Lymphocyte Expansion in Bioreactors: Upgrading Adoptive Cell Therapy

Garcia-Aponte et al.
Journal of Biological Engineering (2021) 15:13

https://doi.org/10.1186/s13036-021-00264-7
REVIEW Open Access
Lymphocyte expansion in bioreactors:

upgrading adoptive cell therapy
Oscar Fabian Garcia-Aponte, Christoph Herwig* and Bence Kozma
Abstract
Bioreactors are essential tools for the development of efficient and high-quality cell therapy products. However,
their application is far from full potential, holding several challenges when reconciling the complex biology of the
cells to be expanded with the need for a manufacturing process that is able to control cell growth and
functionality towards therapy affordability and opportunity. In this review, we discuss and compare current
bioreactor technologies by performing a systematic analysis of the published data on automated lymphocyte
expansion for adoptive cell therapy. We propose a set of requirements for bioreactor design and identify trends on
the applicability of these technologies, highlighting the specific challenges and major advancements for each one
of the current approaches of expansion along with the opportunities that lie in process intensification. We conclude
on the necessity to develop targeted solutions specially tailored for the specific stimulation, supplementation and
micro-environmental needs of lymphocytes’ cultures, and the benefit of applying knowledge-based tools for
process control and predictability.
Keywords: Adoptive cell therapy, ATMP, Bioreactor, Expansion, Lymphocyte, Rocking motion, Stirred reactor,
Perfusion reactor, NK cell, T cell
Background characterized as a “living” treatment that can be en-

In the process of understanding cancer, clinical research hanced by means of gene modification because cells
has developed a resourceful toolbox of treatment options continue to function in vivo after they have been infused
ever increasing in complexity. From surgery and radi- back into a patient [4]. ,To date, many cells have been
ation therapy, going through chemotherapy and bio- used for ACT, including Lymphokine-Activated Killer
logics, we have arrived to the field of Cancer (LAK) cells, Tumor-Infiltrating Lymphocytes (TILs),
Immunotherapy [1], an approach that merges with the Cytotoxic T Lymphocytes (CTLs), Cytokine-Induced
innovative area of Advanced Therapy Medicinal Prod- Killer (CIK) cells, γδ T cells, Regulatory T (TReg) cells,
ucts (ATMPs) to develop the specialty of Adoptive Cell Natural Killer (NK) cells, engineered T cells (T-Cell Re-
Therapies (ACT). ceptor (TCR T) cells and Chimeric Antigen Receptor
This branch of immunotherapy is defined as the intra- (CAR) T cells) [2, 5, 6].
venous administration of ex vivo expanded immune ef- Unfortunately, these cells remain as a limited thera-
fector cells that are capable of selective cytotoxicity. It peutic option that is only applied to a small number of
exploits the immune system’s ability to distinguish be- patients. Partly because of significant knowledge gaps on
tween pathologic and healthy tissue [2, 3]. ACT has been their clinical effectiveness and cost/benefit ratio and a
strong dependency on highly specialized methods, mate-
rials and equipment, therefore the number of products
* Correspondence: christoph.herwig@tuwien.ac.at
Research Area Biochemical Engineering, Institute of Chemical, Environmental approved for commercialization is reduced [7, 8]. As the
and Bioscience Engineering, TU Wien, Gumpendorferstraße 1a, 1060 Vienna, last decades saw progress in the understanding of
Austria
© The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if
changes were made. The images or other third party material in this article are included in the article's Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons
licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the
data made available in this article, unless otherwise stated in a credit line to the data.
Garcia-Aponte et al. Journal of Biological Engineering (2021) 15:13 Page 2 of 34
lymphocyte biology and different companies are devel- Expansion’s ubiquity highlights its importance for ACT’s
oping high throughput systems for ACT manufacturing optimization, relying on the application of Quality by
[9], it is expected that this field will experience a quick Design (QbD) principles for sound bioprocess under-
clinical and technical expansion, that requires process standing. However, optimizing for a process focused
intensification and innovative solutions from engineers. only on high cell output could narrow the Critical Qual-
Hence, there will be a future push to technologize ACTs, ity Attributes (CQAs) down to the productivity issue. In
from hospital-oriented to industrially relevant manufac- that sense, ACT would not benefit from an integrative
ture processes. clinical view, able to compensate for regulatory and en-
The manufacturing of an ACT product usually begins gineering constraints [17] in a broader context that con-
with a mixed lymphocyte population from a patient’s bi- siders yield, cell purity and product functionality.
opsy, or from apheresed Peripheral Blood Mononuclear The aim of this review is to give a comparative over-
Cells (PBMCs) (Fig. 1). It can also be started by differen- view of lymphocyte expansion in bioreactors, assessing
tiating a cell subset from Hematopoietic Stem Cells their ability to generate sufficient, functional and cor-
(HSC) and lymphoid progenitors generally obtained rectly differentiated cell populations, with considerations
from Umbilical Cord Blood (UCB). After cell acquisition, to process flexibility, controllability and scale. We ex-
several workflows can be followed depending on the plore the manufacturing of lymphocytes primarily from
intended application. In upstream, most of the protocols PBMCs and biopsies, summarizing the outcomes from
include cell selection, enrichment, purification, activa- the diverse expansion processes but taking the compar-
tion, stimulation, gene modification and expansion, ability issues arising from the wide range of stimulation
while downstream processes include pooling, further en- and supplementation strategies into the picture, apart
richment, formulation and cryopreservation [10–13]. In- from the selected bioreactor technology. Lymphocyte
dependently from the workflow, and because ACT doses manufacturing from stem cells is excluded from this re-
composed of high cell numbers generally produce more view as it adds an extra layer of complexity to the com-
desirable therapeutic outcome [14, 15], the cell expan- parison exercise. We first provide a context on the
sion process is a common factor in any ACT protocol, general culturing requirements for lymphocytes, later
being subjected to the greatest research efforts and the discussing the challenges of transitioning to technolo-
most significant body of user experience [16]. gized manufacturing. Given that context, a set of
Fig. 1 General upstream and downstream steps of a cell therapy product from autologous or allogenic source. The graph shows the contribution
of the different unit operations to final cell yield (red), functionality (green) and purity (blue). This review focuses on the expansion process
requirements for bioreactor design and comparison for available for the activation of immune cells, including
allogenic and autologous ACT is presented. We then re- cell-based activation, bead-based activation, and
view and categorize the available bioreactor technologies antibody-based activation. Antigen Presenting Cells
based on published results on process yield, cell purity (APCs), as cell-based activators, are endogenous agents
and product functionality. Finally, we propose further that provide an in vivo-like stimulation but they are ex-
knowledge intensive approaches that could be useful to pensive to use in a GMP environment, difficult to re-
take advantage of the data intensive environment that move from the final cell population, variable in their
bioreactors bring to the field of ACT. potential to induce activation and may be scarce when
isolated from donor samples [11]. Traditionally, immune
The complexity of lymphocyte expansion cell expansion has also relied on the supplementation
Compared to small molecules and biologics, living with animal or human serum. However, the use of
cells are much more complex: they sense their sur- serum may generate safety risks of infusion and in-
roundings, react to their environment and express creases process variability due to batch-to-batch differ-
varied and adjustable behaviors [18]. Furthermore, ences [11, 26, 27]. Besides antigen-induced activation,
they have some unique features [19], including the stimulation with cytokines is another factor that influ-
ability to specifically distinguish, bind and kill abnor- ences the composition, quality and phenotype of the
mally growing cells by selectively switching metabolic final cell product. T cells are generally produced by IL-2,
pathways to enhance the production of cytotoxic sub- IL-7 and/or IL-15 stimulation [28], while most current
stances [20]. Because of this complex biological set- NK cell expansion protocols include the use of IL-2 and
ting, any small change in the culture environment IL-15 [29, 30]. Complex, precisely scheduled cytokine
may result in the alteration of product quality [21], a cocktails for culture stimulation can also be used under
concept that acquires a greater dimension, as it be- certain expansion protocols.
comes associated with information on cell state, Through the usage of these stimulation agents, the
phenotype, functionality and identity [22]. expanded cells undergo frequent metabolic changes.
