The Plateau Isn't a Tolerance Problem

Key Finding The most common explanation people reach for is tolerance; the best-documented explanation is energy balance — a smaller body at the same intake reaches a new equilibrium — while no trial has tested adding an agent to a stalled protocol.

The stall arrives on a schedule, which is why it feels like the drug stopping rather than the situation changing. Weeks one to twenty move. Then the scale answers to nothing, and the question becomes: is this the compound failing, or the arithmetic arriving?

It is one of the most common problems in this space, and almost all of the advice around it skips the part that is actually measurable.

What the arithmetic predicts

Body weight does not respond to a change in intake at a constant rate. It responds to a dynamic imbalance, and modelling of human metabolism puts the half-time of that response at around one year (PMID 21872751).

Two consequences fall out of that model, and both look exactly like a stall:

The larger the starting body mass, the larger the expected loss for the same change in intake — so as weight comes down, the same intake produces less movement.
Weight approaches a new steady state slowly, so progress decelerates long before it stops.

That is a system reaching equilibrium, not a receptor desensitising. It predicts a decelerating curve, and a decelerating curve is what most people experience — the question is whether the deficit that produced the earlier movement is still there.

The part that is usually assumed

A deficit is not a property of a vial. It is the difference between what goes in and what is spent, and both halves move.

Intake changes when appetite changes, which is precisely what these compounds do, and it changes without anyone recording it. There is also a well-documented asymmetry in how intake is measured: self-reported intake tends to be lower than measured intake, most sharply in people carrying more body fat. A plateau therefore has at least two candidate explanations that look identical from outside — physiological adaptation, and a deficit that quietly stopped being a deficit.

Both mechanisms are real and are covered in the review literature on weight regain (PMID 37696920, PMID 33461826).

The tolerance story, tested where it can be

If the plateau were tolerance, stopping should not produce much change, and increasing the dose should restore the earlier rate. The best long-term data we have points somewhere else.

In the extension of the STEP 1 trial, 327 participants who had lost a mean of 17.3% of body weight over 68 weeks were followed for a year after treatment (withdrawal). They regained 11.6 percentage points — about two-thirds of the prior loss — leaving a net loss of 5.6%. Cardiometabolic improvements reverted toward baseline alongside it. The authors’ reading was not that the drug had stopped working, but that the condition is chronic and treatment has to continue to hold the result (PMID 35441470).

A 2025 systematic review and meta-analysis of metabolic rebound after GLP-1 discontinuation reaches the same place (PMID 41399474).

One mechanistic result is worth sitting with, because it cuts against a popular framing. In people with obesity, tirzepatide did not change metabolic adaptation; what it increased was fat oxidation. In calorie-restricted mice it reduced the fall in energy expenditure that pair-fed controls showed (PMID 40203836).

So the human finding is not “it repairs a broken metabolism.” It is closer to: the drug shifts intake and fuel use, and the underlying arithmetic still governs the result.

Where the stacking argument is right, and where it is not

This is the part that has to be said precisely, because both sides of the argument are usually overstated.

Combination products have real trial evidence. REDEFINE 1 randomised 3,417 adults to a co-administered cagrilintide–semaglutide combination or comparators over 68 weeks. The combination produced a mean weight change of −20.4% versus −3.0% with placebo — an estimated difference of 17.3 percentage points — and the trial was active-controlled, meaning it included a semaglutide-alone arm rather than only a placebo arm (PMID 40544433). That is a genuine demonstration that adding an amylin component to a GLP-1 produces more loss than the GLP-1 alone. Later analyses of the co-administration strategy point the same way (PMID 42009015).

Ad-hoc stacking at a plateau is a different claim, and it has no trial behind it. No randomised study takes people who have plateaued on one agent and adds a second one at that moment to see whether the plateau breaks. The evidence gap is total, and it runs in both directions: nobody has shown it works, and nobody has shown it cannot.

What follows from the arithmetic is narrower than either camp usually claims. If intake and expenditure are unchanged, an added agent that does not change either one has nothing to act on. If it does change either one, then that is what moved the number — not the act of stacking. Those are different explanations, and they make different predictions about what happens if the item is removed again.

What this suggests if you are stuck

Re-measure the deficit before re-labelling the drug. The model says movement slows as the body gets smaller; it does not say the deficit survived.
Look for the deficit that was never recorded: intake and expenditure both drift, and the drift is invisible from the inside.
Treat the tested combination evidence as what it is — evidence about a co-administered product studied from the start, not a licence to add items to a protocol mid-stall.
If a change is added at a plateau, decide in advance what endpoint would tell you it worked, and what you would remove first to test it.

The reference side carries the published schedules for the components involved — cagrilintide and the CagriSema blend — as references to read, not as a recommendation to add.

What’s next in the Lean Log

The next entry takes the arithmetic seriously in one direction only: what happened when the deficit was measured honestly instead of assumed, with everything else left alone. That includes the uncomfortable part — the weeks where nothing moved because the deficit still was not there.

What this is: one problem examined in research context. N=1 accounts appear in this series only when a contributor’s own log supports them. Not medical advice · no causality claimed · 18+. This series does not recommend protocols; schedule details live on Research Protocols.