№ xlii The Almanac of GST · EN IT

Enrico·rubbo.li

Tech · Longevity · Markets · Opinions ※Enrico Rubboli, propr. ※Dubai, UAE
essay November 17, 2026 16 min

The Weight the Scale Cannot See

In the early 1990s, an endocrinologist at the Bronx VA named John Eng became curious about a strange fact: the Gila monster, a venomous lizard that eats a handful of times per year, keeps its metabolic machinery in flawless order between meals. From its salivary glands he isolated a peptide, exendin-4, that mimics a human gut hormone called GLP-1 but survives far longer in the bloodstream. The drug industry turned that lizard molecule into a class of therapies that now has a market problem nobody would have predicted: it works too well.

The drugs that grew out of that discovery, semaglutide and tirzepatide, the compounds sold as Ozempic, Wegovy, Mounjaro, and Zepbound, produce weight reductions of 15 to 24 percent, numbers that were unimaginable outside of surgery a decade ago. By 2026, 11 percent of American adults reported taking one.[[2]](#ref-2) They work. Anyone who disputes that has not read the trials.

But the drugs have a shadow problem, and it is the one I have covered twice on this site: a substantial fraction of the weight they strip off is not fat. It is lean tissue, mostly skeletal muscle, and along with it goes strength. The muscle loss on Ozempic and its cousins has been measured repeatedly in trials, estimated carefully in reviews, and flagged in editorials asking whether it is clinically relevant. What has never existed, until this month, is a randomized trial testing whether the obvious countermeasure actually works when stacked on top of the strongest drug in the class.

That trial was published in Nature Medicine on November 9. It is called T-REx, and its result is the cleanest piece of evidence yet for a claim this site keeps making: the drug and the barbell solve different halves of the problem, and you need both.

What the drugs actually take

The first generation of data came from the registration trials. In STEP 1, once-weekly semaglutide produced roughly 15 percent body weight reduction over 68 weeks;[[3]](#ref-3) in SURMOUNT-1, tirzepatide pushed past 20 percent.[[4]](#ref-4) When body composition was measured with DXA scans, the pattern under the weight loss was consistent: total fat dropped by a quarter to a third, and total lean soft tissue dropped by 11 to 14 percent.[[5]](#ref-5) Roughly a quarter of everything lost was lean.

That number sounds abstract until you convert it. Lean soft tissue is mostly skeletal muscle, water, and organs, and muscle is the tissue that walks you upstairs, carries your groceries, and absorbs glucose after a meal. An MRI substudy of tirzepatide in type 2 diabetes confirmed that skeletal muscle volume itself shrinks on the drug, not just water weight.[[6]](#ref-6)

The mechanism is not mysterious, and it is not the drug attacking muscle directly. These compounds work by quieting appetite. A large caloric deficit burns through both fuel depots and structural tissue, exactly as any starvation diet does. The drugs are not the villain here; they are a faster and more reliable calorie deficit than most humans have ever managed voluntarily. The question is what else you do while the deficit runs, which is the question every crash diet has always asked and almost nobody has ever answered with a randomized trial.

I covered the pharmacology and the fat-to-muscle trade in the shot does not train. This article is about the trial that finally put numbers on the answer.

The T-REx trial

The design is simple, which is why it matters. One hundred twenty adults with obesity or overweight with comorbidities, mean age 57.7, half women, were randomized to two arms. Both received tirzepatide for 40 weeks, titrated up to the maximum 15 milligram weekly dose. One arm was told to go about its usual activities. The other added three supervised resistance training sessions per week, from the first week of the drug to the last.

Everything was measured with the right instruments: DXA scans for body composition every eight weeks, and actual one-repetition-maximum strength tests for the bench press, the lat pulldown, and knee extension. This last detail deserves emphasis, because nearly all previous drug trials counted kilograms of tissue and never asked whether the remaining body could do anything. The trial, run by Daniel Green’s group and published in Nature Medicine, is the first randomized test of resistance training combined with an incretin mimetic from the outset of therapy.[[1]](#ref-1)

Ninety-three percent of participants finished. Adherence to the training sessions averaged 91 percent. The trial lasted long enough, 40 weeks, for the drug to reach full dose and full effect, which matters because the interesting biology shows up over months, not weeks.

The scale told both groups the same story

First, the boring result, which is the important one to state precisely: the scale cannot tell the groups apart.

Weight loss was statistically identical in both arms. Fat mass fell similarly; the between-group difference was less than a kilogram, not significant. Waist circumference and BMI fell the same. Blood pressure dropped by fifteen to seventeen millimeters of mercury in both groups. Fasting glucose, fasting insulin, HbA1c, total cholesterol, and triglycerides all improved to the same degree. HDL cholesterol did not move in either group.

