Clinical trial phases are the stages through which a candidate medicine is tested in people, and in drug development they function as decision points as much as scientific milestones. Each phase is designed to generate a specific body of evidence, and at the end of each the sponsor must decide whether the evidence justifies the substantially larger commitment the next phase requires.
Viewed this way, the phase structure is a mechanism for managing uncertainty: it limits how many people are exposed to a compound whose properties are still poorly characterized, and it limits how much is invested before the key questions have been answered.
Before the Phases: What Must Be in Place
Human testing cannot begin until preclinical pharmacology and toxicology support a proposed starting dose, clinical-grade material can be manufactured to a defined specification, a protocol has been written, an ethics committee has approved it, and the regulator has authorized the study — via an Investigational New Drug application in the United States or a clinical trial authorisation in European Union member states.
A clinical development plan is normally drafted at this point, setting out the intended indication, the target product profile describing what the finished medicine should achieve, the endpoints that will be used, and the evidence the eventual marketing application will need. Working backwards from the regulatory requirement is standard practice; phases designed in isolation frequently generate data that do not support the submission.
Phase 1: Establishing Human Safety and Exposure
Question answered: can this compound be administered to humans, at what dose, and what does the body do with it?
Phase 1 enrols small numbers of participants, usually healthy volunteers, with patients enrolled instead where the compound is expected to be toxic — the standard situation in oncology. Typical designs include single ascending dose and multiple ascending dose studies, with escalation between cohorts permitted only after safety review.
The evidence generated covers tolerability and the nature of dose-limiting effects, pharmacokinetics describing absorption, distribution, metabolism and excretion, and where possible pharmacodynamic markers indicating that the compound is engaging its intended target. Target engagement evidence is valuable: without it, a later failure cannot be interpreted, because it is impossible to distinguish a wrong target from insufficient exposure at the target.
Decision at the gate: is there a dose range that is tolerable and that achieves exposure sufficient to affect the target?
Phase 2: Proof of Concept and Dose Selection
Question answered: does the compound produce a meaningful effect in patients with the disease, and at what dose?
Phase 2 enrols patients with the target condition. Programmes often divide it: an earlier exploratory stage examining biological effect and dose-response, and a later stage using randomized, controlled designs with a clinical or well-validated surrogate endpoint.
Two outputs matter most. The first is proof of concept: credible evidence that the mechanism produces the intended clinical effect in humans. The second is dose selection, since carrying an incorrect dose into confirmatory trials is a recognized cause of late-stage failure — too low and efficacy is missed, too high and tolerability undermines the programme.
Phase 2 is where the largest share of candidates is discontinued. That concentration of failure is intentional: it is far preferable to stop here than after a confirmatory trial.
Decision at the gate: is the effect real, large enough to matter clinically, and achievable at a tolerable dose in a population that can be recruited?
Phase 3: Confirmation for Regulatory Submission
Question answered: does the medicine deliver a clinically meaningful benefit, and what is the balance of benefits and harms?
Phase 3 trials are randomized, controlled, blinded where feasible, and conducted across many sites and countries. They are powered on a prespecified primary endpoint and sized to detect a difference judged clinically important, while also accumulating enough exposure to characterize less common adverse effects.
Design choices carry regulatory weight. The comparator determines what claim can be made; a placebo-controlled result does not establish advantage over standard care. Superiority and non-inferiority designs answer different questions, and non-inferiority margins must be justified in advance. Endpoint selection is negotiated with regulators, often through formal scientific advice meetings held before the trial begins.
Independent data monitoring committees review accumulating data against prespecified rules and may recommend stopping for harm, clear benefit, or futility.
Decision at the gate: does the evidence package support a favourable benefit-risk assessment for a defined indication and population?
The Studies That Sit Outside the Phase Numbers
A complete dossier requires evidence that does not fit neatly into the numbered phases:
- Drug-drug interaction studies, examining effects on and from other medicines likely to be taken concurrently.
- Special population studies in patients with impaired renal or hepatic function, in older adults, and — where appropriate and following defined pathways — in children.
- Cardiac safety assessment, evaluating effects on cardiac repolarization.
- Bioavailability and bioequivalence studies, particularly when the formulation used in trials differs from the intended commercial product.
- Immunogenicity assessment for biologics, since patients may develop antibodies that reduce efficacy or cause reactions.
- Quality-of-life and patient-reported outcome measures, which increasingly inform both labelling and reimbursement decisions.
Phase 4 and Post-Authorization Evidence
Phase 4 studies follow authorization. They address long-term safety and effectiveness, populations under-represented in earlier trials, comparisons against alternative treatments, and rare adverse effects detectable only in large-scale use. Regulators may mandate specific post-authorization safety or efficacy studies as a condition of approval, and these obligations are recorded publicly.
Phase 4 research is distinct from routine pharmacovigilance, which continuously collects and assesses spontaneous adverse event reports and can trigger label changes or restrictions independently of any formal study.
How the Phase Model Adapts to Different Products
- Vaccines use the same phase labels but with immunogenicity and efficacy endpoints and very large confirmatory trials, since the events being prevented are uncommon in any individual and rare safety signals matter greatly in healthy recipients.
- Biologics require immunogenicity assessment and manufacturing comparability evidence, because the product is defined partly by its process.
- Biosimilars follow an abbreviated route centred on demonstrating similarity to an approved reference biologic through analytical, pharmacokinetic, and targeted clinical comparison rather than repeating full efficacy programmes.
- Generic medicines generally require bioequivalence to the reference product rather than new efficacy trials, since the active substance is already established.
- Cell and gene therapies often involve small single-arm studies in serious rare conditions, long-term follow-up requirements, and manufacturing that is inseparable from the clinical product.
- Seamless and adaptive designs combine phases under one protocol with prespecified transition rules, shortening timelines when the rules are set in advance.
Sources
- U.S. Food and Drug Administration — clinical research phases; guidance on exploratory studies and expedited programmes
- European Medicines Agency — clinical development guidance; scientific advice procedures; biosimilar guidelines
- International Council for Harmonisation — E8 general considerations for clinical studies; E9 statistical principles; E14 cardiac safety
- ClinicalTrials.gov — phase classification and study records
- World Health Organization — vaccine evaluation and prequalification standards
