Efficacy and Safety of GLP-1 RAs in Children and Adolescents With Obesity or Type 2 Diabetes: A Systematic Review and Meta-Analysis.

Methods

Literature Search

A systematic search was conducted in PubMed, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL) to identify relevant RCTs published from inception until February 28, 2025. The search included studies evaluating the efficacy and safety of approved anti-obesity drugs in children and adolescents with obesity, overweight, prediabetes, or T2D. The search used controlled vocabulary (eg, MeSH in PubMed, Emtree in Embase) and free-text keywords related to obesity, T2D, children, adolescents, and GLP-1 RAs. Terms for obesity, T2D, age groups, GLP-1 RA components or brand names, and RCT filters were combined. Full strategies are provided in eTable 1 in Supplement 1.

We included RCTs of approved GLP-1 RAs in individuals younger than 18 years with obesity, overweight, prediabetes, or T2D assessing GLP-1 RAs alone or with lifestyle modification vs placebo. We excluded post hoc analyses and studies lacking both efficacy and safety outcomes. A study author (P.K.) deduplicated results before importing into Covidence, which helped minimize inconsistencies in study exclusion criteria.24,25

Data Screening and Extraction

Data screening and extraction were performed by 4 study authors (P.K., W.H., Y.Y.Y., and V.M.N.) Title and abstract screening of all search results was conducted independently by 2 reviewers using Covidence. Full-text screening of eligible studies was also done by 2 independent reviewers, with a third resolving any conflicts. Systematic reviews meeting the inclusion criteria were examined for relevant component studies.

A prespecified set of variables streamlined data extraction. Extracted data included first author, year, sample size, population, GLP-1 RA type, treatment duration, follow-up, country(ies), and key efficacy or safety measures. The efficacy measures included hemoglobin A1C (HbA1C, in percentage points); fasting plasma glucose (FPG, in milligrams per deciliter); weight outcomes (body weight [in kilograms], body mass index [BMI, calculated as weight in kilograms divided by height in meters squared], and BMI z scores); blood pressure, reported as estimated treatment difference with 95% confidence intervals; and lipid profile, reported using estimated treatment ratio (ETR) with 95% confidence intervals. Metabolic dysfunction–associated steatohepatitis or metabolic dysfunction–associated steatotic liver disease (MASH/MASLD) and OSA were assessed as exploratory efficacy outcomes.

Suicidal ideation and behavior were assessed prospectively in the RCTs using the Columbia-Suicide Severity Rating Scale (C-SSRS), in alignment with the US FDA guidance for industry.26 The guidance classifies this into 3 categories: suicidal ideation, suicidal behavior, and self-injury without intent. Ideation includes 5 severity levels; behavior includes preparatory actions, aborted or interrupted attempts, attempts, and completed suicide.The trials report these 3 groups together and will be hereafter be referred to as suicidal ideation or behaviors.

All safety outcomes, including those related to exploratory efficacy, such as MASH/MASLD and OSA, were classified according to the Medical Dictionary for Regulatory Activities (MedDRA) terminology by the RCTs and accounted for throughout each trial’s follow-up period. Safety measures were reported as total number of events in each group and included gastrointestinal adverse events, hepatobiliary disorders, infections, suicidal ideation or behaviors, depression, hypoglycemia, and discontinuation due to adverse events.

One reviewer compiled data into a standardized format, capturing study details, inclusion and exclusion criteria, follow-up duration, and reported outcomes. If outcome data were unavailable in the published article, data were retrieved from ClinicalTrials.gov. A second reviewer cross-checked and validated the extracted data, resolving discrepancies through consensus discussions.

