|
Abstract:
|
With the success and rapid development of Immuno-Oncology (IO) therapies in cancer treatment and the enhanced understanding of the role of the immune system and immune tumor microenvironment, the ability to identify immune-related biomarkers to predict the efficacy of therapy is becoming a priority. In addition, pre-existing immunological features of both the host and the tumor such as TMB, immune TME, and TILs, among others, are known to be prognostically associated with patient outcomes. In this setting of multiple inter-related continuous biomarkers, it is often necessary to determine appropriate cut-off values based on their relationship to clinical efficacy if they are to have practical applications in clinical decision making and trial design. In this work, we present the results of our simulation studies that compared the ability of several different approaches used to find simultaneous optimal cut-off values for multiple continuous biomarkers viz; ROC based, regression-based, and machine-learning based. We also provide the pros and cons of each method under various scenarios that commonly arise when more than one continuous biomarker is being investigated simultaneously.
|
