NCT04982575 compared protocol-defined cagrilintide–semaglutide co-administration with each component alone in 92 adults with type 2 diabetes. The primary HbA1c comparison, combination versus semaglutide, was not statistically significant. The secondary HbA1c comparison versus cagrilintide and supportive body-weight comparisons nominally favored the combination. The posted statistical analysis plan specified no confirmatory hypothesis testing; the favorable secondary findings do not convert the null primary comparison into a positive one.[1–3][5]
These findings concern the sponsor's separately co-administered test articles and a 32-week treatment period. They do not prove synergy, assign a precise contribution to either component, or validate a premixed blend, different formulation, or separately sourced research material.
Study at a glance
| Field | Trial record |
|---|---|
| Design | Phase 2, multicenter, randomized, parallel-group, double-blind, active-controlled |
| Population | 92 adults with type 2 diabetes, BMI at least 27, and background metformin with or without an SGLT2 inhibitor |
| Groups | Combination: 31; semaglutide alone: 31; cagrilintide alone: 30 |
| Sites | The paper and registry recruitment description report 17 conducting U.S. sites; the current registry directory lists 23 locations |
| Duration | 32 weeks of treatment followed by 5 weeks of follow-up |
| Primary objective | HbA1c change to week 32: combination versus semaglutide |
| Secondary objectives | HbA1c versus cagrilintide; body weight, fasting plasma glucose, and other supportive outcomes |
| Identification | NCT04982575; PMID 37364590; DOI 10.1016/S0140-6736(23)01163-7 |
| Sponsor | Novo Nordisk A/S |
| Current registry | Completed; actual enrollment 92; results first posted July 27, 2023; last updated December 23, 2025 |
The group descriptions identify research comparisons, not instructions for use. Registry status describes the trial, not approval of a product.[2]
Design, population, and masking
Central randomization assigned participants 1:1:1 and was stratified by background SGLT2-inhibitor use. Mean age was 58 years; 59 participants were men. Eligibility required type 2 diabetes and elevated HbA1c; women were restricted to those of non-childbearing potential. Important exclusions included substantially reduced kidney function and uncontrolled or unstable diabetic retinopathy. This selected population does not represent everyone with obesity or diabetes.[1][5][6]
The trial used separate component test articles with matching placebos to preserve masking, but had no placebo-only comparison group. The paper describes participant and investigator masking. The protocol and SAP permitted a limited sponsor interim-analysis team to be unblinded and excluded from daily trial activities until database lock; that planned exception should not be replaced by a claim that all sponsor staff were necessarily masked throughout.[3][5][6]
The accepted manuscript states that no diet or exercise counselling was provided. Lifestyle-program details from a different cagrilintide trial should not be imported into this study.[5]
Primary HbA1c result and secondary comparison
The paper reported these estimated mean HbA1c changes at week 32 under its primary trial-product analysis:
| Group | Estimated HbA1c change |
|---|---|
| Combination | −2.2 percentage points |
| Semaglutide alone | −1.8 percentage points |
| Cagrilintide alone | −0.9 percentage points |
The primary-objective comparison, combination versus semaglutide, was −0.4 percentage points, with a 95% confidence interval of −0.8 to 0.0. The published abstract gives nominal p=0.075; the accepted-manuscript table reports p=0.07. It did not demonstrate a statistically significant difference. The interval is rounded in the paper; a nonsignificant result does not establish equivalence or absence of any effect.[1][3][5]
The HbA1c comparison against cagrilintide was a secondary objective: −1.3 percentage points, 95% confidence interval −1.7 to −0.8, nominal p<0.0001. Keeping that comparison's place in the outcome hierarchy visible prevents a favorable secondary result from being mistaken for primary success.[1][3]
All p-values and confidence intervals here are nominal, not multiplicity-adjusted confirmatory findings. The SAP describes an exploratory trial and bases sample-size considerations on precision rather than a confirmatory superiority test.[3] The guide to confidence intervals and p-values explains why effect size, uncertainty, and the tested question belong together.
