Considerations on Bayesian Dose Window Escalation: Part II

• Activity Number: 127
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Xue Lin*+
• Companies: FDA
• Keywords: Bayesian ; dose window escalation

Abstract

Different from traditional dose escalation plan, some cell therapy treatments do not have fixed doses but rather adjacent dose windows. This special feature constitutes a challenge for identifying optimal dose, since there is no statistical method that is readily available to be applied directly to this situation. Naïve extension of fixed dose escalation method, such as CRM (continuous reassessment method), MCRM (modified CRM), and BLRM (Bayesian logistic regression model), to dose window can under-estimates the toxicity which may result in an aggressive escalation plan that could risk patients' safety. In our previous talk, we have demonstrated using simulation studies that actual cell dose should be used in the model instead of dose cap to update the dose toxicity curve to minimize the risk of overdose. And interestingly, when the assumed dose toxicity curve under-estimates the toxicity, there is good chance of under-dose when actual cell dose is used in the model. In this talk, we further explore the various aspects of dose window escalation, including the choice of dose window size, the optimal number of subjects in each dose window and the incorporation of efficacy information