A Joint Model for Tumor Burden and Progression-Free Survival

• Activity Number: 423
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Aparna B. Anderson*+ and Ye Shen and Ritwik Sinha
• Companies: Bristol-Myers Squibb and Yale University School of Public Health and Hewlett Packard Laboratories India
• Keywords: longitudinal data ; progression-free survival ; mixed effects ; oncology

Abstract

In oncology, overall survival is the ideal measure of treatment benefit. However, the mechanistic and biologic effects of a therapeutic agent are generally characterized as changes in tumor burden (TB) measured repeatedly at specified time points. In clinical trials that support marketing approval, TB is usually analyzed as a categorical variable (e.g., applying RECIST criteria) to summarize objective response rate or progression-free survival (PFS). To address the loss of information due to categorization as well as potentially informative missing TB, a joint model (Henderson, et al., 2000) for the longitudinal tumor burden process and PFS is considered. The operating characteristics of this model are evaluated under different tumor burden distributions and missing data mechanisms. Simulations indicate that joint modeling reduces estimation bias when data are missing not at random and leads to minimal loss of efficiency when conditions are consistent with linear mixed effects modeling assumptions.