Estimands and missing-data methods
The primary trial-product estimand asked what the difference would be if randomized participants continued their assigned treatment without rescue medication. Its mixed model for repeated measures used on-treatment-without-rescue observations, treatment and SGLT2-inhibitor stratification factors, and baseline HbA1c. The posted SAP excludes retrieved-dropout week-32 observations from this primary analysis. This is a hypothetical question about all randomized participants, not simply a claim that every participant remained treated or had an observed final measurement.[3][5]
An additional treatment-policy estimand asked about assignment regardless of treatment discontinuation or rescue medication. The SAP specifies analysis of covariance with retrieved-dropout multiple imputation, with simplifications and a model-based fallback if the planned imputation was infeasible. Those are specified methods and contingencies, not evidence that every fallback was used.[3]
The accepted manuscript reports both estimands. Its additional treatment-policy HbA1c comparison against semaglutide is rounded to −0.3 points, with a 95% confidence interval of −0.8 to 0.2 and nominal p=0.23. The current registry provides the more precise corresponding estimate, −0.30 points, interval −0.79 to 0.19, nominal p=0.2284.[2][5] This is not the primary −0.4 estimate with a differently rounded p-value: the estimand and analysis differ.
Descriptive observed changes, model-estimated changes, and adjusted group contrasts should not be interchanged or reconstructed by subtracting whichever means appear in a registry table. The public record includes several observation periods and denominators. Neither source's additional analysis should be pooled with the primary analysis. See estimands and missing data.
The protocol is version 2, June 16, 2021. The posted SAP is version 1, March 12, 2022, signed March 14; its November 1, 2022 attachment is a safety-search appendix, not a later statistical-plan amendment. The SAP lists changes from protocol methods, including removal of a planned week-32 interim analysis and added modeling contingencies. Access to these documents does not independently establish when every decision was made relative to access to unblinded data.[3][6]
Body-weight and fasting-glucose outcomes
The primary trial-product analysis reported estimated body-weight changes of −15.6% with the combination, −5.1% with semaglutide, and −8.1% with cagrilintide at week 32. Both combination-versus-component comparisons had nominal p<0.0001 in the published abstract. These were supportive secondary findings under a plan with no confirmatory testing.[1][3][5]
Estimated fasting plasma glucose changes were −3.3 mmol/L, −2.5 mmol/L, and −1.7 mmol/L, respectively. The comparison against cagrilintide had nominal p=0.001; the comparison against semaglutide had nominal p=0.10.[1][5] A favorable weight comparison does not make the primary HbA1c or fasting-glucose comparison against semaglutide significant.
These short-term metabolic outcomes do not establish cardiovascular benefit, long-term durability, or effects after treatment stops. For context on interpreting a nonsignificant result without overstating it, see negative and null study results.
Harms, treatment stops, and trial completion
All 92 randomized participants received trial treatment and were included in the full analysis and safety sets. The manuscript's adverse-event table uses an on-treatment safety period; the registry describes safety from week 0 to week 37, including follow-up. These source descriptions should remain explicit rather than assuming every table has an identical observation window.[2][3][5]
| Participants affected | Combination, n=31 | Semaglutide, n=31 | Cagrilintide, n=30 |
|---|---|---|---|
| At least one adverse event | 21 (67.7%) | 22 (71.0%) | 24 (80.0%) |
| Gastrointestinal disorder | 18 (58.1%) | 10 (32.3%) | 10 (33.3%) |
| At least one serious adverse event | 0 | 2 | 4 |
| Adverse event leading to treatment withdrawal | 0 | 1 | 0 |
Most reported adverse events were mild or moderate. The combination had a higher gastrointestinal-event proportion despite a lower proportion reporting any adverse event. The four cagrilintide participants with serious adverse events experienced five serious events; participant counts and event counts are not interchangeable. One semaglutide participant stopped treatment because of diarrhea. These descriptive observations do not establish a causal or statistically reliable safety difference between the groups.[2][5]
No deaths or level 2 or 3 hypoglycemia were reported. Level 1 hypoglycemia occurred in two combination participants and two cagrilintide participants, so “no hypoglycemia” would be inaccurate. Small groups and limited follow-up cannot exclude rare or delayed harms.[5]
Seven participants discontinued trial treatment: four combination, three semaglutide, and none cagrilintide. Four withdrew from the trial itself: two combination and two semaglutide. Registry completion was therefore 29/31, 29/31, and 30/30, respectively. Treatment discontinuation is not the same as study withdrawal, and completing follow-up does not mean all scheduled on-treatment efficacy observations were available.[2][5]
The primary model and additional imputation analysis address missing data under assumptions. A high study-completion proportion does not remove the uncertainty introduced by treatment stops, rescue medication, or missing measurements.
