Most clinical trials performed in drug development contain multiple endpoints to assess the effects of the drug and to document the ability of the drug to favorably affect one or more disease characteristics. When more than one endpoint is analyzed in a single trial, the likelihood of making false conclusions about a drug’s effects with respect to one or more of those endpoints could increase if there are no appropriate adjustments for multiplicity. As discussed below, while not every study design that includes multiple endpoints requires a multiplicity adjustment, there are other study design considerations that must be appropriately adapted to accommodate multiple endpoints (e.g., sample size calculation).
To help sponsors manage these issues in designing their clinical trials and assessing the data, FDA has issued the final guidance “Multiple Endpoints in Clinical Trials,” which provides FDA’s thinking about the problems posed by multiple endpoints in the analysis and interpretation of study results, and about how these problems can be managed in clinical trials for drug development. It finalizes draft guidance from January 2017, and builds on the International Council for Harmonisation (ICH) guidance “E9 Statistical Principles for Clinical Trials,” which was released in September 1998. The final guidance provides standard background information on statistical best practices, and describes various strategies for grouping and ordering endpoints for analysis of a drug’s effects and applying some well-recognized statistical methods for managing multiplicity within a study to control the chance of making false conclusions about a drug’s effects.
In the final guidance, FDA advises trial sponsors to group measurements into a “hierarchy” of endpoints, with the highest group being primary endpoints that are required for approval. Secondary endpoints, meanwhile, are those that may support the primary endpoints or demonstrate additional clinical effects. A third category, “exploratory endpoints,” may apply to all other endpoints, including those that may not impact labeling yet be helpful for additional research or for developing new hypotheses.
There are also different types of primary endpoints, the final guidance explains:
- Co-primary endpoints: Used when a treatment effect on all of two or more clinical features is critically important to demonstrate efficacy.
- The chance of making an erroneous conclusion of efficacy based on co-primary endpoints is reduced compared to the use of any single endpoint.
- However, the chance of failing to detect beneficial effects may be increased.
- Since success is required on both endpoints, a multiplicity adjustment is not required. However, even in cases where a multiplicity adjustment is not required, the introduction
- Multiple primary endpoints: Where a treatment effect on any one of these endpoints is sufficient to support a conclusion of effectiveness.
- Here, the guidance cautions that it is critical to develop a prospective plan to address the chance of erroneously detecting an effect that is not present (i.e., a Type I error).
- A variety of approaches to address multiplicity are described in the guidance.
- Composite endpoints: Important clinical outcomes are combined into a single primary endpoint.
- When a single statistical test is performed on the composite endpoint, there is no multiplicity problem.
- For these endpoints, choice of (and effects on) the components should be examined with care.
- Multi-component endpoints: Within-patient combination of two or more components.
- Each patient gets an overall rating based on observation of all specified components according to specified rules.
- May be efficient if within-subject component effects generally trend in the same direction.
- The chance of detecting a beneficial effect may be adversely affected if there is limited concordance among component endpoints.
- Evaluation of components involves similar considerations as for composite endpoints.
For trials that do include secondary endpoints – such as survival rates after a cardiovascular drug has met the primary endpoint of heart failure-related hospitalizations – the guidance advises that trial sponsors should include secondary endpoints in a Type 1 error control plan to avoid the error of concluding there is a treatment effect when none actually exists.
As compared to the draft version of the guidance, the appendix of the final version is lengthier because it consolidates discussion of available statistical methods: content that was previously scattered throughout the body of the draft guidance.
In the 20 public comments on the draft guidance, there was debate about the role of secondary endpoints, partially because FDA allows additional claims based on the evaluation of secondary endpoints, but the European Medicines Agency does not. FDA’s conclusion is that the inclusion of multiple endpoints is a valuable clinical trial design option; however, when considering efficacy claims based on the evaluation of secondary endpoints, the guidance makes it clear that FDA will be wary in assessing efficacy due to possible multiplicity issues. Note, multiplicity problems for safety analyses are not covered by this guidance.
FDA published alongside the guidance a three-page “Snapshot” infographic summarizing the guidance, as well as a podcast with Dr. John Lawrence, a statistician in CDER’s Office of Biostatistics, who offered highlights from the guidance. Dr. Lawrence emphasized caution for drug sponsors analyzing more than one endpoint in a single trial, and warning that the likelihood of making false conclusions about a drug’s effects can increase due to “multiplicity.”
Next steps
In sum, this latest FDA publication spotlights how critical it is for sponsors of drugs and biologics to consider efficacy endpoint issues when thinking about the scientific questions they are trying to answer in their clinical trials, as these issues carry significant import for the design and analysis of those trials. We recommend consulting with FDA early on regarding the type of endpoints being considered for a trial, and on the types of conclusions about efficacy that will be able to be drawn from the trial data. Adjusting for multiplicity upfront and planning statistical analyses is crucial for successful trials to support a medical product’s indication and benefit claims.
If you have any questions about how to categorize the endpoints in a clinical trial or the ability to make conclusions from multiple endpoints in a clinical trial, or with clinical trials more generally, please do not hesitate to contact any of the authors of this alert or the Hogan Lovells attorney with whom you generally work.
Authored by Heidi Gertner and Blake Wilson
