reduction was only 3.6% greater than lifestyle interventions (mean [SD] change from baseline in these groups was close to zero [0.15]), less than the 5% or 10% goals often cited as a marker for meaningful weight loss. Similarly, weight change was 3.77 kg in studies limited to 6 months. Studies that lasted less than 6 months also showed a benefit, but the maximum effect was seen after 6 months of treatment. As with many weight-loss strategies, this weight-reduction benefit did not appear to be sustained over longer periods because weight and BMI changes were not clearly different between groups in studies with treatment durations of 1 year, although there were only 2 such studies (N = 77 and N = 173). Evaluating treatments for obesity in children is complex and must consider the impact of normal growth and the effects of puberty, age, and sex differences. Our analysis of subgroups indicated that the beneficial effects of metformin may be smaller in those whose baseline BMI was below 35, in studies with more girls or higher mean age (adolescents), in those of Hispanic ethnicity, in those with acanthosis nigricans, and in those who have tried and failed diet and exercise programs in the past. However, for some of
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
jamapediatrics.com
Metformin in Treating Obesity
Review Clinical Review & Education
these important subgroups, the ability to stratify more accurately would greatly improve the clarity of the findings. To complete such analysis, individual patient data or a very large trial with a diverse population would be necessary. For example, while our analysis by age indicated that younger children (mean age ⱕ12 years and more likely to be prepubertal) had a better reduction in BMI (−1.56) compared with adolescents (age ⱖ14 years, assumed to be pubescent, and BMI loss of −0.96), there is certainly contamination in our groupings, reducing the accuracy of the results. For other subgroups (eg, acanthosis nigricans), larger sample sizes are needed. The risk of giving children a medication for potentially long periods is concerning. In this case, metformin is a drug that has been used for multiple decades and has been used long term in children with type 2 diabetes mellitus. In these studies, it does not appear that children discontinued treatment owing to adverse effects. Gastrointestinal adverse effects accounted for most events reported, with twice as many patients reporting at least 1 event. While not well documented, several trial reports commented that these adverse effects resolved with time or dose reduction and titration. Potential limitations of our study included the lack of having searched for studies not published in English and having imputed as standard deviation those variance measures that were not specifically identified as standard deviations or standard errors. Limitations of the evidence included that the efficacy measures included were continuous variables, where 14% of randomized patients did not provide data for the analyses. If all of these patients experienced less benefit, then our results may be an overestimation of effect. As previously noted, while the populations included were of mixed sex, race, and ethnicity, sufficient informa-
ARTICLE INFORMATION Accepted for Publication: August 8, 2013. Published Online: December 16, 2013. doi:10.1001/jamapediatrics.2013.4200.
tion on effects in specific subgroups is mostly lacking. Finally, there is a suggestion of publication bias, where studies that did not show metformin to be beneficial may not have been published. Inclusion of more negative studies is likely to shift the point estimate toward the null. Searches of ClinicalTrials.gov, the Food and Drug Administration website, or general Internet searches did not identify unpublished studies.
Conclusions While our results indicate that some obese children and adolescents may benefit from short-term treatment with metformin combined with lifestyle interventions, these benefits were very modest, not achieving a 5% reduction in BMI based on mean response. Also, it was not clear that there was any benefit from longer-term treatment. Although these findings are based on statistically significant, moderate-strength evidence, the clinical benefit of such a small reduction in BMI is certainly questionable. Subgroup analyses suggest that there may be children who benefit more, for example, those with BMI greater than 35, age 12 years or younger, and who have not failed lifestyle interventions previously. Current evidence is insufficient to fully examine such subgroups or to fully account for some major potential confounding factors (eg, puberty). To determine whether there are specific patients who may have a clinical, and not just statistical, benefit from treatment with metformin, a large scale trial is needed. Such a trial should have adequate statistical power to allow examination of potential confounders and subgroups who may benefit more than the average, adequate duration to further examine the optimal duration of treatment, and an appropriate control group.
7. Inge TH, Krebs NF, Garcia VF, et al. Bariatric surgery for severely overweight adolescents: concerns and recommendations. Pediatrics. 2004;114(1):217-223.
and was paid by the university for her editing assistance as part of her regular salary. REFERENCES
Author Contributions: Dr McDonagh had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: McDonagh, Ozpinar, Foley. Acquisition of data: McDonagh, Ozpinar, Foley. Analysis and interpretation of data: McDonagh, Selph, Ozpinar. Drafting of the manuscript: McDonagh, Ozpinar, Foley. Critical revision of the manuscript for important intellectual content: McDonagh, Selph, Ozpinar. Statistical analysis: McDonagh, Selph, Ozpinar. Administrative, technical, or material support: McDonagh, Foley. Study supervision: McDonagh. Conflict of Interest Disclosures: None reported. Additional Contributions: We thank David Do, BS, Gabriel Edwards, BA, Leah Fletchall, BS, Erika Sohlberg, BA, Nicole Stanley, BS, and Dallas Swanson, BS, for assistance with topic development and data collection. We also thank Leah Williams, BS, for assistance with editing and formatting of the manuscript. Ms Williams is affiliated with Oregon Health & Science University
1. Atabek ME, Pirgon O. Use of metformin in obese adolescents with hyperinsulinemia: a 6-month, randomized, double-blind, placebo-controlled clinical trial. J Pediatr Endocrinol Metab. 2008;21(4):339-348. 2. Ogden CL, Carroll MD, Curtin LR, Lamb MM, Flegal KM. Prevalence of high body mass index in US children and adolescents, 2007-2008. JAMA. 2010;303(3):242-249. 3. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity and trends in body mass index among US children and adolescents, 1999-2010. JAMA. 2012;307(5):483-490. 4. Rosenbloom AL, Joe JR, Young RS, Winter WE. Emerging epidemic of type 2 diabetes in youth. Diabetes Care. 1999;22(2):345-354. 5. Whitlock EP, O’Connor EA, Williams SB, Beil TL, Lutz KW. Effectiveness of weight management interventions in children: a targeted systematic review for the USPSTF. Pediatrics. 2010;125(2):e396-e418. 6. Gogakos A, Tzotzas TC, Krassas GE. Recent concepts of pharmacotherapy and bariatric surgery for childhood obesity: an overview. Pediatr Endocrinol Rev. 2009;7(2):3-14.