Combination attribution and sponsor role
Component comparators allow direct contrasts with the combination as tested. They do not identify the exact fraction of response attributable to each component, establish a greater-than-additive interaction, or show that a different ratio, formulation, source, or population would behave the same way. Separate co-administration in this trial should not be presented as validation of a commercial premixed blend.[5][6]
Novo Nordisk funded the trial and participated in design, monitoring, data collection, analysis, and interpretation. All authors had access to study data, contributed to manuscript drafts, and were responsible for the decision to submit. Sponsor-funded medical-writing and editorial assistance was acknowledged. Three authors were sponsor employees and shareholders, and other authors disclosed industry relationships.[1][5]
These roles warrant clear reporting, not automatic acceptance or rejection of the findings. This page is an editorial evidence summary, not an independent statistical replication.
Current status and what the trial does not establish
As checked on September 2 in Edmonton / September 3 UTC, FDA states that cagrilintide is not a component of an FDA-approved drug, has not been found safe and effective for any condition, and cannot be used in compounding.[4] Separate semaglutide product approvals do not transfer to this combination or to differently sourced materials.
A completed registry record, published result, or familiar ingredient name does not authenticate a lot, establish product quality, or show clinical equivalence. The approved-drug versus research-material guide and trial and FDA status-checking guide explain those boundaries. Related evidence is organized in the metabolic, mitochondrial, and redox research hub.
Frequently asked questions
What was the primary comparison?
Change in HbA1c at week 32 for the combination versus semaglutide. The HbA1c comparison versus cagrilintide was a secondary objective.[2][3]
Did the combination significantly improve the primary HbA1c outcome?
Not in the comparison against semaglutide: the primary analysis had nominal p=0.075. That nonsignificant finding is not proof that the groups are equivalent.[1]
Was the weight reduction greater with the combination?
The paper reported larger estimated reduction than with either component at week 32. Those supportive secondary comparisons were nominal, not confirmatory tests.[1][3]
Why are there two HbA1c estimates against semaglutide?
The primary hypothetical trial-product and additional treatment-policy analyses answer different questions and handle post-discontinuation or rescue observations differently. Both are reported in the accepted manuscript; the registry gives additional precision for the policy analysis.[2][3][5]
Does a lower any-adverse-event percentage mean the combination was safer?
No. The combination had more participants with gastrointestinal events, and approximately 30 participants per group cannot establish comparative safety or exclude uncommon harms.[5]
Did everyone who completed the trial remain on treatment?
No. Seven participants discontinued treatment, while four withdrew from the trial. Follow-up completion and continued treatment are different measures.[2][5]
Does this trial validate another CagriSema-labeled blend?
No. Its findings belong to the sponsor's separate test articles, population, and protocol. A name does not establish identity, formulation, quality, synergy, or clinical equivalence.
Primary and official sources
- Published paper record and abstract: PMID 37364590
- ClinicalTrials.gov: NCT04982575 and posted results
- Posted statistical analysis plan: version 1, March 12, 2022
- FDA: Concerns with unapproved GLP-1 drugs used for weight loss
- University of Leicester repository: accepted manuscript of the trial paper
- Posted protocol: version 2, June 16, 2021
Review and corrections
Sources checked September 2, 2026 (America/Edmonton) / September 3, 2026 (UTC). The review used the published abstract, complete 26-page accepted manuscript, posted protocol and SAP, official registry results, and current FDA statement. The separate journal appendix, actual interim-analysis charter, and participant-level data were not obtained; no full supplementary-material or independent reanalysis claim is made. PubMed's retrieved relationships included an accompanying commentary, not a correction notice; this is not a guarantee that no notice exists elsewhere.
Serum Research is responsible for this editorial summary. No independent scientific, clinical, statistical, analytical, or regulatory review is claimed. This is not medical advice, a use recommendation, or a conclusion about product quality or equivalence. Read how we review evidence, browse the study index and research library, or report a correction.