以動差基礎標準統計評估基改作物之研究

Abstract

根據歐盟委員會 (European Commission)的定義，生物體經由人為的基因轉植之後，使其擁有新的特性或性質者，稱為基因改造生物 (genetically modified organisms, GMOs)。現今，隨著基改作物的發展，基改作物的安全性評估成為一大重要議題。評估的內容包含基改作物的成分、基因型以及外表型。 歐盟食品安全管理局 (European Food Safety Authority, EFSA) 的基改作物委員會 (GMO Panel) 提出了針對基因改造作物安全性的統計考量評估方法。建議同時利用差異性檢定 (proof of difference) 以及對等性檢定 (proof of equivalence) 來進行評估，並建立其信賴區間 (confidence interval) 與對等性限度 (equivalence limit) 。 歐盟食品安全管理局 (European Food Safety Authority, EFSA) 對於基改作物的安全性評估之科學統計考量建議，基改作物與其傳統作物之平均差異值應落於商業品種之自然變異中以證明兩者間的生物相等性 (bioequivalence)。依據此準則，我們提出一項新的標準，亦即利用基改作物與其相對應之傳統作物的平均平方差異，與商業品種變異的比例，來進行基改作物安全性之對等性評估。此外，我們發現在真實資料中，商業品種變異之值時常很小，容易導致檢定結果過於保守，因此，我們給定商業品種變異之下限，取兩者之最大值作為檢定之尺度。 我們利用大樣本修正法 (Modified Large-Sample Method, MLS)進行變異數的估計，在混和效應的模式 (mixed-effects model) 之下建立線性化的95% 信賴上限，以此作為基改作物與其相對應傳統作物的安全性對等性評估之檢定方法。最後以一個例子介紹本方法之應用，並利用SAS進行模擬，以經驗型Ⅰ錯誤率和經驗檢定力來評估此一方法的效率。

According to the European Commission, genetically modified organisms (GMOs) are organisms whose genetic characteristics are being modified artificially in order to give them a new desirable property. However, with the development of genetically modified organisms, the assessment of GMOs for safety is an important issue. The components, genotypes and phenotypes of GM plants should be identified and evaluated. The Panel on GMOs of the European Food Safety Authority (EFSA) issued a guideline of Scientific Opinion on the Statistical Considerations for Safety Evaluation of GMOs in 2010. The GMO Panel of the EFSA indicated that both the proof of difference and the proof of equivalence are required for evaluation of the safety of GMOs. The EFSA’s scientific opinion on statistical considerations for the safety evaluation of GMOs suggests that the difference in average between the GM crop and its conventional counterpart should lie within the natural variability which can be estimated from the commercial varieties, in order to prove that the GMO crop is equivalent to its conventional counterpart. Therefore, we propose a new criterion for assessment of equivalence of the safety profile between the GM crop and conventional crop which is the scaled square mean difference between the GM crop and its conventional crop with the variance of commercial crops as the scaled factor. However, from the real data, we found that the variance of some commercial crops can be very small which would make the testing results too conservative. Thus, we propose to set a lower limit for the variance of commercial crops, and use the maximum value of two variances as the scaled factor. We applied the modified large-sample (MLS) method to estimate variance, and derive the 95% upper confidence limit of the linearized mixed-scaled criterion under the mixed-effects model. Then use this limit as the testing procedure for evaluation of equivalence in the safety profile between the GMO and its conventional crop. A numerical example illustrates the applications of the proposed method. A simulation was conducted with SAS to investigate the performance, in terms of size and power, for the proposed procedure.