The drug moves the weight; your plate protects what’s underneath
On a GLP-1, food is not what drives the scale down — the medication does that. Your job at the table is narrower, and more important: protect the muscle the drug will otherwise strip, and keep carbohydrate quality high so insulin, the engine behind PCOS, stays low.
That reframes every “PCOS diet” rule you’ve read. Here is what the evidence actually supports, and where clinical judgment is filling the gaps.
What the molecule can actually do
In STEP 1 (NEJM 2021, 1,961 adults with obesity and no diabetes), semaglutide 2.4 mg weekly produced −14.9% mean body weight at 68 weeks versus −2.4% on placebo. SURMOUNT-1 (NEJM 2022, 2,539 adults) pushed further: tirzepatide reached −20.9% at 15 mg versus −3.1% on placebo. Neither trial enrolled women for PCOS — these are obesity populations — so treat the numbers as the ceiling of what the molecule can do, not a PCOS-specific promise.
The PCOS relevance is mechanistic. GLP-1 receptor agonists slow gastric emptying, act on appetite circuits in the hypothalamus and hindbrain, and improve insulin secretion; tirzepatide adds GIP receptor agonism. Insulin resistance is the core driver of PCOS androgen excess, so lowering insulin demand — through weight loss and a lighter carbohydrate load — works directly upstream of the problem.
The protein number, and why it isn’t optional
Here is the part the glossy plans skip. DXA substudies of GLP-1 trials show a meaningful fraction of the weight lost is lean mass, not just fat. The standard countermeasure is resistance training plus protein in the range clinicians commonly target: 1.2–1.6 g/kg/day. On a drug that blunts appetite hard, hitting that takes deliberate effort — protein first at every meal, before fullness closes the window.
Be honest about the limit: the long-term functional consequences of that lean-mass loss are not yet well characterized. We protect muscle because the physiology says to, not because a trial has shown who gets hurt if we don’t.
Carbohydrate: quality first, quantity second
Lowering the carbohydrate load lowers post-meal insulin — settled physiology, and the single most useful lever in PCOS eating. The target most clinicians land on is roughly 75–125 g/day, weighted toward vegetables, legumes, oats, and whole fruit, with refined sugar and juice cut hard. Treat that as clinical practice, not a trial result — it is the number that works in clinic, not one we can pin to a named study here.
One position worth stating plainly: very-low-carb and keto are usually overkill in PCOS. Most of the metabolic win comes from the first cut, from “standard” to “lower.” Going below ~100 g buys little extra and is far harder to sustain — and on a GLP-1 that already suppresses appetite, an aggressively restrictive diet is how people undereat protein and lose more muscle. Fiber (aim ~30 g/day) plausibly helps by supporting insulin sensitivity and possibly sex-hormone binding; that androgen-recirculation link is real but less settled — file it under low-cost, probably helpful.
Don’t confuse the plate with the drug
The most important thing nutrition cannot do is hold your weight if you stop the medication. In STEP 4 (JAMA 2021), people who stopped semaglutide at week 20 regained substantially; those who continued kept losing. SURMOUNT-4 (JAMA 2024) was starker — after 36 weeks on tirzepatide, patients switched to placebo regained about 14% of body weight over the next year, while those who stayed on lost another ~5.5%. Diet and training determine the quality of the weight you keep; the drug determines whether you keep it. Plan the medication as ongoing, not a 12-week sprint.
A note on the ramp: nausea is the most common side effect and tracks dose escalation, because it follows directly from delayed gastric emptying — it fades as titration slows. If it is rough, that is a reason to slow the steps with your clinician, not to power through or quit. GLP-1 sits alongside, not instead of, first-line PCOS tools like metformin and spironolactone; semaglutide is one option a clinician titrates to you.
Takeaways
- Evidence: GLP-1s drive 15–21% weight loss in obesity trials, and stopping reverses it; protein plus resistance training is the standard defense against lean-mass loss.
- Consensus and practice: roughly 75–125 g carbs/day, protein 1.2–1.6 g/kg/day, about 30 g fiber, a Mediterranean-style pattern — sustainable beats strict.
- Our opinion: keto is unnecessary for most PCOS; the first cut does the work.
The open question is not which diet — it is that no GLP-1 trial has yet been powered on PCOS-specific endpoints like ovulation rate and androgen levels. Until that trial runs, the weight and insulin data are strong and the reproductive claims are inference. That is the study worth watching for.
