Chimeric antigen receptor (CAR) T cell therapy has advanced into commercially available treatments that have driven an influx of companies into the immunotherapy landscape. In a standard workflow, a patient’s (autologous) or donor’s (allogeneic) genetically engineered T cells must be expanded ex vivo for clinical use. Typically, T cell expansion is the longest and one of the most critical phases in the CAR T workflow.

Engineered T cells are sensitive to their microenvironment and growth conditions. These responses can lead to potential undesirable cellular changes, which pose production and quality control challenges that could ultimately delay patient treatment.

One of the main objectives during the expansion phase is to maintain “younger”, less differentiated memory cell phenotypes. Figure 1 illustrates the spectrum of T cell differentiation with the less differentiated, most therapeutically desirable central memory T cells (TCM) on the left and less desirable, more mature effector memory T cells (TEM) and effector T cells (TEFF) on the right. Surface markers such as CD62L, CCR7, and CD28 are not present when T cells differentiate and transition toward the TEFF cell populations and become less efficacious.

Figure 1. Younger is better. Maintenance of early TCM cells (left) is critical during ex vivo expansion as this phenotype is associated with higher efficacy treatment of patients due to their proliferation and renewal potential.

T cell viability and quality following expansion has a strong influence on treatment efficacy. Fortunately, relative factors such as media and supplement use, cell density, and the culture platform have been identified and improved upon to optimize the conditions needed for a successful and robust workflow. Sampling and analysis of CAR T cells is required during the expansion process to assess cell quality and ensure safety and potency of the product.

Allogeneic vs. autologous expansion

Early cell therapy work placed an emphasis on autologous workflows, where a diseased patient’s own T cells are isolated, activated, genetically modified, expanded, and finally infused back into the patient. Key benefits to an autologous workflow are that it allows for personalized treatment and minimizes the risk of immunorejection and transfer of other diseases or viral infections. Autologous cell therapy poses several challenges, as the patient’s cells often demonstrate slower growth profiles and display more mature T cell phenotypes, most likely due to the nature of the patient’s illness and the in vivo cell environment [1]. According to a 2019 review article, research has demonstrated these inherent patient T cell deficiencies causally relate to the slower CAR T cell expansion, persistence, and lower cytotoxicity observed in autologous therapies [2]. These cell deficiencies directly correlate to autologous cell expansion and quality issues, and most importantly, to loss of therapeutic efficacy and slower turnaround time for patient treatment.

Allogeneic CAR T therapy has shown strong potential to change and improve the therapeutic landscape. An allogeneic workflow, which uses starting material derived from healthy donor cells, can provide more efficacious and timelier “off-the shelf” treatment options that offer standardized therapeutic product for multiple patients. Using healthy donor cells can help overcome many of the expansion and quality issues posed by patients T cells [3,4]. It also provides the potential for re-dosing or delivering a combination of CAR T cells directed against different therapeutic targets. Despite the benefits, allogeneic cell therapy can pose a significant risk to patients by causing life-threatening graft-versus-host disease or elimination by the host immune system. Currently, these issues are a high priority, and research is underway to develop genetic CAR modifications that could mitigate host rejection risks [5].

Since allogeneic therapies require a greater quantity of cells for production of multiple doses, allogeneic workflows tend to be of larger scale than autologous workflows and require a longer timeline, with the expansion phase generally lasting 12–18 days. The additional culture time can negatively impact cell differentiation and function of the therapeutic product. As previously discussed, unwanted differentiation of T cells results in the loss of the younger TCM populations, which can lead to decreased patient therapeutic responses and treatment efficacy. However, successful mitigation strategies to address this include controlled isolation and activation of TCM cells, in conjunction with considerations towards cell density and supplementation throughout the expansion process. Ultimately, with successful mitigation, a potential scalable allogeneic workflow could reduce the overall cost of T cell therapy and provide greater treatment accessibility [6].

Activation to expansion

In cell therapy workflows, white blood cells are collected from donors in a process called leukapheresis. Selected T cell phenotypes from the white blood cells are isolated and activated to support gene transfer and the reprogramming of T cells to express CARs. With allogeneic workflows, the sourcing of T cells from healthy donors dramatically increases the probability of isolating a more desirable, early memory T cell population, which can result in higher cellular output and overall increased treatment efficacy.

