Advanced methods for dose finding during drug development: Summary of the EMA /EFPIA workshop on dose finding (London 4-5 December 2014)

Aims: The most commonly used methods for defining the dose rationale during early stage drug development i.e. pairwise comparison of different doses have shown their limits. Moreover, the exploratory development is often ill conceived by using a stepwise approach across different phases of drug development: each stage focusing on answering specific questions rather than building a core of knowledge on the disease and drug of interest. In this respect the challenge of the dosing rationale is not about selecting "The dose" but characterizing the dose exposure response (DER) and covariate effects on this relationship and providing this information to clinical teams and regulators to select the dose that fits the individual patient. This informed decision should be ideally made ahead of confirmatory trials in order to be prospectively tested. The aim of this presentation to convey the message that dose finding and selection should be a multidisciplinary team decision and that there are many methods available to make sufficiently informed decisions.

Methods: In an effort to shift the current paradigm in dose selection and highlight the availability of well-established tools, EMA hosted with EFPIA a multi-stakeholder workshop on dose finding (London 4-5 December 2014). Some methodologies were presented that could constitute a toolkit for the well informed drug developer and regulator. The added values of these methods as well as limitations and the challenges for their implementation will be presented.

Results Methods presented during the workshop can be divided in two categories: (1) methods for data analysis and (2) methods for design optimization. Four different methods have been presented during the workshop for analysis of data generated during early phases of drug development: 1. Pairwise comparisons. 2. Empirical Regression methods. 3. Pharmacology-physiology guided methods, dedicated to describe the link between the (systemic and/or effect site) exposure to the drug and the clinically relevant endpoints based on the available knowledge on drug pharmacokinetics (PK) and pharmacodynamics (PD) and subsequently derive the optimal dose for confirmatory studies. These methods include pharmacometrics (modeling and simulations) and system pharmacology. 4. Integrative methods aiming to select the dose based on an objective weighing of a set plausible candidate models/related-parameters, very often based on acceptable fitting of the available data. MCP-mod, and model averaging can be included in this category.

Two methods have been described for study design optimization: adaptive dose ranging and Bayesian adaptive dose ranging. These two methods will be contrasted with other existing options for study design optimization including clinical trial simulations and optimization methods based on the population Fisher Information Matrix.

Conclusion: Different methods were described the workshop which can power the decision making for confirmatory trials were presented and discussed. Each of the presented methods has strengths and limitations. There was an agreement during the workshop on that selection of dose for Phase 3 is an estimation problem and should not be addressed via hypothesis testing. Model-based dose finding methods are more efficient than pairwise comparisons approaches. Proper characterization of dose response for both efficacy and safety should be a requirement in drug development programs to inform phase 3 dose selection and product labelling.