Semaglutide: The Evidence Is Not the Problem

Most compounds in this reference have the same shortage. Interesting mechanism, promising rodent studies, almost no human data.

Semaglutide has the opposite situation. The human evidence runs past 25,000 people in outcome trials alone, published in the biggest journals in medicine. Whether it works is settled.

So this entry spends its time somewhere else.

What it is

Your gut already makes a hormone called GLP-1. It tells your brain you have eaten. The problem is it lasts about two minutes before your body destroys it.

Semaglutide is that hormone, rebuilt to survive. Three changes to its structure stretch two minutes into roughly seven days, which is why it is a weekly shot instead of a constant drip.

It works in three places at once. In the pancreas it helps release insulin, but only when blood sugar is actually high. In the brain it turns down appetite. In the stomach it slows how fast food leaves.

The weight loss is mostly that middle one. You are not blocking calories or burning more. You eat less because you want less.

One detail that matters later: it is made by growing it in engineered yeast and then attaching a fatty acid chain chemically. That is a real manufacturing process, not something you can do in a back room.

Does it work

Yes, and here is the size of it.

The main weight trials are called STEP. Percentages are body weight lost over roughly 16 months. For a 230 pound person, 15% is about 34 pounds.

Trial	Who	People	Result
STEP 1	No diabetes	1,961	Lost 14.9%, vs 2.4% on placebo
STEP 2	With type 2 diabetes	1,210	Lost 9.6%, vs 3.4%
STEP 3	Plus 30 coaching sessions	611	Lost 16.0%, vs 5.7%
STEP 5	Ran two years	304	Lost 15.2%, vs 2.6%
STEP 8	Against an older drug	338	Lost 15.8%, vs 6.4%

Three of those rows are more interesting than the headline.

Diabetes cuts the effect by about a third. Same drug, same dose, 9.6% instead of 14.9%. That is a real and consistent difference, and it is almost never mentioned when people quote the big number.

Look at STEP 3's placebo group. They lost 5.7% without any drug at all, because they got thirty counseling sessions. In STEP 1, where placebo meant almost no support, that number was 2.4%. Coaching is doing more than people give it credit for, and it narrows what the drug adds on top.

And STEP 4 is the one to actually sit with. Everyone took the drug for 20 weeks. Then half kept going and half quietly switched to placebo. The ones who continued lost another 7.9%. The ones who switched gained back 6.9%. A separate follow-up tracked people for a year after stopping and found roughly two thirds of the lost weight returned.

This is not a course of treatment you finish. The trials are clear that it works while you take it.

It does more than weight

This is the part that moves it from "an effective diet drug" to genuinely established medicine.

SELECT followed 17,604 people with obesity and heart disease for over three years. Heart attacks, strokes and cardiovascular deaths dropped by 20% compared to placebo. That was the first real proof that treating obesity with a drug prevents heart events. (PMID 37952131)

FLOW followed 3,533 people with diabetes and failing kidneys. Kidney decline and kidney death dropped 24%. They stopped the trial early because the benefit was clear enough that continuing to give people placebo was hard to justify.

STEP-HFpEF tested it in a type of heart failure with very few good treatments. Symptom scores improved roughly twice as much as placebo.

ESSENCE looked at fatty liver disease. Liver inflammation resolved in 63% of the treatment group versus 34% on placebo.

Now the part nobody can answer for you

Every number above came from real pharmaceutical product. Manufactured under inspection, assayed, released.

No trial has ever used research-grade or compounded semaglutide. Not one.

That matters more than it sounds. Those results are evidence about a specific manufactured drug. They are not a promise about a vial that arrived in the mail.

And somebody checked what is in those vials. A 2024 analysis bought semaglutide from three online sellers and put it through a lab. (PMID 39509151)

10.3 % · What was measured

What was measured

What was claimed

0100 %

The three samples came in at 14.37%, 8.97% and 7.7% pure. All three were sold as 99%. Purity here means how much of the powder is actually semaglutide.

Now here is the finding that surprised me, because it is backwards from what almost everyone assumes.

Those vials had more semaglutide in them than the label claimed, not less. Between 29% and 39% more, across all three.

Low purity and too much drug sound like they contradict each other. They do not. The powder was mostly other stuff, and there was still more actual semaglutide in there than advertised. Both things were wrong at once.

The people who ran the analysis said plainly what the risk is: incorrect dosing or overdose, when somebody mixes and injects this at home believing the label.

They also found bacterial toxin in all three samples. Nothing living was growing in them at the time of testing. Checked against 22 standard pharmacy quality criteria, the samples failed 13 and 14 of them.

The street wisdom is that grey-market product is underdosed and weak. In the only published testing I could find, it was strong, dirty, and inconsistent.

About microdosing

You will see it claimed that no trial has ever tested doses below 0.25 mg per week. That is not true, and being accurate about it matters.

A 2016 trial randomized 415 people with type 2 diabetes and included groups at 0.1 mg and 0.2 mg weekly. (PMID 26358288)

What it found was straightforward: less drug, less effect. That is why those doses were dropped before the big trials.

What is still true is narrower and worth stating properly. Nobody has run a trial using tiny doses as an ongoing maintenance strategy for weight. The data we have at those doses is twelve weeks, in diabetics, measuring blood sugar.

What it does to people

The common problem is your gut. In STEP 2, 63.5% of people on the drug reported nausea, diarrhea or similar, mostly mild. About 4.5% quit over it, compared to 0.8% on placebo.

The rarer problems come from a study comparing 4,757 new users against people taking a different weight drug (PMID 37796527):

• Pancreatitis, roughly nine times the risk. Read that with real caution, because the study's own margin of error runs from "barely elevated" to "66 times," which means the true number is genuinely uncertain.
• Bowel obstruction, about four times
• Gastroparesis, a stomach that stops emptying properly, about 3.7 times
• Gallbladder disease, raised but not enough to rule out chance

The label also carries a boxed warning about thyroid tumors, based on rats. Whether that transfers to humans is undetermined.

What nobody knows

Whether small maintenance doses do anything useful over years.

What is in any particular grey-market vial, since the one published analysis found all three wrong in different ways simultaneously.

What ten years of continuous use does. The longest randomized data is two years.

Whether the heart benefit comes from the weight loss, the drug itself, or both.

Regulatory status

Approved by the FDA as Ozempic for diabetes, Wegovy for weight, and Rybelsus as a pill. It is a prescription drug.

The compounding window that opened during the shortage has closed. FDA declared the shortage over and the deadlines have passed, so compounded semaglutide is generally no longer lawful.

Anything sold as "research use only" semaglutide is an unapproved prescription drug. On 7 April 2026 the FDA published warning letters to seven named online sellers, calling that disclaimer a ruse.

Banned in sport by WADA.

My take

The strange thing about semaglutide is that the science is finished and the practical questions are wide open.

It works. It works better without diabetes than with it. It works less impressively against real coaching than against nothing. And it works only while you are taking it. All three of those come from the trials themselves, and all three usually get left out.

The risk here is not that the compound is fake or useless. It is that the version most people in this space can actually get has never been tested by anyone, and the one time somebody did test what was being sold, all three samples were out of spec and stronger than the label said.

That is a completely different problem from the one BPC-157 has, and it deserves a different kind of caution.

---

This content is for educational and informational purposes only and is not medical advice. Peptides discussed are research compounds and may not be approved for human use. Nothing here should be used to diagnose, treat, cure, or prevent any disease. Always consult a qualified healthcare professional before starting any peptide, supplement, or protocol. Individual responses vary. Do not self-administer compounds without proper medical supervision.