From extractables to exposure data: Sensitivity analysis of extrapolation algorithms with focus on USP 〈665〉

Abstract

We evaluated algorithms designed to extrapolate extractables data for predicting process equipment-related leachables (PERLs) and assessing PERL exposure in single-use systems (SUSs) and assemblies. The robustness and sensitivity of these algorithms were tested against variations in input data, including extrapolation algorithms for both short and long contact time extractables data obtained from the standardized extractables protocol provided in USP 〈665〉. Our findings demonstrate that extrapolated data for SUS and assemblies are suitable for safety assessments. Extrapolated and aggregated data do not systematically underestimate potential PERL exposure values, provided that the extractables data originate from experiments with a higher surface area to contact liquid volume ratio and/or a low liquid to material volume ratio compared to the use scenario. The algorithms are non-sensitive to deviations in input data, as these deviations are propagated decreasingly into extrapolated data and parameters. The quality and significance of PERL exposure calculations can be enhanced by incorporating extractables study data from experiments using a semipolar organic solution, such as ethanol.