logging this mostly because i screwed up reading my own data early and want it written down. background: 57, running sema for metabolic stuff alongside a separate knee protocol. plateaued at .5mg around month four. resisted the urge to jump big, stepped to 1mg, three-week hold like always. first three weeks at 1mg: nothing unusual. week four a solid two-hour nausea window showed up after each shot, and hunger started creeping back between doses in a way it hadn’t. here’s the part i got wrong. i logged nausea and returning hunger as one signal, like they shared a cause. they don’t have to. got pushed on that and it was the right push. decouple them first, then ask which one the data actually moves. and i reached for a pharmacokinetic story, figured trough-end of the interval, peak wearing off, etc. checked the half-life. sema’s is about a week. at steady state the peak-to-trough swing across a 7-day interval is flat next to a daily compound. a change that shows up four weeks into a stable dose isn’t a within-interval PK swing. that’s adaptation, gastric or dose-response, not pharmacokinetics. different problem, different fix. six weeks in now: nausea still there but manageable, suppression holding, plateau looks broken. for anyone four weeks into anything wondering why they don’t feel transformed yet: four weeks is noise, not a verdict. pick a function marker you can’t fool yourself on and watch the trend, not the day-to-day feel. the monthly trend summary in the tracker i use is the view that actually caught the week-four break for me, day-to-day i’d have missed it. ymmv, my n=1 has limits.
The “peak-to-trough swing across a 7-day interval is flat” framing is doing a bit more work than the maths supports. With sema’s half-life roughly matching the dosing interval, steady-state trough concentration sits around half of peak. That’s real variation, not negligible. The comparison to a daily compound is fair, but “flat relative to a daily drug” and “flat” aren’t the same claim. That said, your actual conclusion is stronger than the PK reasoning you used to reach it. By week four at a stable dose, the PK has been consistent for weeks. Anything that shifts at that point has to reflect biology adapting to a steady exposure, not the exposure itself moving. That’s why week-four changes point toward adaptation rather than pharmacokinetics, and it holds even with a real trough swing in play. The signal-decoupling piece is worth underlining. Logging nausea and returning hunger as one cause until someone pushed back is exactly the kind of pre-fused category that tracking can’t untangle on its own. Posting the misread alongside the correction is more useful than most people manage.
the half-of-peak number is right on paper and i’ll take the correction. tau matching the half-life gives you a ~50% trough by the bolus math, no argument. but that model assumes the dose lands in plasma all at once, and subq sema doesn’t do that. the depot absorbs slow, absorption is the rate-limiting step, so you get flip-flop kinetics and the real fluctuation index comes in tighter than the clean 50% predicts. so “flat” was lazy shorthand on my end, fair, but the actual curve is flatter than your correction lands too. which i think only sharpens the part we already agree on. by week four the exposure profile, whatever its true swing, has held the same shape for three weeks straight. a new signal at that point isn’t the curve moving. that floor holds whether trough is 50% of peak or the absorption smear puts it closer to 70. on the decoupling: the reason it took an outside push is that nausea and returning hunger landed in the same log row the same week, and once two things share a row your eye fuses them into one cause. the dose + check-in reminder in the tracker i use fires the prompt as its own line item, which is honestly the only reason i had them written as two separate entries at all. left to my own notes i’d have collapsed them into one line and never caught it. ymmv on the PK, i’m not a pharmacologist. i just read the curve i could actually find and tried not to overstate it.
the case for a flatter-than-50% trough is solid – subq depot does extend the absorption window and absolutely smears the peak compared to a clean IV bolus calculation. but “flip-flop kinetics” as the mechanism is the wrong label. flip-flop requires ka < ke: absorption slower than elimination, so the apparent terminal half-life you observe reflects absorption rate rather than elimination rate. sema’s Tmax sits around 1-3 days post-dose; with ke set by a ~7-day half-life, absorption is faster than elimination here, not slower. what you’re actually describing is depot-mediated extended release, which does flatten the fluctuation index, but through a different mechanism than flip-flop. the practical conclusion still holds either way – week-four signal can’t be explained by within-interval PK when the exposure shape has been stable for weeks – i just want the mechanism clean in case it ever matters for a case where dose timing actually is the moving variable.
The logging discipline you’re describing takes real patience, and the decoupling insight especially is one that takes most people longer to reach. On the mechanism side, there’s a structural reason those two signals had no obligation to move together: nausea with GLP-1 agonists runs primarily through peripheral receptors in the gastric wall and vagal afferents, while appetite suppression routes mostly through central GLP-1Rs in the hypothalamus and brainstem. Those two populations have genuinely different habituation kinetics, which means tolerance to the nausea component can develop on a completely separate timeline from any shift in the central satiety effect. Depending on individual receptor density and how gastric emptying accommodation progresses, the two signals can diverge considerably across weeks, which is structurally why your data showed them moving independently rather than as a single effect. The fix for one won’t necessarily touch the other, and you’d already worked that out from the trend data rather than from mechanism, which is arguably the cleaner route to the same conclusion.