jamapediatrics.com
8. Tsai WS, Inge TH, Burd RS. Bariatric surgery in adolescents: recent national trends in use and in-hospital outcome. Arch Pediatr Adolesc Med. 2007;161(3):217-221. 9. Catoira N, Nagel M, Di Girolamo G, Gonzalez CD. Pharmacological treatment of obesity in children and adolescents: current status and perspectives. Expert Opin Pharmacother. 2010;11(18):2973-2983. 10. Dunican KC, Desilets AR, Montalbano JK. Pharmacotherapeutic options for overweight adolescents. Ann Pharmacother. 2007;41(9): 1445-1455. 11. Greydanus DE, Bricker LA, Feucht C. Pharmacotherapy for obese adolescents. Pediatr Clin North Am. 2011;58(1):139-153, xi. 12. Wald AB, Uli NK. Pharmacotherapy in pediatric obesity: current agents and future directions. Rev Endocr Metab Disord. 2009;10(3):205-214. 13. Hsia Y, Dawoud D, Sutcliffe AG, Viner RM, Kinra S, Wong ICK. Unlicensed use of metformin in children and adolescents in the UK. Br J Clin Pharmacol. 2012;73(1):135-139. 14. Kostev K, Richter H. Unlicensed use of metformin in children and adolescents in Germany and France. Br J Clin Pharmacol. 2012;73(2):307-308.
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
183
Clinical Review & Education Review
Metformin in Treating Obesity
15. Park MH, Kinra S, Ward KJ, White B, Viner RM. Metformin for obesity in children and adolescents: a systematic review. Diabetes Care. 2009;32(9):1743-1745. 16. McDonagh MS, Jonas DE, Gartlehner G, et al. Methods for the drug effectiveness review project. BMC Med Res Methodol. 2012;12(1):140. 17. Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines: 1, introduction: GRADE evidence profiles and summary of findings tables. J Clin Epidemiol. 2011;64(4):383-394. 18. Burgert TS, Duran EJ, Goldberg-Gell R, et al. Short-term metabolic and cardiovascular effects of metformin in markedly obese adolescents with normal glucose tolerance. Pediatr Diabetes. 2008;9(6):567-576. 19. Clarson CL, Mahmud FH, Baker JE, et al. Metformin in combination with structured lifestyle intervention improved body mass index in obese adolescents, but did not improve insulin resistance. Endocrine. 2009;36(1):141-146. 20. Evia-Viscarra ML, Rodea-Montero ER, Apolinar-Jiménez E, et al. The effects of metformin on inflammatory mediators in obese adolescents with insulin resistance: controlled randomized clinical trial. J Pediatr Endocrinol Metab. 2012;25(1-2):41-49.
184
21. Freemark M, Bursey D. The effects of metformin on body mass index and glucose tolerance in obese adolescents with fasting hyperinsulinemia and a family history of type 2 diabetes. Pediatrics. 2001;107(4):55.
masked, placebo-controlled clinical trial for controlling childhood obesity. World J Pediatr. 2010;6(4):317-322.
22. Kay JP, Alemzadeh R, Langley G, D’Angelo L, Smith P, Holshouser S. Beneficial effects of metformin in normoglycemic morbidly obese adolescents. Metabolism. 2001;50(12):1457-1461.
27. Srinivasan S, Ambler GR, Baur LA, et al. Randomized, controlled trial of metformin for obesity and insulin resistance in children and adolescents: improvement in body composition and fasting insulin. J Clin Endocrinol Metab. 2006;91(6):2074-2080.
23. Lavine JE, Schwimmer JB, Van Natta ML, et al; Nonalcoholic Steatohepatitis Clinical Research Network. Effect of vitamin E or metformin for treatment of nonalcoholic fatty liver disease in children and adolescents: the TONIC randomized controlled trial. JAMA. 2011;305(16):1659-1668.
28. Wiegand S, l’Allemand D, Hübel H, et al. Metformin and placebo therapy both improve weight management and fasting insulin in obese insulin-resistant adolescents: a prospective, placebo-controlled, randomized study. Eur J Endocrinol. 2010;163(4):585-592.
