Phase I Clinical Trial — First-in-Human Dose-Escalation Study
- Introduced by
- Regulatory and clinical pharmacology community; formalized in U.S. FDA IND regulations (1963) and ICH guidelines
- Year
- 1960s (formal regulatory framework established ~1963–1970s)
- Type
- Interventional clinical study design
- Family
- Clinical / epidemiology
What does it do?
A Phase I clinical trial is the first stage of human testing for a new drug, biologic, or intervention.
Its primary objective is to evaluate safety, tolerability, pharmacokinetics (PK), and pharmacodynamics (PD) rather than therapeutic efficacy. Small cohorts of participants — typically healthy volunteers or patients with advanced disease — receive sequentially increasing doses to identify the maximum tolerated dose (MTD) and the dose-limiting toxicities (DLTs) that define the boundary for subsequent trials.
The intuition behind the method
Think of a Phase I trial as a careful exploration of unknown territory. Before you can ask whether a new drug works, you must ask whether it is safe enough to give to humans at all, and at what dose. Participants are enrolled in small groups (cohorts), each group receiving a slightly higher dose than the last. After every cohort, safety data are reviewed — if serious toxicities appear, escalation stops. The goal is to find the highest dose that the body can tolerate without unacceptable harm, so that Phase II can test whether that dose also achieves the intended clinical effect.
History
Formal Phase I trials as a regulatory category emerged from the 1963 Kefauver-Harris Amendment to the U.S. Federal Food, Drug, and Cosmetic Act following the thalidomide crisis, which required proof of safety before efficacy testing. The 3+3 cohort escalation schema became the de facto standard through the 1970s–1990s oncology literature. A landmark methodological advance came with John O'Quigley's continual reassessment method (CRM), published in Biometrics in 1990, which introduced model-based dose finding. ICH E6 Good Clinical Practice guidelines (1996, revised 2016) harmonised Phase I conduct internationally.
When to use it
Use a Phase I design when a new agent or combination has demonstrated sufficient preclinical efficacy and safety to justify first human exposure, and when the core question is safety, tolerability, PK/PD, and dose selection — not efficacy. It is the mandatory first step in the clinical development pathway for any new drug before proceeding to Phase II. Do not use a Phase I design when the agent already has established human PK data at the dose of interest, when you want to assess comparative efficacy (use a randomised controlled trial), or when the primary endpoint is patient-reported outcomes rather than toxicity.
Limitations and considerations
- Small sample sizes (typically 15–60 participants) provide limited power to detect rare adverse events or subgroup differences.
- The traditional 3+3 design is statistically inefficient — it does not use all accumulating data and tends to underestimate the true MTD.
- Phase I findings in healthy volunteers or end-stage patients may not generalise to the target patient population in Phase III.
- PK/PD may differ substantially across patient subgroups (renal/hepatic impairment, drug-drug interactions), requiring separate Phase I substudies.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Provides the essential human safety and PK data required before broader exposure, protecting subsequent trial participants.
- Identifies DLTs and the MTD, giving Phase II a principled dose basis rather than an arbitrary one.
- Modern adaptive designs (CRM, mTPI-2) improve dose-finding efficiency and reduce the number of participants treated at subtherapeutic doses.
- Yields early PK/PD data that can support mechanistic understanding and biomarker development.
- Regulatory frameworks (FDA, EMA, ICH E6) are well-developed, providing clear operational guidance.
Applications
- First-in-human evaluation of novel oncology agents (small molecules, biologics, antibody-drug conjugates).
- Safety and PK characterisation of new antivirals or antibiotics before Phase II efficacy studies.
- Dose-finding for combination regimens where individual agents are known but interaction toxicity is uncertain.
- Immunotherapy and CAR-T cell therapy trials where cytokine release syndrome defines the DLT.
- Paediatric dose extrapolation studies conducted after adult Phase I data are available.
- Vaccine dose-ranging studies to identify the optimal immunogenic and tolerable antigen dose.
How it works
How it works
A safe starting dose is derived from preclinical data — typically 10% of the severe toxic dose in the most sensitive animal species (NOAEL or STD10), converted to a human equivalent dose (HED). This calculation is mandated by regulatory guidance (e.g., FDA 2005 Estimating the Maximum Safe Starting Dose) to minimise risk to the first participants.