The unpredictable behavior of lymphocytes during cul- They can move into quiescence or active status, start
ture causes noticeable variations in expansion rates amid the division cycle, enter apoptosis or differentiate.
manufacturing [23]. This inherent variability hinders any Knowing what process is triggered in which cells is
comparison between expansion protocols in order to con- important, yet most expansion results just consider
clude and organize best practices. At the core of this issue the overall expansion rate of a given subset of cells.
relies donor heterogeneity as differences related to age, Furthermore, metabolism is not only relevant as a de-
gender, health issues or ethnicity are frequent [24]. Donor scriptor of cell growth. There is a growing body of
variability is also linked to process performance and evidence that shows immune cell metabolism to be
lymphocyte sensitivity to process parameters [25]. Model- essential to cell functionality. For example, glycolysis
ing for process predictability, associated with a thorough and oxidative metabolism have been shown to modu-
characterization of raw materials to compensate for late classical anti-tumor effector functions of NK cells
source’s variability can improve process understanding, [31]. Thus, positive and negative modulation of cer-
accelerating the establishment of new cellular therapies tain metabolic triggers could be used to control
[14]. To make it even more complex, lymphocytes can ex vivo expansion and direct cell functionality. Amino
tune their communication with the environment by modi- acid modulation is another tool that may enhance cell
fying their receptor/ligand repertoire, changing cellular expansion, because some of them, such as glutamine,
sensitiveness to external substances and surfaces [19]. arginine and tryptophan, have been found to influence
These aspects often generate an undesired outcome: when lymphocyte proliferation [32].
subjected to extensive cultivation, cells are prone to de- Summarizing, lymphocytes could be portrayed as deli-
velop phenotypic changes (e.g. differentiation, senescence cate cells requiring very meticulous culturing. Their be-
or immunogenicity) or genetic changes (e.g. mutations, havior can be unpredictable to some extent, because of a
gene deletions or chromosomal aberrations) that can se- combination of factors that include donor and cell popu-
verely undermine their safety and efficacy profiles. There- lation heterogeneity, frequent metabolic changes, high
fore, higher yield due to prolonged expansion often sensitivity to culture environment and strong depend-
correlates with the selection of more proliferative cell sub- ency on an accurate stimulation strategy that mimics
populations, which can be less efficient for their designed typical in vivo conditions. This complexity demands an
function [14]. expansion process that is sensitive and flexible enough
Additionally, immune cells must be stimulated by to compensate for variability. This is offered by various
carefully integrating selection and activation steps during bioreactor systems that were proven to be applied for
the expansion process. There are several technologies lymphoid cultures.
From static cultures to intensified processes yet to be optimally explored with lymphocyte cultures.
Despite of the tight control needed for efficient ACT To profit on these abilities, several bioreactor designs
manufacturing, immune cells are still frequently ex- were already tested for lymphocyte culturing. These dif-
panded in static systems equipped with limited monitor- ferent bioreactor configurations (Fig. 2) are generally
ing capacity [10, 19, 23]. These platforms (plates, flasks suitable for a specific field of ACT (either allogenic or
and bags) depend on incubators and are restricted to a autologous applications). However, as the cultured cells
batch-and-split mode which periodically divides and re- have in principle the same needs, a general set of re-
fills the culture with medium to cope with the cells’ quirements towards maximizing bioreactor capabilities
metabolic activity and stimulation requirements, there- can be formulated, guiding the transition from static cul-
fore these cultures are highly susceptible to contamin- tures to intensified processes.
ation as multiple open vessels are needed to create a
single product [33]. Furthermore, the medium renewal Requirements of bioreactors for lymphocyte
cycles cause frequent nutrient and metabolite fluctua- culture
tions that may trigger high phenotypical variability [19]. Although every cell therapy process has unique ele-
As a result, ACT cells are still manufactured through ments, it is not practical to design specialized devices for
processes and methods that have been characterized as each specific product. Instead, ACT products should be
“archaic, scarcely controlled and incomparable” [34]. Be- grouped on shared process characteristics, defining strat-
cause of their simplicity, cell therapy companies may ini- egies and technologies that fit better for each category as
tiate clinical trials using static systems, requiring further a whole [40]. In that regard, ACT can be performed
assessment as key differences in parameters such as using two general principles: autologous and allogenic.
shear stress, culture conditions, and cell-to-cell interac- In the autologous setting, a batch is individually pro-
tions may cause a divergent biological profile as the cells duced from a patient’s biopsy, isolating and culturing
are moved to a bigger scale dynamic set-up [35]. the cell population of interest. In the allogenic workflow,
Quality testing, which includes complex functionality cell source is a universal donor platform with highly ex-
assays, should be carried in a timely manner, as ACT pandable cells that have similar scale requirements as
products are generally used or preserved briefly after pro- the manufacturing of cell derived proteins and the cell
duction, increasing the risk of uncertainty and therapeutic product may target multiple patients [25]. Process-wise,
mistakes [36]. This implies that Process Analytical Tech- increasing vessel scale and ensuring culture performance
nology (PAT) alone is not able to provide robust informa- (scale-up) is related to allogeneic approaches, while
tion to address most quality questions. Because of that, parallelizing several independent units (scale-out) is gen-
discrete in-process characterization of cell status during erally the goal in optimizing autologous therapy [22]. An
manufacture is generally out of phase with properties con- autologous batch size is not expected to exceed more
tinuously monitored using PAT tools, which are inferen- than a few liters volume, because of the limited amount
tial in nature (e.g. DO, pH, glucose consumption or cell of starting material and the time sensitivity of the cells
density) [37]. However, our comprehension of cell status, to retain their functionality. Thus, scaling up autologous
including metabolomics, clonogenicity and cell cycle regu- is not useful and scaling out for multiple batches still re-
lation is significantly improving [38]. quires a thorough assessment of technical capacities
Most of the bioreactors used for the cultivation of [35]. This delicate setting for autologous cell therapy
therapeutic cells originate from vessels and technology drives bioprocess development towards automation [11,
created for upstreaming bacteria or yeast [14]. However, 25], as the ideal autologous platform should compensate
it is important to note that these systems do not focus for the effects of varying culture conditions on CQA’s
on cell integrity and functionality but on maximizing performance [40]. The allogenic set up, on the other
yield, thus requiring refitting to face the challenge of hand, requires appropriate inoculation levels with min-
generating a healthy and functional cellular product imal seed adaptation to maximize the expansion out-
[38]. Bioreactors allow process scale up with high come. Therefore, the possibility of having a set of vessels
standardization and reproducibility, while enabling the geometrically and dynamically comparable is highly rele-
evaluation of the influence of process parameters on cul- vant [41]. In the same way, achieving consistent process
ture performance [39]. In the same way, process intensi- reproducibility is necessary for a standardized and safe
fication through the implementation of mechanistic allogenic platform, thus, allogenic bioprocess develop-
modeling and PAT tools, along with the use of auto- ment is mostly driven towards process control than
mated culturing techniques, facilitates to reach better workflow automation. To harness a bioreactor’s full po-
control over cell expansion [14] (Fig. 2). tential, its design and application should be fitted to the
A bioreactor’s capability to monitor and control crit- challenges of cultivating lymphocytes and the supple-
ical process parameters is a highly valuable characteristic ments necessary for their growth. These are, in the view
Fig. 2 Main characteristics of static and dynamic culture vessels and their influence on process variability. Typical trends of viable cell density (red
line), nutrient concentration (green line) and dissolved oxygen (blue line) for each culture vessel type
of the authors, and based on previous frameworks of re- GMP compliance

quirements [14, 19, 25, 34, 35, 42], the main standards To avoid cross contamination (between different batches
to be fulfilled by a culturing platform for ACT. or patients) and microbiological contamination, closed
systems (bags, expansion sets, flasks), incubators and
hoods should be used [36]. Bioreactors should guarantee
Suitable vessel size and scalability sterility by keeping a closed system [19]. Each manipula-
Cell-based therapies often require the application of vast tion step (e.g. inoculation, activation, transduction,
quantities of cells (108–1010) to patients therefore the media changes, stimulation, sampling, washing) creates a
space required for their growth is a practical limitation. risk for error and contamination that may lead to a
Assuming a culture density of 106 to 107 cells/mL (a failed run [36]. For that purpose, single-use, closed, dis-
high value for ACT), it would demand a volume starting posable cell production “kits” may represent a desired
from few milliliters up to tens of liters during culture design strategy for patient-specific cell therapy manufac-
[43]. The available bioreactor scale must be flexible turing protocols [44], particularly if such kits can be de-
enough to fully accommodate the range of cell growth signed for simplicity [43].
