Long-term survivors in clinical trials are those patients who are ‘cured’ after receiving treatments and have no chance to reach the primary endpoints, such as disease progression or death due to the underlying diseases. According to the formal definition of censoring in the survival analysis, the event, such as disease progression, will happen after the censoring time. Censoring long-term survivors at the end of trial isn’t consistent to this definition and may introduce potential bias in the survival analysis.

When long-term survivors are present in the clinical trials, it will prolong the trial enrollment duration and may end up underpowered due to insufficient events. In this roundtable discussion, we will focus on following issues • How to detect if there exist long-term survivors • Is the follow-up sufficient to detect long-term survivors • If long-term survivors are not avoidable, how to determine the sample size or treatment duration • What analyses are appropriate if long-term survivors exist?

Our ultimate goal is to come up with a general guidance or consensus in the clinical trial design.

References for this discussion include the book by Ross Maller and Xian Zhou “Survival Analysis with long-Term Survivors”, (1996) by John Wiley & Sons, and other recent publications in the literature.