Data Synthesis and Analysis

Fasting glucose concentrations in millimoles per liter were converted to milligrams per deciliter by multiplying by 18.0182.27 If standard deviations at follow-up were missing, they were imputed using reported baseline standard deviations.28 Effect estimates for the meta-analysis were based on the difference between treatment groups in change from baseline for continuous variables. In case of zero safety events, a 0.5 continuity correction was applied. Missing effect estimates or confidence intervals were imputed using available baseline, follow-up, or change from baseline data for both intervention and placebo groups following Cochrane guidelines.28 Pooled standard deviation and standard error were calculated using formulas from Cochrane guidelines and RevMan Web (Cochrane).28,29 The number of events was used to calculate rate ratios for safety events, as these were more commonly reported than the number of patients experiencing those events.30

All outcomes of GLP-1 RA vs placebo were summarized for the type 2 diabetes or prediabetes population, the obesity population, and overall.

All measures were assessed using random-effects inverse variance-weighted models that allow for heterogeneity among individual study effect estimates.30

We carried out a sensitivity analysis following the leave-one-out approach for the main efficacy outcomes of interest (ie, HbA1C, FPG, BMI, and BMI-SDS) to assess the robustness of our findings.

Risk of Bias Assessment

The risk of bias for RCTs was assessed using the Cochrane Risk of Bias in Randomized Trials version 2 (RoB2) tool by 2 reviewers.31 Each study’s risk of bias scores were compiled, and if consensus was not reached between the 2 reviewers, the final classification was determined by a third reviewer (P.K. or J.G.)

Results are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines.32 The study protocol is registered on PROSPERO (CRD42024583356).33 Statistical analyses were performed using Stata version 16.0 (StataCorp).

Results

Study Characteristics

A total of 18 (11 in obesity, 6 in T2D, and 1 in prediabetes) randomized placebo-controlled trials with 1402 participants (838 GLP-1 RA users and 564 placebo) investigating the efficacy and safety of GLP-1 RAs met the inclusion criteria (Figure 1). Across the 18 trials, the mean (range) age of included participants was 13.7 years (6–17), with 59.3% of participants being female (range, 31%−83.3%), 67.9% being White (range, 20.5%−93.2%), a median (IQR) treatment duration of 0.51 years (0.25–1.00), and a median (IQR) follow-up duration of 0.19 years (0.05–0.49). A total of 5, 15, and 6 trials reported the use of either dose-stabilized metformin or basal insulin at baseline, concomitant lifestyle intervention during the trial, and treatment adherence, respectively. The characteristics of the included trials are summarized in the Table. All trials (except Zhou and colleagues49 and Fox and colleagues51) were sponsored by pharmaceutical companies.34–51

The overall heterogeneity remained high across most outcomes when accounting for both populations (ie, T2D and obesity together) (Figures 2, 3, and 4).

Overall, of 18 trials most showed some concern for the risk of bias (8 trials), followed by low concerns (6 trials) and high concerns (4 trials). The categorization of bias by category for all included trials can be found in eTable 2 in Supplement 1.

Efficacy Outcomes

Glycemic Outcomes

GLP-1 RAs demonstrated significant reductions in HbA1C compared to placebo across trials. Overall, treatment with GLP-1 RAs resulted in an absolute reduction of HbA1C by −0.44% of absolute HbA1C (95% CI, −0.68% to −0.21%), with T2D trials showing a greater absolute reduction of −0.94% (95% CI, −1.25% to −0.62%) and obesity trials showing a comparatively smaller but significant absolute reduction in HbA1C (−0.10%; 95% CI, −0.18% to −0.02%).

The reduction was also notable in another glycemic parameter (ie, FPG) in the GLP-1 RA group, with significant reductions in the overall population (−9.92 mg/dL; 95% CI, −16.20 to −3.64) (to convert from milligrams per deciliter to millimoles per liter, multiply by 0.0555) and the T2D subgroup (−22.56 mg/dL; 95% CI, −35.33 to −9.78), whereas the obesity trials showed a nonsignificant reduction in FPG (−1.39 mg/dL; 95% CI, −3.10 to 0.31) (Figure 2; eFigure 1 in Supplement 1).

