An Evidence Tier for the Peptides Your Patients Are Asking About

The most common failure in peptide conversations is not credulity. It is flattening. A patient asks about peptides, and the answer arrives as a single verdict covering compounds whose evidence bases have almost nothing in common. Semaglutide and epitalon end up in the same sentence, which serves nobody.

Peptides are a chemical class, not a therapeutic category. Sorting them properly takes two axes rather than one: how strong the human evidence is, and what the regulatory position happens to be. Those two axes do not move together, which is precisely why a single verdict fails.

The grid

Compound	Human evidence	Regulatory position (US, mid-2026)	Reasonable clinical stance
Semaglutide	Large randomised trials with cardiovascular and renal outcome data	FDA approved	Prescribe on label with standard monitoring
Tirzepatide	Large randomised trials, head-to-head weight loss superiority	FDA approved	Prescribe on label with standard monitoring
Tesamorelin	Randomised trials in HIV-associated lipodystrophy	FDA approved for a narrow indication	On-label use is well supported; off-label use is a judgement call
Thymosin alpha-1	Human trials, mostly in hepatitis and immune-compromised populations, quality variable	Approved in several countries, not in the United States	Discuss honestly as an unapproved agent with a real but narrow literature
Sermorelin	Historical human data as a growth hormone secretagogue	Original approved product discontinued	Weak modern evidence base; treat claims cautiously
CJC-1295 with ipamorelin	Pharmacokinetic and short-term endocrine studies; no long-term outcome data	Not approved; compounding position contested	Explain that the mechanism is plausible and the outcome evidence is absent
BPC-157	Extensive rodent data, negligible controlled human data	Not on the 503A Bulks List	See our review of the current regulatory position before any prescribing conversation
TB-500	Preclinical only	Removed from Category 2, not listed	Same posture as BPC-157
MOTS-c	Mechanistic and preclinical work; early human observational studies	Under review	Genuinely interesting biology, insufficient human data for clinical use
Epitalon	Long-running Russian clinical literature, methodologically inconsistent	Under review	Worth reading, not yet worth prescribing
Semax	Approved and used in Russia; limited Western replication	Under review	Same posture as epitalon
Melanotan II	Case reports of adverse events including melanocytic change	Not approved; associated with unregulated supply	Advise against, and document the advice

Reading the grid honestly

The top three rows carry outcome data from trials designed to answer clinical questions. Everything below them carries something weaker, and the drop is steeper than the layout suggests. Moving from tesamorelin to thymosin alpha-1 means moving from randomised outcome evidence to a heterogeneous literature with real methodological limits.

The bottom half of the grid shares a common pattern. Mechanism is well described, animal data is often striking, and the human evidence that would tell you whether the mechanism translates has not been generated. Rodent tendon healing is a legitimate scientific finding. It is not a reason to expect the same effect in a fifty-two-year-old with a chronic rotator cuff problem, and saying so is not conservatism.

The conversation that actually works

Patients who arrive asking for a specific compound have usually already encountered enthusiastic content about it. Meeting that with a flat refusal tends to move the request elsewhere rather than resolve it.

What works better is describing the shape of the evidence. Most patients understand the difference between “this was tested in thousands of people and the results were measured for years” and “this worked in rats and the human studies have not been done.” Presented that way, a good number reconsider on their own. Those who do not at least leave understanding what they are choosing, which is a materially different clinical situation than one where they simply source it privately.

Two questions are worth asking in every one of these conversations. What outcome does the patient expect, stated concretely, and how would they know whether it had happened. Vague expectations produce indefinite courses of unproven agents. A patient who says they want to return to overhead pressing without pain has given you something measurable, and a timeframe after which continuing makes no sense.

Where synthesis tools help, and where they mislead

Clinicians increasingly reach for a language model before reaching for PubMed, and for good reason. The volume of material in this space is unmanageable by hand. The risk is that general-purpose models summarise the enthusiastic content and the rigorous content in the same register, because both exist on the open web and both read as confident.

The design question we care about at Longevitix is not whether a system can summarize. It is whether the summary tells you which tier it came from. Every recommendation the platform surfaces carries its evidence hierarchy alongside it and traces back to the underlying source, so a clinician can see immediately whether they are looking at guideline-level material, an emerging signal, or an area where the honest answer is that nobody knows yet. A summary that hides that distinction is worse than no summary, because it launders uncertainty into apparent consensus.

Regulatory positions in this table move on a scale of months. The evidence tiers move more slowly, though several of the compounds in the lower half have trials registered, and a genuine human dataset would change where they sit.