Two GLP-1 pills went head to head. One won on blood sugar and lost on side effects.

Most of what is known about GLP-1 receptor agonists comes from trials against placebo or against older glucose-lowering classes. Direct comparisons between two drugs of this class are less common, and this one is unusual in that both arms took a pill.

The Lancet published on February 26 the results of ACHIEVE-3, which randomised adults with type 2 diabetes to orforglipron or to oral semaglutide and followed them for 52 weeks [s1].

Why the comparison is interesting

Orforglipron is described by the trial investigators as a novel non-peptide GLP-1 receptor agonist designed for daily oral administration without food or water restrictions [s1]. Oral semaglutide, its comparator here, is a peptide, and the trial's framing of orforglipron's dosing flexibility implies what the peptide formulation requires by contrast [s1].

That difference is the reason the trial matters beyond its endpoints. A non-peptide molecule that matches or beats a peptide on glycaemic control is a different proposition to manufacture and distribute.

The trial

ACHIEVE-3 was a 52-week, randomised, open-label, active-controlled, multicentre, multinational phase 3 study [s1]. It enrolled adults aged 18 or older with type 2 diabetes inadequately controlled on metformin at 1,500 mg per day or more, with HbA1c between 7.0% and 10.5% (53–91 mmol/mol) and BMI of at least 25 kg/m², from 131 medical research centres and hospitals in Argentina, China, Japan, Mexico, and the USA [s1].

Participants were randomised 1:1:1:1 to orforglipron 12 mg or 36 mg, or semaglutide 7 mg or 14 mg — all administered orally once per day, all with an up-to-four-week lead-in and a 52-week treatment period [s1].

The primary objective was non-inferiority of orforglipron 36 mg versus semaglutide 14 mg and of orforglipron 12 mg versus semaglutide 7 mg for mean change in HbA1c at week 52, with a non-inferiority margin of 0.3%, in the intention-to-treat population [s1]. Superiority testing was prespecified hierarchically, to be performed only after non-inferiority was met [s1].

From September 22, 2023 to August 22, 2025, 1,698 participants were recruited: 424 on orforglipron 12 mg, 423 on orforglipron 36 mg, 426 on semaglutide 7 mg, and 425 on semaglutide 14 mg [s1].

The efficacy result

From a baseline HbA1c of 8.3%, mean changes at week 52 under the treatment regimen estimand were [s1]:

• Orforglipron 12 mg: −1.71% (SE 0.07)
• Orforglipron 36 mg: −1.91% (SE 0.08)
• Semaglutide 7 mg: −1.23% (SE 0.05)
• Semaglutide 14 mg: −1.47% (SE 0.06)

Estimated treatment differences [s1]:

• Orforglipron 12 mg vs semaglutide 7 mg: −0.48% (95% CI, −0.65 to −0.31; p<0.0001)
• Orforglipron 36 mg vs semaglutide 14 mg: −0.44% (−0.62 to −0.26; p<0.0001)
• Orforglipron 12 mg vs semaglutide 14 mg: −0.24% (−0.41 to −0.072; p=0.0050)
• Orforglipron 36 mg vs semaglutide 7 mg: −0.68% (−0.85 to −0.50; p<0.0001)

Non-inferiority was met, and both orforglipron doses showed superiority to both semaglutide doses — including the cross-comparison of the lower orforglipron dose against the higher semaglutide dose [s1]. That last comparison is the one that carries the most information: 12 mg of orforglipron outperformed 14 mg of semaglutide, the maximum dose in the trial.

The tolerability result

The same drug that won on glycaemia lost on side effects.

Gastrointestinal adverse events, the class's signature problem, were reported in 249 (59%) of 424 on orforglipron 12 mg and 245 (58%) of 423 on orforglipron 36 mg, against 157 (37%) of 426 on semaglutide 7 mg and 193 (45%) of 425 on semaglutide 14 mg [s1]. Most were mild to moderate in severity [s1].

Discontinuation followed: 37 (9%) on orforglipron 12 mg and 41 (10%) on orforglipron 36 mg stopped treatment because of adverse events, against 19 (4%) on semaglutide 7 mg and 21 (5%) on semaglutide 14 mg [s1].

Mean pulse rate rose more on orforglipron — 3.7 bpm on 12 mg and 4.7 bpm on 36 mg, against 1.0 bpm on semaglutide 7 mg and 1.5 bpm on 14 mg [s1]. Heart-rate increase is a known GLP-1 class effect; what the trial establishes is that it is larger with this drug than with its oral comparator, and what it does not establish is whether that difference has any clinical consequence.

Four deaths occurred during the study: one in each orforglipron group and two in the semaglutide 7 mg group [s1].

The design caveats that matter

It was open-label. Nobody was blinded. For an objective laboratory endpoint like HbA1c that matters relatively little; for symptom reporting, adverse-event attribution, and the decision to discontinue, it matters considerably — and the tolerability comparison is precisely where expectations could bite.

Dose is not equivalence. Comparing 12 mg and 36 mg of one drug against 7 mg and 14 mg of another is comparing the doses the sponsor chose, not equipotent doses. A different dose pairing would produce a different balance of efficacy and tolerability.

Titration schedules are not reported in the abstract. GI tolerability in this class is heavily dependent on how slowly a dose is escalated.

The population is narrow. Adults with type 2 diabetes inadequately controlled on metformin, with BMI at least 25. Not people with obesity and no diabetes; not people already on other glucose-lowering agents.

The funder is the manufacturer of the winning drug. The trial was funded by Eli Lilly [s1].

The trade the data describe

Stripped down: at the doses tested, orforglipron lowered HbA1c by roughly 0.4 to 0.5 percentage points more than oral semaglutide, and produced gastrointestinal side effects in between 13 and 22 percentage points more participants, with roughly double the rate of discontinuation for adverse events [s1].

Whether that is a good trade is not a question a trial answers. It depends on how far a given patient's HbA1c is from target and how much nausea they will tolerate to get there.

An accompanying comment by Nauck and Horowitz, published alongside the trial, frames the emerging oral GLP-1 field as a competition for efficacy and tolerability [s2].

What to watch

Cardiovascular outcome data, weight-loss endpoints from the obesity programme rather than the diabetes programme, and whether a slower titration narrows the tolerability gap. The trial is registered on ClinicalTrials.gov as NCT06045221 and is completed [s1].

This article describes trial results, including doses and adverse events, for informational purposes only. It is not medical advice and not a recommendation about any medication.

Sources

• [s1] Rosenstock J, Yabe D, Cox D, et al; ACHIEVE-3 Investigators. Efficacy and safety of once-daily oral orforglipron compared with oral semaglutide in adults with type 2 diabetes (ACHIEVE-3). The Lancet, published online 2026-02-26.
• [s2] Nauck MA, Horowitz M. Oral GLP-1 receptor agonists: competition for efficacy and tolerability. The Lancet, published online 2026-02-26.