A weekly injection against a well-titrated daily one, in people already failing on metformin. 1,444 patients, 52 weeks — the insulin arm gained weight while tirzepatide's highest dose cut nearly 13kg.
SURPASS-3 was a 52-week, randomized, open-label, phase 3 trial comparing tirzepatide (Mounjaro) against titrated insulin degludec as add-on therapy in adults with type 2 diabetes already on metformin, with or without an SGLT2 inhibitor. It enrolled 1,444 insulin-naive participants across 13 countries — the largest of the head-to-head SURPASS comparisons published so far.
Average baseline characteristics: age 57, 44.2% women, 8.4 years since diabetes diagnosis, HbA1c 8.17%, BMI 33.5 kg/m², with about a third of participants already on an SGLT2 inhibitor. Participants were randomized to tirzepatide 5mg, 10mg, or 15mg once weekly, or insulin degludec once daily.
The comparator wasn't a strawman. Insulin degludec was titrated to a fasting glucose target of 90mg/dL, starting at 10 units per day and reaching a mean dose of 48.8 units per day (about 0.5 U/kg/day) — in line with prior treat-to-target insulin studies in similar populations. The trial's lead investigator noted this made it a fair, realistic comparison rather than an under-dosed control arm.
The primary endpoint was noninferiority of tirzepatide 10mg and/or 15mg versus insulin degludec in HbA1c change at week 52. This is a trial about starting injectable therapy in an insulin-naive population — a different clinical question from head-to-head trials between two already-injectable GLP-1-class drugs, like SURPASS-2. The study ran under ClinicalTrials.gov identifier NCT03882970, sponsored by Eli Lilly.
Every tirzepatide dose beat insulin degludec on HbA1c, clearing the noninferiority margin and going on to show statistical superiority at every dose (P<.0001). The estimated treatment difference ranged from −0.59 to −1.04 percentage points in tirzepatide's favor.
The weight difference cut in opposite directions. All three tirzepatide doses produced weight loss, while the insulin degludec group gained a mean of 2.3kg — a well-documented side effect of exogenous insulin therapy. That's not a small footnote: the estimated treatment difference between the extremes was 9.8 to 15.2kg.
Hypoglycemia was far less common with tirzepatide. Clinically significant (level 2) hypoglycemia — a glucose reading below 54mg/dL — hit 7% of the insulin degludec group versus just 1% to 2% across the tirzepatide doses. That gap reflects a structural difference between the drug classes: tirzepatide's insulin-releasing effect is glucose-dependent, while injected insulin isn't.
The tradeoff was gastrointestinal. Nausea, diarrhea, decreased appetite, and vomiting were the dominant tirzepatide side effects, mostly mild to moderate and concentrated during dose titration. Discontinuation due to adverse events was more common in the tirzepatide groups than with insulin degludec. Five participants died during the 52-week study; none of the deaths were judged by investigators to be related to either treatment.
Over 52 weeks in insulin-naive adults with type 2 diabetes stepping up from metformin, every dose of tirzepatide lowered HbA1c more than a well-titrated insulin degludec regimen — and did it while patients lost weight, up to 12.9kg, instead of gaining it, with roughly a third the rate of significant hypoglycemia.
Sources: Ludvik B, et al. Once-weekly tirzepatide versus once-daily insulin degludec as add-on to metformin with or without SGLT2 inhibitors in patients with type 2 diabetes (SURPASS-3): a randomised, open-label, parallel-group, phase 3 trial. The Lancet, Aug 14 2021;398(10300):583-598 (doi:10.1016/S0140-6736(21)01443-4). Lancet abstract · ClinicalTrials.gov NCT03882970.
This is a "which injectable do I start on" trial, not a switching trial. Everyone enrolled was insulin-naive. If you're already on a GLP-1-class drug and weighing a switch, SURPASS-2 (tirzepatide vs. semaglutide) is the more relevant comparison — SURPASS-3 speaks to the decision between starting basal insulin or starting tirzepatide.
Insulin-associated weight gain is real and measurable. A mean 2.3kg gain over one year in the degludec arm isn't an outlier — weight gain is a known, expected effect of exogenous insulin therapy as glucose control improves. Anyone comparing an insulin-based regimen against a GLP-1/GIP-based one should weigh that tradeoff, not just the HbA1c number.
Lower hypoglycemia risk doesn't mean zero risk. Tirzepatide's hypoglycemia rate was much lower than insulin's, but not absent — especially in anyone also taking a sulfonylurea or already on background insulin, which this trial's insulin-naive design doesn't address.
The GI side effects follow the same pattern seen across the tirzepatide trial program. Nausea and GI symptoms cluster early, during dose escalation, and tend to ease with time — which is why prescribers titrate slowly rather than starting at a target dose.
This article summarizes a published, peer-reviewed clinical trial for general education. It is not medical advice. Individual results vary, and any decision about starting, switching, adjusting, or stopping any GLP-1, GIP/GLP-1, or insulin therapy should be made with a prescriber.