The tremendous potential and promise of cellular therapies for treatment of cancer are tempered by the challenges associated with production of cell products in therapeutic doses under GMP requirements for consistency, sterility and quality. In this report, we show that the Quantum system is a promising expansion platform for deployment in the cell therapy space, where it could be utilized in manufacturing processes in which end users would validate their entire process for their intended use. Cell culture expansion was started directly in the Quantum system with no requirement for a flask culture intermediary, thus reducing the risk of contamination. The risk was further reduced by the use of functionally-closed tubing systems in which cell inlet bags and media bags were sterile-welded directly on to the Quantum system disposable sets.
The Quantum system has previously been described as a robust platform for expanding adherent cells such as adult stem cells [4,5,6,7,8] and, more recently, preliminary results for expansion of non-gene-modified T-cells were reported [9]. We now show that the Quantum system supports consistent expansion of T-cells from apheresis collections from multiple healthy donors, yielding numbers of CD3+ T-cells in clinical dose ranges (approximately 13 × 109 to 25 × 109 T-cells per expansion run depending on numbers of cells seeded) in a short time frame of 8 to 9 days and with consistent viability, fold expansion and doubling times. The specific protocols and instrument settings employed to achieve these high cell numbers have been previously described in detail and include accelerated feeding for improved nutrient supply, increased oxygen supply and increased lactate removal [9]. Previously reported T-cell expansion in the Quantum system was 500-fold; the present study demonstrates 760- and 1300- mean fold expansions depending on the starting cell numbers. The reason for the difference in expansion efficiency may be the use of different medium formulations: PRIME-XV [FUJIFILM Irvine Scientific] was used in the current report whereas TexMACS™ [Miltenyi Biotec] was used in the earlier report. Despite rapid and high expansion, all harvested products retained functionality upon restimulation as measured by IFN-γ, IL-2 and TNF-α secretion, although there were some differences among the donors in the T-cell phenotypes generated as discussed further below. There were also some differences between T-cell products from low and high PBMC seeding numbers. These different starting cell numbers were used to investigate the versatility of the Quantum system, and overall products of low and high seedings were functional and had similar frequencies of T-cell memory subsets. Expansion occurred from input cell numbers as low as 7 × 106 CD3+ T-cells in a total of 30 × 106 PBMCs (see Table 1, donor 5). This is an important consideration for expansion of T-cells from patients, particularly pediatric patients, who may have low numbers of circulating CD3+ T-cells and for whom apheresis blood collection may be restricted or challenging. We are continuing to investigate the lower limits of cell numbers that can be used to initiate T-cell expansion in the Quantum system. For high seed cultures, nutrients may have been limiting toward the end of the T-cell expansions, as suggested by the more rapid decline of lactate generation rates in high seed cultures at later timepoints. Increasing the feed rates in the Quantum system beyond those applied here could have the potential to result in greater T-cell harvest yields, a subject which we are currently investigating.
Current autologous CAR T-cell dose ranges are lower than the T-cell yields reported here for the Quantum system (maximum adult doses for axi-cel and tisagenlecleucel are 2.0 × 108 and 6.0 × 108 cells respectively) [YESCARTA™/axi-cel and KYMRIAH™/tisagenlecleucel prescribing information]. This suggests that the Quantum system may have the potential to reduce the culture time required for preparation of autologous CAR T-cell doses because of the rapid cell expansion that the system can support (see Fig. 2). This may be crucial for both the stage of T-cell differentiation required for efficacy and for expedient treatment of patients with rapidly-advancing disease [10]. Since cell numbers can be monitored during expansion via sterile sampling from the IC loop, cultures can potentially be harvested upon attaining target cell numbers with the desired phenotype, similarly to methodology that has been described for axi-cel [3].
The CD4/CD8 ratios of harvested T-cells differed somewhat among individual donors and differed between cell expansion products from the same donor initiated with either low or high seeding numbers, although total numbers of doublings were very similar. In general, CD8+ T cells expanded preferentially in low seed cultures and CD4+ T cells expanded more in high seed cultures. This may suggest that IL-2 was limiting for the high seed cultures. Lower exogenous IL-2 availability could bias towards CD4+ T-cells, which generally secrete higher levels of IL-2 than CD8+ T-cells. Although there is currently no definitive evidence that there is a preferred ratio of CD4+ to CD8+ T-cells for optimal therapeutic efficacy and safety of CAR T-cell products [11,12,13,14], defined ratios have been shown to elicit superior efficacy in preclinical models [15]. However, in the ZUMA-1 clinical trial of axi-cel, averages of 57% CD8+ and 43% CD4+ T-cells comprised the product [14], and for tisagenlecleucel, similar ratios of CD4+/CD8+ T-cells were present in the products for both responders and non-responders, with expansion of CAR T-cells in vivo occurring independently of this ratio [13]. These clinical data suggest that the CD4/CD8 ratios in T-cell products produced here in the Quantum system would be applicable for immunotherapy.
CD8+ T-cells harvested from the Quantum system displayed less-differentiated phenotypes of TN and TCM and CD4+ T-cells were predominantly TCM. These phenotypes are currently believed to be the most potent subsets for adoptive immunotherapy in hematologic cancers, with superior durability in vivo and enhanced proliferative capacity [15,16,17] although unfractionated CAR T-cells are being perfused in most clinical trials [18], and in patients treated with tisagenlecleucel [13]. In addition, less-differentiated T-cell memory phenotypes had strong correlations with responses to therapy in chronic lymphocytic leukemia (CLL) patients, and late memory and TE phenotypes were correlated with poor outcomes in the same patient group [12].
