Skip to content
GLP Loss
← Research
Evidence

REDEFINE 2 Explained: What CagriSema Does in Type 2 Diabetes

REDEFINE 2 randomized 1,206 adults with type 2 diabetes three to one against placebo and returned a weight difference of 10.4 percentage points — against 17.3 points for the same combination in people without diabetes, and 11.6 points for tirzepatide in the matching diabetes trial.

Owen Castellanos8 min read
The same injection, minus a diagnosisCagrilintide 2.4 mg plus semaglutide 2.4 mg, 68 weeks, both trialsWith type 2 diabetes (n = 1,206)10.4 pointsWeight fell 13.7% against 3.4% on placeboWithout diabetes (n = 3,417)17.3 pointsWeight fell 20.4% against 3.0% on placeboOnly one arm was compared against904 on the combination, 302 on placebo, three to oneNeither ingredient was given on its ownThe biggest gap is the glucose oneReached 6.5% or less: 73.5% against 15.9%No results are posted to the registry. No discontinuation rate exists.

REDEFINE 2 is the trial that took a two-drug injection into the population where two-drug injections have the most to prove, and then removed the only measurement that could have judged it. It ran cagrilintide and semaglutide together against placebo in adults with type 2 diabetes for 68 weeks, and it ran nothing else — no semaglutide-alone arm, no cagrilintide-alone arm, no active comparator of any kind.[1] What the combination is and where the program stands is set out in the CagriSema article. This page is about what this particular trial can and cannot be asked.

Who was enrolled, and against what

The design was phase 3a, double-blind and placebo-controlled, conducted in 12 countries. Eligibility required a body-mass index of 27 or more, a glycated hemoglobin of 7 to 10%, and an existing diagnosis of type 2 diabetes. A total of 1,206 patients were randomized in a 3:1 ratio — 904 to once-weekly cagrilintide 2.4 mg with semaglutide 2.4 mg, and 302 to placebo — with lifestyle intervention in both groups for 68 weeks.[1]

Three quarters of the enrollment went to the arm whose result was never in doubt. That allocation is efficient if the question is whether the combination beats nothing, and useless if the question is whether the amylin analog earns a place next to a drug that already works. The sibling trial in people without diabetes at least attempted the second question, sizing its two monotherapy arms at 302 participants each; how that allocation played out is the subject of the REDEFINE 1 article. REDEFINE 2 did not attempt it.

The coprimary endpoints

Both primary endpoints compared the combination against placebo: the percent change in body weight from baseline to week 68, and the proportion of patients reaching a weight reduction of 5% or more. Estimated mean body weight changed −13.7% on the combination against −3.4% on placebo — an estimated difference of −10.4 percentage points (95% CI, −11.2 to −9.5; P < 0.001). More patients on the combination reached 5% or more, and the same held at thresholds of 10%, 15% and 20%, each at P < 0.001 for the last comparison.[1]

Effect estimates used the treatment-policy estimand, consistent with the intention-to-treat principle, which means the headline figure already carries everyone who stopped injecting or started something else.[1] That is the conservative analysis and the right one to publish. No on-treatment figure appears in the abstract for comparison, and no threshold counts are published beyond the statement that each comparison cleared its test.

The column the trial was not named for

The largest contrast in REDEFINE 2 is not the weight one. The proportion of patients reaching a glycated hemoglobin of 6.5% or less — the threshold at which the diagnostic criterion for diabetes is no longer met on that measure — was 73.5% on the combination against 15.9% on placebo.[1] That is a 57.6-point spread against a 10.4-point spread on the endpoint that gave the trial its name.

No confidence interval, odds ratio or p-value is attached to that proportion in the abstract, and with no posted results there is nowhere else to read one. It is a contrast rather than a tested comparison, and the distinction between glucose control and weight control — two indications, two labels, two prices — is the subject of the two-indications article.

Placing 10.4 points next to what is already for sale

Because nothing inside the trial answers the increment question, the only way to size it is across trials, and cross-trial arithmetic is a weaker instrument: separate protocols, separate placebo arms, separate years and separate background therapy. With that caveat carried in full, three figures are worth putting side by side, all of them placebo-adjusted and all of them in adults with type 2 diabetes and a body-mass index of 27 or above.

