COMBINE 4: IcoSema vs insulin glargine

The Lancet Diabetes & Endocrinology published COMBINE 4 on 13 August 2026^[1]. The trial tested IcoSema, a fixed-ratio combination of insulin icodec and semaglutide, against once-daily insulin glargine U100. A fixed-ratio combination means both drugs sit in one pen at a set ratio, so you cannot titrate one without moving the other. Insulin icodec is a basal insulin engineered for once-weekly injection rather than daily. Over 40 weeks, the combination lowered HbA1c further than daily basal insulin did.

What the trial actually tested

COMBINE 4 ran for 40 weeks across 97 sites in nine countries^[1]. Novo Nordisk sponsored and ran it. The registry record lists an open-label parallel-group design with no masking, enrolment of 485 participants, a start date of 15 February 2024 and completion on 8 July 2025^[2]. Participants had to be insulin-naive, with type 2 diabetes diagnosed at least 180 days earlier, an HbA1c of 8.0% or above, a BMI of 40 or below, and one to three oral glucose-lowering drugs at stable doses for at least 90 days. That is a narrow entry gate, and it matters when you read the results across to semaglutide users already on an injectable.

Treat-to-target is the design term to understand. Investigators in both arms titrated the dose upward against the same fasting glucose goal, here 3.9 to 5.0 mmol/L (70 to 90 mg/dL)^[1]. The point of that design is to hold glycaemic ambition constant so the arms differ on what it costs to reach the target: hypoglycaemia, weight, injections. It is the standard way insulin trials are built. The primary endpoint was change in HbA1c from baseline to week 40, and the confirmatory secondary endpoint was change in bodyweight over the same window^[2].

The headline numbers

Investigators screened 653 people between February and August 2024. 151 failed screening and 17 withdrew before treatment, leaving 243 randomised to IcoSema and 242 to glargine U100^[1]. The randomised group was 286 men (59%) and 199 women (41%), with a median age of 58 years and a range of 26 to 82. Baseline HbA1c was high: 9.57% in the IcoSema arm and 9.50% in the glargine arm.

At week 40, mean HbA1c fell 3.32 percentage points on IcoSema and 2.44 percentage points on glargine U100. The estimated treatment difference was 0.88 percentage points in favour of IcoSema (95% CI -1.12 to -0.63, p<0.001), which confirmed superiority^[1]. Bodyweight moved in opposite directions. IcoSema participants lost a mean 0.79 kg while glargine participants gained 3.81 kg, a difference of 4.61 kg (95% CI -5.46 to -3.75, p<0.001).

Hypoglycaemia rates also favoured the combination arm. Clinically significant episodes (blood glucose under 3.0 mmol/L, meter-confirmed) or severe episodes ran at 0.29 per person-year on IcoSema against 0.59 on glargine U100, a rate ratio of 0.56 (95% CI 0.32 to 0.97, p=0.04)^[1]. Gastrointestinal disorders were the most frequently reported adverse events on IcoSema, which is what the GLP-1 half of the pen predicts. The same pattern shows up across the semaglutide and tirzepatide trial programmes.

Why the open-label design constrains the reading

Nobody was blinded^[2]. One arm injected weekly and the other daily, so masking was never realistic without dummy injections. The honest consequence is that dose titration involved clinician judgement, and clinicians knew which drug they were adjusting. HbA1c is a lab value and fairly resistant to that pressure. Bodyweight and patient-reported treatment satisfaction, which the registry lists among the secondary endpoints, are softer and more exposed to it.

The p=0.04 on hypoglycaemia is worth flagging separately. It is a real result at a conventional threshold, and the confidence interval (0.32 to 0.97) reaches close to 1.0. Treat that as a signal in one 485-person trial rather than a settled effect size. Anyone comparing this against their local basal insulin practice should also note that the trial enrolled in China, India, Japan, Italy, Greece, Poland, Turkey, South Africa, Puerto Rico and the United States, so background care differs from a single-country standard.

What this does not tell you about weight management

The 4.61 kg gap is a comparison against a drug that reliably causes weight gain. IcoSema participants lost less than a kilogram over 40 weeks. Read the number as insulin weight gain avoided, not as weight loss in the sense that semaglutide is known for in obesity medicine.

COMBINE 2 makes the point sharply. That trial compared IcoSema with once-weekly semaglutide 1.0 mg on its own in 683 adults over 52 weeks^[3]. IcoSema won on HbA1c by 0.44 percentage points. On weight it lost badly: participants gained 0.84 kg on IcoSema and lost 3.70 kg on semaglutide alone, a 4.54 kg difference favouring the GLP-1 by itself (p<0.0001). Adding insulin to a GLP-1 buys glucose control and gives back weight. Neither COMBINE trial studied the higher semaglutide doses used for chronic weight management, and none of this is a basis for choosing a semaglutide product for weight.

Where IcoSema stands right now

IcoSema is not an approved medicine. The trial's own plain-language registry summary says so directly: doctors cannot prescribe it^[2]. Semaglutide on its own is approved and prescription-only. The US label covers glycaemic control in adults with type 2 diabetes plus cardiovascular and kidney risk reduction in defined populations, at weekly doses of 0.25 mg to 2 mg^[5]. Our country pages track that status for the United States, the United Kingdom, Germany and Denmark.

COMBINE 4 is the fourth readout in the programme. COMBINE 1 compared IcoSema with insulin icodec alone in 1,291 adults already on daily basal insulin and reported a 0.66 percentage point HbA1c advantage plus a much lower hypoglycaemia rate^[4]. The gaps that remain are the interesting part. There is no cardiovascular outcomes trial for the combination. Forty weeks says nothing about durability at year three. And no trial has put IcoSema against a basal insulin plus tirzepatide given separately, which is the comparison a prescriber facing this decision in 2027 would want.

What to watch next

Regulatory filings are the near-term signal. After that, watch whether the once-weekly convenience argument survives contact with reimbursement committees, because a combined pen removes the ability to titrate insulin and semaglutide independently. For readers already using an approved GLP-1, the practical takeaway from COMBINE 4 is narrow: it describes people starting insulin for the first time, not people managing weight. Our semaglutide safety notes cover the adverse-event profile in more depth.

This article is educational and journalistic. It is not medical advice and it is not a dosing guide. If you have type 2 diabetes or are considering any change to your treatment, that conversation belongs with a clinician who knows your history.