24. Love-Osborne K, Sheeder J, Zeitler P. Addition of metformin to a lifestyle modification program in adolescents with insulin resistance. J Pediatr. 2008;152(6):817-822.
29. Wilson DM, Abrams SH, Aye T, et al; Glaser Pediatric Research Network Obesity Study Group. Metformin extended release treatment of adolescent obesity: a 48-week randomized, double-blind, placebo-controlled trial with 48-week follow-up. Arch Pediatr Adolesc Med. 2010;164(2):116-123.
25. Mauras N, DelGiorno C, Hossain J, et al. Metformin use in children with obesity and normal glucose tolerance: effects on cardiovascular markers and intrahepatic fat. J Pediatr Endocrinol Metab. 2012;25(1-2):33-40. 26. Rezvanian H, Hashemipour M, Kelishadi R, Tavakoli N, Poursafa P. A randomized, triple
30. Yanovski JA, Krakoff J, Salaita CG, et al. Effects of metformin on body weight and body composition in obese insulin-resistant children: a randomized clinical trial. Diabetes. 2011;60(2):477-485.
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
jamapediatrics.com
Review | COMPARATIVE EFFECTIVENESS RESEARCH
Systematic Review of the Benefits and Risks of Metformin in Treating Obesity in Children Aged 18 Years and Younger Marian S. McDonagh, PharmD; Shelley Selph, MD; Alp Ozpinar, BS; Carolyn Foley, BA
IMPORTANCE Childhood obesity is an important public health problem with increasing preva-
Supplemental content at jamapediatrics.com
lence. Because treatment often has limited success, new approaches must be identified. OBJECTIVE To evaluate the effectiveness and safety of metformin for treating obesity in children aged 18 years and younger without a diagnosis of diabetes mellitus. EVIDENCE REVIEW We included randomized clinical trials identified through searches of MEDLINE, the Cochrane Library, and ClinicalTrials.gov. Our primary outcome measure was change in body mass index (BMI, calculated as weight in kilograms divided by height in meters squared). We assessed study quality, pooled data using a random-effects model, and performed subgroup and sensitivity analyses. FINDINGS Fourteen randomized clinical trials were eligible. For BMI, moderate-strength evidence indicated a reduction of −1.38 (95% CI, −1.93 to −0.82) from baseline compared with control at 6 months. A similar, if less dramatic, effect was observed in studies less than 6 months, but the pooled estimate from studies of 1 year of treatment was not statistically significant. Subgroup analyses indicated smaller, but significant, effects for those with baseline BMI below 35, those of Hispanic ethnicity, those with acanthosis nigricans, those who had tried and failed diet and exercise programs, and in studies with more girls or higher mean age (adolescents). Moderate-strength evidence indicated that with metformin, 26% reported a gastrointestinal event compared with 13% in control groups (relative risk, 2.05; 95% CI, 1.193.54), although there was no difference in discontinuations due to adverse events. No serious adverse events were reported. CONCLUSIONS AND RELEVANCE Metformin provides a statistically significant, but very modest reduction in BMI when combined with lifestyle interventions over the short term. A large trial is needed to determine the benefits to subgroups or impacts of confounders. In the context of other options for treating childhood obesity, metformin has not been shown to be clinically superior. JAMA Pediatr. 2014;168(2):178-184. doi:10.1001/jamapediatrics.2013.4200 Published online December 16, 2013.
C
hildhood obesity is one of the most prevalent and challenging health care concerns in the United States, with 16.9% of children found to be obese (>95th percentile of body mass index [BMI, calculated as weight in kilograms divided by height in meters squared] for age1) in 2008.2,3 Nearly 1 in 3 children are considered overweight (BMI ⱖ85th percentile for age).2,3 The incidence of type 2 diabetes mellitus among overweight adolescents has increased dramatically in the past 20 years.4 Obese children are more likely to become obese adults, develop type 2 diabetes mellitus, and have cardiovascular disease risk factors such as elevated blood pressure and serum lipids.4-7 While diet and exercise are the first-line weight-loss methods used, few patients achieve success.5 Since 2003, orlistat (Xenical) has been approved by the US Food and Drug Administration to treat obesity in children ages 12 to 16 years. In severe obesity, bariatric surgery can result in 178
Corresponding Author: Marian S. McDonagh, PharmD, Department of Medical Informatics and Clinical Epidemiology, School of Medicine, Oregon Health & Science University, 3181 SW Sam Jackson Park Road, Mail Stop BICC, Portland, OR 97239 ([email protected]).
weight loss of 32 to 45 kg, but its use is limited to adults, except in extreme circumstances.6-8 While metformin is approved by the Food and Drug Administration for treating type 2 diabetes mellitus in adults and children older than 10 years of age, it has been used off label in recent years to treat childhood obesity.9-14 The most serious adverse effect associated with metformin is lactic acidosis, with kidney disease, heart failure, and alcoholism as known risk factors. Cases in children were not found in the literature. A previous systematic review15 of this topic included studies published through 2008. We aimed to update this review on the comparative benefits and risks of metformin used to treat childhood obesity. We posed the following critical questions: (1) Does metformin use in overweight or obese nondiabetic children reduce BMI, weight, lipids, blood pressure, or the risk for associ-
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
Author Affiliations: Department of Medical Informatics and Clinical Epidemiology, School of Medicine, Oregon Health & Science University, Portland, Oregon (McDonagh, Selph); School of Medicine, Oregon Health & Science University, Portland, Oregon (Ozpinar, Foley).
jamapediatrics.com
Metformin in Treating Obesity
Review Clinical Review & Education
ated comorbidities? (2) What are the adverse effects of metformin use in overweight or obese nondiabetic children? (3) Are results different across subpopulations of children (eg, comorbidities, races, or ethnicities)?
