A model structure for describing uncertainties in benefit-risk assessment of oncology drug applications

The European Medicines Agency (EMA) uses structured reports to communicate the scientific review of drug applications. A European Public Assessment Report (EPAR) is published at the end of the review detailing the scientific assessment, which includes benefit-risk analysis and uncertainties in favorable and unfavorable effects. Currently, there is no detailed guidance on how to communicate uncertainties in the EPAR. This study aimed to identify uncertainties described in the benefit-risk section of a series of EPARs and derive a possible model structure for communicating them.

A series of approved oncology drug applications that had used the latest EPAR template at the time of analysis was selected. The model structure was adapted iteratively, aiming to identify a small number of key elements, with input from the study team. Finally, the model structure was discussed with two experienced clinical assessors to determine clarity and potential usefulness.

From 64 oncology EPARs published between 2011 and 2017 (26 related to products with orphan designation), 263 uncertainties in the benefit-risk assessment were identified. The final model structure included Cause (what causes the uncertainty), Aspect (what is the uncertainty about, further described as a high-level domain and a specific component), Type (what is the kind of uncertainty, like not enough information or conflicting information), and Strategy (how the uncertainty is addressed). This four-element structure, Cause, Aspect, Type, Strategy (CATS), was discussed with expert assessors and was found to be generally understandable and relevant.

The CATS model structure has been derived as a starting point for communicating uncertainties in benefit-risk assessment of drug applications. To increase relevance, it was derived based on issues raised during actual reviews and discussed with expert reviewers. Limitations include the narrow focus of the current series and the need for validation. If found useful, this structure could eventually be used to further enhance assessment report templates.

The proposed CATS model structure may facilitate communicating uncertainties in the benefit-risk assessment of drug applications. Further refinements, depending on the purpose and validation, aiming at broader applicability, are needed.