Translation of small-scale CAR-T cell manufacturing methods to a clinical-scale production platform

Abstract

Background & Aim CAR-T cells have displayed potent antitumor effects in liquid but not in solid tumors. Variables like technical and immuno-oncological factors contribute to the modest clinical effect of CAR-Ts in solid tumors. To improve their function, manufacturing conditions have been optimized to yield a favorable CAR-T cell phenotype. A product enriched in stem cell- and central memory T cells (TCM) is believed to improve durability of antitumor responses and engraftment of the adoptive cells, where highly differentiated T cell products may be prone to exhaustion and senescence. Finding optimal manufacturing conditions requires testing of media components and selection of appropriate bioreactor system for scale-up. To avoid the substantial costs of multiple bioreactor runs, screening typically relies on flask scale. We set out to determine the predictive ability of the flask-scale screening on the process performance in clinical scale. Glioblastoma-specific CAR-Ts were produced from six healthy donors in various medium compositions and compared to those produced in closed-system bioreactors. Methods, Results & Conclusion Lentivirally-transduced CAR-T cells were cultured in flasks in Miltenyi TexMACS medium with or without human AB serum (HABS) and either IL2 or IL7 and IL15. In the bioreactor process, T cells were purified and transduced in the Miltenyi CliniMACS Prodigy system, then further expanded in IL7 and IL15 containing HABS-free media in a suspension bioreactor. We characterized the cells by CAR expression level, viability, expansion rate, memory phenotype and in vitro functionality. All CAR-T products were highly viable and specifically cytotoxic, though biological differences between donors can cause variation. The main difference between culture conditions was in expansion rates. The phenotype was substantially effector memory T cell-biased in all flask cultures. The presence or absence of HABS rather than cytokine selection appeared to be the main determinant of the product characteristics. In our bioreactor scale pilot the lead media composition (IL7, IL15, no HABS) yielded a markedly different product, seen as superior expansion and significant enrichment of the TCM fraction. Our results suggest that small-scale screening may translate poorly to bioreactor systems. Such factors as medium replenishment, oxygenation, cell density and culture vessel rocking may have an impact on the product. Early-stage scale-up and testing using the final platform is essential for product success.