The most useful sentence in Novo Nordisk's February 23 REDEFINE 4 announcement is not the largest percentage. It is that the trial did not achieve its primary endpoint. Substantial within-group weight reduction and failure of a prespecified comparative objective can both be true.
The original sponsor announcement filed with the SEC describes an 84-week, open-label phase 3 study in 809 randomized people with obesity and at least one comorbidity. Mean baseline weight was 114.2 kg. Participants received either the investigational fixed-dose cagrilintide–semaglutide combination, 2.4/2.4 mg, or tirzepatide 15 mg, once weekly. Those are trial-arm identifiers, not instructions for treatment.
Compare the same analysis with the same analysis
Under the efficacy estimand, Novo reported weight reductions of 23.0% with CagriSema and 25.5% with tirzepatide at week 84. Its definition assumes adherence to treatment according to the protocol, regardless of dose modification. The difference between these rounded reported changes is 2.5 percentage points; that arithmetic is not a substitute for an estimated treatment contrast and confidence interval.
Under the treatment-regimen estimand, which the release defines as the treatment effect regardless of adherence, reductions were 20.2% and 23.6%, respectively. The difference between those reported changes is 3.4 percentage points. It would be misleading to compare CagriSema's 23.0% efficacy-estimand number with tirzepatide's 23.6% treatment-regimen number: that would mix the statistical questions in the same sentence.
What a missed non-inferiority endpoint means
Novo explicitly says CagriSema did not demonstrate non-inferiority to tirzepatide for weight loss after 84 weeks. Non-inferiority is not established merely because both arms lose substantial weight. It requires satisfying the protocol's comparative criterion. This targeted review has not obtained the full protocol margin, confidence-interval analysis or a subsequent complete peer-reviewed report; it therefore does not reconstruct a formal superiority or inferiority test from rounded percentages.
The study was open-label: investigators and participants knew the assigned treatment. That feature matters when interpreting adherence, reporting and other outcomes. It does not erase randomization, but it distinguishes this evidence from a blinded comparison.
The safety paragraph is not a safety table
The release says the most common CagriSema adverse events were gastrointestinal, with the vast majority mild to moderate and diminishing over time. That is the sponsor's characterization. It does not provide arm-level event rates, serious-event counts or discontinuation figures in the safety passage reviewed. The absence of those figures in this release is not evidence of no events, and it cannot support a comparative tolerability ranking.
The release also says CagriSema had been submitted to FDA in December 2025 and that a decision was anticipated later in 2026. A submission or expected decision is not an approval. CagriSema is investigational in the source reviewed; this article does not infer a later regulatory decision from the forecast.
Why the practice-facing message changes
This is direct randomized comparative reporting, not an indirect comparison of different trials. Its main limitation for a citation-ready clinical decision is the headline-release format, open-label design and lack of the detailed analysis and safety tables in this review. SEC hosting preserves the original company communication; it does not turn a sponsor announcement into independent peer review.
A useful patient-facing summary is that an investigational combination produced substantial average weight loss in this study but failed the intended non-inferiority comparison. It is not that one product has been universally proved better for every patient, nor that the combination offers no benefit.
Evidence record
Event date: 2026-02-23. Source review: 2026-10-29. Editorial publication: 2026-10-29. Original source links below retain their publication dates. These dates are not interchangeable. This is targeted source synthesis, not an exhaustive archive or a qualified clinical/legal review.
Revision history
29 October 2026: original analysis added from the specific source passages identified in the evidence audit. Journal abstracts and sponsor reporting are labeled separately; full-paper and supplementary safety assessment remain outstanding. Report a proposed correction through Contact us; do not include patient data.