Evaluating Human Research in Peptide Science

1. Clinical trial phases and their limits

Phases describe common research aims, not a compulsory fixed sequence. Programs may combine or overlap phases, and trial sizes vary by disease and question.

• Phase I: Initial safety, tolerability, pharmacokinetics and dose exploration, often in 20–100 participants; some studies enroll patients rather than healthy volunteers.
• Phase II: Preliminary effectiveness, dose selection and further safety assessment in up to several hundred patients.
• Phase III: Larger confirmatory comparisons, often hundreds to thousands of patients. Study design and duration depend on the indication; phase alone does not guarantee blinding or adequate detection of rare harms.
• Phase IV: Studies after approval. These complement ongoing adverse-event monitoring and other postmarketing research.

Participant ranges are illustrative, not regulatory thresholds. Source: FDA clinical research.

2. Biomarkers, clinical outcomes and effect size

A biomarker measures biology; a clinical outcome measures how patients feel, function or survive. A surrogate endpoint substitutes for a clinical outcome and needs evidence linking it to that outcome in the relevant context. Patient-reported symptoms and function can also be clinically meaningful; outcomes are not limited to mortality.

SELECT enrolled 17,604 adults aged 45 or older with established cardiovascular disease, BMI of at least 27, and no diabetes. Its cardiovascular composite occurred in 6.5% with semaglutide versus 8.0% with placebo over a mean follow-up of 39.8 months. The hazard ratio was 0.80 (95% CI 0.72–0.90): a 20% relative hazard reduction, alongside an observed 1.5-percentage-point absolute difference. These findings do not apply automatically to every population or formulation.

Source: SELECT trial; Cochrane: interpreting effect estimates.

3. Randomization, blinding and missing outcomes

• Randomization and concealed allocation: Reduce selection bias and balance confounders on average; they do not guarantee identical baseline groups in a particular trial.
• Blinding: Can reduce differences in care and outcome assessment arising from knowledge of assignment. Its feasibility and importance depend on the intervention and outcome.
• Analysis by assigned group: Preserves the randomized comparison, but does not by itself solve missing outcomes, nonadherence or selective reporting. Check the analysis population, assumptions and sensitivity analyses.
• Transparency: Compare reported outcomes with the protocol; inspect funding, conflicts, adverse events and uncertainty.

Source: Cochrane risk-of-bias guidance.