Multivariate equivalence testing for food safety assessment

Area of Science:

• Agricultural Science
• Biotechnology
• Food Safety

Background:

• Genetically modified (GM) crops require rigorous safety assessments for market approval, focusing on human health and environmental impact.
• Current safety evaluations by the European Food Safety Authority (EFSA) use univariate statistical equivalence tests, comparing GM crops to reference varieties on a single characteristic at a time.
• The increasing complexity of phenotypic data from molecular phenotyping platforms necessitates methods that can integrate multiple variables for a holistic safety assessment.

Purpose of the Study:

• To introduce a novel multivariate equivalence test designed to assess safety by simultaneously evaluating multiple compositional characteristics of GM crops.
• To extend a recently developed univariate equivalence test into a multivariate framework capable of handling complex, multi-variable datasets.
• To provide a statistically robust method for regulatory bodies to make integrated decisions on the acceptance of GM products.

Main Methods:

• Development of a new multivariate statistical equivalence test.
• Application of the test to analyze compositional data from field studies of GM maize grain.
• Illustration of the test using untargeted metabolomic data from GM potato tubers.
• Performance evaluation of the proposed multivariate test using simulated datasets.

Main Results:

• The proposed multivariate equivalence test effectively integrates multiple phenotypic variables for a simultaneous assessment of equivalence.
• The method was successfully applied to real-world datasets from maize and potato, demonstrating its practical utility.
• Simulations confirmed the test's performance in evaluating equivalence across numerous characteristics.

Conclusions:

• The new multivariate equivalence test offers a more comprehensive approach to GM crop safety assessment compared to traditional univariate methods.
• This method addresses the challenges posed by high-throughput phenotyping data, enabling more integrated regulatory decisions.
• The developed statistical framework supports robust safety evaluations for GM food and feed products.