Give a teenager with obesity a GLP-1 drug, and which drug you pick seems to decide what gets better. Semaglutide moved the scale most. Dulaglutide moved blood sugar most. An older injection moved fasting glucose, the oldest drug of the bunch nudged blood pressure. That is the picture from a network meta-analysis published July 4 in Pharmacological Research ↗, which pooled 17 randomized trials and 1,230 adolescents. The problem is that almost none of these drugs were ever tested against each other in a teenager.
Start with the finding that stands up. Semaglutide ↗ at its 2.4 milligram weekly dose, the same dose sold as Wegovy, produced the largest weight loss by far: 18 kilograms, roughly 40 pounds, with the analysts rating their confidence in that number as high. It also led on body mass index, down 5.9 points, and waist circumference, down about 12 centimeters. On weight, in adolescents, semaglutide is the drug with the most evidence and the biggest effect, and this analysis does not muddy that.
Everything after that is softer. Dulaglutide ↗, the weekly diabetes injection Trulicity, topped the table for HbA1c, the three-month blood-sugar average, cutting it by 1.5 percentage points. Lixisenatide ↗, an older daily shot, led on fasting glucose. Exenatide ↗, the first GLP-1 drug ever approved, showed the largest drop in systolic blood pressure, close to 7 mmHg. No drug moved cholesterol or triglycerides in a way the statistics could tell apart from chance.
Read as a leaderboard, that is four different winners for four different measures, which is a tidy story and a misleading one. A network meta-analysis does not require that every drug be tested against every other. It builds a web: if one trial pits semaglutide against placebo and another pits dulaglutide against placebo, the math infers how the two drugs would compare, using placebo as the shared rung. That inference is called an indirect comparison, and it is only as good as the trials feeding it.
Here the web is thin. The authors say most of the active-drug comparisons were indirect, many rankings rested on a single trial, and the credible intervals, the Bayesian version of a margin of error, were wide. Semaglutide's weight result ran anywhere from 12 to 24 kilograms. Exenatide's blood-pressure edge came from early-phase studies and, by the authors' own note, limited indirect evidence. They put it plainly: the rankings are hypothesis-generating and should not be read as proof that one drug beats another.
That distinction matters more in children than in adults, because the pediatric evidence base is genuinely small. Two GLP-1 drugs, liraglutide and semaglutide, carry FDA and EMA approval for adolescents 12 and up; the rest are used off the strength of adult data and a handful of pediatric trials. We covered one of those trials, oral semaglutide's first pediatric type 2 diabetes readout ↗, earlier this year. A meta-analysis can make that sparse landscape look more settled than it is, arranging a few small trials into a confident-looking ranking. The value of this one is the opposite of a ranking: it shows that beyond semaglutide and weight, nobody has run the head-to-head studies that would tell a pediatrician which drug to reach for and why.
All four drugs pull the same lever, the GLP-1 receptor ↗, so the differences the analysis reports are unlikely to be about the target and more likely about dose, trial design, and who happened to be enrolled. Until adolescents are randomized head-to-head, the differential profile is a prompt for trials, not a guide for prescriptions.