EMA Can Do More With Its Guidance On Multiplicity Issues In Clinical Trials, Says EFPIA
Executive Summary
EU research-based industry association EFPIA has made several suggestions to improve the European Medicines Agency’s draft guideline on addressing multiplicity issues in clinical trials. Multiplicity can have a substantial influence on the rate of false-positive conclusions, which may affect approval and labeling of an investigational drug.
The European Medicines Agency’s draft guideline on how companies should deal with multiplicity issues in clinical trials to control the rate of false-positive conclusions covers the relevant key principles, but does not provide enough examples of methods that can be used to control multiplicity, said the European Federation of Pharmaceutical Industries and Associations.
The draft guideline, which is intended to provide guidance on how to deal with multiple comparison and control of type I error in the planning and statistical analysis of clinical trials, provides only “limited” details on and/or examples of methodologies that can be used for control multiplicity, according to EFPIA.
EFPIA’s comment is in response to a public consultation of the draft document that was launched in April this year.
The inclusion of methodologies and/or examples would help illuminate the key principles of multiplicity, EFPIA suggested. In the interest of leaving room for further methodological developments, these methodologies and/or examples “should not be construed as a regulatory recommendation” and should not be introduced in any technical detail, the trade group added.
The draft guideline was developed following the publication of a concept paper in 2012, which stated that multiplicity issues, if not handled properly, could present an elevated probability of false positives when performing multiple analyses and hence result in unsubstantiated effectiveness claims for a drug. (Also see "EMA planning new guidance to address "multiplicity issues" in drug clinical trials" - Pink Sheet, 8 Jun, 2012.) The guideline was issued to explain how companies should deal with multiple comparison and control of type I error in the planning and statistical analysis of clinical trials.
EFPIA also suggested expanding the guideline to introduce the concept of “estimand” – i.e., that which is to be estimated in a statistical analysis. It noted that choosing appropriate estimands was being addressed by the International Council for Harmonisation’s new guideline on statistical principles for clinical trials (E9(R1).
If the EMA were to explain how companies should address multiplicity in terms of estimands (if more than one estimand has been defined), then this could potentially simplify and clarify some of the sections of the guideline, the association said. Referring to the section in the agency’s guideline on “analysis sets,” in which the EMA advises performing robustness assessments on different subsets of patients, EFPIA said that as per the anticipated language of the ICH E9(R1) guideline, “this amounts to testing different estimands, rather than robustness testing of the same estimand.” Quoting ICH E9 working group members, EFPIA said the introduction of the estimand concept would likely reduce the need for wider sets of sensitivity analyses with unclear contribution to robustness of the primary analysis.
EFPIA also called on the EMA to consider addressing additional topics related to multiplicity issues in clinical trials that are not covered by the draft guidance, such as:
- Endpoint analysis based on data pooled from two pivotal trials.
- Different regional requirements on the choice of the primary endpoint in global clinical trials. For example, would an “agency-wise” error rate control be sufficient for the EMA?
- Multiplicity considerations in combined non-inferiority/superiority trials, possibly for more than one endpoint or dose.
There is also inconsistency in the use of certain terminologies, EFPIA noted. For example, the terms “variable/endpoint,” “trial/study,” and “subject/patient” have been used in an exchangeable way, the trade group explained, adding that it would be helpful to agree on one term in each case and use it consistently.
In addition, EFPIA noted that according to the EMA’s guideline, “secondary endpoints require multiplicity control if they aim to provide confirmatory conclusions, but not if they only provide supportive evidence.” This approach, it says, is different to the one that the US Food and Drug Administration prescribes in its draft guidance on Multiple Endpoints in Clinical Trials, which seems to suggest that all secondary endpoints require multiplicity adjustments. (Also see "Multiple Endpoints In Clinical Trials: US FDA Advises How To Avoid False Positives" - Pink Sheet, 14 Jan, 2017.) On this front, EFPIA requested clarification and international harmonization to ensure consistent practice.
From the editors of Scrip Regulatory Affairs.