Semaglutide 2.4 mg on its own, in 1,210 patients over the same 68 weeks, produced −9.6% against −3.4%, an estimated treatment difference of −6.2 percentage points (95% CI, −7.3 to −5.2), with 68.8% of 388 reaching a 5% reduction against 28.5% of 376 (odds ratio 4.88; 95% CI, 3.58 to 6.64).[4] Tirzepatide 15 mg, in 938 patients over 72 weeks, produced −14.7% against −3.2%, a difference of −11.6 percentage points (95% CI, −13.0 to −10.1), with 79% to 83% reaching 5% against 32%.[5] That trial is covered in the SURMOUNT-2 article and the semaglutide one in the STEP 2 article.

So the two-drug combination sits above semaglutide alone by roughly four points and below a single-molecule dual agonist by roughly one, in a population all three trials defined the same way. None of that is a head to head and none of it carries a confidence interval for the comparison itself. What it does establish is that the combination is not obviously the strongest option in diabetes, which is a different sentence from the one the headline figure invites.

Where the margin actually is

The comparison runs the other way without diabetes. In the sibling trial, 3,417 adults without diabetes on the same combination at the same doses for the same 68 weeks lost 20.4% against 3.0% on placebo, an estimated difference of −17.3 percentage points (95% CI, −18.1 to −16.6).[3] Tirzepatide 15 mg in an obesity population without diabetes returned −20.9% (95% CI, −21.8 to −19.9) against −3.1%, roughly 17.8 points.[6]

Read across the four trials, the pattern is consistent and awkward. Where diabetes is absent, the combination is level with the strongest single-molecule competitor. Where diabetes is present, its placebo-adjusted margin falls by roughly 40% and lands behind that competitor. A second injected molecule added to semaglutide buys the most in the population that needs the least convincing, and the least in the population where the class is hardest to beat. The trial does not explain that and was not designed to. The expected weight-loss tool applies trial averages to a personal starting weight, and none of these averages was produced in a diabetes population and an obesity population at once.

Tolerability, and a number that does not exist

Gastrointestinal adverse events were reported by 72.5% of the combination group and 34.4% of the placebo group, described as mostly transient and mild or moderate in severity.[1] Against semaglutide alone in a comparable population, where the same category ran 63.5% against 34.3%,[4] the placebo arms are close and the treated arms are nine points apart. That is a cross-trial comparison too, but it is at least a comparison between two arms that received a placebo-controlled protocol in the same indication.

How many patients stopped the injection is not published anywhere. The abstract gives no discontinuation rate. The full text is paywalled with no PubMed Central record, so its own table cannot be opened. And the registry record carries no posted results at all: actual primary completion is recorded as 28 January 2025 and actual overall completion as 3 February 2025, with the record last updated a year later and the results section still empty.[2] There is no participant-flow column, no per-arm adverse-event table and no withdrawal count to read.

That absence deserves to be stated as an absence. A 72.5% gastrointestinal rate with no attrition figure beside it is half a tolerability picture, and the missing half is the one that tells a buyer how many people in their position gave up.

What REDEFINE 2 did not establish

It did not establish that adding cagrilintide to semaglutide helps anyone, because it never ran semaglutide alone. It did not compare the combination with tirzepatide, with any oral agent or with insulin. It measured 68 weeks and nothing longer, withdrew no arm, and so says nothing about what happens when the injections stop — the evidence on that is in the discontinuation article. It adjudicated no cardiovascular, kidney or mortality endpoint, so the outcome benefits that carried semaglutide beyond weight remain undemonstrated for this combination.

It also did not publish the substudy its own abstract advertises. The stated purpose includes patients undergoing continuous glucose monitoring, and no glucose-monitoring result appears in the abstract at all — no time in range, no variability measure, no subgroup size.[1] A subgroup named in the objective and absent from the results is not a finding in either direction.

What was in the injection

Every figure above belongs to manufacturer-supplied investigational drug, titrated under protocol to 2.4 mg of each molecule, injected weekly for 68 weeks with lifestyle support, in adults whose diabetes was already being treated. No product containing cagrilintide is approved for sale, so nothing here describes a purchase anyone can currently make. Sellers listed on the compounded semaglutide board dispense compounded preparations of the molecule that is approved, and compounded drugs are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed.

