Clinical Trial Phases: From First-in-Human to FDA Approval

Why Most Drugs Never Reach Patients

Of every 10,000 compounds screened in early drug discovery, approximately 250 make it to animal testing, 5 reach human trials, and only 1 ultimately gains regulatory approval. The clinical trial system — a structured sequence of phases designed to establish safety and efficacy — is the filter through which all approved medicines must pass. It is deliberately slow, deliberately expensive (average cost per approved drug: $1.3 billion), and deliberately demanding. This is by design.

The Investigational New Drug Application

Before a pharmaceutical company can administer an experimental compound to human subjects in the United States, it must file an Investigational New Drug (IND) application with the Food and Drug Administration (FDA). The IND compiles all preclinical data: in vitro pharmacology, animal toxicology studies (acute, subacute, chronic), genotoxicity assays, and proposed manufacturing protocols. The FDA has 30 days to place the IND on clinical hold or allow it to proceed. Most INDs are not formally approved — they simply are not stopped.

Institutional Review Board Oversight

Every clinical trial site must have its study protocol reviewed and approved by an Institutional Review Board (IRB), also called an ethics committee. IRBs are independent bodies — typically composed of clinicians, researchers, ethicists, and community representatives — tasked with protecting the rights and welfare of human research participants. They evaluate study design, risk-benefit ratio, the adequacy of proposed informed consent procedures, and ongoing monitoring plans.

Informed consent is a foundational principle: every participant must be told, in plain language, the purpose of the study, what procedures are involved, what risks and benefits are known, what alternatives exist, and that participation is voluntary and may be withdrawn at any time. No clinical trial in an IRB-governed jurisdiction can proceed without documented informed consent from each participant.

Phase 0: Microdosing Studies

Phase 0 is an optional, rarely used phase introduced by the FDA in 2006. It involves administering sub-therapeutic microdoses (typically <1% of the calculated pharmacologically active dose) to a small number of subjects (10–15) to gather preliminary human pharmacokinetic data. Phase 0 trials do not assess efficacy or tolerability at therapeutic doses. They are primarily useful for selecting among multiple drug candidates before committing to full Phase I development.

Phase I: Safety in Healthy Volunteers

Phase I is the first-in-human study. The primary objective is safety: what is the dose-limiting toxicity? What is the maximum tolerated dose? Secondary objectives include early pharmacokinetic characterization and preliminary pharmacodynamic measurements.

Most Phase I trials enroll 20–80 healthy adult volunteers, though oncology Phase I trials often enroll patients with advanced cancer, because the risk-benefit calculus is different. Dose escalation follows a structured scheme — often a 3+3 design or a Bayesian continuous reassessment method — in which small cohorts of subjects receive increasing doses, and the next dose level is administered only after the preceding level is reviewed for safety.

Approximately 70% of drugs that enter Phase I proceed to Phase II. Failure is usually due to unacceptable toxicity or unfavorable pharmacokinetics.

Phase II: Proof of Concept

Phase II trials are the first tests of efficacy in actual patients with the target condition. They answer two questions: does the drug work, and at what dose? Phase II trials are typically randomized and controlled, often comparing the experimental drug against placebo or an active comparator, and typically enroll 100–300 patients.

Phase IIa trials are dose-finding studies. Phase IIb trials test whether the optimized dose produces a meaningful signal of efficacy. Approximately 33% of drugs entering Phase II ultimately receive FDA approval. The attrition is steep — efficacy failures dominate at this stage, particularly in CNS indications where the disconnect between animal models and human neurobiology is widest.

Phase III: Pivotal Efficacy and Safety

Phase III pivotal trials are the definitive evidence base for regulatory approval. They enroll large patient populations — typically 1,000–3,000, though some trials enroll tens of thousands — in multicenter, randomized, controlled designs intended to detect clinically meaningful differences with statistical power exceeding 80%. Most Phase III programs consist of at least two independent trials, each powered to confirm the primary endpoint.

Randomization and Blinding

Randomization — randomly assigning participants to treatment arms — is the cornerstone of valid causal inference in clinical trials. It distributes both known and unknown confounding variables evenly across groups, making any observed difference attributable to the treatment itself.

Blinding prevents expectation bias from influencing outcomes. In a single-blind trial, the participant does not know their assignment. In a double-blind trial, neither participant nor investigator knows. In a triple-blind trial, the data analysts are also unaware. The placebo effect — the measurable physiological and psychological improvement produced by a dummy treatment — is not trivial; in some pain and depression trials, placebo response rates exceed 30–40%, making blinding essential for valid signal detection.

Adaptive Trial Designs

Traditional trials fix their design before the first patient is enrolled. Adaptive designs allow pre-specified modifications based on interim data. A group sequential design, for instance, includes pre-planned interim analyses that can trigger early stopping for efficacy or futility, reducing the number of patients exposed to an inferior treatment. Response-adaptive randomization allocates more patients to the better-performing arm as data accumulate.

Basket and umbrella trials are adaptive designs developed for oncology. A basket trial tests a single targeted therapy across multiple cancer types that share a specific molecular alteration (e.g., BRAF V600E mutation), regardless of tissue of origin. An umbrella trial tests multiple therapies within a single cancer type, assigning patients to arms based on their tumor's molecular profile. Both designs increase efficiency by testing biological hypotheses rather than disease categories.

Regulatory Submission: NDA and BLA

After successful Phase III trials, the sponsor submits a New Drug Application (NDA) for small-molecule drugs or a Biologics License Application (BLA) for biological products. These submissions contain all clinical trial data, preclinical data, proposed labeling, and manufacturing information — typically running to hundreds of thousands of pages. The FDA review team, supplemented by external advisory committees, evaluates the data and votes on whether the benefit-risk profile supports approval.

Advisory committee votes are non-binding, but the FDA rarely overrides them without documented justification. The surrogate endpoint controversy — using a laboratory measurement such as tumor shrinkage or LDL reduction as a proxy for clinical outcomes — is a recurring advisory committee debate. Accelerated approval pathways use surrogate endpoints, with post-market confirmatory trials required.

Phase IV: Post-Market Surveillance and FAERS

Phase IV begins the moment a drug is approved. The pre-approval clinical trial database is too small, too homogeneous, and too short-duration to detect rare adverse events, events in understudied subpopulations, or long-term effects. The FDA Adverse Event Reporting System (FAERS) is the primary passive surveillance tool: healthcare providers and patients voluntarily report suspected adverse reactions to the FDA, which uses signal detection algorithms to identify novel safety concerns.

REMS (Risk Evaluation and Mitigation Strategies) programs are sometimes required as conditions of approval for drugs with known serious risks. These programs can mandate patient registries, prescriber certification, or pharmacy certification to ensure the drug is used safely within defined parameters. Post-market surveillance has identified signals that led to drug withdrawals — including rofecoxib (Vioxx) for cardiovascular risk and cisapride for fatal arrhythmias — after millions of patient exposures in the real world.

This article is for educational purposes only and does not constitute medical advice. Clinical trial participation decisions should be made in consultation with a qualified healthcare provider.