Co-stimulation through CD3 and a secondary signaling receptor, such as CD28, provides the “wake-up” signal to activate naïve cells. CD3 signaling is indispensable for T cell growth, while agonistic ligation of CD28 contributes to T cell survival and plays a role in cytoskeletal remodeling, production of cytokines, differentiation, and transcription and post-translational changes during expansion [7].

Currently, T cell activation is primarily performed using an antibody-coated magnetic bead (Figure 2) or nanoparticle technology that imitates antigen-dependent signaling with anti-CD3 and anti-CD28 antibodies. These technologies replace traditional home-brew methods for generic activation that used antigen-presenting cells (APCs), mitogens, soluble or plate-bound antibodies, or chemical activators. Additionally, specific cytokines, such as interleukin-2 (IL-2) and interleukin-7 (IL-7) have been shown to support activation and maintenance of the desirable TCM phenotype with a greater expansion capability [8].

Products such as Gibco CTS Dynabeads CD3/C28 allow for isolation and activation in a one-step process for expansion of the desired T cell phenotype [9] (Figure 2). This covalently bonded antibody bead technology does not require feeder cells, antigens, or APCs. The beads can be removed following activation or prior to genetic modification and expansion. It is important to note, both the product and protocol selected for activating the T cells should conform to the application, process, and regulatory requirements, i.e., research use only (RUO) or clinical use application.

A drawing that illustrates how Dynabeads facilitate isolation, activation, and expansion of T cells using antibodies coupled to the beads.
Figure 2. Magnetic bead approach to T cell isolation and activation. Bead technology products, such as CTS Dynabeads CD3/CD28, offer an ex vivo method for isolation, activation, and expansion of T cells. The uniform, inert, superparamagnetic beads are similar in size to antigen-presenting cells and are covalently coupled to anti-CD3 and anti-CD28 antibodies. These two antibodies provide primary and co-stimulatory signals, optimized for efficient T cell activation and expansion.

Following activation and engineering, buffer exchange is performed to transfer the desired T-cells into the expansion medium. This step can be done using counterflow centrifuge in a closed an automated manner. During expansion, release of immunostimulatory cytokines, such as IL-2 and IFN-ɣ at desired levels can allow CD8+ cells to survive as memory T cells during expansion.

Conditions and factors impacting T cell expansion

Expansion platforms

Selecting a platform for T cell expansion is dependent on the end-user’s application, working volume, workflow, and regulatory requirements. Table 1 provides a summary of the different platform options available along with their benefits and disadvantages. Ex vivo expansion can occur in static or dynamic culture systems. Static culture systems can be performed in flasks or gas permeable static culture bags. To maintain a closed system, many static cell culture bags include sealed sterile tubing and connectors for sampling. The temperature and CO2 levels can be controlled by placing static cultures in a controlled incubator that is set typically at 37ºC and 5%, respectively. With this method, gas exchange occurs only through media exchange. The G-Rex device may be an alternative closed static culture system option, designed to enable gas transfer through a permeable membrane on the bottom of the device that can permit relatively larger medium to surface working volumes compared to flasks [10].

Due to limitations in volume, nutrient, and gas exchange, a static culture system may not be the ideal platform for allogeneic workflows. When scaling up T cell workflows, it is not a given that small-scale success will ensure the same with a larger system. For scaling, select an appropriate bioreactor and develop a quality by design strategy.

Closed and automated rocking bioreactor platform systems are applicable and robust for autologous and allogeneic workflows and scaling. The use of bioprocess single-use cell culture container technology along with automated rocking bioreactor systems, allow for monitoring and control of dissolved oxygen (DO), pH, glucose, and metabolites such as lactate and ammonium. The movement and angle of the rocking bioreactor allows for uniform mixing and gentle agitation.

Compatible presterilized, single-use culture bags are armed with multiple sensors, allowing for automated control and reporting. For fed-batch processing, these systems monitor the culture volume with continuous weight measurements. During expansion, typical conditions include pH of 6.6–7.0, a CO2 range between 6.6%–7.5%,and a DO range of 30–50% to ensure viability. The ideal agitation speed and angle is dependent on the working volume and bag size. For a rocking system such as the Thermo Scientific HyPerforma Rocker Bioreactor, a volume of 1.5 L in a 10 L bioreactor (bag) will typically have an agitation of 8 rpm and a rocking angle of 6 degrees. An increase in the agitation and angle is expected to increase with the use of larger T cell expansion volumes [11].

