Request Copyright Permission

Pharmacokinetic Behavior of Fluticasone Propionate and Salmeterol from Advair Diskus: The Consequences of Batch Variability

Burmeister Getz E, Carroll KJ, Mielke J, Jones B, Benet LZ.

RDD Europe 2017. Volume 1, 2017: 25-34.

Abstract:

Two independent clinical studies have demonstrated substantial pharmacokinetic (PK) variability between manufacturing batches of Advair Diskus® 100/50. Batch-to-batch fluticasone propionate (FP) and salmeterol PK ratios of approximately 150% from crossover designs in healthy subjects were reproducibly observed, with 90% confidence intervals outside the 80-125% bioequivalence (BE) region. Within-batch replicate PK ratios were near 100% and confirmed batch as the source of the added variability.

The ubiquitous two-way crossover design for PK BE assessment neglects between-batch variability by allowing each product to be represented by a single batch. Inflation of the Type I error rate (consumer’s risk) results; when batches are different, a single batch provides an uncertain estimate of the product mean. Type I error rate can exceed 25%, and is inflated over a wide range of between-batch variability values. Consequently, the estimate of the test/reference (T/R) product ratio from a two-way crossover is associated with low confidence, and the probability of an incorrect BE conclusion is untenably high. The Type I error rate of a two-way crossover study rises sharply as between-batch variance increases from zero; a between-batch variance of only 5% of residual error variance inflates the Type I error rate to 14% for a 26-subject trial, with more extreme inflation as study size increases. Recommended adaptations to BE methodology for products with substantial batch-to-batch variability include use of multiple batches and extension of the reference-scaling methodology.