變方成份分析於體外生物相等性統計評估之研究

Abstract

由於鼻藥噴霧劑(Nasal aerosols)與鼻腔噴霧器(Nasal sprays)之噴霧藥效皆為局部性作用藥品，不易藉由血液的流動進入人體器官中，因此我們利用藥劑生體可用率(Bioequivalence)與生體相等性(Bioequivalence)的機器實驗來評估兩種藥物的藥效性是否相等。美國食品與藥物管理局(簡稱FDA)(2003b)，提出以95%信賴上界來作為比較兩種藥物的生體相等性(Bioequivalence)之標準。Hyslop, Hsuan和Holder首先在2000年將此標準線性化並提出相對應之統計方法。因在評估噴霧器實驗有批次、生命階段與檢體等不同變因，我們根據二階段巢式隨機模式的變方分析之大樣本修正法(MLS)來尋求線性化準則95%信賴上限。最後利用各種樣本數及不同變異數組合下之模擬資料，以經驗型I誤差和經驗檢定力來評估由延伸之變方分析大樣本修正法(MLS)是否能求得一個較佳的95%信賴上界。最後我們以一個實例介紹本方法在實務上之應用。

For some locally acting products such as nasal aerosols and nasal sprays that are not intended to be absorbed into the bloodstream, the in vitro bioavailability / bioequivalence studies based on machinery experiments are used to measure that whether the two drug products provide the same therapeutic effect. According to the U.S. Food and Drug Administration (FDA) draft guidance (2003b), the evaluation of the in vitro bioequivalence (BE) between two drug products would be determined by comparing the upper 95% confidence limit of with zero. This method was first proposed by Hyslop, Hsuan and Holder (2000) based on a linearized bioequivalence criterion. However, their method does not consider the variations due to different sources. Hence, we apply the method of modified large-sample (MLS) for the variance components under the 2-stage nested random-effects model for construction of the upper 95% confidence limit. The total variance of the 95% upper limit considers the variance components due to lot, life stage and sample which are obtained through the MLS method. The results of empirical size and empirical power from simulation under different sample size and variance combinations are provided to investigate the performance of the extended MLS method. A numerical example illustrates the proposed method.