Weight-Related Outcomes

BMI reduced significantly in the overall population and the obesity subgroup. Overall, GLP-1 RA treatment resulted in a significant decrease in BMI by −1.45 (95% CI, −2.40 to −0.49), with the obesity trials showing a larger effect (−1.71; 95% CI, −2.84 to −0.58) compared to the T2D trials (−0.42; 95% CI, −1.10 to 0.27). BMI percentile also showed a similar trend, with the overall effect being −7.24% (95% CI, −12.97% to −1.51%) and the obesity trials exhibiting a greater reduction (−8.78%; 95% CI, −14.90% to −2.67%), whereas the T2D trials showed a nonsignificant reduction of −0.23% (95% CI, −0.76% to 0.30%). Similarly, BMI-SDS decreased by −0.20 (95% CI, −0.36 to −0.05) overall, with obesity trials demonstrating a more pronounced reduction (−0.26; 95% CI, −0.47 to −0.06) compared to T2D trials (−0.05; 95% CI, −0.13 to 0.04) (Figure 2; eFigure 2 in Supplement 1).

The reduction in waist circumference was significant in the GLP-1 RA group compared to placebo in obesity trials (−3.81 cm; 95% CI, −6.76 to −0.87), with no available data from T2D trials. In terms of body weight, the overall reduction was −3.02 kg (95% CI, −4.98 to −1.06), while the obesity trials saw a greater reduction of −4.72 kg (95% CI, −7.62 to −1.83). The T2D trials showed a nonsignificant reduction of −0.40 kg (95% CI, −1.10 to 0.31) (Figure 2; eFigure 3 in Supplement 1).

Lipid Outcomes

In terms of lipid profile, changes across trials were generally modest when analyzed using ETR. Overall, total cholesterol, low-density lipoprotein (LDL), and high-density lipoprotein (HDL) demonstrated minimal differences in the overall analysis (total cholesterol: 1.00; 95% CI, 0.98–1.02; LDL: 1.00; 95% CI, 0.99–1.04; HDL: 1.00; 95% CI, 0.97–1.04). The trends for total cholesterol, LDL, and HDL remained consistent in obesity and T2D trials. Triglyceride and very low–density lipoprotein levels showed a nonsignificant reduction in the GLP-1 RA group vs placebo, with overall ETRs of 0.95 (95% CI, 0.90–1.01) and 0.97 (95% CI, 0.90–1.04), respectively (Figure 3; eFigure 4 in Supplement 1).

Blood Pressure Outcomes

There was a significant overall reduction of systolic blood pressure (SBP) in the GLP-1 RA group by −2.73 mm Hg (95% CI, −4.04 to −1.43), with a slightly more pronounced effect in 9 obesity trials (−2.85 mm Hg; 95% CI, −4.32 to −1.38) compared to 2 T2D trials (−2.29 mm Hg; 95% CI, −5.14 to 0.56). Diastolic BP (DBP) showed nonsignificant reductions overall by −1.21 mm Hg (95% CI, −2.81 to 0.39), with a nonsignificant reduction in 9 obesity trials by −1.49 mm Hg (95% CI, −3.39 to 0.41). However, the effect in 1 T2D trial was minimal (−0.26 mm Hg; 95% CI, −3.11 to 2.59) (Figure 2; eFigure 5 in Supplement 1).

Exploratory Efficacy Outcomes

GLP-1 RAs demonstrated a reducing nonsignificant trend for OSA outcomes (ie, log[rate ratio] [logRR]): −0.70 (95% CI, −2.75 to 1.35). The information for this outcome was only available from 2 obesity trials. For MASH/MASLD, data were only available from 1 T2D trial, but with 3 dose periods (0.75 mg, 1.5 mg, and pooled), and we summarized the overall estimate, which showed no significant difference in rate between the GLP-1 RA group and placebo (logRR, 0.24; 95% CI, −1.80 to 2.29) (Figure 4; eFigure 6 in Supplement 1).

Safety Outcomes

Gastrointestinal Adverse Events, Hepatobiliary Disorders, and Infections

Gastrointestinal adverse events showed a significantly increasing trend with GLP-1 RA treatment vs placebo overall (logRR, 0.73; 95% CI, 0.38–1.07) across 14 RCTs. In T2D trials, no meaningful effect was observed (logRR, 0.31; 95% CI, −17 to 0.78). However, obesity trials showed a notable increase in gastrointestinal adverse events in the GLP-1 RA group compared to the placebo group (logRR, 1.02; 95% CI, 0.65–1.40) (Figure 4; eFigure 7 in Supplement 1).

