IcoSema is a once-weekly injection that combines basal insulin icodec with semaglutide. It is built for type 2 diabetes that daily basal insulin no longer controls. In its first phase 3a trial it lowered HbA1c 1.55 points, against 0.89 on weekly insulin alone [2].
That is a glucose drug with a GLP-1 inside it. It is not a weight-loss product, and one of its own trials shows why. The closer comparison is swapping mealtime insulin for a GLP-1, which aims at the same patient.
Three trials, three comparators
The COMBINE program tested it three ways, each against a different standard. The comparator decides what each result means.
| Trial | Compared with | HbA1c difference | Weight |
|---|---|---|---|
| COMBINE 1, n=1,291 | weekly insulin icodec | −0.66 points | not in abstract |
| COMBINE 2, n=683 | semaglutide 1.0 mg | −0.44 points | +0.84 kg vs −3.70 kg |
| COMBINE 3, n=679 | basal-bolus insulin | −0.06 points | −6.72 kg difference |
COMBINE 1 set it against the insulin half alone. Adding semaglutide won by 0.66 points, and superiority was confirmed [2].
COMBINE 2 set it against the semaglutide half alone, in people already on a GLP-1. IcoSema won on HbA1c by 0.44 points [3]. Weight went the other way: up 0.84 kg on IcoSema, down 3.70 kg on semaglutide.
COMBINE 3 set it against basal-bolus therapy, the several-injections-a-day regimen. HbA1c came out level, a difference of 0.06 points, which met the non-inferiority test [4]. Weight favored IcoSema by 6.72 kg.
Why the pooled number reads 98%
A 2026 meta-analysis pooled COMBINE 1 and COMBINE 3, 1,970 people [1]. It left out COMBINE 2 for asking a different question.
The pooled HbA1c difference was −0.37%, interval −0.95 to 0.21. Heterogeneity was 98%.
That heterogeneity comes from the comparators. One trial beat weekly insulin alone by 0.66 points. The other tied full basal-bolus therapy. Average a clear win with a tie, and the pooled interval here crosses zero.
So the pooled figure does not say IcoSema fails on glucose. It says two different questions were averaged. The same trap sits in comparing two drug classes by inference.
What moved beyond glucose
Against insulin-based intensification, the pool found weight down 6.10 kg. Systolic pressure fell 2.45 mmHg. LDL and triglycerides each came in at a ratio of 0.94.
Hypoglycemia fell sharply against insulin. Severe or clinically significant episodes ran 0.14 per person-year against 0.63 on icodec. Against basal-bolus they ran 0.21 against 2.23. Against semaglutide there was no significant difference.
That pattern matches where low blood sugar comes from: the insulin, not the GLP-1.
What it would change for a patient
One injection a week instead of three to five a day. Less weight than insulin intensification. Fewer low-sugar episodes.
None of the four papers prices it. The nearest comparison is the add-on arithmetic in CagriSema's diabetes trials. The wider view is the map of what these drugs affect.