So if your entire model of health is a bathroom scale and an annual lipid panel, resistance training did nothing here. The paper’s own framing supports the reading that the lifting did not amplify any of the conventional wins of the drug. The conventional wins arrived in full regardless.

Everything interesting happened underneath those numbers.

The bodies told two different stories

The drug-only group lost 4.67 kilograms of total lean soft tissue over 40 weeks, about 11 percent of what they started with, which matches the range from earlier trials. The lifting group lost 3.57 kilograms, a statistically significant 1.09-kilogram difference. The appendicular lean tissue, the arms and legs, the muscle that actually moves you through the world, showed the same pattern: 3.52 versus 2.94 kilograms lost.

Then the strength data, which is where the trial stops being interesting and becomes startling.

The drug-only group got weaker while losing weight. Bench press strength fell 8.4 percent, lat pulldown 6.5 percent, knee extension 9.0 percent. The lifting group got dramatically stronger: bench press up 10.7 percent, lat pulldown up 15.7 percent, knee extension up 17.5 percent. The net difference between the two groups after 40 weeks was roughly 19 to 23 percent on every measure.

Read that again with the identical scale results in mind. Two people step off the same machine, down the same twenty-odd kilograms, with the same improved blood panel. One of them is now weaker than when they started, in every direction they can push or pull. The other is stronger by a sixth to a fifth in absolute terms. The scale reported the same event. The bodies describe opposite trajectories.

There is a further number in the paper that I think is the one people will remember. Muscle mass declines naturally at roughly 0.5 to 1 percent per year from middle age onward.[[8]](#ref-8) The lean loss in the drug-only group, converted from DXA soft tissue to actual skeletal muscle using published prediction models, comes to about 2.5 kilograms.[[7]](#ref-7) At the natural rate of decline, that is what happens to a body over a decade or more of ordinary aging. The trial compressed years of sarcopenia into 40 weeks, in a group whose average age was under sixty. And that was the group that took the drug with no countermeasure, in a supervised, well-fed, modern clinical setting.

One more pattern worth noting: the lean mass loss was not linear. It fell steeply for 24 weeks and then plateaued. The body defends what it can. The lifting group paid less at every timepoint.

The protein question

Here is the detail that makes the trial intellectually honest, and slightly uncomfortable for people who, like me, have been repeating the protein gospel. The researchers collected food diaries, and the two groups ate the same amount of energy and the same amount of protein.[[1]](#ref-1) The dramatic difference in strength and lean mass happened at equal protein intake.

This does not overturn the protein argument; it locates it. Protein sets the floor. The GLP-1 protein protocol matters because under-supplemented patients on these drugs lose disproportionately more lean tissue, and because muscle cannot be built without raw material. But protein alone, in this trial, preserved nothing: the equal-protein drug-only group still lost a decade’s worth of muscle. The mechanical signal, the thing that tells muscle it is needed rather than expendable, was the differentiator. Nutrition and training are not competitors. One is the substrate, the other is the instruction.

That division of labor is the same logic behind the insulin resistance diet I described last week: the protein target is calculated on desired body weight precisely because the goal is not to weigh less, it is to be less fat and at least as strong.

Why function beats composition

The clinical stakes are well documented, just poorly advertised. Loss of skeletal muscle predicts falls, fractures, and frailty in older adults.[[9]](#ref-9) The ATTICA cohort found that low muscle mass independently tracked with ten-year cardiovascular disease incidence in middle-aged and older adults.[[10]](#ref-10) Muscle is also the body’s dominant glucose sink, which is why the muscle chapters of the metabolic health and metabolic flexibility stories always come back to the same tissue.

The framing I keep returning to is this: weight loss without strength preservation is not a smaller version of the same person. It is a different clinical picture, one that weighs less on the chart and performs worse in the world. We now have randomized evidence that the choice between those pictures is, in large part, elective.

The honest opposing case

The objections to leaning hard on this trial are real, so here they are in full.

The training was supervised, center-based, three times a week, and achieved 91 percent adherence. That is the most favorable possible environment, and nobody’s actual life looks like it. Community-effectiveness versions of this intervention will produce smaller effects and weaker adherence, and it would be surprising otherwise. The result establishes that resistance training can preserve and build strength on these drugs, not that it conveniently will.

It is one trial, in 120 people, over 40 weeks, with a mean age of 57.7. It did not test older, frailer adults, the population where the sarcopenia stakes are highest. It tested tirzepatide, not semaglutide, and while the mechanism generalizes, the numbers are drug-specific. The lean mass measured by DXA is not pure muscle; the paper converts carefully, but the conversion has error bars. And lifting produced zero additional improvement in blood pressure, glucose, or lipids beyond the drug’s own effect: the trial’s cardiometabolic story belongs entirely to tirzepatide.

