Strategies for improving T-cell manufacturing for adoptive immunotherapies

Abstract

Immunotherapy is a novel approach to disease treatment and has shown numerous clinicalsuccesses. In brief, immunotherapy exploits the principles of immunology to generate ananti-viral or anti-cancer response. One of the myriad methods that exist to stimulate animmunotherapeutic response is adoptive T-cell transfer (ACT). ACT is an immunotherapyapproach whereby patient-derived antigen-specific T-cells are expanded in vitro and infusedinto the patient. Thus, a central factor of ACT immunotherapy is the in vitro T-cell stimulationand expansion step. In recent years, T-cell stimulation and expansion methods havefocused on maximising cell yield. These expansion methods rely on the use of high-doseand often repeated cycles of stimulation. While effective in deriving large cell yields, this T-cell expansion strategy appears to compromise T-cell quality. Furthermore, various antigensources are used to stimulate and expand antigen-specific T-cells. The impact thesediffering antigen sources have on T-cell quality is poorly understood. An aspect of this project aims to describe the consequences of antigen stimulationdose on T-cell quality. This project also evaluated common antigen stimulation sources,including Epstein-Barr Virus (EBV) transformed lymphoblastoid cell lines (EBV-LCLs),adenoviral vector-infected antigen presenting cells (APCs), and peptide-pulsed APCs, todetermine their impact on T-cell quality. Of particular interest is the impact of EBV-LCL-based stimulation as this method is widely used for the manufacture of ACT drug products. To this end, we hypothesised that the therapeutic potential of an ACT drug product is heavilyinfluenced by the stimulation and expansion method used. Furthermore, this project aimedto demonstrate the utility of comprehensive ACT drug product assessment.Indeed, our findings revealed that T-cell cultures stimulated with varying stimulationdoses exhibit different phenotypic, transcriptional, and T cell receptor (TCR) repertoireprofiles that reflect functional differences. This analysis can potentially be used to distinguishbetween ACT drug products of superior quality. The comprehensive analysis approach usedalso revealed that the dose of EBV-LCL-based stimulation had a profound impact on theoverall ACT drug product composition. Specifically, it was observed that a high-dose EBV-LCL stimulation promoted a significant increase in natural killer cells but limited expansionof CD4+ T-cells. In contrast, varying stimulation dose of other antigen stimulation sources—adenoviral vector-infected APCs and peptide-pulsed APCs—had less influence on culturecomposition. Overall, these two stimulation approaches resulted in ACT drug products thatpredominately contained T-cells. Additionally, T-cells derived from a high-dose EBV-LCLstimulation showed hallmarks of T-cell exhaustion, via the co-expression of multiple inhibitory receptors and decreased polyfunctionality. High-dose EBV-LCL stimulation alsoinduced a transcriptional signature reflective of a type 2 response, in both CD4+ and CD8+T-cells. Previous experimentation has demonstrated a type 2 response to be undesirablefor anti-viral or anti-cancer immunotherapy. In contrast, a low-dose EBV-LCL stimulationgenerated a gene signature that could predict improved overall survival in patients with solidtumours. Further interrogation involving TCR sequencing revealed that EBV-LCL stimulationdose had a profound impact on the resultant TCR repertoire. In this regard, low-dose EBV-LCL stimulation promoted a more clonal T-cell population—enriched for T-cells with inferredhigh affinity TCRs. In contrast, high-dose EBV-LCL stimulation induced a polyclonal TCRrepertoire, hypothesised to be the result of the increased expansion of low affinity TCRs. Altogether, the data generated as part of this study have important implications in themanufacture of ACT drug products. Our findings demonstrate that factors such as antigensource and stimulation dose significantly impact the quality of an ACT drug product. Notably,these results demonstrate that using adenoviral vector-infected APCs or peptide-pulsedAPCs can generate a therapeutic drug product highly enriched for T-cells. Furthermore,these methods are robust to variations in stimulation dose. In contrast, the stimulation doseof EBV-LCLs significantly impacts T-cell quality and can derive drastically different ACT drugproducts. This is an important finding given the widespread use of high-dose and repeated-dose EBV-LCL-based stimulation within the ACT therapy field. Additionally, these dataclearly demonstrate the utility of comprehensive ACT drug product profiling. This notion isillustrated by TCR sequence analysis revealing the presence of predicted low affinity TCRsas well as transcriptional analysis inferring T-cell status, including T-cell exhaustion and type2 polarisation. Thus, this data demonstrates that comprehensive analysis of ACT drugproducts can aid in determining T-cell function and hence, the therapeutic fitness of a givendrug product. Such analysis is seldom performed; however, this project suggests that thispractise has its place within the manufacturing process in the ACT immunotherapy field.