The specific misuse to watch for with this trial is the transfer of its sibling’s number. Twenty percent of body weight is REDEFINE 1’s figure, produced in people without diabetes. In diabetes the same regimen produced 13.7%, and 3.4 points of that belonged to the placebo arm.

Frequently asked

How much weight did people lose in REDEFINE 2?
Estimated mean body weight fell 13.7% on cagrilintide-semaglutide and 3.4% on placebo over 68 weeks, an estimated difference of 10.4 percentage points with a 95% confidence interval of 11.2 to 9.5 points. Those estimates use the treatment-policy estimand, so they already include everyone who stopped treatment. The trial randomized 1,206 adults with type 2 diabetes in a 3:1 ratio, 904 to the combination and 302 to placebo.
Why is the REDEFINE 2 number smaller than the 20% figure people quote?
The 20.4% figure belongs to REDEFINE 1, which enrolled adults without diabetes. The same combination at the same doses for the same 68 weeks produced 13.7% in the diabetes population, and its placebo-adjusted margin fell from 17.3 percentage points to 10.4. Attenuation in diabetes is a familiar pattern across this drug class, but the two figures are not interchangeable and only one of them describes a person with type 2 diabetes.
Did REDEFINE 2 show that adding cagrilintide helps?
No, because it could not. The trial had only two arms, the combination and placebo, with no semaglutide-alone group and no cagrilintide-alone group. Nothing inside the trial isolates what the second molecule contributed, and the only way to estimate it is by comparing separate trials with separate placebo arms, which is a much weaker instrument than a randomized comparison would have been.
How many people stopped the injection?
That figure is not published. The abstract reports no discontinuation rate, the full text is paywalled with no PubMed Central record, and the registry entry for NCT05394519 carries no posted results despite the trial completing in February 2025 and the record being updated a year later. What is published is that gastrointestinal adverse events affected 72.5% of the combination group against 34.4% of placebo.
What happened to blood sugar in REDEFINE 2?
The glycemic contrast is larger than the weight one. A glycated hemoglobin of 6.5% or less was reached by 73.5% of patients on the combination and 15.9% on placebo. No confidence interval, odds ratio or p-value is attached to that comparison in the abstract, and with no posted results there is no other source for one, so it should be read as a contrast rather than a tested result.
Can this trial support buying a compounded product?
No. Cagrilintide is not approved in any product, alone or combined, so nothing sold today is what this trial administered. Compounded preparations of the approved molecule are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing, and a 68-week protocol with free investigational drug and supervised lifestyle support does not describe a monthly shipment.

Sources

  1. [1] Davies MJ, Bajaj HS, Broholm C, et al. (2025). Cagrilintide-Semaglutide in Adults with Overweight or Obesity and Type 2 Diabetes. N Engl J Med. PMID 40544432
  2. [2] Novo Nordisk A/S (2026). A Research Study to See How Well CagriSema Helps People With Excess Body Weight and Type 2 Diabetes Lose Weight (REDEFINE 2), NCT05394519: registry record, no results posted. ClinicalTrials.gov. Source
  3. [3] Garvey WT, Blüher M, Osorto Contreras CK, et al. (2025). Coadministered Cagrilintide and Semaglutide in Adults with Overweight or Obesity. N Engl J Med. PMID 40544433
  4. [4] Davies M, Færch L, Jeppesen OK, et al. (2021). Semaglutide 2·4 mg once a week in adults with overweight or obesity, and type 2 diabetes (STEP 2): a randomised, double-blind, double-dummy, placebo-controlled, phase 3 trial. Lancet. PMID 33667417
  5. [5] Garvey WT, Frias JP, Jastreboff AM, et al. (2023). Tirzepatide once weekly for the treatment of obesity in people with type 2 diabetes (SURMOUNT-2): a double-blind, randomised, multicentre, placebo-controlled, phase 3 trial. Lancet. PMID 37385275
  6. [6] Jastreboff AM, Aronne LJ, Ahmad NN, et al. (2022). Tirzepatide Once Weekly for the Treatment of Obesity. N Engl J Med. PMID 35658024

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence