How to read a peptide study: five questions that tell the real thing from the sell

Every peptide you get curious about arrives the same way — buried under studies, quoted by someone who wants you to buy. Here’s the good news nobody mentions: you don’t need a biology degree to see through it. You need five questions. Ask them of any paper a seller waves at you, and the honest signal pulls away from the sell almost every time.

That’s the real skill, and it’s worth having. Once you can read the evidence yourself, this field stops being a coin flip between blind hype and blind cynicism. You get to be excited about the peptides that have earned it, skip the ones that haven’t, and know which is which. Nobody’s born able to do this. It’s small, it’s learnable, and peptides are the perfect place to practise — the mistakes cluster in five specific spots, and the same five keep coming back.

Here are the five. Each one comes with a real study from peptides people are actually chasing right now — because the worked examples are what make the question easy to ask the next time.

Question 1 — What molecule did they actually dose?

The most common move in peptide marketing is citing a real trial for the wrong peptide. A whole family of these compounds are cousins: close enough to sound interchangeable, different enough that a result in one tells you almost nothing about another.

Baker and colleagues (2012, Arch Neurol) is the clean example. 152 adults, twenty weeks, a real improvement in cognition at p=0.03. The peptide they used was tesamorelin — an FDA-approved drug. That trial gets quoted in marketing for sermorelin, a different molecule sold off-label, as if it were the proof. It isn’t. Same receptor, same family, different drug, and only one of the two has ever been approved for anything. Read the methods section and check which compound actually went into the people. It takes thirty seconds, and it catches the trick most of the time.

Question 2 — Was it a person, an animal, or a dish?

This is the big one, and it’s the easiest to check. A result in a rat is a reason to run a human trial. It is not a human result.

BPC-157 is the cleanest case of the gap. Its reputation for rebuilding tendon and gut is real and genuinely exciting — and almost all of it comes from rats. We laid that out in full in BPC-157 and injury recovery; the short version is that the animal signal is strong and the human file is nearly empty. Wu and colleagues (2020, Mater Sci Eng C) is the same problem one step further from a person: thymosin β4 grown on a synthetic scaffold, in a dish of human cells, with no animal and no human injected at all. That paper gets cited as evidence that injecting TB-500 repairs a human tendon. Cells in a dish can’t tell you that. Find the sentence that says who or what got the dose — a person, a rat, or a plate of cells — before you let the study carry any weight.

Question 3 — Did the headline result actually pass?

Every serious trial names one main outcome in advance — the thing that decides whether it worked. Everything else is a side finding. Sellers quote the side findings.

Beck and colleagues (2014, Int J Colorectal Dis) ran ipamorelin against placebo in 117 people after bowel surgery. The treated group reached their first proper meal about seven hours sooner. That gap looks like a win — until you read the p-value, 0.15, which means it could easily be noise. The main result was null: no significant difference on the endpoint that mattered. The company shut the programme down after the readout. Quoted gently, the same trial still gets used as evidence the peptide works.

GHK-Cu shows the polished version of the trick. The one randomised human trial of GHK-Cu skincare — Miller and colleagues, 2006, thirteen patients after laser resurfacing — came back with no significant difference on redness, wrinkles, or skin quality. Null on every objective measure. The single number that moved was the patients’ own satisfaction score (P=0.04), the softest and most placebo-friendly reading in the whole study. That is the one that gets quoted. We told the fuller GHK-Cu story in the repair molecule your body stops making. When a study is sold on its satisfaction score, ask what the hard measures did.

Question 4 — How many people, for how long?

A positive direction in two people over three days is not safety, and it isn’t proof of anything else either. The number of participants and the length of the study set the ceiling on what a trial can honestly claim.

Lee and Burgess (2025) is the example worth memorising. Intravenous BPC-157, no adverse effects, no worrying bloodwork — which sounds reassuring until you see the design. Two volunteers. Three days. A low-impact journal. None of the good-looking surface — safe, well-tolerated, no red flags — survives contact with an n of two. It’s a reason to run a bigger study, nothing more. Whenever a peptide is sold as proven safe, the first question is: in how many people, for how long?

Question 5 — Has anyone outside the original lab found the same thing?

A real finding from a single research group is still a single research group. Results that only ever appear inside the lab that discovered them have a habit of looking different once an outside team tries to repeat them.

This is a defining feature of the peptide field, not a rare one. GHK-Cu’s science is largely one man, Loren Pickart, across forty years. BPC-157’s runs through a single lab in Zagreb since the early 1990s. MOTS-c is younger and cleaner, but its human evidence is still one heat-stress trial in nineteen men plus a stack of associations, laid out in the exercise signal in your mitochondria — nobody has injected it and measured what happens. None of that makes the work fake. It makes it unfinished. The question to carry is simple: has anyone with nothing riding on the answer gotten the same result?

One more habit worth building: a review is not a trial. A review paper summarises other people’s studies, so it’s only ever as strong as what it’s summarising — and the honest ones say so. Mendias and Awan (2026, Sports Medicine) is a good one. It surveys the whole grey-market peptide landscape and states plainly that the controlled human safety data is thin and the placebo effect is amplified by social media. A review that admits what’s missing is useful. A review quoted as if it were fresh proof is not.

What a study worth trusting actually looks like

Run the five questions and a credible peptide study answers all of them cleanly: the exact molecule named, given to actual people by the route you’d actually use, in enough of them for long enough to mean something, with the main outcome passing — not a side score — and at least one team outside the original lab landing in the same place.

Most peptide studies in today’s marketing don’t clear all five. The ones that do tend to be trials of already-approved drugs. That isn’t a reason to be cynical about the rest. It’s a reason to be precise. Some of these molecules are genuinely promising and simply haven’t had the trial that would prove it yet — and being able to tell those apart from the ones dressed up in borrowed evidence is the entire point of the skill.

Where the regulators come in

Several of the peptides worth this kind of reading are in front of the FDA’s Pharmacy Compounding Advisory Committee across 2026 and 2027. What a review like that actually decides — and what it doesn’t — is a longer story, told in what a PCAC review actually is. The short version: a committee looking hard at the evidence is doing, at national scale, the same job these five questions do at your kitchen table.

Why this is the most useful thing on the site

Every other article here asks you to trust our reading of the evidence. This one hands you the reading itself. Once you can run these five questions on any study, you don’t have to take our word for a single peptide — you can check us, and check everyone else selling you a vial.

That’s exactly the standard Wolverine Health is being built to meet: a physician who has read the studies as carefully as this checklist asks you to, a licensed US pharmacy behind every batch, and nothing quoted at you that couldn’t survive these five questions. It can’t open until the regulation catches up. Leave your email and we’ll tell you the day it does.