Charter for EFSPI/PSI Special Interest Group (SIG) on Estimands in Oncology Applications

TA draft addendum of the ICH E9 guideline on Statistical Principles for Clinical Trials was released in August 2017 and introduced an estimand framework. The new framework aims at aligning trial objectives and statistical analyses by demanding a precise definition of the population quantity of interest, the estimand. The addendum is anticipated to have a major impact on drug development, as the choice of an estimand will drive the trial design and sample size, conduct, data collection and analysis of clinical trials moving forward. Furthermore, the framework is anticipated to facilitate discussions about relevant treatment effects, interpretation of study results, and added value of the drugs between sponsors, regulators, payers, physicians, and patients. 

The need to amend E9 with a discussion on estimands grew out of the realization of an apparent lack of alignment between the objectives stated in a clinical trial protocol and the accompanying quantification and interpretation of the treatment effect reported in a regulatory submission. While the estimand framework has been developed with different clinical trial settings and endpoints in mind, the examples discussed in publications, at scientific meetings, and in the addendum have largely focused on symptomatic studies and continuous, longitudinal endpoints. However, in oncology primary endpoints in clinical trials are typically of the time-to-event (T2E) type (such as overall, progression-free, or event-free survival). For these endpoints, there are still many open questions concerning endpoint definition, data collection, analysis methodology, interpretation, and use in drug development:
  • How can T2E endpoints be embedded in the addendum framework, and what are key estimands in Oncology? 
  • How do the 5 strategies to handle intercurrent events proposed in the addendum apply to T2E endpoints?
  • How can established methods to analyze questions in oncology, e.g. treatment switching or various censoring schemes, be embedded in the addendum framework?
  • The addendum in its current version does not require a causal interpretation of a proposed estimand. However, are there parts in the drug development lifecycle that require a causal interpretation of an estimand? Being aware that the hazard ratio from Cox regression does not admit a causal interpretation, what are effect measures for T2E endpoints that are amenable to such a causal interpretation?
Furthermore, increased transparency on the treatment effect of interest is considered an important goal of the addendum in order to facilitate discussions between different stakeholders. However, no guidance on protocol implementation has been provided so far, and there is a risk that the same estimand will be described in different ways by sponsors, potentially leading to confusion for regulators, physicians, patients, and possibly even inconsistent labels. Therefore, it is desired to ensure common understanding and consistent definitions for estimands across industry.

The scope of this SIG is to provide some answers to the above questions through a variety of activities. These will include: 
  • Collaborating with experts to refine and possibly extend available methods.
  • Publishing reviews of the available methods, case studies, and white papers with recommendations, for statisticians and non-statisticians.
  • Interacting with regulators and the broader clinical community to obtain a better understanding of their requirements. Work with them to agree on common definitions of estimands in oncology.
  • Providing trainings, workshops and talks.
  • Promoting good practice through templates for study protocols and statistical analysis plans.

  1. Membership click here
  2. Subteams: Five subteams are built that will work on dedicated topics (see table below)
  3. Meetings: 3-weekly teleconferences of the subteams, regular teleconferences of the entire SIG
  4. Information sharing: Via the EFSPI and PSI websites, EFSPI and PSI Newsletters, webinars, presentations at the PSI conference and other conferences, publications in mainstream journals.