Figure 1. Results of Literature Search 54 Records identified from database searches after removal of duplicates
Methods
21 Additional records identified through other sources (ClinicalTrials.gov and reference lists)
75 Records screened
53 Records excluded at abstract level
22 Full-text articles assessed for eligibility
8 Full-text articles excluded 1 With ineligible outcome 4 With ineligible population 3 With ineligible study design
Inclusion Criteria Studies eligible for this review included nondiabetic children or adolescents (aged ⱕ18 years) who were overweight or obese (BMI >25 or had BMI for age ⱖ85th percentile). We included randomized clinical trials of metformin compared with any intervention. Benefit outcomes included change in BMI or BMI-for-age percentile; weight loss (percentage and absolute); change in weight category (obese, overweight, and normal); change in serum lipids, blood pressure, or quality of life; exercise tolerance; cardiovascular events (eg, stroke); and onset of type 2 diabetes mellitus. Adverse event outcomes included discontinuation from study owing to adverse events and the type and incidence of adverse events including lactic acidosis and alterations in height growth. This systematic review did not require institutional review board approval nor patient consent.
14 Randomized clinical trials included
ing of drug intervention, ITT analysis, and loss to follow-up. According to these methods, good studies meet all criteria, poor studies fail to meet combinations of criteria thought to constitute a fatal flaw, and the rest are fair quality. Disagreements were resolved through consensus.
Analysis Literature Search We conducted literature searches in MEDLINE (1996 to September 2012 including in-process and nonindexed citations), the Cochrane Central Register of Controlled Trials, and ClinicalTrials.gov. We used the search term metformin combined with overweight or obese combined with child. Searches were limited to human and English language (owing to resource limitations). Manual searches of reference lists of included studies were conducted.
Study Selection Two reviewers independently screened titles and abstracts and then full-text publications by applying the inclusion criteria just described. Disagreements were resolved through consensus. We excluded studies of adolescent girls with polycystic ovary syndrome because it has been associated with weight gain and obesity and because many studies did not stratify results of teens from the women included.
Data Abstraction The following data were abstracted from each included study: study design, eligibility criteria (eg, BMI), baseline characteristics (eg, failure of prior weight-loss interventions and insulin resistance), intervention (eg, drug, dosage, and length of treatment), control intervention, other permitted concurrent interventions, methods of outcome assessment, population demographics, sample size, loss to follow-up, and results. We recorded intentionto-treat (ITT) results when available. A second reviewer checked abstracted data.
Quality Assessment Internal validity of included trials was independently graded as good, fair, or poor by 2 reviewers based on the methods of the Drug Effectiveness Review Project.16 Elements of internal validity included methods of randomization and allocation concealment, maskjamapediatrics.com
We analyzed results qualitatively and quantitatively. Our primary outcome measure was mean change in BMI. We also analyzed mean change in other continuous variables (eg, change in BMI and blood pressure) and relative risk for events (eg, rates of adverse events). Where we had at least 2 trials reporting the same outcome, we conducted meta-analyses using a Mantel Haenszel random-effects model using Stata (StataCorp). For continuous variables, we analyzed databases on those who had both baseline and follow-up values (ie, not ITT), but for categorical variables, we analyzed based on number randomized (ITT). Statistical heterogeneity was assessed using the Q and the I2 statistics. We analyzed publication bias visually using a funnel plot. We conducted sensitivity analyses to explore the effect of study quality, duration, dose, percentage male, country of study, and presence of acanthosis nigricans.
Strength of Evidence and Applicability We graded strength of the body of evidence for change in BMI, change in lipid and blood pressure parameters, and adverse events based on the GRADE method.17 This approach assesses 4 key domains: risk for bias, consistency, directness, and precision of the evidence. We separately reported the applicability of the body of evidence based on the descriptions of the populations, interventions, comparators, duration of study, and settings of the studies included.
