Statistical Approaches to Outliers Test in Bioassay Analytical Testing Data

• Activity Number: 549
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Lingmin Zeng* and Binbing Yu and Harry Yang
• Companies: MedImmune and MedImmune and MedImmune
• Keywords: Outlier ; 4PL ; ROC ; Bioassay

Abstract

Outlier detection in Bioassay analytical testing method is recommended by USP < 1034> before assessing system suitability and sample suitability. The four parameter logistic model is commonly fitted for Bioassay samples under multiple series dilutions across multiple plates. In practice, some data are known with technical problems (pipetting error, dilution error or contaminated wells, et.al. ). And some data with random outliers are without good explanation. Those outliers can cause failure of system suitability or sample suitability, or even torture the estimated relative potency results. Different outlier test methods are compared in appearance of severity of outliers using simulations. ROC curves are used to evaluate the sensitivity and specificity of those methods. Several Bioassay data are screened by various outlier test methods in fitting the four parameter logistic curve.