Recently, using a low-shear force benchtop stirred bioreactor with a fed-batch or perfusion process has been considered as a T cell expansion platform. If shown successful, this may have a significant impact, particularly on allogeneic therapies, that require high T cell yields. A modular stirred benchtop bioreactor system could potentially help reduce costs since these systems are highly automated and require less handling. In addition, they can be used across a greater range of volumes, from research to commercialization scale (Table 1). This capability reduces or eliminates the need to re-engineer and transition to an entirely different scale-up process, which can be costly and introduce transition error risks. This can be especially important when processes are locked in at clinical stage 2 or 3 trials. Additionally, stirred benchtop bioreactors can reach a higher volumetric mass transfer coefficient (KLa), which supports effective and homogenous oxygen delivery inside the bioreactor.

Table 1. Comparison of T cell expansion platforms.

OptionsTypical working volumeClinical or RUO*AdvantagesDisadvantages
Static bags5 mL–3 LRUO or Clinical
  • Closed system
  • Limited gas transfer and working volume
T-flasks/Static plates100 µL–370 mLRUO
  • Economical Great for screening multiple conditions
  • Open system
  • Small scale
Rocking motion bioreactor300 mL–50 LClinical
  • Closed, automated, scalable
  • Gas, liquid, DO, and pH control and sensing
  • Perfusion capable
  • Digital integration
  • Large footprint
  • High cost
G-Rex8 mL–5 LRUO or Clinical
  • Supports scaling
  • Automated
  • DO and pH control
  • Supports KLa value compared to static culture
  • Difficult to close system
  • Requires training
  • High cost
Stirred tank bioreactor250 mL–2,000 LClinical
  • Closed, automated, scalablea
  • Gas, liquid, DO, and pH control and sensing
  • Perfusion capable
  • Digital integration
  • Supports higher KLa valuesa
  • Large footprint
  • High cost


The choice of media and supplements can significantly influence the growth of the T cell population, differentiation, viability, and the CD8:CD4 ratio during expansion. It is important to select a flexible expansion medium that is compatible with other workflow processes such as T cell isolation and activation, while being amendable to various platforms ranging from static cell culture systems to larger scale dynamic bioreactors. 

As variation in workflows and protocols increases, manufacturers have investigated and developed optimized media and conditions to support a flexible, seamless, and scalable workflow. The media source can be a bottleneck in T cell expansion by presenting challenges including batch-to-batch variability, which can negatively impact the consistency and quality of the product output. Utilizing a chemically defined and serum-free medium from a dependable supplier with strong quality control processes can help reduce this variability. These product attributes will also minimize downstream purification and regulatory risks, as well as lower overall production costs.

The significant influence of media on T cell expansion was demonstrated in a recent study conducted using a novel culture medium that was developed specifically for expansion of human T cells in allogeneic cell therapy workflows [12]. A major challenge in CAR T workflows is the need for a larger number of cells with the preferred younger central memory T cell phenotype that results in more functionality and effective therapeutic outcomes. Modest increases in central memory cells early in the expansion phase result in larger cell yields at harvest. In this study, healthy donor T cells were tested in an 18-day allogeneic type workflow, with the results demonstrating approximately 20% higher cell proliferation by day 10 and nearly a 100% increase by day 17 using the newly formulated medium, when compared to the control medium (Figure 3). A 10% to 20% increase in the size of the desired central memory T cell subset was also demonstrated when normalized to the control medium (Figure 4)[12]. In addition, this boost in cells displaying an early memory phenotype coincided with a higher level of interferon gamma (IFNγ) release when healthy donor cells are used. An average 187% increase in IFNγ production across six patients was demonstrated when normalized to the control medium (Figure 5). The increase observed will likely act as a catalyst to enhance the overall immune response and provide more efficacious patient therapies by stimulation of macrophages, neutrophils, and natural killer cells [12].

Not only does medium impact the expansion phase, but the benefits are dependent and specific to the initial cell source used during the expansion phase. In these studies, a metabolic shift in the T cells allowed for a longer workflow [12].

Proliferation data shows that CTS OpTmizer Pro Serum Free Media supports large scale expansion of T cells
Figure 3. Expansion of T cells in specialized medium. Normalized proliferation with CTS OpTmizer Pro SFM in an 18-day workflow with six healthy donors shows approximately 20% higher fold expansion by day 10 and nearly a 100% increase by day 17, when compared to the control medium.
Data show that CTS OpTmizer Pro maintains the desired T cell population over time.

Figure 4. Expression of TCM markers in cells grown in specialized medium. In an allogeneic workflow, six healthy donor cells grown with CTS OpTmizer Pro SFM showed a 10% to 20% increase in the size of desired TCM population based on analysis of central memory markers CD62L, CCR7 and CD27.