Results Our searches identified 75 studies, from which we included 14 randomized clinical trials (Figure 1). 1,18-30 These small trials enrolled a total of 946 children and adolescents, ranging in age (mean) from 10 to 16 years, with baseline BMIs ranging from 26 to 41 (eTable in the Supplement). Only 3 studies used age-adjusted JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
179
Clinical Review & Education Review
Metformin in Treating Obesity
Figure 2. Change in Body Mass Index With Metformin Compared With Control by Length of Follow-up Effect Size (95% CI)
Source 6 mo follow-up Lavine et al,23 2011
–0.60 (–1.66 to 0.46)
Wilson et al,29 2010
–1.10 (–2.49 to 0.29)
Subtotal (I 2 = 0.0%, P = .58)
–0.79 (–1.63 to 0.06)
Overall (I 2 = 60.0%, P = .003)
–1.16 (–1.60 to –0.73) –5
–1
0
1
5
Mean Change in Body Mass Index
BMI of greater than the 95th percentile for inclusion.25,26,29 Of these, 1 study25 reported that at baseline, the participants were in the 98th percentile and had a mean BMI of 33. Ten studies documented insulin resistance at baseline (4 did not report on this characteristic). Two studies required that patients be normoglycemic or nondiabetic at baseline to enroll,18,29 and 8 reported participants were normoglycemic at baseline (6 did not report fasting glucose levels at baseline). Only 2 studies required that patients have tried and failed prior weight-loss efforts (eg, diet and exercise program) prior to enrolling in the study.26,28 Race and ethnicity distribution was not reported in a consistent manner across the studies, with 3 not reporting these data at all: 1 in Iran, 1 in Turkey, and 1 in Mexico.1,20,26 Three studies reported enrolling more than 90% white children,19,22,28 while the remainder reported a more mixed population including a study from Australia, where 64% were ethnically Indian subcontinent or Pacific Islanders. The balance of males and females ranged from 32% to 81% male across the studies, with a mean of 45% male. Three trials were determined to be good quality,23,29,30 2 were poor quality,19,25 and the remainder were fair quality. The studies that were rated poor quality were open label (everyone involved was unblinded including those making assessments of outcomes), were unclear on the methods of randomization and allocation concealment, and either did not use an ITT analysis or had potentially important differences between groups at baseline. The other 12 studies used a placebo control. The study durations varied, with 4 being of short duration (6
−0.79 (−1.63 to 0.06)
0
1000
−1.09 (−2.50 to 0.32)
64
>1000 to 35
−1.23 (−1.66 to −0.79)
0
≤12
−1.56 (−2.29 to −0.84)
58
13-14
−0.61 (−0.71 to −0.50)
0
>14
−0.96 (−1.59 to −0.32)
0
–6
–4
–2
0
2
4
Effect
Baseline BMI
Age, y
Intensity of control intervention Low
−1.06 (−1.60 to −0.51)
0
Moderate
−1.33 (−1.99 to −0.67)
72
High
−0.85 (−2.31 to 0.61)
18
Country of study United States
−1.01 (−1.44 to −0.58)
0
Outside United States
−1.35 (−2.26 to −0.45)
80
Failed diet/exercise Yes
−0.60 (−0.66 to −0.54)
0
No
−1.34 (−1.79 to −0.90)
33
Acanthosis nigricans Yes
−0.99 (−1.52 to −0.46)
0
No
−1.34 (−1.96 to −0.72)
73
Abbreviation: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared).
blinded follow-up for another 48 weeks, and both groups regressed toward baseline.29 Subgroup analyses (Table) showed greater differences in BMI reduction in patients whose baseline BMI was 35 or more (−1.23; 95% CI, −1.66 to −0.79) than for those with BMI less than 35 (−1.05; 95% CI, −1.81 to −0.29), although both were statistically significant. Studies with greater proportions of girls had smaller decreases in BMI relative to studies with more boys, but both subgroup analyses were statistically significant. While the study reports did not allow an analysis based on pubescent / prepubescent status, analysis based on mean age indicated that younger children (aged ⱕ12 years; −1.56; 95% CI, −2.29 to −0.84) had greater reduction in BMI compared with adolescents (aged >14 years; − 0.96; 95% CI, −1.59 to −0.32). Analysis of metformin dose showed only a small difference in effect with 2000 mg per day compared with lower doses. Analysis of studies including children with insulin resistance or insulin sensitivity or those who jamapediatrics.com
were known to be normoglycemic did not result in any clear differences in effect on BMI. This was most likely because the comparison groups were studies that did not report these characteristics rather than studies of children without them. Children with acanthosis nigricans had smaller, but significant, decreases in BMI. Similarly, analyzing 2 studies of children who failed diet and exercise prior to enrollment resulted in a smaller decrease in BMI compared with 10 studies that did not have this criterion. Studies conducted only in the United States found a smaller mean change in BMI than studies conducted outside the United States, although this analysis resulted in a large amount of statistical heterogeneity. Sensitivity analysis removing the 2 studies rated poor quality changed the overall estimate of effect from −1.16 to −1.09, but it remained statistically significant (95% CI, −1.54 to −0.63). We conducted an exploratory metaregression analysis of subgroup characteristics potentially associated with change in BMI. While there were too few studies to provide statistically significant results, our findings suggested that higher proportions of patients with Hispanic ethnicity or a history of failing lifestyle intervention in the past should be examined more closely in future studies as potential predictors of smaller change in BMI. The funnel plot of our primary outcome measure (difference in change in BMI) indicates a potential for publication bias, given the gap in studies on the side of no effect/favoring placebo at the top of the plot (Figure 3).
