Hierarchical Bayesian Models for Understanding the Pharmacokinetics and Pharmacodynamics of Lorenzo's Oil

• Activity Number: 372
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Cynthia Basu* and Mariam Ahmed and James C. Cloyd and Richard C. Brundage and Bradley P. Carlin
• Companies: University of Minnesota and University of Minnesota and University of Minnesota and University of Minnesota and University of Minnesota
• Keywords: PK-PD ; Hierarchical Bayesian Models ; X-ALD Lorenzo's Oil ; Phase IIa Clinical Trial Design ; Commensurate Prior ; Rare Disease

Abstract

Lorenzo's Oil (LO) is a treatment available for X-linked adrenoleukodystrophy (X-ALD), a fatal neurodegenerative disease. Little has been done to establish its clinical efficacy. We analyze data on 116 male asymptomatic patients who were administered LO, including their fasting fatty acid profile (FAP). We adopt hierarchical Bayesian statistical approaches to understanding pharmacokinetics (PK) and pharmacodynamics (PD) of LO with respect to the FAP. We deal with the limitation of having one observation per cycle of the drug instead of multiple observation per cycle, as is usually used in PK-PD studies. Our action plan is to first link LO dose to the erucic acid concentration in the blood by PK modeling, and then link this concentration to a biomarker (C26, a very long chain fatty acid) by PD modeling. Next we design an adaptive Bayesian Phase IIa to estimate improvements in the biomarker from various LO doses accounting for possible toxicity and a Phase III study linking LO dose to actual improvements in health status. Finally, we look ahead toward using hierarchical commensurate prior models to incorporate relevant historical information in this rare pediatric disease setting.