Response to Comment on Wong et al. Efficacy of GLP-1 Receptor Agonists on Weight Loss, BMI, and Waist Circumference for Patients With Obesity or Overweight: A Systematic Review, Meta-analysis, and Meta-regression of 47 Randomized Controlled Trials. Diabetes Care 2025;48:292–300 Free

We appreciate the insightful comments provided by Huang (1) on our study (2) and are grateful for the opportunity to further elaborate on the points raised. In this reply, we present the results of funnel plots, Egger test, and trim-and-fill analyses for the three study outcomes, mean differences in weight, BMI, and waist circumference, to address the author’s concerns about possible publication bias.

For the outcome of overall mean difference in weight, the funnel plot demonstrated asymmetry, which was consistent with Egger test (P = 0.002). There was no change in the direction or significance of results in the trim-and-fill analysis (−2.71 kg, 95% CI −3.58 to −1.83, I² = 97.9%), which imputed 13 studies. For the outcomes of overall mean difference in BMI and waist circumference, the funnel plots were relatively symmetrical, consistent with Egger test (P = 0.52 and P = 0.92, respectively). In addition, findings of the trim-and-fill analyses (−1.96 kg/m², 95% CI −2.42 to −1.50, I² = 96.6% and −4.55 cm, 95% CI −5.72 to −3.38, I² = 94.2%, respectively) were consistent with the primary analyses. These results indicate that publication bias may have influenced the overall mean difference in weight, but it is unlikely to have an impact on the results for BMI and waist circumference.

However, the high degree of heterogeneity for the mean difference in weight despite adjusting for potentially missing studies via trim and fill suggests that other factors beyond publication bias may be contributing to variations in effect size, as evident in the funnel plot. These may include 1) inclusion of studies that exclusively included individuals with overweight or obesity and those who did not explicitly define status of overweight or obesity but had a mean BMI within 1 SD of the standard criteria for overweight; 2) inclusion of individuals both with and without diabetes; and 3) differences in treatment duration/follow-up between studies. To explore potential sources of heterogeneity, we first conducted sensitivity analysis by subgrouping studies into those that exclusively included individuals with overweight or obesity and those that did not explicitly define a status of overweight or obesity but had a mean BMI within 1 SD of the standard criteria for overweight. We found that reductions in weight were significantly greater in the former subgroup. Second, we performed subgroup analysis for studies including only individuals with diabetes and studies including only individuals without diabetes, along with a meta-regression based on HbA_1c levels at baseline. We observed that weight-lowering efficacy was significantly greater in the studies involving only individuals without diabetes and that it declined with increasing baseline HbA_1c levels. Third, subgroup analysis and meta-regression were used to evaluate the effect of treatment duration or follow-up on weight-lowering efficacy. Our findings revealed that our analyses revealed that the longer the duration of follow-up or treatment, the greater the weight-lowering efficacy (2). Although we are unable to exclude the possibility of publication bias, it is likely that other factors, such as between-study heterogeneity, are contributing to the variations in effect size between studies, as illustrated by the funnel plots. Regardless, we would like to emphasize that our study provides estimates that are generalizable to a broad population. However, their application must be individualized, considering factors such as population differences, prescribing practices, patient compliance and tolerance, and concurrent lifestyle modifications, like exercise. The results of our extensive subgroup analyses and meta-regression, beyond the three aforementioned factors, are intended to assist the reader in making these considerations.

We thank Huang (1) for highlighting the potential impact of glucagon-like peptide 1 receptor agonists (GLP-1 RAs) on thyroid function, such as potentially increasing the risk of thyroid and medullary thyroid cancer (3). However, an assessment of the influence of GLP-1 RAs on the thyroid gland was not a focus of our study. We agree that further research is required to robustly elucidate its effects on GLP-1 RAs.