across all feasible batches, and to compensate for the ex-
pected potential growth variability from the source [25]. Process control
To achieve this, ultra-high cell density cultures or an in- Once the specific requirements for the cells being ex-
dustrial scale production that is able to maintain uni- panded have been defined, process parameters such
form culture conditions are required [39]. Some current as temperature, shear stress, dissolved oxygen (DO)
expansion processes include a preliminary stage where and CO2 and environmental variables like osmolality
cells are activated and rapidly multiplied in static sys- and pH must be kept at optimal values [14]. .Exten-
tems, generating enough cells for bioreactor inoculation. sive, online process monitoring and integrated control
However, enough bioreactor space for the actual expan- is required for adaptation to process changes [45].
sion is still necessary. DO and pH of the medium are typically held
constant to provide a consistent environment sup- mass transfer, suspension of cells and avoidance of het-
porting optimal cell expansion. DO and pH signals, erogeneities that may cause cell inconsistencies [25].
are valuable for assessing the status of the expansion
medium and cell proliferation, triggering a propor- Representative sampling
tional feeding strategy [41], although this is a fairly The designed bioreactor process should stay out of any
limited approach. Some technologies that should be artificial deleterious influences on cell integrity by passa-
considered for ACT process monitoring and control ging and reseeding the cells, as it may decrease total yield
are included in Table 1. The final goal of process [19]. Sampling and harvesting of cells, medium, or both
monitoring should be to find descriptors that can give should be also designed with simplicity in mind. Taking
information about the influence of batch-to-batch or samples has certain drawbacks that need to be mitigated
donor-to-donor variability on the expansion process [35]: to get a representative bioreactor sample, a signifi-
[58]. The best approach for process control develop- cant volume should be drawn, which can impact on yield,
ment would be to use PAT data to facilitate process especially if multiple small scale vessels are used for the
related decisions in real-time, or even predictively. cell expansion. Repeated sampling can also increase the
This can include decision points for transduction, risk of contaminating the bioreactor. Issues that need to
perfusion initiation, harvest point, or even quality be resolved in such cell therapy process development plat-
control release based on minimum viability or endo- forms include deciding on the amount of cells needed to
toxin level. Ideally, such technologies would evolve to reflect heterogeneity and the usage of live cell-based image
measure surface markers expression of key phenotypic analysis and “lab-on-chip” strategies [43].
markers.
Stimulation and supplementation
Media changes in bioreactors are usually done by nutri-
Handling of shear stress ent addition, or by total or partial media replacement, or
Ex vivo expansion of all immune cell types should avoid by perfusion. If a cell culture produces non-damaging
mechanical stress by chaotic, inhomogeneous medium levels of waste products, concentrated levels of nutrients
dynamics [19]. It has been long established that animal can be added over time to feed the growing culture. In-
cells are sensitive to shear, which, above certain levels, evitably, waste metabolites such as lactate and ammonia
compromises their viability. Besides the direct effect that start to accumulate, and either media replacement or
mechanical forces can exert on a cell membrane’s integ- perfusion is required. Perfusion, in which fresh media is
rity, animal cells are adapted to the environment of each gradually fed and old media is removed while the cells
tissue, evolving sensitive mechanisms for detecting shear are retained, is the ideal way to intervene and still main-
changes. To develop an acceptable understanding of tain a stable environment for cell therapy [35]. It also
how these forces influence cell behavior, it is necessary should be noted that cell exhaustion can be induced by
to recreate similar level of shear forces than found in the current activation methods, which generally also demand
body within a bioreactor, allowing for a detailed careful operator attention [32]. Because of that, precise
characterization and control of the mechanotransduction optimization of the feeding of nutrients and cell activa-
process [59] and the direct effects of shear on the cells. tors/stimulants is needed, being able to precisely supply
Importantly, agitation must be designed to manage not them into the culturing medium, allowing for different
only shear exposure of cells, but also the efficiency of feeding profiles.
Table 1 Advanced process monitoring tools for ACT

Tool Application Type Ref.
Raman spectroscopy Metabolite monitoring (glucose, lactate, amino acids). On-line [46]
Total Cell Concentration. On-line [46]
Cell identity determination (phenotype & activation). At-line [47, 48]
Sequential injection capillary electrophoresis Metabolite monitoring (glucose, lactate, amino acids). At-line [49]
Cell concentration. At-line [49]
FT-IR spectroscopy Glucose monitoring. On-line [50]
Electrical impedance Cell-mediated cytotoxicity and cell adhesion. At-line [51–53]
Biosensors for acidification measurement Metabolite monitoring (lactate). On-line [54]
Biosensors – optical Cytokine quantification. Potential [55, 56]
Gas chromatography-mass spectrometry Volatile organic compound (VOC) emissions profiling – metabolic monitoring. On-line [57]
Gas transfer Different reactor configurations may fulfill these re-

Gas transfer happens passively in static systems, which quirements to a varying extent. Given this framework, in
limits oxygen availability in high volume vessels, as the the next chapter we explore the currently available op-
diffusive flux of a gas is inversely proportional to the tions and highlight the most relevant characteristics that
thickness of the liquid that needs to be permeated, ac- stand out from comparison.
cording to Fick’s law and the McMurtrey model of oxy-
gen diffusion [60]. Oxygen transfer may be limited in Comparison of currently available act bioreactor
non-perfused bioreactors because low agitation rates are technologies
required to minimize shear stress on the lymphocytes During the 1980s, the foundational protocols [68–71]
and headspace aeration is also generally preferred for for TILs and LAK therapies were established to be car-
the same reasons. This, on the long run, may hinder the ried out in plates, flasks, bags and roller bottles [72]. At
final expansion output of the system [41]. .Oxygen can the same time, several attempts of culturing lymphocytes
be supplied to a bioreactor either via the headspace or for cytokine production in stirred reactors were being
via a sparger which disperses gas into the medium, how- performed [73–77]. It was Knazek [78], Alter [79] and
ever, sparging has been shown to be possibly detrimental Tanji [80] who in the late 1980s performed the first bio-
for immune cell growth [61]. The physiological oxygen reactor runs intended for cell therapy, using a hollow
concentration is usually lower than the atmospheric. Be- fiber perfusion system. In the 1990s, the use of the
cause of that, establishing culturing protocols that re- hollow fiber technology increased significantly, while
sembles in vivo oxygenation conditions may improve stirred reactors were begun to be used for NK cell ACT
expansion yield and cell functionality [22]. Similarly, the applications [81] and the rotating wall bioreactor was in-
use of CO2 levels representative of the biological fluctu- troduced as a low shear device [82]. Stirred reactors con-
ation threshold could also be beneficial of the process tinued into the 2000s as a solely experimental platform,
outcome. It must be noted that reduced oxygen tension while the rotating wall technology was not used in clin-
results in reduced human T cell proliferation, increased ical applications, focused exclusively in microgravity
intracellular oxidative damage and susceptibility to studies [83–88]. The late 2000s have seen in the usage
apoptosis upon activation, highlighting the importance of the hollow fiber reactor a relative decline compared
of controlling oxygen levels in culture [62]. to the rise in the application of the static culturing G-
Rex device (Wilson Wolf Manufacturing, Saint Paul,
MN) and the dynamic culturing rocking motion reactor.
