473 – Innovations in Survival Analysis: PFS and Other Issues
Competing Risk Analysis of PFS Considering Patients Who Switch Therapy Prior to Progression
Lixia Jiao
Sanofi Aventis
Standard survival analysis methods, such as Kaplan Meier curves, log-rank test and Cox proportional hazard model, are widely accepted tools to compare the cause-speci�c hazards when there is only one event of interest and the time to event and time to censoring are independent. However, competing risks are often encountered in clinical research, where multiple failure types exist and one type of event either precludes the occurrence of another event or fundamentally alters the probability of occurrence of the other event. In the analysis of competing risks data, the standard analysis methods may lead to biased results by treating the competing event as censored at the time this event occurs. This way, it is assumed that the patients failing from a competing risk are no more or less likely to fail from the cause of interest than the patients still at risk beyond this time. The newly developed methods, such as Gray’s test and Pepe and Mori’s method, take into account of the competing risks and provide different clues regarding the effect of a co-variate. Gray proposed a class of generalized linear rank statistics for testing equality of cumulative incidence functions. Pepe and Mori proposed a different class of test statistics, not based on ranks, for comparing cumulative incidence functions and conditional probability functions. In standard progression free survival (PFS) analysis, patients who change their cancer therapy prior to progression will be labeled as censored at the time of stopping randomized treatment. As changing cancer therapy alters the probability of progression, it should be considered as a competing risk event and the newly developed methods apply.