Title: Disease progression-based feature screening for ultrahigh- dimensional survival-associated biomarkers
Authors: Liming Xiang - Nanyang Technological University (Singapore) [presenting]
Mengjiao Peng - East China Normal University, China (China)
Abstract: The increased availability of ultrahigh-dimensional biomarker data and the high demand for identifying biomarkers importantly related to survival outcomes made feature screening methods commonplace in the analysis of cancer genome data. In the presence of progression-free survival (PFS), a surrogate endpoint for overall survival (OS), the correlation between OS and PFS has suggested a high concordance in both survival endpoints; namely, patients with higher PFS would most likely have longer OS. We propose a novel feature screening method by incorporating surrogate information of PFS into the selection of important biomarker predictors for more accurate inference of OS after disease progression. The proposal is based on the rank of correlation between individual features and the conditional distribution of OS given observations of PFS. It is advantageous for its flexible model nature, which requires no marginal model assumption for OS or PFS, and the minimal computational cost for implementation. Theoretical results show its ranking consistency, sure screening, and false rate control properties. Simulation results demonstrate that the proposed screener leads to a more accurate feature selection than the method without considering the prior information about PFS. An application to breast cancer genome data illustrates its practical utility and facilitates disease classification using selected biomarker predictors.