Alternative Approach for Estimating Tumor Response Rate

• Activity Number: 112
• Type: Contributed
• Date/Time: Monday, August 9, 2004 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Xiaolong Luo*+
• Companies: Johnson & Johnson Pharmaceutical R&D, LLC
• Keywords: clinical trial ; cancer ; binomial ; Markov ; interim analysis ; consistency

Abstract

In clinical oncology trials, patients receive treatments of various modalities. During a study, a patient's tumor volume is periodically measured and a confirmed shrinkage, i.e., tumor response, is often used as a study endpoint. The assessment of such tumor response rate is typically based on binomial distribution modeling. This approach is acceptable when the trial is finished and all clinical assessments have been completed. However, it can lead to many practical issues for a trial interim analysis. One dilemma is whether to include ongoing patients. We cannot include them since their endpoints have not been reached and we do not want to exclude them either since they would represent a significant portion of the data. In this paper, we propose a Markov process model for the longitudinal tumor volume measurements and develop an algorithm to estimate tumor response rate. The method would allow us to utilize all ongoing measurements independent of whether final endpoint has been reached. We will use simulation analyses to compare both approaches. In addition, we will derive large sample theory for the new estimator's asymptotic consistency and limiting distribution.