Designs and statistical methods for adaptive clinical dose-finding trials


Designs and statistical methods for adaptive clinical dose-finding trials Evidence Maps

Maps

Further information

Version of the Evidence Map

Current version

Version 1.0 – published 19 September 2025 – literature coverage up to 15 January 2025.

Version control

Version	Date	Description
Version 1.0	19/9/2025	Initial version published concurrently with the original article, incorporating literature up to 15 January 2025

Review strategy

The review strategy for updating the evidence map follows the same methodology as described in the original paper using the following search string:

(exp *Research design/ or exp *Clinical Trials as Topic/)

AND

((adapt* or sequential* or interim or bayes* or seamless or flexible or early-phase or two-stage or multi-stage or platform or modification or (dose adj2 (selection or response or finding or escalation)) or re-estimation or reassessment or "pick-the-winner" or "drop-the-loser" or (adjust adj2 sample) or (adaptive adj3 (response or outcome)) or (design adj3 (modification or enrichment)) or (phase adj (1-2 or I-II))).ti,ab,kf.)

AND

(("Annals of Statistics" or "Bayesian Analysis" or "Biometrical Journal" or "Biometrics" or "Biometrika" or "Biostatistics" or "BMC Medical Research Methodology" or "British Journal of Cancer" or "Clinical Cancer Research" or "Clinical Trials" or "Communications in Statistics: Theory and Methods" or "Computational Statistics & Data Analysis" or "Contemporary Clinical Trials" or "Journal of Biopharmaceutical Statistics" or "Journal of Clinical Oncology" or "Journal of Statistical Planning and Inference" or "Journal of Statistical Software" or "Journal of the American Statistical Association" or "Journal of the Royal Statistical Society. Series B, Statistical Methodology" or "Journal of the Royal Statistical Society. Series C, Applied Statistics" or "Pharmaceutical Statistics" or "PLoS ONE" or "Statistica Sinica" or "Statistical Methods in Medical Research" or "Statistics & Probability Letters" or "Statistics in Biopharmaceutical Research" or "Statistics in Medicine" or "Technometrics" or "The Annals of Applied Statistics" or "trials" or "Controlled Clinical Trials" or "pharmaceutical medicine" or "BMC Medical Research Methodology" or "The International Journal of Biostatistics").jw.)

The authors hold the right to adapt this review strategy and/or search string at any time. Any changes to the original review strategy and/or search string will be documented here.

How to cite the evidence map?

R.Kessels, Y.Long, R.Spijker, et al. A Knowledge Base of Designs and Statistical Methods for Adaptive Clinical Dose-Finding Trials. Pharmaceutical Statistics. 2026; 25, no. 4: e70074, https://doi.org/10.1002/pst.70074.

Glossary

Motivations for introducing the new design or methodology	Explanation
Treatment-related motivations	Methodologies related to characteristics of the treatment.
Combination therapies	Considering methodology to deal with dose escalation for drug combinations.
Courses of treatment	Methods considering multiple treatment cycles during dose-escalation as opposed to using only the first treatment cycle and/or methods identifying the most optimal dose-schedule combination using multiple treatment cycles
Intra-patient dose escalation	Escalation of doses within patients (as part of the treatment strategy) or to collect more data.
Non-monotonicity	Methods considering dose-efficacy profiles that can decrease after certain dose levels.
Model uncertainty	Incorporating methods aimed to reduce the impact of incorrect assumptions about the underlying dose-toxicity/efficacy relationship.
Vaccine setting (low toxicity)	Trial design/method specifically developed for the vaccine setting where low toxicity rates are expected.
Endpoint-related motivations	Related to characteristics of the endpoint.
Delayed outcomes	Accounting for late onset (delayed) outcomes of toxicity and/or efficacy.
Surrogate endpoints	Using surrogate endpoints.
Population-related motivations	Related to characteristics concerned with patients/population.
Group/patient heterogeneity	Accounting for different effects of a drug across different patient groups (i.e. finding subgroup-specific doses).
Pediatrics	Trial design/method specifically developed for the pediatric setting.
Design/methodology-related motivations	Related to the design/methodology.
Convergence to suboptimal dose	Ensuring that the methodology convergences to the most optimal dose. Solving issues with methods that can get stuck on a suboptimal dose.
Informative prior	Specifically focusing on incorporating informative prior distributions.
External information	Incorporating external information observed outside the trial.
Selecting multiple doses	Selecting multiple doses or a dose range to be tested in next study phases.
Adding new dose levels	Allow additional dose levels to be defined when study has already been initiated.
No fixed dose levels	Relaxing restriction to predefined set of dose levels
Varying dose levels for the same cohort	Patients entering the trial within the same cohort may receive different dose levels (as opposed to conventional ways where patients in the same cohort all receive the same dose level)
Large/wide dose range	Design accommodates the possibility to evaluate a wide range of different dose levels (i.e. much more than widely used standard of 6 dose levels).
Dose with continuous infusion	Dose-finding for drugs that are delivered through continuous infusion.
Control arm	Designs that use a (randomized) control arm in the early-phase trial
Lack of software	Developing user-friendly software to evaluate and compare different designs.
Type I error control	Trial specifically focusing on type I error control
Backfilling	Trial designs that include backfilling cohorts in addition to dose-escalation cohorts to collect more data on safety, efficacy and/or pharmacokinetics on doses already deemed safe.
Dose expansion cohort	Trial designs that include dose expansion cohorts in addition to dose-escalation cohorts, where new patients are recruited to collect additional information.
Suboptimal inference in 3+3 design	Improved MTD estimation for the 3+3 design (e.g. use of more data than recent response).
Suboptimal dose-escalation	Dose escalation decisions only based on efficacy data and not on toxicity data or dose escalation decisions only based on data of current and adjacent doses.
Operational/ethical motivations	Related to operational/ethical characteristics.
Trial duration	Solving problems with too long trial durations.
Sample size	Solving problems with (too) small or (too) big sample sizes.
Underdosing	Minimize exposing too many patients to doses that have no or little therapeutic effect.
Overdosing	Minimize exposing too many patients to overly toxic doses (i.e. overdose control).