Participants are enrolled in small cohorts, classically 3–6 individuals per dose level (the '3+3 design'). Each participant within a cohort receives the same dose. Enrollment of the next cohort is held until the current cohort has completed the DLT observation window — usually one treatment cycle (28 days for oncology).
Safety and tolerability are assessed continuously: vital signs, laboratory values, pharmacokinetics (blood sampling over time to model Cmax, AUC, t½), and adverse event (AE) grading using standardised criteria such as NCI CTCAE. A Data Safety Monitoring Board (DSMB) or equivalent reviews accumulating data between cohorts.
If none or fewer than a pre-specified fraction of participants in a cohort experience a DLT, the next cohort is opened at the next higher dose level. If the pre-specified DLT rate threshold is exceeded, escalation stops. The traditional 3+3 rule stops escalation when 2 of 3 or 2 of 6 participants experience a DLT; model-based designs (continual reassessment method, CRM; EWOC; mTPI) update the dose recommendation after each patient rather than by cohort.
The MTD is conventionally defined as the dose level at which fewer than a predetermined proportion of participants (e.g., one-third) experience a DLT, just below the dose level that caused unacceptable toxicity. The recommended Phase II dose (RP2D) may be the MTD or a lower biologically active dose, particularly for targeted therapies where saturation of the target — not maximum tolerated toxicity — drives dose selection.
Common pitfalls
- Starting dose set too conservatively — if the starting dose is orders of magnitude below the biologically active dose, many participants receive a sub-therapeutic dose, increasing trial duration without meaningful safety benefit.
- Using the 3+3 design when model-based alternatives (CRM, mTPI) would more accurately identify the MTD — especially for targeted agents where the dose-toxicity curve is flat.
- Failing to pre-specify the DLT definition precisely — vague definitions of what constitutes a DLT create inconsistent stopping decisions across cohorts.
- Confusing MTD with RP2D in targeted therapies — for immunotherapies and molecularly targeted agents, the biologically effective dose may be well below the MTD, and using the MTD as RP2D risks unnecessary toxicity without therapeutic gain.
- Inadequate PK sampling design — sparse or mistimed blood draws can produce unreliable AUC and Cmax estimates, undermining the dose-escalation rationale.
Frequently asked questions
Are healthy volunteers or patients enrolled in Phase I trials?
It depends on the agent. For most drugs (e.g., antibiotics, vaccines), healthy volunteers are enrolled because the risk-benefit calculation is acceptable for a person without illness. For cytotoxic oncology agents, enrolling healthy volunteers would be ethically unacceptable — patients with advanced cancer who have exhausted standard options participate instead, accepting the risk in exchange for potential benefit and to advance knowledge.
What is the difference between DLT and MTD?
A dose-limiting toxicity (DLT) is a pre-specified adverse event of sufficient severity (e.g., Grade 3–4 by NCI CTCAE) occurring within the DLT observation window that is attributable to the study drug. The maximum tolerated dose (MTD) is the highest dose at which the observed DLT rate does not exceed the pre-specified threshold — conventionally one-third of participants. The MTD is derived from the pattern of DLTs across cohorts, not from a single event.
Why use model-based designs (CRM, mTPI) instead of 3+3?
The 3+3 design makes binary go/no-go decisions using only the most recent cohort's data, ignoring all prior information. Model-based designs continuously update a statistical model of the dose-toxicity curve using all data collected so far, providing more accurate MTD estimates with fewer patients treated at non-optimal doses. Simulation studies consistently show CRM and mTPI-2 outperform 3+3 in accuracy and efficiency.
Does a Phase I trial need a control arm?
Typically no. Phase I trials are single-arm, non-randomised studies focused on tolerability and PK. A placebo or active control is occasionally added when it is important to separate drug-related from disease-related adverse events, or in vaccine trials to blind participants, but this is the exception rather than the rule.
What happens after the Phase I trial is complete?
The RP2D and the safety, PK, and preliminary PD findings inform the design of Phase II trials, which enrol a larger patient population to evaluate preliminary efficacy signals at the recommended dose. Regulatory submissions (IND updates, clinical study reports) documenting Phase I findings are required before Phase II can begin.
Related methods
Sources
Storer, B. E. (1989). Design and analysis of phase I clinical trials. Biometrics, 45(3), 925–937.
Open DOI referenceCopied.International Council for Harmonisation (ICH). (2016). ICH E6(R2) Good Clinical Practice: Integrated Addendum to ICH E6(R1). ICH Harmonised Guideline.
Open publication referenceCopied.