How much of the weight effect is delayed gastric emptying, and is the plateau receptor desensitisation?

Two related mechanism questions I have never seen answered together.

First, how much of the weight loss is attributable to delayed gastric emptying as opposed to central satiety signalling? The gastric effect is easy to demonstrate and easy to feel, and it would be a tidy explanation for reduced intake. But I have also read that the gastric-emptying delay shows tachyphylaxis with continued exposure, which would predict that the weight effect fades, and it does not.

Second, the plateau. Weight loss slows and stops at somewhere around 60 to 80 weeks in the published curves. The explanation I see most often is receptor desensitisation - the receptors downregulate, the drug stops working as well, and the loss stops. That is mechanistically plausible given everything about arrestin recruitment and internalisation, but I do not know whether there is evidence for it or whether it is just an appealing story.

These feel like the same question from two directions: if tolerance develops to the gastric effect, why not to the central effect, and if it develops to the central effect, what stops the weight coming back while still on drug?

Adding a quantitative sanity check on the energy-balance account, because it is testable with arithmetic and the numbers come out sensibly.

Take a participant starting at 105 kg who reaches -15%, so about 89 kg, a loss of roughly 16 kg. Rules of thumb from the metabolic literature put the fall in total daily energy expenditure at roughly 20 to 25 kcal per kilogram of weight lost, once both the reduced tissue mass and the reduced cost of movement are counted.

• Expenditure reduction: 16 kg x 22 kcal/kg = about 350 kcal per day.
• So maintaining 89 kg requires roughly 350 kcal per day less intake than maintaining 105 kg did.
• To have lost 16 kg over roughly 60 weeks, the average daily deficit was about 16 kg x 7700 kcal/kg divided by 420 days = 123200 / 420 = about 293 kcal per day.

Now the point. The average deficit that produced the loss, about 293 kcal per day, is smaller than the expenditure reduction the loss caused, about 350 kcal per day. So by the time you reach that weight, the appetite suppression that was generating a 293 kcal deficit is now fully consumed by the 350 kcal expenditure reduction, and the deficit closes. That is the plateau, arriving at exactly the weight where the two quantities cross, and it requires no change in drug effect whatsoever.

The same arithmetic explains the dose-response: a larger pharmacological offset takes longer to be consumed, so the crossing point occurs at a lower weight. And it explains regain on withdrawal: remove the offset while the 350 kcal expenditure reduction persists, and you have a substantial positive balance until weight returns towards baseline.

The numbers above use standard approximations - 7700 kcal per kilogram of tissue, 20 to 25 kcal per kilogram per day of expenditure - and should be read as order-of-magnitude reasoning rather than measurement. The conclusion is robust to plausible variation in both constants, which is the useful property.

One caveat on the withdrawal-trial argument, since it is doing a lot of work in the accepted answer and it is not airtight.

The inference is that regain on withdrawal proves the drug was still active at plateau. That is sound. But there is a weaker version of the desensitisation hypothesis it does not exclude: partial tolerance. Suppose receptor responsiveness declines by some fraction over a year. The drug would still be doing substantial work - so withdrawal still causes regain - while contributing to the plateau arriving earlier or higher than it otherwise would.

Distinguishing partial tolerance from pure energy-balance equilibrium would need something the published trials do not provide: a dose-escalation challenge at plateau. If you increased the dose in someone who had plateaued and weight loss resumed proportionally, that argues against meaningful tolerance. If it resumed less than the original dose-response predicted, that would suggest some was present. Fixed-maintenance-dose designs cannot answer this, and maximum-tolerated-dose designs confound it with tolerability.

There is indirect evidence against strong partial tolerance: the 104-week data show maintenance rather than erosion, and if responsiveness were declining materially you would expect slow regain at a fixed dose. But "no detectable erosion over two years" is not the same as "no tolerance", and the durations studied are short relative to how long anyone would use these agents.

Worth flagging as a genuine open question rather than a settled one. The strong desensitisation story is refuted; the weak one is unaddressed.