Physiological congruency Both were quickly adopted into clinical practice, stirring
There is no ideal bioreactor that suits all purposes for the debate of high throughput static vs. dynamic lymph-
all cells, but it should be able to replicate in vitro oid cell culturing. In the late 2010s, the hollow fiber re-
many of the conditions experienced in vivo, therefore actor returned to wider usage thanks to the Quantum
it should allow for experimental testing, mechanical System (Terumo BCT, Tokyo, Japan), and a renewed
conditioning and monitoring of living cells in dy- interest in stirred reactors has been perceived from re-
namic conditions [59]. In a close physiological recent publications [89, 90]. The late 2010s also saw the
membrance, immune cells cultured in bioreactors introduction of the Z RP platform [91] (ZellWerk
often require APCs for stimulation, three-dimensional GmBH, Oberkrämer, Germany) and the Prodigy system
culturing, controlled cell-cell contact and undisturbed [92] (Miltenyi Biotec, Bergish Gladbach, Germany). The
local microenvironments [25]. These needs should be latter is an integrated autologous-targeted platform that,
taken into consideration during the design of suitable despite of its novelty, has been extensively used. There is
devices, starting from the fact that hematopoietic cells also high expectation on the Cocoon system [42] (Lonza,
do not require a surface to grow, being anchorage in- Basel, Switzerland) and rotating wheel reactors [11],
dependent [63]. It is true that cells can be adapted to both announced to be capable of lymphoid cell cultur-
a specific bioreactor design as a replacement to en- ing. Given this historical background, the literature re-
gineering the bioreactor itself, but it must be noted view presented here is based on 117 publications
that this approach may not be available to most cell fulfilling the eligibility criteria, of which 73 contained de-
therapies, as cells may become senescent after a cer- tailed descriptions of the expansion protocols and re-
tain amount of doublings [25]. It should also be sults, categorized in Rocking motion reactors (16
noted that some cells may need to be in extensive results), Hollow fiber systems (18 results), Alternative
contact with each other, such as TILs [64] and T perfusion systems (4 results), stirred reactors (10 re-
cells [65, 66], some of them also tend to form aggre- sults), G-rex-device-based processes (14 results) and
gates that must be controlled for optimal growth [67], Prodigy-system-based processes (11 results). From the
usually by mechanical disruption of the clusters. 71 articles, 29 contained actual comparisons, mainly
between a static protocol and a bioreactor culture with process parameters. The rocking rate may differ ac-
the same stimulation/supplementation strategy. cording to the intended application, usually from 5 to
15 rocks per minute (RPM), and a perfusion strategy
Rocking motion bioreactors is generally used, starting from perfusion volume of
In the rocking motion system, a configurable swinging 250 up to 4000 mL/day. The perfusion begins when a
plate conveys a wave-like oscillation to the contents of a certain threshold is reached either by the decline or
culture bag. The continuous agitation ensures proper increase in metabolites such as glucose, glutamine,
oxygen transfer and medium homogeneity, which may ammonia or lactate (especially TIL and T cell cul-
provide a higher kLa than achievable with a stirred re- tures) or by the increasing cell density. Once the per-
actor, resulting in greater maximal cell densities under fusion is started, the pH and nutrients fluctuate
limited oxygen transfer conditions [93]. The agitation within a narrow range with proven positive effects for
pattern is set by the rocking angle and rate, oscillation TILs and T cells [64, 100], while facilitating glycolysis
sequence and culture volume, which translates into a and glutaminolysis.
specific fluid flow, mixing time, residence time and oxy- Despite of the difficulties to compare the outcome of
gen transfer efficiency. This gentle agitation is consid- different studies, several authors have performed com-
ered to be a low shear method [94], which may cause parative analysis between static set-ups and the condi-
lower cell stress even at increased rocking rates, improv- tions provided by a rocking motion reactor (Table 3). In
ing nutrient and oxygen transfer efficiency and promot- relation to expansion yield, although initially observed as
ing cell growth without exerting detrimental mechanical detrimental for growth [94], it has been shown that
conditions to the culture [95, 96]. In contrast to a static rocking conditions do not induce significant changes in
system, where cells lay closely together, the continuous the total fold of the expansion in case of T cells [66] and
oscillation reduces the time that cells may spend in con- NK cells [65, 97, 104], while boosting the growth of TILs
tact with each other, which may not be optimal for cul- [107, 108], DCs and CIK [97]. However, one study found
tures requiring close cell-to-cell contact, such as TILs no statistical difference in TIL expansion for static bags
[64] and T cells [65, 66], or adherent cells. Because of compared to a rocking motion bioreactor [64], possibly
that, most cultures performed in this reactor include a because of differences in the conditions of media ex-
static phase prior to the transfer to the rocking platform. change [108]. Similarly, non-perfused T cell cultures has
Current rocking motion devices can execute fully au- been found to lose viability as low as 80% by the end of
tomated perfusion cycles, optimizing medium and sup- cultivation [100] because of critical deprivation of stimu-
plements consumption thus, decreasing the overall lants and metabolites. Contrary to stirred cultures, the
process cost. Additionally, perfusion enables cells to be use of shear protectant additives has been explored in
expanded above 107 cells/mL, supporting high volume rocking motion systems, where attempts to expand TIL
cultures to be carried out in a single bag with a signifi- in the absence of a surfactant (Pluronic F68), derived in
cantly reduced volume (some bioreactor cultures need significant cell damage and consequent decrease in cell
about half the amount of media to harvest 1010 cells, as count [64].
compared to static conditions). Since bags are single use Although NK cells’ expansion fold in a bioreactor is
there is no need for cleaning validation, they provide a the same as in a static system, the proportion of NK
ready-to-use closed system decreasing turnaround time cell subpopulations have consistently shown to be
and resource requirements, significantly reducing costs enriched under rocking conditions [65, 94, 97, 104].
in GMP operations [97]. Consequently, this platform is Reactor-generated products contain fewer CD3+ T
frequently used academically and industrially during cells and higher ratio of CD56 + CD3- NK cells than
phase 1 and 2 clinical trials [12]. The system also has in static set-ups, perhaps because T cells could prefer
some disadvantages, including a difficult transition from non-dynamic conditions [65]. In the same way,
research scale to full scale GMP expansions. As it is ne- clinical-scale activated CD56+ cells in a rocking mo-
cessary to purchase ancillary equipment additional to tion reactor have similar phenotype and function as
the bioreactor [64], it has been argued that rocking mo- those derived from static cultures [105]. Unfortu-
tion bioreactors are an ideal solution for scaling the nately, the available studies are not clear about the ef-
manufacture up from 1 L to 1000 L, but do not econom- fect of rocking on cell subpopulations in TIL cultures:
ically scale out from one patient to 1000 patients. the phenotype of TIL and genetically modified PBL
The rocking motion bioreactor has been successfully expanded in static bags and in a rocking motion bio-
used for T, NK, NKT and TILs expansions (Table 2). reactor have been found to differ [64]. However,
Unfortunately, the results of these protocols are not eas- under a different protocol, the numbers of CD4+ and
ily comparable due to differences in cell stimulation CD8+ populations in a TIL culture were reported to
strategy, media composition, starting material and be similar under dynamic and static conditions [107].
Table 2 Summary of culture characteristics with rocking motion bioreactors
Protocol features Starting material Reactor configuration
Author –Year IL-2 Stimulation Medium Serum and Source Seed Bag Culture System Rocks Air Rocking Feeding strategy
[IU/ strategy supplements concentration volume volume per flow Angle [°]
mL] [cells/mL] [L] [mL] minute [L/
min]
T CELLS
Hami [98] - Useda CD3 and Xvivo 15 – CD3+ T – 20 – Wave Bioreactor – – – Perfusion
2004 CD28 beads cells System 20XE
from
PBMC
Tran [66] - 50 Irradiated RPMI 10% FBS; CD4+ T – – Wave 10–12 0.1 4 Perfusion: Media to keep GLN at 2 mM
