ISSUE: Partitioned survival analysis (PartSA) with standard parametric distributions is widely used in disease areas such as oncology, as it is perceived to provide a transparent way of extrapolating and quantifying survival benefits beyond the trial duration. A myriad of methods are available for extrapolating survival curves within a PartSA framework, but all are based on the same structural assumptions. With the introduction of immunotherapies such as checkpoint inhibitors and CAR T-Cell therapies, a new era of oncology therapies is upon us, bringing the potential for prolonged survival and possibly even cure. The question is, do these new therapies warrant new methods for extrapolating survival? Alternative methods such as mixture cure models, splines and responder-based approaches may provide benefits over the traditional PartSA approach, but also introduce new limitations. Furthermore, and perhaps most importantly, what is the acceptability of these new methods for different stakeholders such as policy makers and payers?

OVERVIEW: This panel will debate the application of different survival extrapolation approaches. The debate will be focussed around three different perspectives: a methodological perspective, industry perspective, and policy maker perspective. After 10-minute presentations introducing each of the different perspectives, the panellists will have a plenary discussion in two rounds, covering: merits and limitations of different survival extrapolation approaches, and adoption of new approaches. Each round will be governed by a central question, on which the audience will be polled both at the start and end of the round. Participation of the audience in the discussion, either in-person or via the app, is highly encouraged. The discussion is not meant to involve a technical deep-dive but rather focus on the implications of using the different extrapolation approaches and the adoption of new approaches. The issue panel will therefore be relevant for data analysts, industry representatives, and decision makers alike.