The high failure rate of Phase III trials in oncology presents a major obstacle to the drug discovery process. One likely reason for the high failure rate in Phase III trials is the inaccuracy of predictions of efficacy from the preceding hypothesis generating Phase II trials. While historically Phase II trials have used tumor response rate as the primary endpoint, where response is assessed via the Response Evaluation Criteria in Solid Tumors (RECIST), the concerns over RECIST-based response metrics (i.e. best response and confirmed response) as endpoints to assess trial success are multifold. The overall goal of this project is to identify and validate tumor-measurement based Phase II trial endpoints that are distinct from RECIST-based response metrics for predicting survival outcomes. Specifically, we propose to use the unique resource of the RECIST 1.1 data warehouse, which includes cycle-by-cycle and lesion-by-lesion tumor measurements for 7,976 patients in breast, non-small cell lung, and colon cancers to achieve the following three specific aims:
Specific Aim 1 : To evaluate metrics for predicting time-to-event endpoints based on RECIST-based tumor metrics and other categorical classifications of response. The purpose of this aim is to evaluate the cut points used in RECIST and categorical definitions of response, specifically Response vs. No Response;Response vs. Stable vs. Progression;and Progression vs. No Progression.
Specific Aim 2 : To develop and validate metrics for predicting time-to-event endpoints based on continuous summaries of longitudinal tumor measurements.
This aim will address the concern over RECIST as a categorization of inherently continuous measurements by developing metrics that capture key features of the tumor trajectory.
Specific Aim 3 : To conduct re-sampling based simulation studies to understand and assess the clinical utility of the metrics developed in Aims 1 and 2 as Phase II endpoints for predicting Phase III results. The purpose of this aim is to study the clinical utility of the validated metrics in predicting Phase III results and in particular, in addressing the high failure rate of Phase III trials. This research will provide a better understanding of the current RECIST-based response metrics and a potentially improved metric based on longitudinal tumor measurements for the prediction of time to event endpoints. Our proposed metrics are clinically relevant and can be readily adopted into clinical trials for prospective validation.
The purpose of this project is to identify alternative tumor measurements based Phase II trial endpoints that predict survival outcomes better than the current RECIST-based response metrics, using the unique resource of the RECIST 1.1 data warehouse, which has individual patient data on 7,976 patients with three tumor types.