Research
Peptide side effects: where the numbers come from
A side-effect percentage is only worth reading if you can see three things beside it: how many people it counted, what the comparison group did, and the cutoff below which nothing was printed. Most peptides sold today have no document that carries any of the three.
One question, two completely different answers
"What are the side effects?" is the most reasonable question anyone can ask about a peptide. The answer depends almost entirely on whether the compound has an FDA-approved product behind it.
For the handful that do, there is a document with a section built to answer exactly that, filled with numbers that came out of the trials the approval rested on. For the rest, there is no such document, and what fills the gap is marketing copy and forum reports.
This article is about telling those two situations apart, and about reading the numbers when they exist. It does not tell you whether any compound is safe for you. That belongs with a prescriber who knows your history.
A real side-effect number carries three things
Start with a worked example from an approved peptide drug. Bremelanotide is a melanocortin receptor agonist approved for a specific sexual desire disorder in premenopausal women, and it is the same molecule sold elsewhere under the name PT-141.
Its label reports two identical 24-week randomized, double-blind, placebo-controlled trials in 1,247 premenopausal women. The adverse reaction table lists everything reported in at least 2 percent of the treated group at a rate above placebo. In that table, 627 women received the drug and 620 received placebo. Nausea occurred in 40.0 percent on the drug and 1.3 percent on placebo. Flushing, 20.3 percent against 0.3 percent. Headache, 11.3 percent against 1.9 percent.
So the number carries a denominator, 627. It carries a comparator, the 620 who got nothing. And it carries a threshold, 2 percent, below which nothing was printed at all.
Take any one of those away and the percentage stops meaning what it looks like it means.
The comparator is the part that gets dropped
The most instructive row in that table is not nausea. It is injection site reactions: 13.2 percent on the drug, and 8.4 percent on placebo.
Nearly two-thirds of the injection site reactions happened to people receiving an inert injection. Without the placebo column, 13.2 percent reads as a property of the molecule. With it, most of that number is a property of injecting.
This is the single most useful habit to carry into a product page. When a seller quotes a side-effect rate with no comparison group, the number is describing the experience of being in a study, not the effect of the compound.
The same label also reports what the numbers cost people in practice. Serious adverse reactions occurred in 1.1 percent of treated patients and 0.5 percent on placebo, and 18 percent of the treated group stopped taking it because of a side effect, against 2 percent on placebo.
The threshold is a decision, and the label prints it
Every adverse reaction section states the frequency cutoff it used. Semaglutide's injectable label says the most common adverse reactions reported in at least 5 percent of treated patients are nausea, vomiting, diarrhea, abdominal pain and constipation. Tesamorelin's label uses the same idea at above 5 percent, listing arthralgia, injection site erythema, injection site pruritus, pain in extremity, peripheral edema and myalgia.
Those lists are short because the cutoff is high, not because nothing else happened. A five-item list at a 5 percent threshold and a fifteen-item list at a 1 percent threshold can describe the same drug.
Sample size sets what a threshold can even detect. The semaglutide placebo-controlled pool reflects 521 patients with a mean exposure of about 33 weeks. A rare event that shows up once in ten thousand people will not appear in 521 of them, which is why labels also carry a separate postmarketing section for reports that arrived after approval.
The structure of the label itself is covered in more detail elsewhere on this site. What matters here is that the threshold is never hidden, and a list quoted without it is a list quoted without its scale.
Some of the strongest warnings are about what nobody knows
A label is not only a record of what was observed. It also states, in the same authoritative voice, where the observation runs out.
Tesamorelin works by prompting the pituitary to release growth hormone, which raises insulin-like growth factor 1. Its label says the drug increases serum IGF-1, that the effects of prolonged elevations in IGF-1 levels are unknown, and that IGF-1 should be monitored during treatment, with discontinuation considered if elevations persist.
That is a regulator writing "unknown" into a warning section rather than leaving it out. The same label also quantifies a risk that is known. From baseline to week 26, patients with an elevated HbA1c at or above 6.5 percent ran 5 percent on the drug and 1 percent on placebo. Hypersensitivity reactions occurred in 4 percent of treated patients.
A page that quotes a compound's benefits from a label and skips this half of the same document is not citing the label. It is quarrying it.
What the record holds when there is no label
Most compounds in this market have no approved product anywhere, so none of the above exists for them. Two published documents describe what stands in its place, and both are worth reading in the original.
The first is FDA's own answer on compounded medicines. Compounded drugs are not FDA-approved, and the agency states that it does not verify the safety, effectiveness or quality of compounded drugs before they are marketed. That is a statement about the process, not an accusation about any pharmacy.
The second is the agency's Category 2 table, which lists bulk substances it has identified as potentially presenting significant safety risks and prints a rationale beside each one. The peptide entries there are worth reading closely, because they say different things.
For KPV, FDA writes that it has not identified any human exposure data on drug products containing KPV administered by any route, and that it lacks important information about whether the substance would cause harm. For BPC-157, that it has identified no or only limited safety-related information for the proposed routes, and therefore lacks sufficient information to know whether the drug would cause harm.
For CJC-1295, the entry says something different again: FDA has identified serious adverse events associated with the substance, including increased heart rate and a systemic vasodilatory reaction, and that available clinical data are limited. For ipamorelin acetate, that a study published in the literature identified serious adverse events including death when it was given intravenously to improve gastric motility.
