生物相似性基於區間估計的設計與分析

Abstract

近年來，生物仿製藥的市場正蓬勃發展，因此，如何評估生物仿製藥和原廠生物藥的相似性成為一個重要議題。評估生物相似性的試驗通常包含生物仿製藥與原廠生物藥，如果兩組藥物反應平均數差異之100(1-2α)%雙邊信賴區間，落在一個預先決定的範圍內，我們便可得到生物仿製藥與原廠生物藥為生物相似的結論。此外，我們假設兩組生物藥反應為常態分佈且變異數不同，眾所周知，這便是所謂貝倫斯-費雪問題。在文獻中，多種方法像是薩特斯韋特的近似、科克倫-考克斯的近似、豪的近似均可用來解決此問題，這些近似可以被應用於建構相對應之信賴區間，進以利用評估生物相似性，我們並使用這些信賴區間計算生物仿製藥試驗所需要之樣本數。模擬研究顯示我們提供的樣本數計算法則，得以提供一個充分且一致的評估力。於本篇論文中，我們亦提供一個例子來闡釋如何使用信賴區間設計並分析生物相似性。

In recent years, the biosimilar market is flourishing. To assess the biosimilarity between biosimilar product and the innovative product is an important issue. In this thesis, a biosimilar trial is conducted for assessing biosimilarity between a biosimilar product and the innovative product based on a continuous efficacy endpoint. We will develop an approach based on confidence interval to assess biosimilarity. More specifically, biosimilarity is concluded if a 100(1-2α)% two-sided confidence interval for mean difference between a biosimilar product and the innovative product falls within a predetermined criterion. In addition, normal distribution and unequal variability of the primary endpoint for both groups are assumed. This is known as the Behrens-Fisher problem. In the literature, several approaches such as Satterthwaite’s approximation, Cochran-Cox’s approximation, and Howe’s approximation have been used to solve the problem. These approximations can be applied to construct the confidence interval for assessing biosimilarity as well as sample size determination for the biosimilar trial. Simulation studies show that our proposed method is able to provide a sufficient and consistent biosimilar probability. An example was represented for the illustration of our proposed approach.