There is no evidence base for GLP-1 medication in people who are not overweight. A query of 354 registered semaglutide and tirzepatide trials found not one weight-management efficacy trial with an entry threshold below BMI 23, and the global programme floor is 27 (registry query 2026). That absence is the whole point of this article, and it is compounded by a second finding: the leaner a person already is, the greater the share of any further weight they lose that comes from lean tissue rather than fat. For the last few kilograms, the tools with evidence behind them are training and protein.

Fat-free mass: everything in the body that is not fat — muscle, bone, organs and water. When an article says weight loss "cost lean tissue", this is the quantity that fell.

Has anyone actually studied this?

No, and the search was specific. Of 354 registered interventional trials of semaglutide or tirzepatide in obesity, overweight or weight loss, only 18 mentioned any BMI figure below 25 in their eligibility text — and on inspection those were phase 1 pharmacokinetic studies in healthy volunteers or trials of other drugs. The lowest entry threshold in any efficacy trial was BMI 23, in a phase 4 study in India. The Asia-specific phase 3 trials used local floors: 28, or 24 with a comorbidity, in China (registry query 2026).

The registration programme reads the same way. Every weight-management trial in the US semaglutide label enrolled people with a BMI of 30 or above, or 27 and above with a weight-related condition (US label).

So a person at BMI 22 asking what this medication would do to them is asking a question that has never been put to a trial. Not answered unfavourably — never asked. What can be said is that nobody has measured it, and anyone who says otherwise is extrapolating from a population they are not in.

The thresholds, and why Singapore runs two of them

Singapore's registered product information for semaglutide states the weight-management indication in exact terms: an initial BMI of "≥30 kg/m² (obesity), or ≥27 kg/m² to <30 kg/m² (overweight) in the presence of at least one weight-related comorbidity" (NDF Singapore). Tirzepatide's current registrations carry a weight-management indication with the same BMI thresholds; the 2023 injection registration SIN16718P covers type 2 diabetes only, with no BMI threshold anywhere in it.

Singapore's own national obesity guideline classifies overweight from BMI 23.0 and obese from 27.5 for the local population (HPB-MOH 2016), following the WHO expert consultation that proposed additional action points at those numbers because risk rises in Asian populations below the standard international threshold (WHO 2004). The reasoning behind those lower numbers is set out in why Asian BMI thresholds are lower.

Both scales are correct and they are not the same scale. The label's 27 and 30 are the international BMI cut-offs; they have not been adjusted for Asian bodies. A Singaporean can therefore be classified obese under the national guideline while sitting below the registered label's threshold for an uncomplicated prescription. That gap is real, and it is not resolved by picking a number. Whether medication is appropriate for a particular person is a clinical decision made by the doctor at the consultation, taking the whole picture into account.

Why leanness changes the arithmetic

The relevant relationship was described by Forbes from human underfeeding, overfeeding and twin-pair data, and the sentence that matters reads: "In humans, there is an inverse curvilinear relationship between initial body fat content and the proportion of weight loss consisting of lean tissue" (Forbes 2000). The same paper adds a second lever: "as energy intake is reduced, lean tissue makes up an increasing fraction of the total weight loss".

Read plainly, that is two multipliers stacking in the same direction for exactly the person this article is about. Someone already lean is at the wrong end of the first relationship. Someone chasing a stubborn last few kilograms usually cuts harder, which puts them at the wrong end of the second — sharper still when there is a date on it, which is what deadline weight loss before a big day works through.

Two honest qualifications belong with it. Forbes describes a modelled relationship derived from observational and experimental body-composition data, not a randomised trial, and it predates these medicines entirely — it is not GLP-1 evidence. And the numeric figures in that paper are for overfeeding and for twin-pair differences, not for underfeeding, where Forbes states the relationship qualitatively rather than with a number. So the direction is well described; a specific percentage for a specific lean person is not available.

Later mathematical modelling checked the equation against published human data and confirmed it for modest weight changes, while showing that for large weight losses the original equation understates how much fat-free mass is lost (Hall 2007). The correction runs in the less comfortable direction.

The general dieting literature agrees on the deficit half independently. Across 16 dietary studies, the median share of weight lost as fat-free mass was about 14% on standard low-calorie diets and about 23% on very-low-calorie diets, and the degree of caloric restriction correlated with the proportion lost as fat-free mass (Chaston 2007). Harder deficit, worse composition — measured in people who were not lean to begin with.

What it means for protein

Protein requirements move in the same direction, and this is documented separately from Forbes. A systematic review in lean, resistance-trained athletes during caloric restriction concluded that protein needs are likely "2.3-3.1g/kg of FFM scaled upwards with severity of caloric restriction and leanness" (Helms 2014).

Two things about that figure. The denominator is fat-free mass, not body weight — using body weight would overstate the target substantially, and the mix-up is common enough to be worth checking whenever the number appears. And the population is lean resistance-trained athletes, which for once is closer to the reader of this article than to a typical clinic patient.