Weight Moderate-strength evidence, based on 8 studies, showed a statistically significant greater weight loss with metformin compared with control (−3.26; 95% CI, −4.23 to −2.30; I2, 0%).1,18,20,22,23,25,27,30 The largest difference in weight loss was seen at 6 months (−3.77; 95% CI, −5.03 to −2.51), while a year-long study found no significant difference between groups (0.70; 95% CI, −3.62 to 5.02; P = .75).23 The difference between groups was inversely associated with the intensity of the control intervention (low, medium, and high) but statistically significant for all 3 subgroups (data not shown). Similarly, greater differences in weight loss were found for patients whose BMI was 35 or more (−3.45; 95% CI, −4.23 to −2.30) than for those whose baseline BMI was less than 35 (−2.98; 95% CI, −4.61 to −1.36), although both were statistically significant. Stratifying by the percentJAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
181
Clinical Review & Education Review
Metformin in Treating Obesity
Figure 4. Change in Total Cholesterol After Metformin Treatment Effect Size (95% CI)
Study Favors Metformin Favors Control Atabek and Pirgon,1 2008
–3.10 (–12.70 to 6.50)
Bugert et al, 2008
–8.60 (–18.03 to 0.83)
Freemark and Bursey,21 2001
–0.30 (–23.00 to 22.40)
18
Kay et al, 2001 22
–18.00 (–34.86 to –1.14)
Lavine et al,23 2011
0.80 (–10.34 to 11.94)
Wiegand et al, 2010 28
0.00 (–15.18 to 15.18)
Yanovski et al,30 2011
–4.54 (–12.54 to 3.46)
Overall (I2 = 0.0%, P = .60)
–4.65 (–8.90 to –0.41) –35
–5
0
5
25
Mean Change in Total Cholesterol, mg/dL
age male or by metformin dose did not indicate important differences across groups.
Lipids and Blood Pressure Pooled analysis of 7 trials showed slightly greater decrease in total cholesterol with metformin compared with control (−4.65 mg/dL; 95% CI, −8.90 to −0.41; I2, 0%; Figure 4).1,18,21,23,25,28,30 A small but significantly greater change in triglycerides was found with metformin (−17.42 mg/dL; 95% CI, −33.68 to −1.16); however, owing to statistical heterogeneity (I2, 81%), we rated this as low strength of evidence.1,18,19,21-23,25,28-30 Other lipid outcomes were not different between groups. Only 4 studies reported changes in blood pressure, but these changes were inconsistent across the studies, and pooled analyses did not result in statistically significant differences.1,18,25,28
Adverse Events and Discontinuations In these studies, metformin was relatively well tolerated, with discontinuations owing to adverse events occurring in 1.6% of metformin patients overall and in 1.7% of control group patients. Evidence on the specific risks of metformin relative to placebo treatment was poorly reported in these studies. Ten of 14 reported limited details on specific adverse events stated during the trials, but none adequately reported how these events were ascertained. In these studies, there were no serious adverse events related to treatment and no cases of lactic acidosis reported. Likewise, alterations in height growth were not reported. The most commonly reported adverse events were gastrointestinal in nature. Moderate-strength evidence indicated that in metformin groups, 26% of patients reported some type of gastrointestinal event compared with 13% of those in the control groups (relative risk, 2.05; 95% CI, 1.19 to 3.54; I 2 , 35%). The types, frequency, and severity of events reported varied, and it was not clear that adverse events were ascertained equally in control group patients in all studies. Several studies noted that gastrointestinal adverse events generally resolved with time or dose reduction (often followed by slower dose increase). Other adverse events, such as increases in liver function tests, were reported inconsistently, but among those studies that did report values, differences between groups were not seen. 182
Discussion During the past decade, there has been interest in the potential benefits of metformin in obese children and adolescents, as evidenced by 14 trials examining various metabolic outcomes. Our focus was on the potential benefit in a primary health outcome for these children: reduction in BMI. Because the studies were generally small, ranging from 24 to 173 participants, a pooled analysis is ideal for determining the effectiveness with adequate statistical power. Our analysis indicated that metformin combined with lifestyle interventions is efficacious in helping obese children aged 10 to 16 years reduce their BMI and weight as compared with lifestyle interventions alone in the short term. However, the magnitude of change was small relative to known reductions needed to impart long-term health benefits. Our pooled analysis indicated a maximum reduction in BMI of 1.4 relative to lifestyle interventions alone in studies with durations of 6 to 12 months. Given that the mean BMI at study outset was 33, this reduction was only 3.6% greater than lifestyle interventions (mean [SD] change from baseline in these groups was close to zero [0.15]), less than the 5% or 10% goals often cited as a marker for meaningful weight loss. Similarly, weight change was 3.77 kg in studies limited to 6 months. Studies that lasted less than 6 months also showed a benefit, but the maximum effect was seen after 6 months of treatment. As with many weight-loss strategies, this weight-reduction benefit did not appear to be sustained over longer periods because weight and BMI changes were not clearly different between groups in studies with treatment durations of 1 year, although there were only 2 such studies (N = 77 and N = 173). Evaluating treatments for obesity in children is complex and must consider the impact of normal growth and the effects of puberty, age, and sex differences. Our analysis of subgroups indicated that the beneficial effects of metformin may be smaller in those whose baseline BMI was below 35, in studies with more girls or higher mean age (adolescents), in those of Hispanic ethnicity, in those with acanthosis nigricans, and in those who have tried and failed diet and exercise programs in the past. However, for some of
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
jamapediatrics.com
Metformin in Treating Obesity
Review Clinical Review & Education
these important subgroups, the ability to stratify more accurately would greatly improve the clarity of the findings. To complete such analysis, individual patient data or a very large trial with a diverse population would be necessary. For example, while our analysis by age indicated that younger children (mean age ⱕ12 years and more likely to be prepubertal) had a better reduction in BMI (−1.56) compared with adolescents (age ⱖ14 years, assumed to be pubescent, and BMI loss of −0.96), there is certainly contamination in our groupings, reducing the accuracy of the results. For other subgroups (eg, acanthosis nigricans), larger sample sizes are needed. The risk of giving children a medication for potentially long periods is concerning. In this case, metformin is a drug that has been used for multiple decades and has been used long term in children with type 2 diabetes mellitus. In these studies, it does not appear that children discontinued treatment owing to adverse effects. Gastrointestinal adverse effects accounted for most events reported, with twice as many patients reporting at least 1 event. While not well documented, several trial reports commented that these adverse effects resolved with time or dose reduction and titration. Potential limitations of our study included the lack of having searched for studies not published in English and having imputed as standard deviation those variance measures that were not specifically identified as standard deviations or standard errors. Limitations of the evidence included that the efficacy measures included were continuous variables, where 14% of randomized patients did not provide data for the analyses. If all of these patients experienced less benefit, then our results may be an overestimation of effect. As previously noted, while the populations included were of mixed sex, race, and ethnicity, sufficient informa-
ARTICLE INFORMATION Accepted for Publication: August 8, 2013. Published Online: December 16, 2013. doi:10.1001/jamapediatrics.2013.4200.
