A weight-loss graph can fall for months and then flatten while treatment continues. The puzzle is what that flattening means: has the medicine stopped contributing, or has the balance between food intake and energy use changed? A 2025 analysis of tirzepatide trials documented plateaus among people who remained on treatment and had already lost weight.[1] A separate 2026 mathematical model explored how changing energy dynamics could produce a plateau.[2] These sources answer different parts of the question.
A changing body changes the energy balance
Energy balance compares energy consumed with energy expended. Sustained weight loss involves a deficit: expenditure exceeds intake. As weight falls, the body's energy requirements change, and physiological adaptations can encourage hunger and reduce expenditure. The gap that initially produced weight loss can therefore become smaller even during continued treatment.[1]
Appetite feedback is part of this process. The observed plateau paper describes shifts in hunger and satiety signals after weight loss, including increased ghrelin, a hormone involved in hunger. It also discusses changes in resting energy requirements and adaptive thermogenesis—adjustments in energy expenditure accompanying weight loss.[1]
These mechanisms explain why extending the early downward line indefinitely is unrealistic. A treatment can continue influencing appetite while the overall energy balance moves toward neutral. However, the SURMOUNT plateau analysis did not directly measure the relevant energy-regulation mechanisms. Its weight records cannot determine how much any particular participant's plateau reflected intake, expenditure or another factor.[1] Biology provides a framework for interpretation, not a diagnosis from a graph.
What counted as a plateau in SURMOUNT
Horn and colleagues examined selected participants receiving injected tirzepatide, which acts at GIP and GLP-1 receptors, in SURMOUNT-1 and SURMOUNT-4. These trials enrolled adults with obesity, or overweight with a weight-related complication, without diabetes. The analysis was post hoc: an exploratory examination of existing trial data.[1]
Selection matters. Included participants had received at least 75% of scheduled study treatment, achieved at least 5% weight loss by the study endpoint, reached their assigned or maximum tolerated treatment level by week 24, and had no subsequent reduction through the endpoint. Required endpoint weight measurements further narrowed the sample to 1,438 participants in SURMOUNT-1 and 259 in SURMOUNT-4.[1]
In SURMOUNT-1, plateau meant less than 5% weight change over a 12-week interval and every subsequent assessed interval, starting with weeks 24–36. SURMOUNT-4 used a comparable definition but included one 16-week interval because of its visit schedule.[1]
Less than 5% is not zero. Someone could continue losing weight within that definition. Plateau timing was also constrained by scheduled measurements and identified retrospectively. These findings describe adherent responders under a particular research definition, not a universal deadline for everyone who begins treatment.[1]
The observed timelines are useful, with boundaries
In the selected SURMOUNT-1 sample, average weight reduction was greatest during the first 24 weeks. Most participants met the study's plateau definition by week 72, although approximately 12% had not. Results in SURMOUNT-4 were broadly similar.[1]
| Starting body mass index category | Median time to plateau in SURMOUNT-1 | Meeting plateau definition by week 72 |
|---|---|---|
| Overweight | 24.3 weeks | 90.2% |
| Class I obesity | 26.0 weeks | 88.9% |
| Class II obesity | 36.1 weeks | 87.6% |
| Class III obesity | 36.1 weeks | 87.8% |
These are descriptive results within the selected sample and the study's less-than-5% definition.[1] Body mass index, or BMI, relates weight to height; these categories group starting body size rather than determine an individual's response.
Although the table suggests differences by starting BMI, BMI and waist circumference were not independently associated with plateau timing in the SURMOUNT-1 multivariable analyses, which considered other characteristics together. Younger age and female sex were associated with later plateaus; later plateaus were associated with greater overall loss.[1] Associations do not provide a personal timetable.
SURMOUNT-4 also needs a design qualification: randomization occurred only after a 36-week open-label tirzepatide lead-in. Its continued-treatment data do not represent an unselected group starting treatment at randomization.[1]
What the mathematical model contributes
The 2026 paper by Hubert and colleagues applied Hall's human metabolism model to a representative phenotype—a modeled body profile based on clinical-trial averages. It simulated 176 weeks of treatment. This was a secondary mathematical analysis, not a new trial observing people through those weeks.[2]
In the simulation, weight loss peaked at 24.0% around week 96, followed by a plateau lasting roughly 78 weeks despite continued treatment. Modeled intake fell sharply initially, then rose toward expenditure; expenditure also declined and stabilized. The two approached balance around week 98.[2]
That sequence illustrates a plausible explanation for flattening: the energy deficit closes. It does not establish that every patient follows those numbers, or that the model measured a change in appetite in real participants. Its variability estimates also depend on its assumptions and inputs.[2]
Only the original abstract was available for this article. It describes Monte Carlo methods for representing variability, but does not provide enough detail here to independently assess all assumptions or validation. Its simulated week-96 peak should not be treated as a contradiction of the earlier SURMOUNT plateau timings: the methods, populations represented and meaning of plateau differ.[1][2]
A trajectory to investigate, with safety still in view
A plateau warrants interpretation alongside the wider clinical picture. It is neither automatic proof of drug failure nor proof that every treatment goal has been reached. The observed study supports considering weight maintenance as a phase of treatment, while acknowledging that response varies and no universal plateau definition exists.[1]
Safety remains relevant during continued treatment. The Zepbound label warns about severe gastrointestinal reactions, dehydration-associated kidney injury, gallbladder disease and acute pancreatitis. Its boxed warning concerns thyroid C-cell tumors in rats; whether tirzepatide causes these tumors in humans is unknown.[3] A slower weight trajectory does not resolve those risks or justify self-directed treatment changes.
Nor does it justify consuming laboratory research materials. FDA warns that unapproved GLP-1 products lack its review for safety, effectiveness and quality, and that some sellers falsely use research-only labeling while marketing for human use.[4] Legitimate reagents may be analytically characterized, but chemical identity or purity cannot establish clinical suitability, injectable sterility or safety. Reconstitution does not convert laboratory evidence into a medicine's evidence base. Research materials should not be consumed or self-administered; they should remain within their intended laboratory use.[4]
Source note: The full Horn paper was reviewed; Hubert et al. was read as its original abstract, not the full modeling paper. Relevant current Zepbound warnings and FDA safety information were retrieved October 23, 2026. Observed selected-participant results are kept separate from simulations. This is editorial source review, not medical review or a Chudik experiment.