Read as a group, the entries share one technical concern — risk of immunogenicity from aggregation and peptide-related impurities — and then split into two kinds of statement. Some report a gap. Some report an event. Those are not the same finding and should never be summarized as one.
"Well tolerated" and "no reported side effects" are not measurements
Both phrases appear constantly on peptide product pages, and neither carries a denominator, a comparator or a threshold.
"Well tolerated" is a conclusion. In a trial report it follows a table; on a product page it usually replaces one. Ask what the table said and how many people were in it, and the phrase either resolves into numbers or it does not.
"No reported side effects" is a claim about reports, not about events. Reports need a route to travel down. An approved drug has a manufacturer with a reporting line, a national reporting system, and a label section that gets updated when postmarketing reports accumulate. A vial bought as a laboratory chemical has none of that attached, so a low report count is what the arrangement produces regardless of what happened to anyone.
Volume of anecdote does not repair this. A thousand people saying they felt fine is not a denominator, because the people who stopped and did not post are not in it.
Six checks to run on any safety claim
Ask whether an approved product exists for the compound at all. If it does, the answer to your question is in the label's adverse reactions section, not on the page you are reading.
Find the denominator. A percentage with no number of people behind it can be one person in three.
Find the comparator. If there is no placebo column, you cannot tell what the compound did from what the study did.
Find the threshold. A short list is often a high cutoff rather than a clean drug.
Check that the source studied the same molecule, form and route as the product on sale. A rate measured after intravenous infusion does not describe a subcutaneous injection.
Separate a gap from a finding. "Nobody has measured this" and "this was measured and looked bad" are both worth knowing, and they are not interchangeable.
Key takeaways
- A usable side-effect figure carries a denominator, a comparator and a stated threshold.
- On the bremelanotide label, injection site reactions ran 13.2 percent on the drug and 8.4 percent on placebo — most of that rate belongs to the injection.
- A short adverse reaction list is often a high frequency cutoff rather than a clean drug.
- Labels also state what is unknown: tesamorelin's says the effects of prolonged IGF-1 elevation are unknown and directs monitoring.
- FDA states that it does not verify the safety, effectiveness or quality of compounded drugs before they are marketed.
- In FDA's Category 2 rationales, some peptide entries report a gap in information and others report specific serious events; those are different findings.
- "Well tolerated" and "no reported side effects" carry none of the three things a real number carries.
Frequently asked questions
What are the side effects of peptides?
There is no single answer, because "peptides" is a category rather than a drug. For the few with an FDA-approved product, the label reports measured rates from the approval trials. Above a 5 percent cutoff, semaglutide's injectable label lists nausea, vomiting, diarrhea, abdominal pain and constipation. Tesamorelin's label lists arthralgia, injection site reactions, pain in extremity, peripheral edema and myalgia. For compounds with no approved product, no such measured list exists, and any figure offered came from somewhere other than a reviewed trial.
Why does a placebo group matter so much?
Because it separates the compound from everything else that happens in a study. The bremelanotide label reports injection site reactions in 13.2 percent of treated patients and 8.4 percent of those receiving placebo. Most of that rate belongs to being injected rather than to the molecule. A side-effect number quoted without its comparison group hands the reader the whole figure and lets them assume the drug caused all of it.
Does "no reported side effects" mean a peptide is safe?
No. It is a statement about how many reports exist, and reports need somewhere to go. An approved drug has a manufacturer reporting line, a national reporting system and a label section that gets revised as postmarketing reports build up. Material sold as a laboratory chemical sits outside all of that, so a low count is a product of the arrangement rather than evidence about what happened to the people who used it.
What has FDA actually said about the safety of compounds like BPC-157 or KPV?
It publishes a rationale for each substance it has placed in Category 2. For KPV it writes that it has not identified any human exposure data by any route and lacks important information about whether the substance would cause harm. For BPC-157, that it has identified no or only limited safety-related information for the proposed routes and therefore lacks sufficient information to know. Those are statements about missing information, not findings that either compound is dangerous, and the difference is on the page in the agency's own wording.
Are compounded peptides checked for safety before they are sold?
Compounded drugs are not FDA-approved, and the agency states plainly that it does not verify the safety, effectiveness or quality of compounded drugs before they are marketed. Compounding is lawful and serves real patient needs, and a licensed pharmacy carries obligations to its state regulator. What it does not carry is the premarket review that produces an adverse reaction table with a denominator in it.
A seller says its peptide is "well tolerated" in studies. Is that useful?
Only if you can get to the studies. The phrase is a conclusion that normally sits underneath a table, and on a product page it usually stands in for one. Ask how many people were in the study, what the comparison group experienced, at what frequency the report cut off, and whether the compound was given to humans at all. If those answers do not exist, the phrase is describing an absence of data rather than a favorable result.
Sources
Each document below is named as it names itself, with the date printed on that document rather than the day it was read.
- VYLEESI (bremelanotide) injection — FDA-approved prescribing information — National Library of Medicine, DailyMed, November 2025
- OZEMPIC (semaglutide) injection — FDA-approved prescribing information — National Library of Medicine, DailyMed, June 2026
- EGRIFTA WR (tesamorelin) for injection — FDA-approved prescribing information — National Library of Medicine, DailyMed, July 2026
- Compounding and the FDA: Questions and Answers — U.S. Food and Drug Administration, September 2025
- Certain Bulk Drug Substances for Use in Compounding that May Present Significant Safety Risks — U.S. Food and Drug Administration, April 2026