tion on effects in specific subgroups is mostly lacking. Finally, there is a suggestion of publication bias, where studies that did not show metformin to be beneficial may not have been published. Inclusion of more negative studies is likely to shift the point estimate toward the null. Searches of ClinicalTrials.gov, the Food and Drug Administration website, or general Internet searches did not identify unpublished studies.
Conclusions While our results indicate that some obese children and adolescents may benefit from short-term treatment with metformin combined with lifestyle interventions, these benefits were very modest, not achieving a 5% reduction in BMI based on mean response. Also, it was not clear that there was any benefit from longer-term treatment. Although these findings are based on statistically significant, moderate-strength evidence, the clinical benefit of such a small reduction in BMI is certainly questionable. Subgroup analyses suggest that there may be children who benefit more, for example, those with BMI greater than 35, age 12 years or younger, and who have not failed lifestyle interventions previously. Current evidence is insufficient to fully examine such subgroups or to fully account for some major potential confounding factors (eg, puberty). To determine whether there are specific patients who may have a clinical, and not just statistical, benefit from treatment with metformin, a large scale trial is needed. Such a trial should have adequate statistical power to allow examination of potential confounders and subgroups who may benefit more than the average, adequate duration to further examine the optimal duration of treatment, and an appropriate control group.
7. Inge TH, Krebs NF, Garcia VF, et al. Bariatric surgery for severely overweight adolescents: concerns and recommendations. Pediatrics. 2004;114(1):217-223.
and was paid by the university for her editing assistance as part of her regular salary. REFERENCES
Author Contributions: Dr McDonagh had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: McDonagh, Ozpinar, Foley. Acquisition of data: McDonagh, Ozpinar, Foley. Analysis and interpretation of data: McDonagh, Selph, Ozpinar. Drafting of the manuscript: McDonagh, Ozpinar, Foley. Critical revision of the manuscript for important intellectual content: McDonagh, Selph, Ozpinar. Statistical analysis: McDonagh, Selph, Ozpinar. Administrative, technical, or material support: McDonagh, Foley. Study supervision: McDonagh. Conflict of Interest Disclosures: None reported. Additional Contributions: We thank David Do, BS, Gabriel Edwards, BA, Leah Fletchall, BS, Erika Sohlberg, BA, Nicole Stanley, BS, and Dallas Swanson, BS, for assistance with topic development and data collection. We also thank Leah Williams, BS, for assistance with editing and formatting of the manuscript. Ms Williams is affiliated with Oregon Health & Science University
1. Atabek ME, Pirgon O. Use of metformin in obese adolescents with hyperinsulinemia: a 6-month, randomized, double-blind, placebo-controlled clinical trial. J Pediatr Endocrinol Metab. 2008;21(4):339-348. 2. Ogden CL, Carroll MD, Curtin LR, Lamb MM, Flegal KM. Prevalence of high body mass index in US children and adolescents, 2007-2008. JAMA. 2010;303(3):242-249. 3. Ogden CL, Carroll MD, Kit BK, Flegal KM. Prevalence of obesity and trends in body mass index among US children and adolescents, 1999-2010. JAMA. 2012;307(5):483-490. 4. Rosenbloom AL, Joe JR, Young RS, Winter WE. Emerging epidemic of type 2 diabetes in youth. Diabetes Care. 1999;22(2):345-354. 5. Whitlock EP, O’Connor EA, Williams SB, Beil TL, Lutz KW. Effectiveness of weight management interventions in children: a targeted systematic review for the USPSTF. Pediatrics. 2010;125(2):e396-e418. 6. Gogakos A, Tzotzas TC, Krassas GE. Recent concepts of pharmacotherapy and bariatric surgery for childhood obesity: an overview. Pediatr Endocrinol Rev. 2009;7(2):3-14.
