Model-Based Tolerance Intervals Derived from Cumulative Historical Composition Data: Application for Substantial Equivalence Assessment of a Genetically Modified Crop

Abstract

Compositional analysis is a requisite component of the substantial equivalence framework utilized to assess genetically modified (GM) crop safety. Statistical differences in composition data between GM and non-GM crops require a context in which to determine biological relevance. This context is provided by surveying the natural variation of key nutrient and antinutrient levels within the crop population with a history of safe use. Data accumulated from various genotypes with a history of safe use cultivated in relevant commercial crop-growing environments over multiple seasons are discussed as the appropriate data representative of this natural variation. A model-based parametric tolerance interval approach, which accounts for the correlated and unbalanced data structure of cumulative historical data collected from multisite field studies conducted over multiple seasons, is presented. This paper promotes the application of this tolerance interval approach to generate reference ranges for evaluation of the biological relevance of statistical differences identified during substantial equivalence assessment of a GM crop.