A trial can meet its primary endpoint and still hand a buyer a difficult number. REIMAGINE 2 compared cagrilintide-semaglutide against semaglutide alone in 2,713 people with type 2 diabetes, and the combination won on HbA1c by 0.16 percentage points [1]. That is statistically superior, with a confidence interval from -0.27 to -0.05 and a p value of 0.0035, and it is 0.16 points.
The arms fell 1.91 and 1.75 percentage points respectively over 68 weeks from a baseline HbA1c of 8.2%. Both are large. The gap between them is what a second molecule bought, and any seller eventually pricing this combination will be pricing that gap — a question this desk has put to the existing combination in the earlier combination data.
Tolerability moved against the combination. Adverse events were reported in 86.9% of the combination group and 81.2% of the semaglutide 2.4 mg group, with gastrointestinal disorders most common in every active arm. Placebo ran 70.5%. So the second molecule costs roughly six percentage points of additional adverse events, plus whatever it eventually costs in dollars, for 0.16 HbA1c points.
One design detail is worth noting. Randomization ran 8:8:2:8:8:1:1, which deliberately made the cagrilintide-alone arm (n=152) and the pooled placebo arm (n=149) much smaller than the combination and semaglutide arms. That is an efficient design for the comparison the sponsor cared about, and it means the cagrilintide-alone numbers rest on a fraction of the trial’s participants — the kind of denominator this desk checks first, as in the sellers publishing two prices.
Nothing here is purchasable yet. When it is, the arithmetic a buyer needs is the one this desk applies to every add-on: what the second component costs per month against what it adds. On the endpoint this trial was built to test, the answer so far is sixteen hundredths of a point, which is a hard number to build a price around — the problem set out in whether a GLP-1 is worth its price.