Going back through the STEP 1 extension paper this week (the one tracking participants for a year after semaglutide and lifestyle intervention were withdrawn) and a few things stand out that I don’t see discussed much. The headline most people remember is that participants regained roughly two-thirds of the lost weight by week 120. Fine. But the curve underneath that average is the interesting part. Regain wasn’t linear; the steepest slope was in the first ~20 weeks post-withdrawal, then it flattened. That shape matters because it suggests the rebound isn’t purely a slow drift back to setpoint, it looks more like a faster physiological correction followed by a new, lower equilibrium for some subset of people. The cardiometabolic markers tracked the weight pattern but not perfectly. Some improvements (lipids, A1C in non-diabetics) regressed faster than weight did, others lagged. I’d want to see that decoupling unpacked more carefully before anyone uses this trial to argue for permanent therapy in everyone. What’s missing, and what I keep wishing the extension had captured, is any structured measure of appetite or food noise during the off-drug year. Self-reported regain mechanisms would tell us a lot about whether the rebound is mostly hunger reasserting itself or something more metabolic. The study showed regain. The press release said it proves you need the drug forever. Those aren’t the same sentence.
That flattening curve detail matters so much. It’s the difference between “the drug works” and “your body found room to stabilize differently.” I’m still in the window with tirz. What I watch for is whether hunger noise returns or something actually shifted underneath. Regain tells us what happened. Mechanisms tell us what it means. You’re right.
“new, lower equilibrium for some subset of people” is doing a lot of interpretive work for a conclusion drawn from an averaged curve. a flattening group mean is consistent with two completely different stories: some people genuinely stabilizing at a lower setpoint, or fast-regainers plateauing while slow-regainers keep climbing. you can’t separate those without individual-level trajectories, which the extension didn’t report. the food noise gap you’re flagging is real and i’d love that data too. but the subset equilibrium point needs individual data to hold, not just an aggregate slope that slows down.
eta: one more thing
The “regain tells us what happened, mechanisms tell us what it means” framing is rhetorically clean but I’d push on it a little. Mechanisms aren’t actually a separate layer you uncover after the weight data, they’re inferred from imperfect proxies (self-reported hunger, food noise scales, occasionally ghrelin or PYY if you’re in a study that bothered). And those proxies are noisy enough that two people with the same regain trajectory can give you opposite mechanism stories. So I don’t think we get to “what it means” cleanly even with better instrumentation; we get to a slightly less ambiguous version of the same question. The other thing I’d flag, gently, is that being in-window with tirz is the worst vantage point for distinguishing “hunger noise returning” from “something shifted underneath.” On-drug, GLP-1/GIP agonism is suppressing the appetite signal so loudly that you can’t read what your underlying setpoint is doing. The interesting data is in the first 8 to 16 weeks AFTER discontinuation, when the pharmacology has cleared and you’re seeing your actual physiology again. Half-life on paper, vibes in practice, but tirz has a long tail and people often misattribute the post-drug stretch. None of which is a reason not to track. Just that the mechanism question is probably unanswerable from inside the cycle, and possibly unanswerable from self-report at all.
I hear the caution about reading your own physiology while suppressed. The ‘unanswerable from self-report’ frame might be overstating it though. Most people I’ve talked to report pretty similar sequences - appetite returns first, then physical cues lag for weeks, then real surprise at how fast the old patterns resurface. That’s messy but definitely not random. And I’d push back on dismissing the on-drug timeline entirely. Learning what fullness actually feels like while suppressed, even if it’s muted and chemically mediated, shapes how you read your own body after. That’s not noise either. It’s actually part of how people build literacy around what normal means to them.
the food noise gap is real, but even structured self-report over a year post-withdrawal would’ve been noisy without a validated instrument. i track food noise daily (1-10) and the number alone doesn’t tell you much - the journal field tied to each entry in CareClinic is what makes retrospective patterns legible, bc a “4” on day 3 reads totally different than a “4” on peak suppression day. aggregate recall from participants a year out would flatten exactly the mechanistic signal you’re pointing at.
anyone look at whether the plateau people actually reported feeling fine? because that changes everything. if half the cohort stabilized at two-thirds loss and genuinely felt okay, that’s not “permanent drug” that’s “got to a new setpoint and stopped fighting.” i’m seeing something similar with sema right now (month 4 area). there’s the aggressive drop, then you hit something that just feels like equilibrium and your body stops fighting. my knees don’t hurt on stairs anymore, but that’s partly lower load and partly months of different movement patterns rewiring how i carry weight. their regain could be exactly that. not hunger reasserting, just the body relearning how to exist at its new weight. and neither story needs permanent therapy.