Background: Assuring that cell therapy products are safe before releasing them for use in patients is critical. Currently, rapid sterility tests for bacteria and fungi take weeks. The goal of this work was to develop an approach for the rapid and sensitive detection of microbial contaminants at low abundance from low volume samples during the manufacturing process of T-cell therapies. Methods: We developed a Nanopore-based MinION long read sequencing methodology using 16S and 18S amplicon sequencing on USP<71> organisms and other microbial species. Reads are classified metagenomically to predict the microbial species. We used an extreme gradient boosting machine learning algorithm (XGBoost) to first assess if a sample is contaminated and second, determine whether the predicted contaminant is genuine. The model was used to make a final decision on the sterility status of the input sample. Results: An optimised experimental and bioinformatics pipeline starting from engineered spiked samples through to sequenced reads allowed for the detection of microbial samples at 10 CFU using metagenomic classification. Conclusion: Machine learning can be coupled with long read sequencing to detect and identify sample sterility status and microbial species present in T-cell cultures, including the USP<71> organisms to 10 CFU.

While adoptive cell therapies have revolutionized cancer immunotherapy, current autologous chimeric antigen receptor (CAR) T cell manufacturing face challenges in scaling to meet patient demands. CAR T cell production still largely rely on fed-batch, manual, open processes that lack environmental monitoring and control, whereas most perfusion-based, automated, closed-system bioreactors currently suffer from large footprints and working volumes, thus hindering process development and scaling-out. Here, we present a means of conducting anti-CD19 CAR T cell culture-on-a-chip. We show that T cells can be activated, transduced, and expanded to densities exceeding 150 million cells/mL in a two-milliliter perfusion-capable microfluidic bioreactor, thus enabling the production of CAR T cells at clinical dose levels in a small footprint. Key functional attributes such as exhaustion phenotype and cytolytic function were comparable to T cells generated in a gas-permeable well. The process intensification and online analytics offered by the microbioreactor could facilitate high-throughput process optimization studies, as well as enable efficient scale-out of cell therapy manufacturing, while providing insights into the growth and metabolic state of the CAR T cells during ex vivo culture.