Japanese cohort finds 2.5 mg and 5 mg tirzepatide produce similar six-month weight loss

A prospective study from Japan published July 28, 2026 in Diabetes, Obesity and Metabolism reports that non-diabetic adults with obesity who stayed on 2.5 mg of weekly tirzepatide lost about as much weight over six months as those who moved up to 5 mg, while reporting fewer side effects ^[1]. It is one of the few published attempts to compare low maintenance doses head to head rather than treating 2.5 mg only as a starting step.

What the study did

Researchers at Hiroshima Cardiovascular Clinic, Onomichi General Hospital and Hiroshima University ran a multicenter, prospective, non-randomized cohort study in Japanese adults aged 18 to 65 with a body mass index of 30 kg/m² or higher, or 27 kg/m² or higher with weight-related health conditions ^[1]. Everyone started tirzepatide at 2.5 mg once weekly. After four weeks, each person either stayed at 2.5 mg or increased to 5 mg based on shared decision-making with their clinician — not by random assignment ^[1]. All participants also got standardized diet and exercise counseling ^[1].

The primary outcome was the share of people who reached a BMI under 25 kg/m² or lost more than 15% of their body weight at six months ^[1].

The numbers

Of 112 participants analyzed, 58 stayed on 2.5 mg and 54 went to 5 mg. Mean baseline BMI was 30.8 kg/m² and the authors describe baseline characteristics as well balanced ^[1].

At six months, the primary outcome was met by 62.1% of the 2.5 mg group and 63.0% of the 5 mg group (hazard ratio 0.93, 95% confidence interval 0.58 to 1.49; p = 0.772) ^[1]. Mean percent weight loss was −15.3% with 2.5 mg and −16.1% with 5 mg (p = 0.563) ^[1]. Reductions in BMI, skeletal muscle mass and bone mass were similar between the groups ^[1].

Side effects went the other way. Adverse events occurred in 50% of the 5 mg group versus 35% of the 2.5 mg group, mostly gastrointestinal symptoms ^[1]. Adherence to the diet plan was 96.4% in both groups, while exercise adherence was only 37.5% ^[1].

The authors conclude that the two doses produced similar weight loss at six months with better tolerability at the lower dose, and that this "supports the potential utility of low-dose tirzepatide strategies in real-world practice" ^[1]. They report no conflicts of interest ^[1].

Why it matters for patients

Tirzepatide is sold as Mounjaro for type 2 diabetes and Zepbound for weight management. Cost, supply and side effects all push some people to ask whether a smaller dose could do most of the job. This study is the closest published evidence to that question — but it is not the kind of evidence that settles it.

The biggest limitation is the design. Because the dose was chosen through shared decision-making rather than randomization, people who felt fine and were losing weight may have been more likely to stay at 2.5 mg, while others moved up. That can make the lower dose look better than a randomized comparison would ^[1].

The population also differs from a typical US clinic. Mean BMI was 30.8 kg/m², at the lower end of obesity, and dietary adherence was reported at 96.4% — far higher than most real-world settings ^[1]. All participants were Japanese adults under 66 without diabetes ^[1]. Whether the same pattern holds in heavier, more diverse populations or in people with type 2 diabetes is not addressed by this study.

Finally, the follow-up was six months ^[1]. Whether weight loss at 2.5 mg continues, plateaus or reverses beyond that point is not reported here. The study also does not report costs, insurance coverage, or any comparison with higher tirzepatide doses such as 10 mg or 15 mg.

What happens next

The paper appears in Volume 28, Issue 10, pages 9395 to 9402, first published online July 28, 2026 ^[1]. The authors say the underlying data are available from the corresponding author on reasonable request but are not public because of privacy and ethical restrictions ^[1]. No randomized trial comparing 2.5 mg against 5 mg as a maintenance strategy is described in the sources, and no regulatory or labeling change is proposed by the authors ^[1].