Three-year SURMOUNT-1 data show a 93% relative drop in progression to type 2 diabetes

Adults with obesity and prediabetes who took tirzepatide for 176 weeks — about three and a half years — were far less likely to develop type 2 diabetes than those on placebo, according to results from the SURMOUNT-1 trial published in the New England Journal of Medicine ^[1]. In the tirzepatide groups, 1.3% received a type 2 diabetes diagnosis, compared with 13.3% in the placebo group, a hazard ratio of 0.07 (95% confidence interval, 0.0 to 0.1; P<0.001) ^[1]. That works out to roughly a 93% lower risk over the treatment period.

Tirzepatide is the molecule sold as Mounjaro for type 2 diabetes and Zepbound for weight management. SURMOUNT-1 was a phase 3, double-blind, randomized, controlled trial that enrolled 2,539 participants with obesity, of whom 1,032 also had prediabetes ^[1]. Participants were assigned in a 1:1:1:1 ratio to once-weekly tirzepatide at 5 mg, 10 mg, or 15 mg, or to placebo ^[1]. This analysis focused on the 1,032 people with both obesity and prediabetes, who stayed on their assigned treatment for 176 weeks and were then followed for a 17-week period off treatment ^[1].

The weight numbers

At week 176, mean weight change from baseline was −12.3% with the 5-mg dose, −18.7% with 10 mg, and −19.7% with 15 mg, compared with −1.3% with placebo (P<0.001 for all comparisons) ^[1]. Those figures are notable because they come three years into treatment, not at the usual 68- or 72-week mark. An earlier SURMOUNT-1 analysis had reported substantial weight reduction over 72 weeks ^[1]; this is the longest randomized follow-up of tirzepatide published to date.

The trial also looked at what happened after treatment stopped. After the 17-week off-treatment period, 2.4% of participants who had received tirzepatide and 13.7% of those who had received placebo had type 2 diabetes (hazard ratio, 0.12; 95% CI, 0.1 to 0.2; P<0.001) ^[1]. In other words, the gap narrowed somewhat once the drug was withdrawn, but a large difference remained at 193 weeks.

On safety, the abstract reports that other than COVID-19, the most common adverse events were gastrointestinal ^[1]. Most were mild to moderate and happened mainly during the dose-escalation period in the first 20 weeks ^[1]. The investigators reported no new safety signals over three years ^[1]. The trial was funded by Eli Lilly, which makes tirzepatide (ClinicalTrials.gov number NCT04184622) ^[1].

Why it matters for patients

Prediabetes is common in people with obesity, and a large share of people with prediabetes go on to develop type 2 diabetes. These data suggest that in a randomized setting, sustained treatment with tirzepatide sharply reduced how often that happened over three years, alongside weight loss that held up well past the first year and a half ^[1].

The results also underline a point that comes up often with GLP-1 and dual GIP/GLP-1 medicines: the effect is tied to continued use. When treatment stopped for 17 weeks, more diabetes diagnoses appeared in the tirzepatide group — 2.4% versus 1.3% at week 176 ^[1]. The abstract does not report how much weight participants regained during that off-treatment stretch, so that is not yet known from this source.

A few other limits are worth noting. The diabetes rates reported here pool all three tirzepatide doses, so the abstract does not break out how many people on 5 mg versus 15 mg progressed to diabetes ^[1]. The analysis covers only the 1,032 participants who had prediabetes at baseline, not all 2,539 people in the trial ^[1]. And these are trial conditions, with regular visits and structured dose escalation, which may not match real-world use.

What happens next

The findings drew published correspondence in the New England Journal of Medicine on June 26, 2025, including letters from Birkenfeld and colleagues and from Shukla and Bhavya, with a reply from the SURMOUNT-1 investigators Jastreboff, Bunck, and Ahmad ^[1]. Whether these data change labeling, coverage, or clinical guidelines for diabetes prevention is not addressed in the source material.