Marker Methodology With Focus On Hierarchical Outcomes

Abstract

Statistical methods that can be used for the selection and evaluation of biomarkers which later will be validated to graduate into potent surrogate markers are in dire demand. The selected biomarkers need to go through rigorous testing procedures both statistically as well as biologically to finally end up being reliable replacements for the clinically relevant endpoint. The competitive nature of the pharmaceutical industry and the impending societal demand for urgent drugs for some of the chronic diseases threatening human life add to the need for these methods. In this thesis we have revised some existing methods, introduced some new once and assessed their merit through simulation studies. The thesis is organized in to two parts. The first part dealt with surrogate marker validation and the second one is devoted to the selection and evaluation of biomarkers mainly genomic biomarkers from microarray experiments. This however is not the natural ordering of the events. In practice, we first identify a biomarker which later will be promoted to a surrogate endpoint. However, the aim here is to introduce a set of methods that have been designed to validate a surrogate endpoint for the selection and evaluation of biomarkers.