Data show that T cells grown in CTS OpTmizer Pro Serum Free Media show increased interferon gamma production.
Figure 5. Increased IFNγ production in cells grown in specialized medium. Six healthy donor cells grown with CTS OpTmizer Pro SFM demonstrated an average 187% increase in IFNγ production when normalized to the same cells grown in control medium.  


In addition to medium, supplements also play an important role in T cell expansion. L-glutamine is an essential carbon source for metabolism and differentiation that needs to be supplemented throughout the cell therapy manufacturing process. 

Serum is commonly used in research and can aid in supporting basal activation of T cells as well as cell growth. While serum use may be acceptable for research, it poses regulatory and supply concerns for use in clinical applications and can demonstrate variability from batch to batch, which can impact T cell differentiation and the overall product output. To address these issues, serum replacement alternatives are available, which is a defined supplement shown to support comparable T cell CD8+/CD4+ ratios while maintaining high cell expansion and viability [13,14].  


Perfusion is a bioprocessing technique that involves continuous exchange of spent media with fresh media, while retaining cells within the culture vessel. This method refreshes nutrients while preventing the buildup of toxic metabolic waste products that could negatively impact culture performance and impact product quality. This technology allows the culture to reach much higher cell densities and fold expansion within a smaller footprint and can lead to increased productivity compared with traditional batch or fed-batch processes. Perfusion has been applied to both autologous and allogeneic T cell expansion, but it is essential for most allogeneic workflows, which require a higher quantity of cells. Achieving greater density and expansion of cells in a shorter time frame can reduce the population of exhausted cells and increase the population of younger TCM cells, which are known to support more efficacious treatments.

A side-by-side T cell expansion experiment comparing fed-batch and perfusion workflows in rocking motion bioreactors was used to evaluate the impact of perfusion on cell growth, cell viability, key metabolites, and growth factors [15]. Perfusion supported a higher density of viable T cells by reducing the accumulation of lactate and ammonia during expansion. As with other cell types, these metabolites are toxic to T cells and can induce arrest of cell growth or apoptosis, which directly impacts cell viability and expansion.

Another key to allogeneic workflows is having a sufficient IL-2 concentration to support T cell survival and proliferation. Through perfusion, a sufficient level of IL-2, a key growth factor for T cell survival and proliferation, was maintained during expansion. On the contrary, the non-perfused culture showed the IL-2 concentration dropping to undetectable levels by day 12, which prevented cells from reaching comparable cell viability [15]. To fully understand the role of perfusion and IL-2, an additional experiment was setup to analyze and compare cell growth and viability with perfusion and non-perfusion setups. The amount of IL-2 injected into the test non-perfusion bioreactor was equivalent to what was delivered to the perfusion control bioreactor.Despite the daily equivalent IL-2 injections, by the end of the T cell expansion phase (day 14), the non-perfusion culture expansion was only 63% of the perfusion control [15].

While perfusion has its advantages, the high volume of perfusion medium required can be costly, and it is imperative that manufacturers optimize their process and perfusion rates to avoid medium waste. That said, the benefits of perfusion technology are undeniable for T cell expansion; it is a readily scalable solution that can help maintain a favorable culture environment for longer, more productive workflows specifically important for allogeneic cell therapy.

Maintaining cell density on expansion

Additional cell culture parameters, such as maintenance cell density, have been shown to influence T cell expansion. The results of a recent evaluation of the impacts of maintaining different T cell densities demonstrated several direct effects on the quantity and quality of the cellular products. Many classical T cell expansion protocols call for the maintenance of cells at 0.5 X106 cells/mL; however, the results of this study demonstrated that a lower cell density of 0.25 x 106 cells/mL correlated with higher-fold cell expansion and improved viability. Conversely, increasing the maintenance cell density to 0.75 x 106 cells/mL demonstrated lower-fold expansion and slightly lower cell viability. These results suggest that maintaining a lower T cell density can positively impact the quantity and quality of the cellular output by exposing the T cells to more nutrients and help maintain a larger central memory phenotype [6].


An ongoing shift toward supporting viability of the desired T cell phenotypes has created interest in understanding the durability of the response elicited by activation and reactivation in the expansion phase. In a recent study, secondary “restimulation” during the expansion phase was evaluated using CTS Dynabeads CD3/CD28 [6]. The goal was to better understand the effects of restimulation of the T cells, how it affected the therapeutic cell output and whether it had any impact on the T cell manufacturing process.

