CagriSema and REDEFINE: What Adding Amylin Agonism Bought Over Semaglutide Alone

Adding cagrilintide to semaglutide bought about five and a half percentage points of additional body-weight change over semaglutide alone at 68 weeks. In REDEFINE 1, under the treatment-policy estimand, mean body-weight change was 20.4% on the fixed combination, 14.9% on semaglutide alone, 11.5% on cagrilintide alone and 3.0% on placebo 1. The margin is real. It is also smaller than the phase 2 programme suggested, and in REDEFINE 4, an open-label comparison with tirzepatide, the combination did not meet its primary endpoint of non-inferiority 4.

The amylin pharmacology behind the combination — a separate receptor system with a separate entry point into the central circuitry — is covered on this site and not repeated here. This article is the readout: the design that made REDEFINE 1 a genuine test of the addition, the result under both estimands, the dosing flexibility that complicates it, REDEFINE 2 in type 2 diabetes, the head-to-head miss, tolerability, and where the regulatory file stands.

Design: a four-arm trial that could isolate the addition

REDEFINE 1 is the right shape for the question in this article's title. A trial comparing a combination only with placebo cannot say what either component contributes. REDEFINE 1 included both components alone, alongside the combination and placebo. It was a phase 3a, double-blind, placebo- and active-controlled trial run across 22 countries in adults without diabetes who had a body-mass index of 30 or above, or 27 or above with an obesity-related complication 1.

With 2,108 participants on the combination and 302 on each single agent, the trial was built chiefly to characterise the combination against placebo, and it is less precise about the combination's margin over each component. That is a reasonable design for a registration trial. It means the comparison this article is about carries wider uncertainty than the headline one 1.

The result, under two estimands

The treatment-policy estimand counts everyone as randomised, including those who stopped treatment. The trial-product estimand models full adherence and gives the larger number. The placebo-adjusted difference for the combination was 17.3 percentage points on the first and 20.4 on the second 12. Under the trial-product estimand, 60.2% of combination participants reached at least 20% body-weight change, 40.4% reached at least 25% and 23.1% reached at least 30% 1.

What the addition bought can be read off the treatment-policy column. The combination exceeded semaglutide alone by about 5.5 percentage points and cagrilintide alone by about 8.9 1. A further calculation is instructive. Over placebo, semaglutide alone contributed about 11.9 points and cagrilintide alone about 8.5; if the two pathways added fully, the combination would sit about 20 points above placebo. It sat about 17 above 1. The combination is better than either part and less than their sum. Physiological floors make perfect additivity unlikely for any pair, so this is not evidence against independent pathways. It is a measure of how much of the theoretical gain was realised.

The flexible protocol, and why it complicates the comparison

REDEFINE 1 allowed investigators to keep participants below the top maintenance level where that was judged best. At 68 weeks, 57.3% of combination participants were on the top level, against 70.2% of those on semaglutide alone and 82.5% of those on cagrilintide alone 2. The combination arm therefore received, on average, less of its intended exposure than either comparator.

Two readings follow, and they point in opposite directions. The combination's margin may understate what it achieves at full exposure. Or the lower share reflects poorer tolerability at full exposure, in which case the achieved margin is the honest figure, because a combination that fewer people can take at its top level has a real ceiling. The published data do not separate these readings. Either way, the phase 2 result that set expectations for the combination came from a small trial without this kind of flexibility, which is one reason the phase 3 figure landed lower.

Tolerability

Gastrointestinal adverse events were reported in about 80% of combination participants against about 40% on placebo 6. Nausea affected about 55% of the combination arm, constipation 30.7% and vomiting 26.1%. Discontinuation because of adverse events was 6.0% on the combination against 3.7% on placebo 1. Those rates are the price of the added five and a half points. Judging whether the price is worth paying requires the adverse-event rates in the single-agent arms, which sit in the full paper rather than in the summaries reviewed here, and this article does not restate them.

For scale, cagrilintide alone had already produced clinically meaningful body-weight change in a 26-week phase 2 dose-finding trial, with a gastrointestinal adverse-event profile dominated by nausea 7. The combination did not introduce a new kind of adverse event. It concentrated two sources of the same kind.

REDEFINE 2: type 2 diabetes

REDEFINE 2 randomised 1,206 adults with overweight or obesity and type 2 diabetes to the combination or placebo, in a three-to-one ratio, for 68 weeks 36. Body-weight change was 13.7% on the combination against 3.4% on placebo under the treatment-policy estimand, and 15.7% against 3.1% under the trial-product estimand. On glycaemia, 74% of combination participants reached a glycated haemoglobin of 6.5% or below, against 15.9% on placebo 6.

Two points limit the trial's contribution to this article's question. Smaller body-weight change in people with diabetes is the pattern across the incretin class, so the lower figure is expected rather than informative. And REDEFINE 2 had no semaglutide-alone arm, so it cannot say what the amylin component added in this population 3.

REDEFINE 4: the head-to-head that missed

REDEFINE 4 randomised 809 adults with obesity and at least one comorbidity to the combination or to tirzepatide, open-label, for 84 weeks. It did not achieve its primary endpoint of non-inferiority on body-weight change 4. The sponsor described the combination as having a safe and well-tolerated profile, with gastrointestinal events mostly mild to moderate and diminishing over time 4.

A failed non-inferiority test is not formally a demonstration of inferiority, but both estimands point the same way, and the gap widened under the treatment-regimen analysis, which counts people who stopped. The open-label design is a limit, and it matters more for tolerability reporting than for a measured endpoint such as body weight. The strategic meaning is plain. The combination's case rested on exceeding what single-pathway incretin agents achieve. In the one direct comparison, a single dual-agonist molecule did better.

Regulatory status

Novo Nordisk submitted a new drug application to the FDA in December 2025 5. The sponsor's February 2026 announcement stated that an FDA decision was anticipated by late 2026 4. If approved, it would be the first once-weekly combination of a GLP-1 receptor agonist and an amylin analogue 5.

Limits

• The single-agent arms in REDEFINE 1 were small, so the combination's margin over each component is estimated less precisely than its margin over placebo.
• The flexible protocol left fewer combination participants at the top level than in either single-agent arm, confounding the comparison in a direction that cannot be determined.
• REDEFINE 2 had no single-agent arm.
• REDEFINE 4 was open-label.
• No cardiovascular outcome data for the combination have reported, and durability after stopping is not described here.
• Body composition is not part of the results summarised here.

What the combination added, stated plainly

REDEFINE 1 answered its question cleanly: adding amylin receptor agonism to semaglutide increased body-weight change, by about five and a half percentage points at 68 weeks on the conservative estimand, at the cost of more gastrointestinal adverse events and with fewer participants reaching the top maintenance level 12. The addition was less than additive, and the combination fell short of tirzepatide in the one head-to-head trial 4. The idea that a second, independent satiety pathway adds something is supported. The idea that it adds enough to lead the field is not.