高个体差异药物的生物等效性和生物类似性评价(英文)

Bioequivalence and biosimilarity of highly variable drugs

对于具有较高受试者个体间差异的药物进行生物等效性(BE)和生物类似性评价时,使用通常的双单侧检验(TOST)法既十分困难,甚至也不可能;除非不考虑伦理规范,而进行大样本的人体试验。因此,美国食品药品管理局(FDA)和欧盟药品管理局(EMA)等监管机构对高个体差异药品的生物等效性评价分别颁布了替代方法。这2个监管机构的替代方法依据相同的原理,但是关键细节略有不同。FDA建议当受试者个体间差异超过30%时,使用"标化均值生物等效性"(scaled-average BE,SABE)评价;该方式使用已有计算机软件通过线性转换来计算等效性的95%上限;并要求采用第2级标准评价:两产品间相关参数的几何均值之比的点评估(point estimate)在0.80到1.25之间。而EMA则建议采用"带扩展限度的平均生物等效性"(average BE with expanding limits,ABEL)进行高个体差异药品的评价;该方式与SABE方式相关,但可以使用简单的双单侧检验方法进行评估;EMA也要求采用相同的第,2级标准评价,且要求这2个标准仅适用于受试者个体间差异不超过50%的情形。这2个机构采用了不同的监管常数(等效性评价指标)。FDA建议的计算指标会使生物等效限不连续,且所需样本量大,并且在CV=30%附近有很高的I型误差。而EMA的评价指标不会产生这种不连续性,I型误差也很低。总之,EMA的评价方法更好。

The determination of bioequivalence （BE）and biosimilarity by the usual two one-sided tests（TOST）pro- cedure is difficult, and can even be impossible,when the within-subject variation is high unless an unethically large number of subjects is used. Therefore, regulatory authorities such as the US FDA and the EU EMA,introduced alter- native approaches for the evaluation of BE of highly variable （HV） drug products. The two approaches rely on the same principle but differ in important details. The US FDA recommends that scaled average BE （SABE）be used when the within-subject variation exceeds 30%. It can be evaluated by a linearizing transformation for which the upper 95% bound can be calculated by a published computer program. A secondary criterion is also required：it ex- pects that the point estimate should be between O. 80 and 1.25 for the ratio of geometric means, for relevant parame- ters, between the two drug products. On the other hand, the EU EMA suggests that average BE with expanding limits （ABEL） be applied. It is related to SABE but can be evaluated by the simple TOST procedure. The European agen- cy also requires the secondary criterion and permits both criteria only up to a variation of 50%. The two agencies ex- pect differing regulatory constants. The value suggested by US FDA gives rise to discontinuities of the BE limits and the required sample size as well as to very high Type I error around CV = 30%. The EU EMA condition results in no discontinuities and much lower Type I error. Altogether, the approach of EU EMA should be favored.