The direction of travel is the useful part: the leaner someone is and the harder they cut, the more protein it takes to defend the tissue they are trying to keep. That is the same conclusion Forbes reaches from body-composition data, arrived at independently from nitrogen balance and trial evidence.

The one study in people below BMI 25

There is a single real-world analysis of incretin therapy in this weight range, and it is not about weight management. Of 7,942 patients prescribed a GLP-1 or GLP-1/GIP medicine in one US health system, 100 had a BMI under 25, and 45 had at least six months of therapy and complete records. All of them were being treated for type 2 diabetes, with dosing optimised for blood glucose. Average weight fell 1.61 kg; the change in BMI was not statistically significant (Carson 2025). The authors' own conclusion calls for further study of safety and efficacy in normal-weight and underweight people.

Forty-five people, retrospectively, treated for a different purpose. It cannot tell anyone what these medicines do to a lean person seeking weight loss, and it should not be quoted as if it could.

What "no evidence" does and does not mean

It does not mean the medication has been shown to be unsafe or ineffective in lean people. Nothing has been shown either way, and claiming harm without evidence is the same error as claiming benefit without evidence.

It also does not mean the labels have quietly permitted it. A full-text search of the US semaglutide label found no occurrence of "eating disorder", "anorexi", "malnutrition" or "underweight", and no warning about use in people without excess adiposity (US label). The control in a prescribing label is the indication itself, not a warning further down — so silence there is not reassurance.

The nearest thing to a professional-society caution comes from a 2025 joint advisory by four US clinical and nutrition societies, which recommends screening for binge eating disorder, anorexia nervosa, bulimia nervosa and night eating disorder before starting, and states in a footnote that "restrictive eating disorder is a general contraindication to GLP-1 use", with referral to an eating-disorder specialist for anyone who screens positive or has a history (joint advisory 2025). That is professional-society guidance rather than a regulatory contraindication, and the advisory itself says the underlying evidence is not well established. It is the right conversation to have with a doctor before starting anything, at any weight.

What the last few kilograms actually respond to

Two things, both with a real evidence base in exactly this situation.

Resistance training is the intervention with demonstrated effect on lean mass during a deficit. Across 114 trials and 4,184 people, lean mass was statistically unchanged where resistance training accompanied caloric restriction, and resistance training alone increased lean mass by about 0.8 kg versus untrained controls (Lopez 2022). The dose required is smaller than most people assume: in resistance-trained men, a single hard set per exercise two to three times a week produced significant strength gains over 8 to 12 weeks, which the authors describe as a floor rather than a target (Androulakis-Korakakis 2020). What that floor looks like in practice is covered in the minimum effective resistance training dose.

Protein is the second, at a total that rises with leanness and deficit severity (Helms 2014). And for someone whose weight has stopped moving but whose shape has not settled, the goal worth examining is whether the remaining problem is weight at all — is losing weight the same as losing fat describes the case where the scale barely moves and the result still arrives.

For a person who is already lean, that is where the evidence sits.

Common questions

Has GLP-1 medication been studied in people who are not overweight?

No. A registry query of 354 registered semaglutide and tirzepatide trials found no weight-management efficacy trial with an entry threshold below BMI 23, and the global programme floor is 27 (registry query 2026). Nobody has measured what these medicines do in people at a healthy weight.

Why does losing the last few kilograms cost more muscle?

Because of a relationship Forbes described from human underfeeding and overfeeding data: the leaner a person is at the outset, the greater the proportion of any weight lost that comes from lean tissue, and a deeper deficit raises that proportion further (Forbes 2000). It is a modelled relationship derived from observational and experimental data, not a randomised trial result.

What BMI do you need for GLP-1 medication in Singapore?

Singapore runs two different scales and they do not line up. The registered semaglutide product information sets the weight-management indication at BMI 30 or above, or 27 with a weight-related condition (NDF Singapore), while Singapore's own national obesity guideline classifies overweight from 23 and obese from 27.5 (HPB-MOH 2016). Whether medication is appropriate for a particular person is a clinical decision made by the doctor at consultation.

Is it dangerous to take GLP-1 medication if you are already lean?

Nobody knows, and the labels do not answer it — a full-text search of the US semaglutide label found no warning, limitation or contraindication addressing use without excess adiposity (US label). Silence in a label is not reassurance. The one professional-society caution that touches this area is a 2025 joint advisory stating that restrictive eating disorder is a general contraindication to GLP-1 use (joint advisory 2025).

What actually works for the last few kilograms?

Resistance training and enough protein. Across 114 trials and 4,184 people, lean mass was statistically unchanged where resistance training accompanied caloric restriction (Lopez 2022), and protein needs scale upward with leanness and with the severity of the deficit (Helms 2014). Those two have an evidence base for this situation; medication does not.