Risk-Adjusted Phase I Clinical Trial
Also known as: risk-stratified Phase I trial, risk-adaptive dose-escalation study, covariate-adjusted Phase I study, risk-based dose-finding trial
A risk-adjusted Phase I clinical trial is a first-in-human or dose-finding study that explicitly incorporates patient-level risk covariates — such as organ function, prior therapy, or genetic markers — into the dose-escalation model. Rather than treating all enrolled participants as homogeneous, the design accounts for individual differences in tolerance, allowing the recommended dose to vary by risk stratum. This approach is especially common in oncology, where patients with impaired renal function or heavily pre-treated disease may tolerate lower doses than the broader population.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use a risk-adjusted Phase I trial when (1) the patient population is clinically heterogeneous with respect to drug tolerability — such as when organ impairment, prior therapies, or pharmacogenomic factors are expected to meaningfully affect the MTD; (2) there is mechanistic or pharmacokinetic rationale supporting differential dosing across subgroups; and (3) the trial is large enough to enrol adequate numbers within each stratum. Do not use this design when the patient population is homogeneous, when there is no prior evidence or mechanistic basis for risk stratification, when sample size is too small to estimate stratum-specific MTDs reliably, or when simpler designs (standard 3+3 or single-stratum CRM) are sufficient and more operationally feasible.
Strengths & limitations
- Explicitly accounts for patient heterogeneity, reducing excess toxicity in higher-risk subgroups while avoiding under-dosing of lower-risk patients.
- Borrowing strength across strata through a shared model component improves MTD estimation efficiency compared with running fully independent trials for each subgroup.
- Aligns with precision medicine principles by generating stratum-specific dosing recommendations that can directly inform Phase II eligibility and dosing criteria.
- Bayesian implementations update dose recommendations in real time as data accumulate, reducing the number of patients treated at sub-therapeutic or toxic doses.
- Regulatory guidance increasingly supports covariate-adjusted dose-finding designs, making trial approval more straightforward when the rationale is well-documented.
- Requires larger total sample sizes than a single-stratum Phase I trial, making it slower and more resource-intensive to complete.
- Model specification — choice of covariate, link function, and priors — must be justified before data collection; post-hoc covariate selection inflates error rates.
- Operational complexity increases: real-time model updating, stratum-specific dosing, and blinded safety monitoring require specialized biostatistical support infrastructure.
- Stratum boundaries must be fixed a priori; if the clinical variable does not cleanly stratify tolerability, the design gains little over a standard Phase I.
Frequently asked
How is this different from a standard Phase I trial?
A standard Phase I trial treats all patients as equivalent and finds a single MTD for the entire population. A risk-adjusted design explicitly models patient subgroup membership as a covariate in the dose-toxicity relationship, producing subgroup-specific dose recommendations and allowing dose-escalation rules to differ across strata.
Can the 3+3 rule be used for risk-adjusted Phase I trials?
Technically yes, by running separate 3+3 cohorts per stratum, but this forfeits the efficiency gains from borrowing information across strata. Model-based methods (CRM with covariates, BOIN with stratification) are preferred because they share information across strata through shared model parameters, reducing total sample size.
How many risk strata can the trial accommodate?
Practically, two to three strata are typical. With more strata the per-stratum sample size becomes too small to estimate stratum-specific MTDs reliably, even with model-based borrowing. If more subgroups need dosing guidance, a model with continuous covariate adjustment may be more appropriate than discrete strata.
What statistical software supports risk-adjusted CRM?
R packages such as dfcrm, crmPack, and BOIN support covariate-adjusted and stratified dose-finding models. FACTS (a commercial platform) and custom Stan/JAGS models are also used for Bayesian CRM with risk covariates. Simulation of operating characteristics under plausible true dose-toxicity scenarios is required before trial launch.
Does FDA accept risk-adjusted Phase I designs?
Yes. FDA guidance documents on adaptive designs (2019) and complex innovative trial designs acknowledge model-based Phase I methods including covariate-adjusted approaches. Pre-IND or end-of-Phase I meetings are advisable to discuss the specific model, simulation results, and stopping rules with regulatory reviewers.
Sources
- Iasonos, A., Wilton, A. S., & Gonen, M. (2008). A review of stochastic dose-finding methods. Statistics in Medicine, 27(25), 5031–5046. link ↗
- O'Quigley, J., Pepe, M., & Fisher, L. (1990). Continual reassessment method: A practical design for phase 1 clinical trials in cancer. Biometrics, 46(1), 33–48. link ↗
How to cite this page
ScholarGate. (2026, June 3). Risk-Adjusted Phase I Clinical Trial. ScholarGate. https://scholargate.app/en/epidemiology/risk-adjusted-phase-i-clinical-trial
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Adaptive Phase I Clinical TrialEpidemiology↔ compare
- Bayesian Phase I clinical trialEpidemiology↔ compare
- Dose-Response AnalysisEpidemiology↔ compare
- Phase I Clinical TrialEpidemiology↔ compare
- Randomized clinical trialEpidemiology↔ compare