Reply to S. Barni et Al and M. Sun et Al.

Abstract

The comments of Barni and Petrelli and Sun et al raise additional considerations and speculation that some yet unintended consequences could result from the study design that served as the basis for the study by Escudier et al, with its purpose of helping clinicians to understand patient and physician preferences for pazopanib versus sunitinib. I expressed my comments relating to the robustness of the overall interpretation of the study results in an accompanying editorial. We agree that the assessment of response was not made available to patients in the study by Escudier et al at the time of their preference choice. Given that the overall response rate in this study favored sunitinib, lack of this information could have certainly confounded the results. Whether the dropout of nearly a third of the patients in an otherwise small study constitutes a significant bias can and will continue to be debated. The groups commenting on this study now raise another spectrum of real and thorny issues that clinical investigations into patient preferences and outcomes must face. First, there is no precedent for switching a systemic therapy from one agent to another in still responding patients. Second, both letters emphasize that the subsequent response and the potential development of resistance mechanisms to an agent that was transiently used (as part of the study) is unknown, yet this unknown component of the study was subordinated to the determination of a patient preference outcome. Third, as Sun et al point out, the tolerability of tyrosine kinase inhibitors may improve with time, making the assessment of patient preference at arbitrarily selected earlier time points problematic. These are all real concerns and must be further investigated before a so-called maintenance-switch study design is used in future patient preference studies. Perhaps a more optimal future design could sequentially assess patient preferences (with appropriately designed instruments) up to the time of disease progression, and then perform similar assessments of the second-line therapy. We would have felt more comfortable had a study been performed in which patients were initially randomly assigned to either pazopanib or sunitinib until disease progression, with accompanying study of patient preferences. At progression, cross-over to the alternate agent would occur and similar assessments could be made. Had this been done, important information about comparable tolerability would be known, and subsequent responsivity and resistance characteristics would be better understood. These unresolved issues should not deter future research into patient and physician preferences of one pharmaceutical agent over another, but must be studied in more rigorous and better designed investigations that provide answers to the question at hand, with no concerns that either patient safety or clinical efficacy has been compromised.