Euro Roundup: EMA seeks feedback on the use of Bayesian methods in clinical development

The European Medicines Agency (EMA) has begun a consultation into the use of Bayesian methods in the analysis of clinical trial data.

Bayesian methods are one of the main approaches to statistical inference. The methods incorporate prior information into the analysis and can calculate the probability that a hypothesis is true. Frequentist methods, another main approach to statistical inference, have traditionally been the standard technique for analyzing clinical data. The methods are driven solely by the data and give a p-value that shows the probability of observing results at least as extreme as the analyzed dataset.

EMA has published a draft concept paper on Bayesian methods. In the document, EMA sets out the problems with the current guidance and outlines plans to write a reflection paper about when to use Bayesian methods, the information and justifications needed to support the choice, and the importance of sponsors engaging early with regulators when considering using the approach.

The International Council for Harmonisation’s (ICH) guideline on statistical principles for clinical trials supports the use of Bayesian methods when the reasons for doing so are clear and the resulting conclusions are “sufficiently robust.” Other ICH and EMA guidelines also reference Bayesian methods.

Yet EMA has identified lack of clarity on when Bayesian methods can be accepted in the confirmatory setting and the methodological requirements needed to address potential regulatory concerns. Specific areas of uncertainty include when companies must justify their use of Bayesian methods and how to handle the increased methodological complexities of the approaches, particularly when using external data.

EMA’s assessment of the uncertainties reflects an increase in the number of proposals it has received about enriching clinical trial data. Applicants have proposed using Bayesian methods to borrow historical or external data “to draw conclusions in the same or a related population or to draw conclusions in populations where adequately powered trials are not possible,” EMA said.

The agency also noted the use of Bayesian methods to generate supportive evidence, rather than the confirmatory evidence used to seek approval, and to inform internal decisions during early-phase trials. Even if the analyses are only used internally, Bayesian methods are part of scientific advice submissions and may be included in assessments of marketing authorization applications. EMA named using Bayesian methods to fit pharmacometrics models and in early-phase dose-finding as examples of supportive work.

EMA is accepting feedback on the draft until 30 April. The agency aims to release a draft reflection paper for consultation next year and finalize the document in 2028.

Reflection Paper

EMA finalizes reflection paper on the qualification of non-mutagenic impurities

EMA has finalized its reflection paper on the qualification of non-mutagenic impurities, significantly expanding and revising the draft it released for consultation last year.

The agency first released a draft reflection paper on the topic in 2018. However, the agency never finalized that version, choosing instead to start again with a consultation on a new draft in 2025. EMA created the draft to build on ICH guidelines for qualifying non-mutagenic impurities, recognizing the need for an adequate safety evaluation and suggesting alternatives to animal testing. The ICH rules on qualifying impurities have led to dedicated animal studies, EMA said.

EMA has retained the draft’s high-level objectives but made extensive changes to the details, increasing the length of the paper from 30 pages to 42 pages in the process. Changes include a total rewrite of an introductory section that defines the scope of the document.

“The principles and methodologies discussed are intended to support the qualification of new or elevated impurity levels, which are specified at levels that exceed existing ICH Q3A/Q3B [qualification thresholds], particularly when such impurities are identified after completion of non-clinical toxicology studies,” EMA said. “These new or elevated impurity levels may result from any type of [drug substance] or [drug product] related changes, such as changes in manufacturing processes.”

EMA added that the paper does not reflect further on the “As Low As Reasonably Possible,” or ALARP, principle for the limits of impurities. The agency expects companies to generally base specification levels on batch data and not only be guided by toxicologically qualified levels.

Unlike in the draft, surrogate and grouping approaches to qualifying impurities are not classed as “new approach methodologies” in the final paper. EMA has separated the surrogate and grouping read-across methods from a set of new approach methodologies that, in text added to the final paper, it said are still under development.

The developmental approaches to impurity qualification include computational toxicology, in vitro approaches, and hazard characterization and quantitative risk estimation. Those three approaches were in the draft. EMA has added a subsection on adverse outcome pathways, which were previously grouped with computational toxicology, to the final set of new approach methodologies.

Reflection Paper

Swissmedic updates guidance on reporting safety signals for human medicines

The Swiss Agency for Therapeutic Products (Swissmedic) has comprehensively revised its guidance on drug safety signals for human medicines.

Swissmedic said it fundamentally redesigned the guidance structure to ensure a better overview and readability. At the same time, the agency made major changes to the content of several sections of the text, more clearly defining the start and end of the reporting obligation and deleting redundant content on signal correspondence.

“Applications to change the medicinal product information should no longer be submitted at the same time as the signal report as in future, risk-minimization measures from signals from authorities will be implemented entirely within the scope of the signal evaluation procedure,” Swissmedic said.

The agency added a section on signal evaluation procedures. Swissmedic used the section to explain that it will contact marketing authorization holders up to six months after the signal conclusion report if it considers measures to be appropriate or requires additional information.

Swissmedic Notice

European Commission posts harmonized standards on medical devices, sterilization

The European Commission has published harmonized standards for multiple types of medical devices and for the sterilization of healthcare products.

Under the medtech regulations, European standardization organizations develop harmonized standards for medical devices and in vitro diagnostics (IVDs) based on requests from the Commission. Last week, the Commission published one medical device standard and one IVD standard. The publications added new entries to standards first released in 2021.

The device standard covers areas including neurosurgical implants, the biological evaluation of medical devices, clinical investigations of medical devices for human subjects, and non-active surgical implants. The IVD standard covers healthcare product sterilization and information supplied by the manufacturer.

Like other standards, the texts refer to external documents to define the approach to the products. The IVD standard refers to six documents published by the International Organization for Standardization about topics such as the sterilization of healthcare products. The device standard adds 12 documents.

Implementing Decision, More

Other News:

EMA has started a consultation into the clinical evaluation of diagnostic agents. In the draft concept paper, the agency explains the need to update its current advice on the topic, which it adopted in 2009. The concept paper sets out EMA’s intent to revise the guideline in light of changes in the scientific landscape and regulatory thinking since the agency adopted the current document. Concept Paper

Euro Roundup: EMA seeks feedback on the use of Bayesian methods in clinical development

Related topics

Recommended content

Welcome to the new RAPS Digital Experience