GLP-1 RAs demonstrated a nonsignificant trend for hepatobiliary outcomes (logRR, 0.07; 95% CI, −0.58 to 0.72). For infections, the overall analysis across 9 RCTs suggested a possible increased rate (logRR, 0.05; 95% CI, −0.13 to 0.23), but the effect was not statistically significant. No significant difference was observed between the treatment groups in T2D trials (logRR, −0.03; 95% CI, −0.40 to 0.33), as well as in obesity trials (logRR, 0.26; 95% CI, −0.15 to 0.66) (Figure 4; eFigure8 in Supplement 1).

Psychological Outcomes

There was no significant difference in depression in the GLP-1 RA group compared to placebo, with an overall logRR of 0.02 (95% CI, −1.54 to 1.58). Similarly, no significant effect was observed in T2D trials (logRR, 0.27; 95% CI, −2.13 to 2.67) or obesity trials (logRR, −0.16; 95% CI, −2.20 to 1.89).

Only 5 of 18 trials mentioned the use of the C-SSRS at baseline and follow-up; it remains unclear if C-SSRS was used in other trials. GLP-1 RAs were associated with a trend toward a decreased rate of suicidal ideation or behaviors, with an overall logRR of −0.46 (95% CI, −1.42 to 0.49), although this was not statistically significant. Similar trends were observed in T2D trials (logRR, −0.62; 95% CI, −2.07 to 0.82) and obesity trials (logRR, −0.34; 95% CI, −1.62 to 0.94), but neither reached statistical significance (Figure 4; eFigure 7 in Supplement 1).

Hypoglycemia

GLP-1 RA therapy showed a trend toward increased rates of hypoglycemia across 7 RCTs overall (logRR, 0.51; 95% CI, −0.07 to 1.08), but the effect was not statistically significant. The risk of hypoglycemia showed a higher trend but remained nonsignificant in the GLP-1 RA group vs placebo in T2D trials (logRR, 0.40; 95% CI, −0.71 to 1.51) and obesity trials (logRR, 0.51; 95% CI, −0.07 to 1.08) (Figure 4; eFigure 7 in Supplement 1). Diene and colleagues37 and Tamborlane and colleagues38 reported 1 serious hypoglycemia adverse event each, 1 in the treatment group (due to age) and placebo group, respectively.

Treatment Discontinuation Due to Adverse Events

Discontinuation due to adverse events showed a trend toward an increased rate overall (logRR, 0.39; 95% CI, −0.37 to 1.14) across 7 RCTs, although this was not statistically significant. No notable effect was observed in T2D trials (logRR, −0.01; 95% CI, −1.13 to 1.10); however, obesity trials showed a nonsignificant increasing trend (logRR, 0.93; 95% CI, −0.38 to 1.14) (Figure 4; eFigure 7 in Supplement 1).

Leave-One-Out Sensitivity Analysis

We performed a leave-one-out sensitivity analysis by excluding the Roth and colleagues48 trial, as it included a small proportion of patients aged between 18 and 25 years.48 The results for the efficacy outcomes of HbA1C, FPG, BMI, and BMI-SDS remained consistent with the main analysis, with significant reductions in the GLP-1 RA group compared to placebo (eFigure 9 in Supplement 1).

Discussion

This extensive meta-analysis of 18 clinical trials evaluating the use of GLP-1 RAs in the pediatric population with diabetes or obesity offers a thorough synthesis of the benefits and risks associated with GLP-1 RAs in this underresearched group and may contribute to informed clinical decision-making.