2007 PBMCs HEPES, GLN cells bioreactor and glucose > 2 g/L.
(APCs) and Glucose from system 2/10 EH
PBMC
Hollyman [99] 100– Preb - CD3/ Xvivo 15 5% AB CD3+ – 2 – WAVE EHT 6–15 – Perfusion: Volume increased over 24 h
Garcia-Aponte et al. Journal of Biological Engineering
- 2009 500 CD28 beads from Bioreactor periods (200–1600 mL/day)

PBMCs
Janas [100] - 20 Pre - CD3/ Xvivo 10 5% HS; GLN T cells 5.0 × 106 – 1000 W25 or W5 Xuri 15 – 6 Perfusion: < 2 × 106 cells/mL–0 mL/day;
2015 [ng/ CD28 beads from cell expansion 2–10 × 106 cells/mL–500 mL/day; 10–
mL] PBMCs system 15 × 106 cells/mL–750 ml/day; > 15 ×
106 cells/mL–1000 mL/day
(2021) 15:13
Vavrova [101] - 20 Pre - Ag- RPMI 5% AB; GLN, T cells – 2 – WAVE 6 – 6 Media fed: maintain cell concentration
2016 mDCs Ne AAs, BME, from bioreactor 2/10 in the 0.5–1 × 106 cells/mL range
(APCs); CD3/ Pyruvate PBMCs system
CD28 beads
O’hanlon [102] 200 Pre PHA IMDM 10% AB; GLN T cells 2.5–5.0 × 105 2 1000 Wave 2/10 (Xuri 5–10 – – Perfusion: Culture days 1–2: 250 mL;
- 2017 and glucose from W5) Culture days 3–5: 500 mL
PBMCs
McCartney 350 – Xuri T 5% AB T cells – 2 1000 Xuri Cell 10 – 6 Perfusion: < 2 × 106 cells/mL–0 mL/day;
[57] - 2019 CEM from Expansion 2–10 × 106 cells/mL–500 mL/day; 10–
PBMCs System 15 × 106 cells/mL–750 ml/day; > 15 ×
Smith [103] - 200– Pre anti Xuri T 5% HS Tcells 1.0 × 106 2 – Xuri Cell 10 – 6 Perfusion: < 2 × 106 cells/mL–0 mL/day;
2019 500 CD3/CD28/ CEM from Expansion 2–10 × 106 cells/mL–500 mL/day; 10–
CD2 PBMCS System W25 15 × 106 cells/mL–750 ml/day; > 15 ×
NK CELLS
Sutlu [94] - 500 OKT3 SCGM 5% HS PMBC 2.0 × 106 – from Wave Bioreactor 6 0.1 6 Media fed: 300 mL/day when 3 e6
2010 800 System 2/10 cells/mL when 7 e6 cells/mL, 500 mL/
day; 1 e7 cells/mL, 750 mL/day; 2.5 e7,
1 L/day.
Spanholtz Useda GM-CSF. G- GBGM 10% HS CD34+ 1.0 × 106 – from WAVE 10 0.1– 6 Media addition to adjust cell density
[104] - 2011 CSF, IL-6, IL- UCB 250 Bioreactor 0.2
7, IL-15 System 2/10
and BIOSTATH
Page 9 of 34
Table 2 Summary of culture characteristics with rocking motion bioreactors (Continued)
Protocol features Starting material Reactor configuration
Author –Year IL-2 Stimulation Medium Serum and Source Seed Bag Culture System Rocks Air Rocking Feeding strategy
[IU/ strategy supplements concentration volume volume per flow Angle [°]
mL] [cells/mL] [L] [mL] minute [L/
min]
CultiBag RM
Rujkijyanont 500 CD56- SCGM 5% AB CD56+ 1.0 × 106 2–20 from WAVE 5–9 – – Media addition to adjust cell density
[105] - 2013 (APCs); IL15 from 200 Bioreactor
OKT3 PBMCs
Lapteva [106] - 500 K562-41BBL- SCGM 10% FBS CD56+ 2.0 × 105 – – WAVE 6 – 6 –
2014 mbIL-15 from Bioreactor
(APCs) PBMCs
Meng [97] - Useda Pre OK-432 Xvivo 15 1% AP PBMC – 3 3000 GE Xuri W25 7 – 6 Perfusion (parameters not specified)
2018
TILs
Sadeghi [107] - 600 PBMCs RPMI 5% AB; GLN TIL 5–10.0 × 107 2 1000 Wave Bioreactor 10 – 6 Perfusion: 350–1000 mL/day to
2011 (APCs); OKT3 12 mM, 25 (TOTAL) System 2/10 maintain glucose and GLN in a range
mM HEPES; (GE) of 1.5–2 g/L and ~ 2–4 mM,
BME respectively.
Somerville 3000 Pre – AIMV 5% AB; 0,02% TIL/PBL – – 1500 WAVE 7 – 6 Perfusion: maintain glucose
(2021) 15:13
[64] - 2012 allogenic pluronic bioreactor 2/10 concentration at ~ 170 mg/dL.

APCs, anti- system
CD3
Donia [108] - 6000 Pre - AIMV Pluronic TIL – 10 WAVE 10 0.2 6 1000–4000 mL/day
2014 allogeneic bioreactor 2/10
APCs, system
antiCD3
a
Used: stimulant was used but the amount was not specified. bPre: the stimulant was added to the culture prior to bioreactor expansion
Page 10 of 34
Table 3 Summary of the results of comparative studies about static and rocking motion cultures
Expansion yield Purity Functionality changes
Author - Static Static Reactor Culture Static Reactor
Year system folda fold days
(vs.)
T cells
Tran [66] - Bags 247– 200–800 14 – > 98% CD4+ Markers of cell activation increased. No detectable Treg cells
2007 1340 produced. Cytokines are produced normally.
NK cells
Sutlu [94] - Bags 530 77 20 31% NK 38% NK; Degranulation and cytotoxic activity are greater in bioreactor
2010 14% NKT cultures.
Spanholtz Bags 759– 1435– 42 71 ± 9% 92% ± 2% Higher expression of activating receptors in bioreactor cultures
[104] - 1770 2657 CD56 + CD56 + 27% degranulation in reactor vs 14–18% in static cultures
2011 CD3- CD3-
Lapteva G-rex No difference 9 Fewer CD3+ T and a Potency is similar (phenotype and in cytotoxicity assays)
[65] - 2014 higher CD56 + CD3- NK
cells in reactor culture
Meng [97] Bags No difference 15 Reactor improves the There is no significant modulation of the cells’ secretome.
- 2018 percentage of NK cells Cytotoxicity is significantly higher for bioreactor cultures.
TIL
Sadeghi Bags 72 ± 11 228.8 ± 14 No difference in CD8+ No difference in Phenotype
[107] - 17.1 and CD4+ percentage
2011
Somerville Bags 1259 ± 1130 ± 14 Lower CD8 and higher Increased IFN-γ release to cognate peptide in reactor culture
[64] -2012 137 127 CD4 in reactor Significant phenotype differences
Donia Bags 1433 ± 5576 ± 14 – > 97% CD3+ –
[108] - 887 1677
2014
a
Fold = Harvested cells / Seeded cells
In addition to the improvement of the proportion into bioreactor bags, and most rocking reactors collect
of target cell subpopulation, the functionality of NK data from single-use DO and pH probes, which can be
cells expanded in rocking motion bioreactors has used, with some limitations, as surrogate measures of
been found superior than in static systems. Cells cul- VCD to decide on perfusion and DO control, eventually
tivated in bioreactors show higher expression of acti- decreasing the frequency of sampling. Alternatively, dif-
vating receptors such as CD314 (NKG2D) and NCRs, ferential digital holography imaging devices allows for
which correlates with a higher degranulation capacity the assessment of cell morphology features and culture
of bioreactor-expanded NK cells (27%) towards K562 characteristics such as cell density, size and viability [11].
cells compared to the 14–18% reached by NK cells in Recently, measurements of cellular downstream volatile
static bag cultures [104]. This higher degranulation organic compound (VOC) emissions were made from
profile was also found in a different study [94], as the the gas exhaust lines in a rocking motion reactor, using
consequential increase in cytotoxicity [97]. T cell cul- Headspace Sorptive Extraction (HSSE) and Stirbar Sorp-
tures in rocking motion bioreactors have shown in- tive Extraction (SBSE) coupled with GC–MS. Unique,
creased expression of cell activation markers as total VOC profiles correlated well to cell densities over
compared to pre-cultures [66]. the course of 8 days. The majority of the relevant VOCs
Because of the versatility and successful application of decreased during cell expansion that opens the possibil-
the rocking motion reactor, several studies have been ity to monitor the nutrients in the media by VOCs and
performed using this platform (Table 4). There have adjust perfusion rates accordingly [57].
