GLP-1 Drugs Get a Reality Check: What Tirzepatide, Semaglutide, and Liraglutide Are Actually Doing

Here's the line most people get wrong at the squat rack: "It works because you eat less, so you lose weight, so your blood sugar drops." Half right. A 2025 mediation analysis pooled three SURPASS randomized trials — 2,831 participants on tirzepatide versus placebo or semaglutide — and asked how much of the drug's HbA1c reduction was actually caused by the weight loss. The answer, depending on the comparator and the background therapy: surprisingly little. Against placebo as monotherapy, only roughly 12–27% of the A1c difference was mediated by weight loss; on a background of insulin, it climbed to 25–45%. The rest is the molecule doing molecule things — incretin signaling, insulin sensitivity, hepatic glucose handling.

Against semaglutide, the picture flips: 54–71% of tirzepatide's incremental A1c advantage was weight-loss dependent, which fits the clinical impression that dual GIP/GLP-1 agonism just drags more fat off than GLP-1 alone. Translation for the curious lifter: tirzepatide isn't just a better appetite suppressant. It's a better glucose drug independent of what the scale says — and most of its edge over semaglutide is the extra pounds.

Type 1 is a different beast. Beta cells are gone or going; insulin is the floor, not the ceiling. So the question of whether GLP-1s belong in the T1D toolkit has been live for years. A 2025 systematic review pulled together 25 RCTs and gave the cleanest answer yet. Versus placebo on top of insulin, GLP-1 RAs delivered a mean A1c reduction of 0.23%, a 3.93 kg weight drop, and a 5.74 unit/day cut in total insulin dose. Time-in-range didn't budge meaningfully, and the odds of severe hypoglycemia were similar between groups — important, because that's the failure mode everyone fears.

The honest read: this is a modest, real benefit for the right T1D patient, not a beta-cell rescue. The same analysis noted that adding a GLP-1 did not prevent progressive C-peptide loss, and glucagon counter-regulation to hypoglycemia stayed intact. You're not regrowing islets. You're trimming weight, shaving insulin, and nudging A1c — useful, not miraculous. And this is a clinician's decision, not a DIY one.

Here's the part of the conversation that matters in our world. If a drug strips 15–20% body weight, what's coming off — and what's staying on? A 2025 meta-analysis specifically zoomed in on T2D patients with sarcopenia across nine RCTs and 1,089 participants. The fat-side findings were unambiguous: GLP-1 RAs significantly reduced BMI, fat mass, and body fat ratio. Visceral fat moved in the right direction. Body weight overall trended down without reaching the threshold the authors set.

The muscle picture, per the authors' framing, is more Dionysus-and-Apollo than victory parade: signals are mixed, heterogeneity is real, and the dataset is small enough that anyone shouting either "GLP-1s spare muscle" or "GLP-1s gut muscle" is overstating what we know. The practical takeaway for the gym-literate reader: if a peptide is doing the calorie-deficit heavy lifting, the resistance training, the protein floor, and the sleep all matter more, not less. None of that is new science. It's just the part the influencers skip.

Two more 2025 entries pushed the GLP-1 story further off-label. The first: a retrospective database analysis of more than 2.5 million diabetic patients comparing metformin, GLP-1 agonists, sulfonylureas, and short-acting insulin for Alzheimer's risk. The headline result raised eyebrows — metformin use was associated with a statistically significant increased likelihood of AD diagnosis compared to GLP-1 use (HR 2.228), undercutting the long-running assumption that metformin was the neuroprotective champ. Diabetes itself raised AD risk versus non-diabetics. Retrospective data can't prove causation — channeling bias and confounding-by-indication are live concerns — but the GLP-1-favoring signal lines up with mechanistic work on incretin signaling in the brain. Worth watching, not worth prescribing on.

The second: a mechanistic mouse study showed liraglutide preserved endothelium-dependent vasodilation in ophthalmic arteries of septic mice, blunting oxidative stress markers and NOX2 overexpression. This is preclinical, in a single vascular bed, in a sepsis model — about as far from a clinical recommendation as research gets. But it adds to a coherent picture: GLP-1 receptor activation has vascular and anti-inflammatory effects that aren't explained by glucose or weight. Whether that translates to humans, in any tissue that matters, is the next decade's question.

The GLP-1 class is doing more than suppressing appetite — that part of the hype is grounded. Tirzepatide's glycemic muscle isn't all about the scale. The T1D add-on case is real but modest. The muscle question deserves serious tracking, not panic. The Alzheimer's and vascular signals are intriguing and early. None of it makes these drugs casual supplements. They're prescription medications with real effects, real side effects, and real unknowns — and the smartest thing a lifter can do with this evidence is read it carefully, train hard, eat enough protein, and bring any actual clinical question to an actual clinician.