Comparison of duration of response endpoint in oncology

Details

Supervisors

Legrand, Catherine ; Lambert, Alexandre

Faculty

Faculté des sciences

Degree label

Master [120] en statistique, orientation biostatistiques, à finalité spécialisée

Abstract

In clinical trials, the efficacy of a new treatment can be measured using a wide range of endpoints. The proportion of patients responding to a treatment and the duration of their response are generally assessed, as they represent a direct measure of the anti-tumor activity of new agents. Comparing two treatment groups (e.g. experimental versus control) for proportion of responders is straightforward in a randomized trial, but comparing duration of response presents more difficulties. This master thesis aims to examine, compare and apply three methods of response duration analysis: Duration Of Response, Time In Response and Probability of Being in Response. A simulation study is performed to explore the influence of response rate and duration of response on the three methods and an illustration with real data aims to provide a better practical understanding of the interpretation of the conclusions drawn according to each method.