Dose-formulation risk gradients of gastrointestinal adverse events with subcutaneous and oral semaglutide in type 2 diabetes: a systematic review and dose-response network meta-analysis

Objective To quantify formulation- and dose-dependent risk gradients of gastrointestinal adverse events (GI AEs) with subcutaneous and oral semaglutide in adults with type 2 diabetes mellitus (T2DM) and assess follow-up duration as a potential moderator, to inform hypothesis-generating clinical considerations. Methods PubMed, Embase, and Web of Science were searched from inception to June 2025 for randomized controlled trials (RCTs). Risk of bias was assessed with the Cochrane RoB 2.0 tool. Descriptive single-arm random-effects pooling summarized within-dose GI AE proportions without comparative inference; all between-treatment comparisons relied on a frequentist network meta-analysis (NMA) generating odds ratios (ORs), with treatments ranked by the surface under the cumulative ranking curve (SUCRA). Meta-regression on follow-up duration was post-hoc and exploratory. The review was not prospectively registered. Results Twenty RCTs enrolling 18–014 participants were included. Pooled single-arm proportions were similar for subcutaneous semaglutide 0.5 mg (0.40, 95% CI 0.33–0.46) and 1.0 mg (0.42, 95% CI 0.37–0.46); follow-up duration was positively associated with logit-transformed 0.5 mg incidence (β = 0.0082 per week, 95% CI 0.0020–0.0143; 58.1% of heterogeneity explained). Pooled single-arm proportions for oral semaglutide were 0.25 (95% CI 0.11–0.46) at 3 mg, 0.42 (95% CI 0.26–0.61) at 7 mg, and 0.49 (95% CI 0.37–0.62) at 14 mg. NMA ranked subcutaneous semaglutide 2.0 mg (SUCRA = 90.1%), subcutaneous 1.0 mg (86.5%) and oral 14 mg (82.4%) highest for GI AE risk. Oral 14 mg did not differ significantly from subcutaneous 1.0 mg (OR = 0.96, 95% CI 0.67–1.36, P = 0.804), nor from oral 7 mg (OR = 1.28, 95% CI 0.88–1.84, P = 0.196), although clinically meaningful differences could not be excluded. Conclusion GI AE risk with semaglutide is formulation- and dose-dependent: tolerability was comparable across standard subcutaneous maintenance doses, whereas oral risk increased monotonically with dose, with 14 mg ranking among the highest. Clinical suggestions remain hypothesis-generating; interpretation is limited by under-representation of Asian participants and sparse 2.0 mg evidence.