jamapediatrics.com
8. Tsai WS, Inge TH, Burd RS. Bariatric surgery in adolescents: recent national trends in use and in-hospital outcome. Arch Pediatr Adolesc Med. 2007;161(3):217-221. 9. Catoira N, Nagel M, Di Girolamo G, Gonzalez CD. Pharmacological treatment of obesity in children and adolescents: current status and perspectives. Expert Opin Pharmacother. 2010;11(18):2973-2983. 10. Dunican KC, Desilets AR, Montalbano JK. Pharmacotherapeutic options for overweight adolescents. Ann Pharmacother. 2007;41(9): 1445-1455. 11. Greydanus DE, Bricker LA, Feucht C. Pharmacotherapy for obese adolescents. Pediatr Clin North Am. 2011;58(1):139-153, xi. 12. Wald AB, Uli NK. Pharmacotherapy in pediatric obesity: current agents and future directions. Rev Endocr Metab Disord. 2009;10(3):205-214. 13. Hsia Y, Dawoud D, Sutcliffe AG, Viner RM, Kinra S, Wong ICK. Unlicensed use of metformin in children and adolescents in the UK. Br J Clin Pharmacol. 2012;73(1):135-139. 14. Kostev K, Richter H. Unlicensed use of metformin in children and adolescents in Germany and France. Br J Clin Pharmacol. 2012;73(2):307-308.
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
183
Clinical Review & Education Review
Metformin in Treating Obesity
15. Park MH, Kinra S, Ward KJ, White B, Viner RM. Metformin for obesity in children and adolescents: a systematic review. Diabetes Care. 2009;32(9):1743-1745. 16. McDonagh MS, Jonas DE, Gartlehner G, et al. Methods for the drug effectiveness review project. BMC Med Res Methodol. 2012;12(1):140. 17. Guyatt G, Oxman AD, Akl EA, et al. GRADE guidelines: 1, introduction: GRADE evidence profiles and summary of findings tables. J Clin Epidemiol. 2011;64(4):383-394. 18. Burgert TS, Duran EJ, Goldberg-Gell R, et al. Short-term metabolic and cardiovascular effects of metformin in markedly obese adolescents with normal glucose tolerance. Pediatr Diabetes. 2008;9(6):567-576. 19. Clarson CL, Mahmud FH, Baker JE, et al. Metformin in combination with structured lifestyle intervention improved body mass index in obese adolescents, but did not improve insulin resistance. Endocrine. 2009;36(1):141-146. 20. Evia-Viscarra ML, Rodea-Montero ER, Apolinar-Jiménez E, et al. The effects of metformin on inflammatory mediators in obese adolescents with insulin resistance: controlled randomized clinical trial. J Pediatr Endocrinol Metab. 2012;25(1-2):41-49.
184
21. Freemark M, Bursey D. The effects of metformin on body mass index and glucose tolerance in obese adolescents with fasting hyperinsulinemia and a family history of type 2 diabetes. Pediatrics. 2001;107(4):55.
masked, placebo-controlled clinical trial for controlling childhood obesity. World J Pediatr. 2010;6(4):317-322.
22. Kay JP, Alemzadeh R, Langley G, D’Angelo L, Smith P, Holshouser S. Beneficial effects of metformin in normoglycemic morbidly obese adolescents. Metabolism. 2001;50(12):1457-1461.
27. Srinivasan S, Ambler GR, Baur LA, et al. Randomized, controlled trial of metformin for obesity and insulin resistance in children and adolescents: improvement in body composition and fasting insulin. J Clin Endocrinol Metab. 2006;91(6):2074-2080.
23. Lavine JE, Schwimmer JB, Van Natta ML, et al; Nonalcoholic Steatohepatitis Clinical Research Network. Effect of vitamin E or metformin for treatment of nonalcoholic fatty liver disease in children and adolescents: the TONIC randomized controlled trial. JAMA. 2011;305(16):1659-1668.
28. Wiegand S, l’Allemand D, Hübel H, et al. Metformin and placebo therapy both improve weight management and fasting insulin in obese insulin-resistant adolescents: a prospective, placebo-controlled, randomized study. Eur J Endocrinol. 2010;163(4):585-592.
24. Love-Osborne K, Sheeder J, Zeitler P. Addition of metformin to a lifestyle modification program in adolescents with insulin resistance. J Pediatr. 2008;152(6):817-822.
29. Wilson DM, Abrams SH, Aye T, et al; Glaser Pediatric Research Network Obesity Study Group. Metformin extended release treatment of adolescent obesity: a 48-week randomized, double-blind, placebo-controlled trial with 48-week follow-up. Arch Pediatr Adolesc Med. 2010;164(2):116-123.
25. Mauras N, DelGiorno C, Hossain J, et al. Metformin use in children with obesity and normal glucose tolerance: effects on cardiovascular markers and intrahepatic fat. J Pediatr Endocrinol Metab. 2012;25(1-2):33-40. 26. Rezvanian H, Hashemipour M, Kelishadi R, Tavakoli N, Poursafa P. A randomized, triple
30. Yanovski JA, Krakoff J, Salaita CG, et al. Effects of metformin on body weight and body composition in obese insulin-resistant children: a randomized clinical trial. Diabetes. 2011;60(2):477-485.
JAMA Pediatrics February 2014 Volume 168, Number 2
Copyright 2014 American Medical Association. All rights reserved.
Downloaded From: by a University of Toledo Libraries User on 11/02/2017
jamapediatrics.com
