Dose escalation refers, in early clinical trials (typically Phase I), to the stepwise increase of the administered dose in order to derive a dose that is both safe and potentially effective for further development phases. The aim is to systematically investigate dose-limiting toxicities (DLTs) and to establish the maximum tolerated dose (MTD) or a recommended Phase II dose (RP2D).
Objectives and endpoints of dose escalation
Safety and tolerability data are the central focus. Predefined DLT criteria are frequently assessed within an observation window (e.g. the first treatment cycle). In addition, pharmacokinetic and pharmacodynamic parameters are gaining importance, such as exposure (AUC, Cmax), target engagement or biomarker changes. Particularly for oncology products and biologics, the RP2D is no longer determined exclusively via the MTD but via an integrated benefit-risk profile.
Statistical and clinical interpretation is also important: an “acceptable” toxicity probability is defined in advance (e.g. a 20–30% DLT rate), from which the target region for dose finding is derived. In addition, dose escalation can support secondary objectives, such as the selection of dose levels for combination therapies or the planning of dose expansions in biomarker-defined subgroups.
Study designs: from 3+3 to model-based approaches
The classic approach is the 3+3 design: small cohorts receive a starting dose; if no DLTs occur, escalation proceeds to the next dose level, while DLTs trigger expansion or de-escalation. This approach is simple but statistically inefficient and yields an imprecise estimate of the MTD. Modern alternatives are model-based designs such as the CRM (continual reassessment method) or Bayesian approaches, which continuously update the toxicity probability per dose level and thereby more frequently assign doses close to the target toxicity.
Further variants include accelerated titration designs, which initially allow larger dose increments, as well as mTPI or BOIN designs, which bridge rule-based and model-based approaches. In Europe, ethics committees and authorities pay particular attention to ensuring that the escalation pathway is traceable, patient-friendly and practically feasible.
Operational implementation and patient safety
For sponsors and CROs, a clear escalation algorithm is essential: starting-dose derivation, cohort size, escalation steps, DLT definitions, safety review processes and stopping rules must be set out in the clinical study protocol. In practice, a safety review committee or a Data Safety Monitoring Board is often involved, granting approval to escalate after each cohort. Central elements are timely SAE reporting, rapid data availability (EDC) and robust query management, so that decisions are not based on incomplete data sets.
Logistics and IMP handling are also operationally relevant: different dose levels frequently require multiple strengths, possible repackaging, or a specific randomisation/dispensing procedure. Errors in drug accountability can create immediate patient hazards and lead to critical audit findings. Standard operating procedures, training and clear delegation-of-authority logs are therefore essential.
Typical pitfalls and common misunderstandings
A common mistake is excessive fixation on the MTD, even though for targeted agents the optimal biological dose may be lower. Also critical: inconsistent DLT assessment between investigational sites, delayed laboratory data, or protocol deviations in dose adjustments. The definition of the observation window is also relevant: windows that are too short may miss late toxicities, while windows that are too long slow down recruitment. A pragmatic solution is a clear approach to “late-onset” toxicities and a documented procedure for re-escalation following protocol amendments.
Misunderstandings also arise when PK/PD data are not evaluated in a timely manner: escalation may then continue even though exposure is already approaching saturation, or relevant interactions (e.g. CYP inhibition) are underestimated. Equally, an overly conservative escalation can unnecessarily keep many subjects at subtherapeutic doses. Predefined decision criteria and good coordination between the clinical team, biostatistics and pharmacometrics help address this.
Regulatory expectations and documentation
Regulatorily, dose escalation studies are part of clinical trials under EU CTR 536/2014; requirements on risk minimisation, safety reporting and protocol quality apply without restriction. Planning should be guided by ICH E6(R3) (GCP), in particular by risk-based quality management, data integrity and clear responsibilities. For IMPs, requirements on manufacturing and labelling (GMP/GDP) are also relevant, as dose levels may require different strengths or formulations. Changes to the escalation scheme may constitute a substantial amendment and require consistent documentation, including justification in the study report.
For international programmes, it should additionally be noted that requirements for safety reporting and dose finding may vary slightly between regions; the study report should therefore transparently justify the derivation of the recommended dose. Particularly for first-in-human studies, authorities expect a rigorous risk analysis, including the selection of the starting dose, escalation steps and defined stopping criteria.
Practical relevance: transition to expansion cohorts and later phases
Once a suitable dose level has been identified, expansion cohorts are often conducted to collect additional safety data in defined patient groups and to examine early efficacy signals. For the later regulatory strategy, it is important that the derivation of the RP2D is traceable and data-driven: dose-exposure relationships, dose-response, and the clinical plausibility of the selected dose should be presented coherently. The quality of the dose escalation phase therefore directly influences the probability of success of Phase II/III and reduces the risk of costly later replanning.
FAQ: How is the starting dose for a dose escalation determined?
It is usually derived from non-clinical data (e.g. NOAEL, MABEL) with safety factors and a dose-level logic that takes into account both patient safety and meaningful increases in exposure.
FAQ: What is the difference between MTD and RP2D?
The MTD is the highest dose with acceptable toxicity within the defined window; the RP2D additionally takes into account PK/PD, biomarkers and overall clinical assessment and may lie below the MTD.
FAQ: When is a model-based design appropriate?
When many dose levels, complex toxicity profiles or combination therapies are involved, model-based designs typically provide more efficient and ethically more favourable dose allocations than purely rule-based approaches.
Regulatory references: Regulation (EU) No 536/2014 (Clinical Trials Regulation); ICH E6(R3) Good Clinical Practice; supplementary ICH E8(R1) (General Considerations for Clinical Studies).