This meta-analysis found that GLP-1 RAs significantly reduced HbA1C and FPG overall and in T2D trials, with smaller, nonsignificant effects in obesity trials. These findings are consistent with existing findings from limited reviews in this population, as well as findings from the adult population.21,52 Additionally, reduction in weight-related parameters, such as BMI, body weight, waist circumference, BMI percentile, and BMI-SDS, was observed overall, but was significant only in the obesity trials. This was consistent with the GLP-1 RA meta-analysis performed by Wong and colleagues53 in the adult population, which demonstrated greater reduction in patients without diabetes than in patients with diabetes. Overall, the lipid profile demonstrated nonsignificant increases by less than 0.1 ETR, which suggests limited clinical significance. The results were consistent with Rivera and colleagues’ meta-analysis54 comparing lipid profiles between the GLP-1 RA group and the placebo group in adults. The significant reduction in SBP and nonsignificant reduction in DBP across populations was similar to existing reviews in children and adults.22,55

We also observed a reducing trend in OSA in the GLP-1 RA group with obesity. Although this difference was not significant, possibly because of different doses, fewer trials, or heterogeneity among trials, the lowering trend warrants further research in this population. This could be of particular interest, as some studies in adults have shown benefits for OSA, leading to the first approval of GLP-1 RAs for OSA by the US FDA in December 2024.20,56

The risk of gastrointestinal adverse events remained significantly elevated in the GLP-1 RA group compared to placebo; this could be attributed to the mechanism of action of GLP-1 RAs, and the findings were consistent with existing literature.21 There was a nonsignificant increase in the rate of treatment discontinuation due to adverse events in the GLP-1 RA group. Hypoglycemia events showed an elevated nonsignificant risk.

It is noteworthy that FDA guidance for industry requires trials to document any suicide-related behaviors using C-SSRS at baseline, during the trial, and at follow-up.26 However, only 5 of 18 trials document using this in publicly available publications or on ClinicalTrials.gov.36,37,39,46,50 In the trials that assessed and published this safety outcome, the risk of suicide ideation or behaviors was not significantly different between the 2 groups. This is important given the growing concerns about the use of GLP-1 RAs and suicidal behaviors and the limited knowledge of this outcome in the pediatric population.12,13 However, continued surveillance is warranted through RCTs and real-world evidence. There was no significant difference in the risk of depression between the 2 groups. The nonsignificant findings of suicide ideation or behaviors and depression were consistent with a meta-analysis of GLP-1 RA use in adults by Tang and colleagues.57–59

To the best of our knowledge, this is the first meta-analysis capturing 18 trials in the pediatric population assessing GLP-1 RAs and assessing 23 outcomes—specifically, the risks of suicidal behaviors and depression—and exploring OSA and MASH/MASLD across available RCT data in this population.

Limitations

This meta-analysis also has limitations. First, outcomes like suicidal ideation or behaviors, depression, and MASLD/MASH were not primary outcomes in the included trials, possibly leading to underreporting, unclear classifications, or low precision. Second, we used aggregated data rather than individual participant data, limiting subgroup analysis (eg, by comedications) and confounding control. Third, variations in study design, populations, and outcome definitions may have introduced heterogeneity. Additionally, some data were extracted from ClinicalTrials.gov due to the lack of reported outcomes in published literature, potentially resulting in incomplete outcome capture. Some adverse events planned for extraction (eg, eating disorders) could not be extracted due to limited reporting. BMI-related efficacy by lifestyle intervention could not be assessed, as most trials reported concomitant lifestyle interventions at baseline or during the trial. Lastly, the variability in reporting of adherence and adverse events limited stratification by severity or evaluation of adherence effects. These factors may affect the robustness and generalizability of our findings.

Conclusions

In conclusion, this systematic review and meta-analysis provides a comprehensive evaluation of the benefits and risks associated with GLP-1 RAs in the pediatric population, enabling patients, caregivers, and clinicians to make more informed treatment decisions. However, to build on these findings, future clinical trials and real-world studies could assess treatment adherence with respect to patient preferences (eg, problems with self-management related to injectables or cost) and their effect on safety (severity of gastrointestinal and other adverse events) and efficacy/effectiveness outcomes. Moreover, longer follow-up from future RCTs and real-world studies is essential to establish the long-term effects of GLP-1 RAs in children and adolescents.

Supplementary Material