Team: Dr Andrew Titman (Lancaster University), Dr Tom Jacobs (Janssen)
Clinical Advisors: Prof Munir Pirmohamed, a chair in clinical pharmacology at the University Liverpool. He is NHS Chair of Pharmacogenetics and co-director of the MRC centre for drug safety.
Location: Lancaster University (Lancaster, UK)
Progression-free survival is often used as a surrogate endpoint for overall survival in early phase trials of cancer treatments. The extent to which this is effective depends upon the true relationship between progression-free and overall survival. One way of modelling the dependence between the endpoints is to assume a multi-state survival model where patients have some hazard of progression, some hazard of death before progression and a subsequent hazard of death given progression. The effect of a given treatment on overall survival may arise through differing effects on each of these hazards.
This project will consider the design of multi-arm cancer trials with treatment selection. The aim is to utilize the totality of progression-free survival and overall survival information for treatment selection and utilize different assumptions about the association between events and the correspondence in treatment effects on inter-event hazards. The sole use of disease free progression should be optimal in the case where the multi-state model can be assumed to be Markov, where the hazard ratio between death from progressed cancer compared to death before progression of cancer is substantially greater than 1 and where treatments only affect the rate of progression and not any of the other process hazards. If the treatment also affects the hazard of death after progression then there may be benefit in considering methods that also use information on overall survival when it is available. A range of scenarios will be considered in order to determine in what circumstances sole use of data regarding progression-free survival is satisfactory and when incorporating additional information on overall survival may be useful.
Possible approaches to incorporating additional information could include fitting Cox models to each transition hazard with shared or correlated regression coefficients, or by using a pseudo-observations approach2 to consider the effect of treatment on overall survival without making specific assumptions about the effect of treatment on any particular transition.
Meet our Early Stage Researcher: Enya Weber, Lancaster University
I was born on 19.09.1989 near Stuttgart, Germany. In 2009 I started my studies in mathematics at the university in Freiburg, Germany. I achieved her Bachelor’s degree in science of mathematics in 2012 and continued with Master of Science in mathematics. During the studies of master I developed interests for both medical research as well as statistics because of my student assistant job at the “Institute of Biometrical and Medical Statistics Medical Center – Universität Freiburg“. My masters thesis is about survival analysis. I like reading, travelling, watching movies and adventure sports like bungee jumping.
The IDEAS network brings together leading institutions in the training of medical statisticians, large and small sized pharmaceutical companies and is supplemented by clinical experts.