been pre-clinical and clinical assays using the rocking
motion technology for chronic lymphocytic leukemia Hollow fiber bioreactors
[98, 99], metastatic melanoma [109, 110] and prostate A perfusion reactor generally uses a semi-permeable
cancer [101]. It has also been successfully used to intro- membrane to separate cells from the medium. With this
duce NMR markers during the expansion process [102]. technique, culture medium continuously refreshes nutri-
In regard to on-line monitoring and control, bio- ents and removes waste metabolites in a system that al-
capacitance probes have been successfully integrated lows specific flow rates on diverse membrane types,
Table 4 Further applications of the rocking motion bioreactor

Author - Type Cell Disease Target Expansion Functional highlights
Year
Hami [98] Pre- T Chronic T cells from Chronic 400 fold in High in vitro activity and T cell receptor repertoire
- 2004 clinical cell lymphocytic lymphocytic leukemia patients 13 days restored after expansion.
leukemia
Hollyman Pre- T Chronic T cells from Chronic 87–668 fold Transduced and expanded T cells were able to
[99] - clinical cell lymphocytic lymphocytic leukemia patients in 13–18 eradicate the tumors in 90% of a mice population;
2009 leukemia days release criteria were met
Andersen Clinical TIL Metastatic Tumor-Infiltrating Lymphocytes 2856–9975 Tumor regression was achieved and associated with a
[109] - Melanoma from Patients with Metastatic fold in 13– higher absolute number of infused tumor-reactive T
2016 Melanoma 36 days cells
Vavrova Pre- T Prostate Prostate cancer reactive T cell 6 fold in Significantly greater cytotoxicity against LNCaP cells
[101] - clinical cell Cancer effectors 8 days after expansion.
2016
Bjoern Pre- TIL Metastatic Effect of Ipilimumab in – Ipilimumab induced marked changes in T cell infiltrates,
[110] - clinical Melanoma metastatic melanoma derived T which can still be detected despite heavy in vitro
2017 cells expansion.
O’hanlon Research T Non-specific 19F labeling for T cells – Cellular viability was maintained; ∼90% of the T cell
[102] - cell preparation was labeled with reagent
2017
making it suitable for continuous cell culture applica- Furthermore, the cells can grow to high concentrations
tions, including monoclonal antibody production [111]. without the metabolites accumulating in the media.
This perfusion principle can be achieved with many dif- Therefore, cells can achieve the required cell-to-cell
ferent membrane systems. However, the most common proximity for optimal expansion in contact demanding
solution is the capillary-based hollow fiber membrane. cultures such as TILs. Additionally, multiple therapeutic
In this system, separation occurs as the medium diffuses cell doses can be harvested from a single hollow fiber
between the intra-capillary (IC) and extra-capillary (EC) cartridge, enabling periodic use of the bioreactor [78,
sides and, depending on the maximum size allowed by 118]. This is also related to the fact that hollow fiber sys-
the membrane’s molecular cut-off, large macromolecules tems are able to support cell growth at densities greater
such as cytokines or antibodies are permanently retained than 108 cells/mL [113]. While a bag cannot handle opti-
on the side where they were originally added [112]. In mally more than 2 × 109 cells, a hollow fiber reactor
that way, only small molecules such as carbohydrates, could handle at least twenty times that amount [119].
amino acids or small peptides can actually diffuse from The possibility of executing cell transfection while the
and into the compartment where the cells are growing expansion is being performed has also been found ad-
(usually in the EC space), while medium circulates vantageous by some authors [120–122] as it combines
within the IC space [113]. The IC space provides large the process of vector concentration and vector-to-target
surface for gas exchange and the cells are not subject to exposure into a single step.
flow therefore they are protected from shear stress [114]. As previously mentioned, the fluid compartmentalization
The independent flows in the IC and EC spaces are gen- of hollow fiber bioreactors is advantageous due to the de-
erated by a set of pumps and valves that direct the fluid creased physical stress on the cells. However, excessive recir-
through the hollow fiber unit. The basal medium passes culation of medium may still lead to negative effects [115].
through a gas exchange module where sensors are usu- Protein build-up in the EC space does not directly lead to
ally placed to monitor parameters such as pH or DO, growth inhibition but its accumulation might also limit the
and sampling systems are allocated for metabolites’ off- convective flow, creating micro-gradients [123]. Meanwhile,
line analysis. The flow in the EC circuit generally runs reduced microenvironment homogeneity caused by the axial
countercurrent to the IC flow, ensuring homogenous and radial concentration gradients is challenging, as gravity
distribution of nutrients [115]. also influences cell distribution within the bioreactor. Some
Perfusion reduces the need for extensive use of culture counteract measures, such as periodical rotation, are usually
vessels and multiple incubators [116]; around 80% de- implemented to prevent the cells from sticking together, dis-
crease in manual labor and incubator space is possible rupting the formation of significant detrimental gradients
[78]. As the bioreactor uses medium equally or even bet- [123]. Harvesting is also not a straightforward process in
ter than regular static systems [117], up to 30–50 L of some configurations, as the detachment and wash out of the
medium that otherwise would be used for static cultur- cells from the tight pores might require frequent
ing [118] may be economically used for perfusion. optimization [24] to retrieve as much cells as possible
without significantly affecting their integrity and viability; this conflicting results in this regard (Table 6). Cell expan-
problem is solved by using a suspension culture configur- sion is usually in the range of 100 to 200 fold after 1 to
ation, that enables automated sampling and harvesting. In 2 weeks of culturing, but newer technologies have
addition, the usually high cell densities attained in the system reached higher than 500 fold after 8 days [111, 112].
can be problematic if an electrical or other mechanical issue These results should be weighed against the seeding
occurs: the cells do not withstand a decrease in temperature characteristics and the specific stimulation strategy. It
or change in pH as steadily as they do when grown at mod- also seems that culture performance cannot be easily
erate densities in bags (around 107 cells/mL). It also predicted based on viability or inoculum density [129] as
must be noted that an entire bioreactor has to be total medium consumption differs between cell donors
harvested to monitor parameters like cytotoxicity, cell as a function of the metabolic activity of their cells [78].
phenotype or cell count [72]. A representative sample Even when the culture performance is highly variable, a
cannot be periodically obtained from the dense cell lag phase (lasting from 1 to 9 days) is generally observed
culture without performing a major intervention in [117, 119, 129], then the glucose consumption and lac-
the system that disrupts the cellular allocation within tate production rate change exponentially reaching a
the fibers while exerts significant shear to remove plateau or peak after some days of culturing [121, 124].
cells from the thin capillaries, where cell populations More pronounced peaks for lactate generation can be
are more representative than the cells caught in re- seen for high seed cultures, followed by a faster decrease
tention filters. Therefore, culture monitoring is based than observed in low seed cultures [112]. The lactate
mainly on physiochemical parameters, or off-line me- production may start to increase right after inoculation
tabolite analysis to the medium effluent, but not on when a static pre-adaptation is performed [120]. Differ-
actual cell samples from the culture. ent patterns of cell-produced cytokine concentrations
Similar to other expansion systems, it is not easy to can also be observed during T cell expansions, as TNFα,
compare the performance of different hollow fiber pro- IL-6, IFNγ and GM-GSF are proportional to the extent
tocols because of differences in the stimulation and cul- of lymphocyte multiplication, which may depend on cell
turing strategies (Table 5). The reactor allows the use of donor and to the formation of microenvironments, hin-
cytokines and other growth enhancing additives in high dering the supply of nutrients and oxygen to some cul-
concentration, while significantly reducing the use of tures [125].