This study revealed that a single round of activation with the beads was not only sufficient to induce robust cell proliferation and high viability over the entire 20-day workflow, but also provided evidence that restimulation can cause a temporary growth lag and plunge in viability during the following days (Figure 6A and 6B). In addition, cells subjected to secondary activation displayed a lower CD8:CD4 ratio and a sharp downregulation of central memory cell biomarkers. In both cases, with and without restimulation, there is a clear decrease in the central memory population (CD62L+ CCR7+) and an increase in the double-negative effector population (CD62L, CCR7 ,) in the later stages of expansion, which is much more pronounced within the restimulated group (Figure 6C) [6].

Data show that restimulation causes T cells to expand less robustly.
 Click image to enlarge

Figure 6. Effect of restimulation on T cell growth and phenotype. A second round of stimulation with CTS Dynabeads CD3/CD28 during a longer workflow negatively impacts: (A) cell expansion, (B) cell viability, and (C) cell differentiation. (Error bars represent the standard deviation of three donors performed in triplicate).

These results suggest that restimulation catalyzes the transition of early memory T cells into effector T cells. Slowing this transition is important to developing and fine-tuning T cell manufacturing processes. 

Overall study results show the critical importance of choosing the optimal media, supplements, platform, and process parameters that will support robust T cell expansion, while maintaining the desired early central memory T cell phenotype which are key for efficacious T cell therapies.

Regulatory and analytical testing

Knowledge of the regulatory requirements is critical to maintaining approval at each stage to avoid critical issues and delays in progressing to later clinical phases. In addition, regulatory requirements influence many of the analytical quality testing requirements of cell therapy products.

During the expansion phase, robust and reliable analytical tools are required to accurately measure and monitor various cellular characteristics such as proliferation, viability, differentiation status, and other cell phenotype attributes. During and after the expansion process, testing is performed for a range of required quality attributes with regulatory body-approved or in-house validated assays. These tests evaluate quality attributes related to safety, purity, and potency of the cellular product. Safety testing for microbial and fungal sterility, mycoplasma, as well as testing for replication competent viruses and vector copy number are usually required when viral vectors are used in the manufacturing process. Purity testing is often conducted with flow cytometry techniques to assess the proportion of positive cells for respective T cell- and CAR-associated surface markers, as well as safety testing to make sure undesirable cell types are not present. Potency testing is conducted to assess the CAR T cell content with flow cytometry or PCR methods, while cytotoxicity and cytokine secretion is assessed with various in vitro assays [16].

There are several different ways to analytically test any given cell characteristic, therefore, variability in testing presents a known source of variation during the manufacturing process [16].

Vision and concluding remarks

Over the last decade, many advances have been made in the field of cell therapy and there are many more to come. As time is of the essence for patient therapy solutions, efforts to reduce production and product quality release timelines are a critical focus of research and development efforts. The T cell manufacturing process is complex and currently many opportunities for variability exist at each step. Further research is underway to identify and address manufacturing variables that can affect the critical quality attributes of the final product [16]. Implementation of greater automation, closed system manufacturing and real-time characterization will likely be important future developments [17].

Achieving “off-the-shelf” allogeneic and improving upon autologous therapeutic solutions for patients requires substantial investment in product and process development. Current and developing work for T cell expansion involves mitigating risk and developing a robust more standardized and transferable process. Key to making the journey from initial research and development to commercial manufacture as simple and as seamless as possible is a detailed understanding of the product and process development to enable successful scale-up [17]. Equally important will be innovation within the field to explore new approaches and expand upon the currently available technologies.

A note about CAR NK cell therapies

While CAR T therapies have demonstrated success in treatment of circulating blood-related cancers, utilizing engineered T cells for other cancers, such as solid tumors, has proven challenging. An approach under extensive development involves the use of natural killer (NK) cells, a cell subset of the innate immune system, which have been shown to be effective in a number of clinical trials for various cancer treatments. Allogeneic NK cells exert their cytotoxic effect in an antigen-and HLA-independent manner.

Engineering of CAR NKs has facilitated specific targeting to tumor specific antigens enabling them in a solid tumor microenvironment. This aspect combined with iPSC derived CAR NK technologies could lead to an unlimited supply of cells as an off-the shelf solution, bypassing long lead times for patients. Recent studies with CAR-NK-19 for lymphoid tumors (CD19) [18, 19] and CAR-NK-GPC3 for solid tumors of hepatocellular carcinoma (HCC) and ovarian cancers [20] demonstrate the promise of this approach.

