Development of instability analysis for the filling process of human-induced pluripotent stem cell products

Abstract

The filling process is a part of downstream processing in cell manufacturing. In this study, volume (V/VT), particle density (D/D0), and cell potential (P/P0) quality indices were proposed for the outputs of the filling process designed for human-induced pluripotent stem cell products. Instability in the filling process was evaluated by using an algorithm with Isolation Forest outlier detection and Spearman’s rank correlation to categorize fluctuations in the quality indices within a batch as order-dependent, process-time (tP)-dependent, and random fluctuations. The variations between batches were evaluated after categorizing the batches as normal or outliers and calculating their coefficients of variation. Nonparametric statistical techniques were used in the developed algorithm to avoid the influence of outlier measurements on the analysis. V/VT has order-dependent fluctuations within a batch during the filling process with and without a mixing operation. D/D0 has order-dependent and tP-dependent fluctuations within a batch during the filling process without a mixing operation. P/P0 was determined to have a tP-dependent fluctuation within a batch during the filling process with a mixing operation. The variations between batches were the highest for P/P0 and were stabilized by implementing a low-temperature filling process. The developed algorithm can also be considered for the evaluation of instability in other cell manufacturing processes.