serum. Generally, the membranes have a molecular cut- With respect to product purity, cultures grown in
off of 4 to 17 kDa and a very high total surface for opti- hollow fiber bioreactors have shown to consistently
mal diffusion of metabolites. In a typical culture process, achieve high levels of target cell fraction (Table 7). T
cells are seeded at high densities into the EC space, usu- cells grown from TILs and PBMCs do not have statis-
ally after some days of static culture enrichment. But tically significant differences in their CD4+/CD8+ ra-
there is at least one exception: the recently introduced tios [119, 124], however it has been reported that
Quantum system keeps the cells in the IC space [111, CD8+ T cells prefer to expand in low-seed cultures,
112], although it allows different seeding configurations while CD4+ T cells expand more in high-seed cul-
depending on the cells to be cultured. The perfusion tures [112]. Furthermore, certain shifts in T cell sub-
control strategy is based on the monitoring of the viable populations can be higher in bioreactors compared to
cell density by correlating it with non-automated sam- static cultures [115]. The proportions of CD3+ cells
pling of glucose or lactate concentrations: Glucose may also increase throughout TIL expansion pro-
consumption and lactate generation rates exhibit loga- cesses [119]. In general, the stimulation strategy has a
rithmic behavior, correlating with the cells’ doubling greater impact on the cell differentiation profile than
time [78, 127]. The culture status may be inferred based the culturing platform, because the stimulation proto-
on the glucose uptake rates as it reflects the proportion col is specifically designed to induce a specific pheno-
of metabolically active lymphocytes [78]. Although glu- type and may only be enhanced by the direct contact
cose consumption and lactate production rate have been of the cells with the stimulant, which corresponds to
shown to be closely correlated in lymphoid cultures the nature of the system. In the same way, bag and
[125], it is possible that lactate levels are not a good indi- hollow fiber cultures have shown similar surface anti-
cator of growth inhibiting conditions or nutrient exhaus- gen profiles and cytotoxicity [78] with normal cyto-
tion, as some cultures seem to grow independently of kine production profile [129] and no functional
this metabolite [131]. alteration upon re-stimulation as measured by IFN-γ,
Initial studies have not found a significant difference IL-2 and TNF-α secretion [112]. Reactor grown cells
in the expansion yield of cells cultured in the hollow also preserve the same biological properties as those
fiber bioreactor as compared to the classical static cul- grown in static set ups [115] and T-cell products had
turing methods [78]. However, later experiments shown lower abundance of exhaustion markers [112] when
Table 5 Protocol features in Hollow fiber reactors
Author - Pre Cell culture IC Starting material Culture system
Year stimulaiton
Medium Stimulation Serum and Source Inoculation Volume Reactor Fibers Cut Surface Perfusion Control
supplements [cells/ [mL] off area flow [mL/
reactor] [kDa] [cm2] min]
T cells
Lamers IL2, PHA or AIMV IL2 10% AB Glutamine PBMC 109 NP Immuno*star 10,000 10 NP 50 Control glucose
[124] - CD3 mAb and 4000 at 1.5 g/L
1994 glucose +
RPMI
Lamers IL2, PHA AIMV IL2 NP Glutamine PBMC 1.7–3.1 × 108 100 Immuno*star 10,000 10 NP 50 Control glucose
[125] - and 4000 at 2 g/l; control
1999 glucose + lactate levels
RPMI
Liu [126] - Anti CD3/ CCM IL2 1% AB Medium CD4+ and 2–8 × 107 NP Cellco Cellmax NP NP NP NP Control glucose
1999 CD8 mAb CD8+ from at 50–100 mg/dL
PBMC
Trickett IL2, Anti AIMV CD3/CD28 5 to 2% FBS Medium HIV infected 2–3 107 NP Cellmax Quad NP NP NP 50 Keep glucose
[127] - CD3 or PHA beads or CD4+ cells above 50% of
2002 PHA from PBMC the baseline
value
De Bartolo PHA DMEM NP 10% FBS Medium PBMC 8 × 107 24 PEEK-WC-HF NP NP 128 5 to 10 NP
(2021) 15:13
[114] -
2007
Curcio PHA DMEM PHA 10% FBS Medium PBMC 8 × 107 25 Parallel-HFMBR NP NP 128 2–10 Adjust
[128] - concentrations
2012 to number of
cells
Curcio PHA DMEM PHA 10% FBS Medium PBMC 8 × 107 35 Crossed-HFMBR NP NP NP 2–10 Adjust
[128] - concentrations
2012 to number of
cells
Nankervis IL2, Anti NP IL2; IL7 NP Cells PBMC NP 100 Quantum 1st 11,520 17 NP EC: 100 NP
[111] - CD3/CD8 generation IC: 1
2018 beads + IL7 protocol
Nankervis IL2, Anti NP IL2 NP Cells PBMC NP 100 Quantum 2nd 11,520 17 NP Max. 300 NP
[111] - CD3/CD8 generation
2018 beads protocol
Coeshott IL2, Anti PRIME- NP NP Cells PBMC 3.0–8.5 × 107 124 Quantum 11,520 17 21,000 Max. 300 Remove lactate
[112] - CD3/CD8 XV (IC)
2019 beads
TIL
Knazek IL2, Autol. AIMV IL2 OR LAK Glucose and Medium Melanoma NP 50 Cellmax 100 8000 11,000 40–300n Control glucose
[78] - 1990 LAK supern. supernatant. glutamine at
and serum + HS 1–1.5 g/L
Page 14 of 34
Table 5 Protocol features in Hollow fiber reactors (Continued)
Author - Pre Cell culture IC Starting material Culture system
Year stimulaiton
Medium Stimulation Serum and Source Inoculation Volume Reactor Fibers Cut Surface Perfusion Control
supplements [cells/ [mL] off area flow [mL/
reactor] [kDa] [cm2] min]
Hillman IL2, TNFa AIMV IL2 10% AB RPMI Kidney 5–30 × 108 30 IMMUNO*STAR® NP 10 3000 EC: 2 Control glucose
[115] - tumors 1000 Cell IC: 200
1994 Expander
Freedman IL2 AIMV IL2 NP Medium Ascites/ 109 NP Cellmax 100 8000 4 23,000 60–300 Control glucose
[119] - pleural
1994 effusions/
solid tumors
Lewko IL2, Autol. AIMV IL2 NP Medium Tumor 1–2 × 109 NP Cellmax 100 NP 4 23,000 NP Control glucose
[129] - LAK supern. between 1.0–1.5
1994 and serum g/L
Lewko IL2, Autol. AIMV IL2 NP Medium Tumor 1–2 × 109 NP Cellmax 100 NP NP 23,000 NP Control glucose
[117] - LAK supern. between 1.0–1.5

2000 and serum g/L
Freedman IL2 AIMV IL2 NP Medium Tumor 1–2 × 109 NP Cellmax 100 NP NP NP NP NP
[130] - samples
2000
Malone IL2, OKT3 AIMV IL2 Glutamine Medium Tumor 4–6 × 108 100 Celco - not NP 30– 2200 NP Keep lactate
(2021) 15:13
[118] - samples specified 150 below 1000

2001 units/mL
PBL
Pan [120] - IL2, OKT3 AIMV IL2 5% FBS, Medium PBL 5–9.3 × 107 NP Cellmax Quad NP NP NP NP Controlling
1999 glutamine pump station lactate levels
Shankar OKT3 AIMV IL2 5% FBS, Medium PBL 5 × 107 11.4 Cellmax Artificial NP NP NP NP Keep lactate
[121] - glutamine Capillary below 0.5 mg/
1997 mL
Stroncek OKT3 AIMV IL2 5% FBS, Medium PBL 108 11 Cellmax Artificial NP NP NP NP NP
[122] - glutamine Capillary
1999
NP Not published
Page 15 of 34
Table 6 Performance differences between static and perfusion reactor cultures

Author - Expansion Purity Functionality changes
Year
Fold Days Static Fold Days
control
T cell
Lamers 52.6 ± 21.3 14–17 Bag 238.4 ± 14– CD4/CD8 ratio = 0.51 ± 0.23 vs. CD4/CD8 = NP
[124] - 168.7 17 0.44 ± 0.16 static culture
1994
Trickett 53.2 ± 20.1 7–8 Flasks 71.2 ± 7–8 NP NP
[127] - 42.8
2002
Jones [60] 17.7 fold 9 Flask – 9 Treg phenotype 93.7% for flasks versus Reactor cultures had 8-fold greater
2020 higher 97.7% for reactor. interleukin-10 stimulation index
than static
TIL
Knazek 124–1170 14–32 Bag No NP NP Bag and hollow fiber cultures has similar
[78] - 1990 difference surface-antigen profiles; Cytotoxicity was
similar in both systems
Hillman 20–60 7 Plate 3 7 Shift in the T cell subpopulations is more NP
[115] - pronounced in the bioreactor.