There is also a pharmaceutical answer approaching fast. Two trials published this year, bimagrumab with semaglutide and apitegromab with tirzepatide, showed that muscle-preserving drugs can blunt the lean loss pharmacologically.[[12]](#ref-12),[[13]](#ref-13) If those reach approval, the pitch will be to solve the muscle problem with another injection. But those drugs target mass, and mass is not function. The T-REx trial is the strongest existing evidence that the intervention which improves muscle function, not just muscle kilograms, is three sessions a week of picking things up and putting them down, a technology whose total cost is a pair of shoes and whose side-effect profile has been studied for a century.

Earlier work had already hinted at the shape of this result: in the liraglutide era, pairing the drug with exercise improved weight-loss maintenance.[[11]](#ref-11) T-REx upgrades the hint into proof, with a stronger drug and a stronger modality.

What I would take from it

If you are on one of these drugs, or about to start, the trial maps directly onto a protocol that fits in three lines.

Start resistance training in the same week the injection starts, not after the weight plateaus; the first 24 weeks are where most of the lean tissue leaves, which is exactly when the mechanical signal needs to be loudest. Set the protein target at 1.6 to 2 grams per kilogram of desired body weight and treat it as the floor, knowing it is necessary and not sufficient. And track strength, not just weight: a simple log of what you can lift is a better instrument of what these drugs are doing to your body than the bathroom scale will ever be, which is a point the longevity protocol and the resistance training essays both argue in detail.

The trial’s own numbers make the case better than any argument I could add. Same scale. Same labs. Opposite futures.

The drug decides how much weight you lose. The barbell decides what you are left holding.


References

1. Green DJ, Haynes A, Underwood DR, et al. (2026). Resistance exercise to attenuate lean mass loss during tirzepatide therapy for obesity: a randomized controlled trial. Nature Medicine. https://doi.org/10.1038/s41591-026-04752-5 2. Witters D. (2026). In U.S., GLP-1 usage reaches new high. Gallup. https://news.gallup.com/poll/712157/glp-usage-reaches-new-high.aspx 3. Wilding JPH, Batterham RL, Calanna S, et al. (2021). Once-weekly semaglutide in adults with overweight or obesity. New England Journal of Medicine, 384(11), 989-1002. https://pubmed.ncbi.nlm.nih.gov/33452951/ 4. Jastreboff AM, Aronne LJ, Ahmad NN, et al. (2022). Tirzepatide once weekly for the treatment of obesity. New England Journal of Medicine, 387(3), 205-216. https://pubmed.ncbi.nlm.nih.gov/35658024/ 5. Locatelli JC, et al. (2024). Incretin-based weight loss pharmacotherapy: can resistance exercise optimize changes in body composition? Diabetes Care, 47(10), 1718-1730. https://doi.org/10.2337/dci23-0100 6. Sattar N, et al. (2025). Tirzepatide and muscle composition changes in people with type 2 diabetes (SURPASS-3 MRI). Lancet Diabetes & Endocrinology, 13(6), 482-493. https://doi.org/10.1016/S2213-8587(25)00027-0 7. McCarthy C, et al. (2023). Total and regional appendicular skeletal muscle mass prediction from dual-energy X-ray absorptiometry body composition models. Scientific Reports, 13, 2590. https://doi.org/10.1038/s41598-023-29827-y 8. Mitchell WK, Williams J, Atherton P, Larvin M, Lund J, Narici M. (2012). Sarcopenia, dynapenia, and the impact of advancing age on human skeletal muscle size and strength: a quantitative review. Frontiers in Physiology, 3, 260. https://doi.org/10.3389/fphys.2012.00260 9. Yeung SSY, Reijnierse EM, Pham VK, et al. (2019). Sarcopenia and its association with falls and fractures in older adults: a systematic review and meta-analysis. Journal of Cachexia, Sarcopenia and Muscle, 10(3), 485-500. https://doi.org/10.1002/jcsm.12411 10. Tyrovolas S, et al. (2020). Skeletal muscle mass in relation to 10 year cardiovascular disease incidence among middle aged and older adults: the ATTICA study. Journal of Epidemiology and Community Health, 74(1), 26-31. https://doi.org/10.1136/jech-2019-212268 11. Lundgren JR, et al. (2021). Healthy weight loss maintenance with exercise, liraglutide, or both combined. New England Journal of Medicine, 384(18), 1719-1730. https://doi.org/10.1056/NEJMoa2028198 12. Heymsfield SB, et al. (2026). Bimagrumab plus semaglutide alone or in combination for the treatment of obesity: a randomized phase 2 trial. Nature Medicine, 32, 869-882. https://doi.org/10.1038/s41591-026-04204-0 13. Pratley RE, et al. (2026). Apitegromab for lean mass preservation during tirzepatide-induced weight loss: a randomized, double-blind, placebo-controlled phase 2 trial. Nature Medicine, 32, 2673-2678. https://doi.org/10.1038/s41591-026-04440-4