All T-cell products derived from the Quantum system had low frequencies of cells bearing the exhaustion markers LAG-3 and PD-1 and these markers did not increase during expansion. In contrast, Tim-3 expression frequency increased on all products following expansion. Although Tim-3 has been shown to have a role in exhaustion of T-cells, it may also have costimulatory capacity and its exact mechanistic functions are likely not yet fully elucidated [19,20,21]. Moreover, co-expression of Tim-3 with other exhaustion markers on T-cell products was low and, since T-cell dysfunction is generally associated with co-expression of inhibitory markers and receptors [22,23,24], it is unlikely that T-cells harvested from the Quantum systems would be dysfunctional. In addition, following restimulation in vitro, all harvested T-cell products responded potently with Th1-type cytokine secretion, indicating that these populations were functional and not exhausted. The phenotypes and differentiation pathways of T-cells and gene-modified T-cells expanded in vitro are influenced by many factors including the cytokines used for expansion [25], disease status of the patient [10, 25], duration time of cells in culture [10] and composition of the culture medium used [26]. These factors are currently being evaluated in the Quantum system to better understand the system’s versatility and advantages, particularly with regard to the growth of CAR T-cells.
Other semi- and fully-automated expansion technologies are available for growth of T-cells for immunotherapy, e.g. the Xuri™ W25 and WAVE Cell Expansion systems (GE Healthcare Life Sciences) [27, 28], Gas Permeable Rapid Expansion (G-Rex®) devices (Wilson Wolf Corporation, St. Paul, MN) [29, 30] and the CliniMACS Prodigy® (Miltenyi Biotec) [31,32,33,34]. With all these systems, as well as with the Quantum system, the goals are to provide functionally-closed systems for T-cell growth and to obtain high cell yields while minimizing risk of contamination. While these other systems have distinct attributes applicable for different cell expansion settings, published cell yields encompass those presented here. Reported yields from the Prodigy are lower than the Quantum system [31,32,33,34], with ranges up to 6 × 10 9 CAR T-cells when healthy donor cells were used [34]. G-Rex reported yields are similar to Quantum system yields, with up to 20 × 109 T-cell receptor (TCR)-engineered T-cells obtained with products derived from cancer patients as well as healthy donors [30]. Yields for CAR T-cells from the WAVE have been reported to be similar to the Quantum system, with ranges up to 24 × 109 for cells derived from CLL patients [28], although with higher yields of 44 × 109 cells reported for tumor infiltrating lymphocytes (TILs) [27].
A unique feature of the Quantum system is the dual loop system that allows for partitioning of different formulations and volumes of medium to the IC and EC fluid loops, as illustrated herein. In addition, the dual loop design has the potential to reduce use of cytokines and growth factors such as IL-2, since these can be preferentially added to the relatively small volume of IC medium only, with nutrient and gas supply provided from the EC side. For the T-cell expansions in the Quantum system described here, although a total of 20 L medium was consumed in the high seed feeding protocol, which generated an average of 23.5 × 109 T-cells, only 3.6 L of medium on the IC side of the bioreactor contained IL-2 (see Fig. 6), reducing the amount of IL-2 required by 82%. Considering that in expansion of TILs or TCR-engineered cells for autologous immunotherapeutics, use of 3000 U/mL IL-2 has typically been reported with other bioreactor systems [27, 30, 35], it is clear that the use of the Quantum system could result in substantial savings on the costs of this cytokine.
TILs are currently being tested for treatment of solid tumors and have been infused at extremely high doses to date. For example, Tran et al. infused a dose of 1.48 × 1011 TILs to a patient with metastatic colorectal cancer [35] and Zacharakis et al. described infusion of 8.2 × 1010 TILs to a patient with chemorefractory hormone receptor-positive metastatic breast cancer [36]. In both cases, the authors reported durable regression of metastatic disease. With the current configuration, it would require several Quantum system expansions to generate T-cell harvest numbers in dose ranges compatible with requirements for TILs. However, substantially higher T-cell numbers have been harvested using medium supplemented with additional ingredients such as albumin and human serum, indicating the potential for the Quantum system to support cell yields as high as approximately 40 × 109 T-cells (see Additional file 2). Moreover, as discussed above, increasing feed rates to the bioreactor may also result in higher T-cell harvest yields.
Allogeneic universal CAR T-cell products, derived from healthy donors, are also being developed as potential future cancer immunotherapies. This approach has advantages over autologous therapies in terms of standardization in manufacturing processes, removal of inter-patient variability and removal of the impact of patient health on the final product [37, 38]. Although allogeneic cell therapeutic doses are likely to be in similar ranges to autologous products, the ability to expand donated products to very high numbers would be necessary to create master cell banks and working cell banks for off-the-shelf use. The Quantum system has the capacity to support allogeneic cell therapy manufacturing by generation of T-cell products in the range of 23 × 109 T-cells from a single expansion using defined, serum-free medium as reported here. Based on these yields of T-cells, a single apheresis mononuclear cell collection from a healthy donor yielding 4.3 × 109 mononuclear cells (Terumo BCT internal data for the Spectra Optia® apheresis system) could be expanded via scale-out in Quantum systems to allow production of almost 4000 doses of allogeneic T-cells, based on the current maximum dosing for tisagenlecleucel in B-ALL [13]. If dosing were in range for the maximum doses of tisagenlecleucel recently prescribed for DLBCL (6 × 108 CAR+ T-cells; KYMRIAH/tisagenlecleucel prescribing information), approximately 1500 doses could potentially be manufactured from a single donor. In addition, alternative allogeneic effector cells for use as universal cell therapies such as NK cells, NKT cells and γδ T-cells, which do not cause graft-versus-host-disease [39], are likely to be amenable to expansion in the Quantum system following similar protocols to those described here.