The workflow to generate CAR NK cells is similar to that of CAR T cells but involves sufficient ex vivo expansion of NK cell cultures to meet dose and lot size requirements for allogeneic therapies. Typically, each dose requires approximately 5 x 106 cells/kg of body weight, or nearly 500 x106 cells per person. The cell culture platforms that are used for ex vivo expansion of CAR-NK cells include T-flasks, G-Rex vessels, bioreactors or culture bags depending on scale of final clinical product. Recommended culture conditions include use of a xeno-free medium supplemented with IL-2, IL-15 and IL-21. As is the case with CAR T cells, extensive QC requirements precede the use of CAR NK cells in a clinical setting, and often include testing the cell product for its cytotoxic killing abilities on various tumor cell lines in vivo animal tumor models.


  1. Staff (2021) CAR-T Cell Therapies: Allogeneic Workflows - Bioprocess Insider. BioProc Int.
  2. Cheng J, et al. (2019) Understanding the Mechanisms of Resistance to CAR T-Cell Therapy in Malignancies. Front Oncol 9:1237.
  3. Li T, et al. (2019) A Good Response of Refractory Mantel Cell Lymphoma to Haploidentical CAR T Cell Therapy after Failure of Autologous CAR T Cell Therapy. J ImmunoTher Cancer 7(1):52.
  4. Cho JH, Collins JJ. & Wong WW. (2018) Universal chimeric antigen receptors for multiplexed and logical control of T cell responses. Cell 173(6): 1426–1438.
  5. Depil S, et al. (2020). "Off-the-shelf" allogeneic CAR T cells: development and challenges. Nat Rev Drug Discov 19(3):185–199.
  6. Thermo Fisher Scientific (2021) Successful support of longer T cell expansion workflows with CTS Dynabeads CD3/CD28 and CTS OpTmizer T Cell Expansion SFM.
  7. Esensten J, et al. (2016). CD28 costimulation: From mechanism to therapy. Immunity 44 (5):973–988.
  8. Kalia V & Sarkar S. (2018) Regulation of Effector and Memory CD8 T Cell Differentiation by IL-2-A Balancing Act. Front Immunol 9:2987.
  9. Thermo Fisher Scientific (2019) One-step isolation and activation of naive and early memory T cells with CTS Dynabeads CD3/CD28.
  10. Ludwig J & Hirschel M (2019) Methods and process optimization for large-scale car t expansion using the G-rex cell culture platform. Methods in Mol Biol 2086:165–177.
  11. Thermo Fisher Scientific (2020) Optimizing parameters for 10-day and 20-day T cell expansion in rocking bioreactors using CTS OpTmizer T cell Expansion SFM (in house-data).
  12. Thermo Fisher Scientific (2021) Improving Allogeneic Workflow and Cell Proliferation Using CTS OpTmizer Pro SFM.
  13. Thermo Fisher Scientific (2019) CTS OpTmizer T Cell Expansion SFM, No Phenol Red, Displays T Cell Manufacturing Workflow Flexibility.
  14. Thermo Fisher Scientific (2019) The Benefits of Supplementing CTS OpTmizer T Cell Expansion SFM, No Phenol Red.
  15. Janas M, et al. (2015) Perfusion's Role in Maintenance of High-Density T-Cell Cultures.” BioProc Int 13(1):18–26.
  16. Reddy OL, Stroncek DF & Panch SR (2020) Improving CAR T cell therapy by optimizing critical quality attributes. Semin Hematol 57(2):33–38.
  17. Eyles JE, et al. (2018) Cell therapy products: focus on issues with manufacturing and quality control of chimeric antigen receptor T-cell therapies. J Chem Technol Biotechnol 94(4): 1008–1016.
  18. Liu E, et al. (2020) Use of CAR-transduced natural killer cells in CD19-positive lymphoid tumors. N Engl J Med 382:545–553.
  19. Ueda T, et al. (2020) Non-clinical efficacy, safety and stable clinical cell processing of induced pluripotent stem cell-derived anti-glypican-3 chimeric antigen receptor-expressing natural killer/innate lympohoid cells. Cancer Sci 111:1478–1490.
  20. Li Y, et al. (2018) Human iPSC-derived natural killer cells engineering with chimeric antigen receptors enhance anti-tumor activity. Cell Stem Cell 23(2):181–192.e5.