“Studies show” is not a single category. A cell-culture result and a Phase 3 randomized trial are separated by a decade of work and a high failure rate. This guide explains the hierarchy and what each level actually justifies claiming.
Six levels from weakest to strongest — each answers different questions
In vitro (cell culture)
WeakestExperiments on cells or cell lines in a dish — no organism, no pharmacokinetics, no immune system, no tissue context.
What it tells you
A compound can interact with a receptor or pathway. Necessary for initial mechanism work.
What it misses
Whether the compound reaches that receptor in a living organism, at what concentration, and whether the cell line behaves like the tissue of interest.
Example: Most early BPC-157 angiogenesis data is in vitro.
In vivo — rodent models
Moderate (preclinical)Studies in living animals — usually rats or mice, often using disease models (surgically created tendon injuries, chemically induced colitis, diet-induced obesity).
What it tells you
A compound can produce measurable effects in a mammalian biological system, with real pharmacokinetics and immune responses.
What it misses
Whether the effect translates to humans. Rodent metabolism, immune function, and tissue biology differ meaningfully from human. Disease models are often simplified proxies.
Example: The majority of BPC-157, TB-500, and DSIP evidence is rodent-model work.
Phase 1 — human safety/PK
Human safety onlyFirst-in-human trials, typically small (20–100 participants), focused on safety, tolerability, pharmacokinetics, and dose-finding.
What it tells you
The compound is not acutely toxic at studied doses in humans, and provides PK data (half-life, Cmax, AUC). Does not establish efficacy.
What it misses
Whether it works for any indication. Phase 1 is not designed to test efficacy.
Example: Several GH-axis peptides (tesamorelin, CJC-1295) have Phase 1 PK data.
Phase 2 — signal of efficacy
Indicative (not confirmatory)Randomized trials in people with the target condition — typically 50–500 participants — testing whether a signal of efficacy exists at a tolerable dose.
What it tells you
The compound shows a preliminary efficacy signal in the target population. Sufficient to justify a larger trial.
What it misses
Definitive confirmation of benefit. Phase 2 trials are often too small to distinguish true effects from chance, and effect sizes may shrink in Phase 3.
Example: Retatrutide's 24% weight-loss data is from a Phase 2 trial. Semaglutide has Phase 3 data.
Phase 3 — confirmatory efficacy
Strong (registration standard)Large randomized, usually placebo-controlled trials (hundreds to thousands of participants) designed to confirm efficacy and establish safety at scale. The standard for regulatory approval.
What it tells you
The compound produces a reliable benefit in the studied population at a studied dose, and the safety profile is characterized.
What it misses
Real-world heterogeneity (trials select narrow populations). Long-term effects beyond the trial window. Rare adverse events that require post-marketing surveillance.
Example: STEP-1 (semaglutide), SURMOUNT-1 (tirzepatide), SURMOUNT-5 (head-to-head) are Phase 3.
FDA approval + post-market (Phase 4)
Strongest (for approved indication only)Regulatory approval means the Phase 3 evidence met the FDA standard for safety and efficacy for a specific indication. Post-approval surveillance continues.
What it tells you
The compound is reliably beneficial for the approved indication in the studied population. Real-world use data accumulates after approval.
What it misses
Approval is indication-specific. Using an approved drug off-label (different dose, different population) returns you to lower evidence levels for that new use.
Example: Semaglutide, tirzepatide, tesamorelin are FDA-approved (specific indications). BPC-157 has no human trials.
Publication bias
Positive results are more likely to be published than negative results. This inflates the apparent success rate of compounds in the literature — especially true for early-stage peptide research.
Effect size vs statistical significance
A result can be statistically significant (p < 0.05) but clinically meaningless if the effect is tiny. Always check the actual magnitude of change, not just whether the p-value is below a threshold.
Single-lab findings
When the majority of evidence for a compound comes from one research group (Epitalon, some BPC-157 mechanisms), treat findings as preliminary until independently replicated.
Disease model vs clinical disease
A surgically severed tendon in a rat is a controlled model — not the same as chronic tendinopathy in a human with varied activity levels, age, and comorbidities. Model-to-clinic translation frequently fails.
Dose and route
A compound that works at one dose or route (e.g., direct injection into a joint) may not work at another (systemic subcutaneous). Evidence from one context does not automatically extend to another.
| Compound(s) | Highest evidence level | Notes |
|---|---|---|
| Semaglutide, Tirzepatide | Phase 3 + FDA approved | Multiple large RCTs; head-to-head data |
| Tesamorelin | Phase 3 + FDA approved | Approved for specific HIV-related indication only |
| PT-141 (Bremelanotide) | Phase 3 + FDA approved | Approved for HSDD in premenopausal women |
| SS-31 / Elamipretide | Phase 2–3 (investigational) | Clinical trials in mitochondrial disease; not approved |
| Retatrutide | Phase 2 | Phase 2 weight-loss data; Phase 3 ongoing |
| Cagrilintide, CJC-1295 | Phase 1–2 | Human PK/PD data; limited efficacy trials |
| Semax, Selank | Clinical (Russia) | Approved in Russia; limited independent RCT data |
| BPC-157, TB-500, MOTS-c | Preclinical only | No completed controlled human trials |
| Epitalon, DSIP | Preclinical (limited) | Evidence concentrated or mechanistically incomplete |
What does "preclinical" mean?
Preclinical means the evidence comes from studies in cells or animals — not humans. Preclinical findings are hypothesis-generating: they provide a rationale for studying a compound in humans, but they do not confirm that the effect will occur in human beings at tolerable doses. The translation from preclinical to clinical regularly fails — including for compounds with multiple positive animal studies.
Can I trust a study published in a peer-reviewed journal?
Peer review is a quality filter, not a guarantee. The relevant questions are: what was the study design (in vitro vs in vivo vs randomized human trial)? What was the sample size? Was there a proper control group? Was the result replicated? Who funded the study? A peer-reviewed cell-culture study is still a cell-culture study — the level of evidence is determined by the experimental design, not the journal.
Why do so many peptides have animal data but no human trials?
Human trials are expensive, slow, and require regulatory approval. Animal studies can be conducted quickly and cheaply. For peptides without a clear commercial sponsor (i.e., not under patent), there is limited financial incentive to fund Phase 3 trials. This is why most non-approved research peptides have a rich rodent literature and minimal human data — it reflects funding economics, not a neutral assessment of their potential.
What should I look for in a peptide clinical trial?
Key questions: (1) Was it randomized and placebo-controlled? (2) Was it blinded? (3) How large was the sample? (4) What was the primary endpoint, and was it pre-specified? (5) Was the trial registered on ClinicalTrials.gov before data collection? (6) Did the effect size reach clinical meaningfulness, not just statistical significance? (7) Has it been replicated? The ClinicalPulse tool on this platform links to ClinicalTrials.gov registrations where you can check these directly.