1994
Freedman 30.6 ± 5.6 18.2 ± Plate, 303.1 28.9 CD4/CD8 ratios do not have a statistically NP
[119] - 1.7 flask, significant difference; no difference in
1994 bag proportions of CD16+ and CD56
grown in the Quantum System. However, a reduction first platform employed for lymphoid cells culturing. It
in cytolytic activity at the end of the culture has also was initially used for lymphokine production [73–77], al-
been described [125] and an increase in the concen- though it was later replaced as more efficient techniques
tration of cytokines or growth factors in the medium, were available for cytokine manufacturing. After that,
produced by the PBLs has been proposed as a reason diffusion of stirred bioreactor into cell therapy was slow
for overshadowing any inhibitory effects related to the and mainly circumscribed to small-scale experimental
increased lactate levels [122]. There are also alterna- applications. The spinner flask has been frequently used
tives available for perfusion reactors that were not in that regard, as the simplest stirred vessel, having a
discussed here in details but were used previously for couple of side-arm vents for gas and medium exchange
ACT manufacturing (Table 8). and a central stirrer shaft [24]. This reactor is often used
as the first step to adapt new cell types to stirring [89].
Stirred bioreactors Culturing cells for ACT in stirred reactors is mainly use-
While hollow fiber reactors focus on highly efficient and ful in allogenic therapies, where process scale-up is more
compact cultures, rocking motion systems specialize in important, contrary to scale out primacy with patient-
easily scalable platforms, the stirred reactor, as the most specific applications [11].
widespread and classical bioreactor technology, excels in As hematopoietic cells are relatively sensitive to shear,
tight process control and straightforward scale up due to the mechanical stress induced by impellers has become
easy parametrization, ideal for process intensification. a main concern when using a stirred reactor. In that re-
These bioreactors are characterized by a central agita- gard, higher than 75 rpm has been found detrimental for
tion element, which keeps the medium in motion, some T cell cultures [139], but such low shear rates may
thereby maintains cells and stimulants in suspension and unlikely to produce physical damage to the cells and it is
provides homogeneous distribution of gases and nutri- more plausible that the cells actually respond to the
ents [35]. The vessel’s geometry, the shape of the impel- transduction of fluid-mechanics forces at a molecular
ler and the selected mixing and aeration strategy level [63]. Additionally, a decrease in the rate of prolifer-
influence the culture’s yield and cell surface markers ex- ation has also been observed when gas sparging is used
pression [138]. This translates into a versatile system instead of surface aeration, as rupturing bubbles may
with high process control capability [89], that provides subject the cells to hydrodynamic forces that could affect
an efficient mass transfer of oxygen and nutrients, high the expression of the IL-2R receptor [61]. This receptor
robustness, precise process control and outstanding scal- has been frequently found to be downregulated in cul-
ability. These features enabled stirred reactors to be the tures subjected to stirring conditions [139, 140]. In
Table 7 Performance of non-comparative cultures in hollow fiber reactors

Author - Expansion Purity Functionality changes
Year
Fold Days Media
feed
[L]
T cell
Lamers 41–149 15 NR Predominantly CD3+ and Reduction in cytolytic activity at the end of the culture
[125] - CD8+
1999
Liu [126] - 2 × 105–108 50– NR 95–99% CD4 + CD3+ T cells 50–95% of the cells had elevated expression of HLA-DR.; (IL2R)-a chain
1999 70 with virtual elimination of expression was increased; 40 to 90% of CD25 levels higher than freshly
CD8+ cells isolated CD4+ T cells
Nankervis 117,450 (1st 13 2.2–2.4 90.9–98.8% CD3+ NR
[111] - generation)
2018
439–557 10 10.4– 98.8–99.5% CD3+ NR
(2nd 13.9
generation)
Coeshott 543–1079 8 19.9 91.9–94.5% CD3+; T-cell products had had low frequencies of cells bearing exhaustion
[112] - (high CD4+ expanded preferentially markers
2019 seeding)
951–1787 9 13.6 94.2–97.5% CD3+;
(low CD8+ expanded preferentially
seeding)
TIL
Lewko 17.3 22.3 40 96% T cells based on CD2+ Cells produced cytokines normally
[117, 129] reactivity.
- 1994
PBL
Pan [120] 104–187 11– NR NR 57% transduction frequency
- 1999 12
Shankar ~ 100 10 5.5 NR 1–10% transduction frequency
[121] -
1997
Stroncek ~ 200 17 NR NR < 2.5% transduction frequency
[122] -
1999
addition to the proved downregulation of IL-2R, agita- seeding density is usually below 1 × 106 cells/mL and the
tion could include effects such as changes in gene and culture is kept at a low cell density throughout the dur-
protein expression, disturbances in plasma membrane ation of the expansion, implying a very high final culture
permeability and cell cycle and changes in other intracel- volume to attain clinically relevant cell counts on the
lular signal pathways [141]. Due to the enhanced inter- long run. Cell retention by filters has also been applied
action between the cells and the stimulant agent, for stirred vessels [144] but with no remarkable differ-
demonstrated increase in cell expansion and phenotype ences from non-perfused cultures. DO levels are set into
at high stirring levels with cultures that used stimulation a 5 to 70% wide range. Interestingly, hypoxic conditions
beads for cell activation [89]. Although it has been sug- have frequently been found ideal for cell growth [140,
gested before [61], the use of shear protectant additives 145, 146] as the best cell expansion is usually obtained
has not been investigated yet in stirred reactors. These by culturing at the lowest oxygen tension. This
additives may also prove useful in countering the nega- phenomenon could be explained by the low mean O2
tive effects observed on the IL-2R downregulation. tension in the hematopoietic and lymphoid organ tis-
Stirred reactors have been applied for expanding T sues, that is closer to 40 mmHg (or 5% O2 in the gas at-
and NK cells, although cell-to-cell contact-intensive cul- mosphere), while the anatomical architecture of these
tures have not been successfully executed yet. Protocols organs might expose cells to even lower O2 tensions
are different (Table 9) but there are several common [145]. In a similar manner, maximum T cell growth rate
points. As previously mentioned, almost every protocol has been found to increase at 38.5 °C, although most of
use a low stirring range between 50 to 70 rpm. The the published culturing protocols used 37 °C [140].
Table 8 Alternative perfusion reactors for ACT
Culture Results Characteristics
System Cells Medium Stimulation Perfused Inoculated Days Yield Purity Functionality Features Disadvantages Other
strategy medium cells/ml ACT
Aastrom TIL AIMV, RPMI 1640, Irradiated AIM V, 1% 5 × 106 14 up to 5.8 × Populations Activity against Slow medium exchange Scaling is not UCB
[132] HEPES, BME, 10% AB PBMC APCs; human 109 cells nearly identical HLA-A2+ matched rates maintain a tissue-like available; [133];
serum, 6000 IU/mL 6000 IU/mL serum, (1127 fold) to static; 90% tumor lines; microenvironment limited DC
IL-2 IL2, OKT3 Glutamine CD8+ IFNγ secretion equal opportunity [134]
6000 IU/mL or higher than in for in-process
rhIL-2 static cultures monitoring;
low surface
area
generates low
yield
(2021) 15:13
ZRP NK Alpha medium, Specific NP 70 × 106 12– ~ 14 fold NK cell purity > Cytotoxicity did not Directed laminar flow of Efficiency and Other
[91] glucose, 10% HS, proprietary 22 85%; T cells, B exceed 20% medium, which allows an killing NK
glutamine, 1000 IU/ activating cells and NK T expression of undisturbed cell/cell- and capacity are [19]
mL IL2; proprietary cocktail cells were below activating receptors; cell/surface-contact and questioned
activation cocktail. 2% strong IFNγ minimizes cell stress
expression
NK RPMI1640 complete Irradiated NP 106 14 Static Majority of Static culturing
/γδT medium (10% HS K562- culturing expanded NK/ resulted in higher
/CIK and 100 IU/mL IL2) mb15– resulted in γδT/CIK cells cytotoxicity of NK/
[135] 41BBL cells higher cell developed a γδT/CIK than in
counts than CD56 bright dynamic culturing
Z®-RP phenotype
Packed Tonsil OPTI-MEM; NP OPTI-MEM 107 NP Tissue NP NP Architectural features typical NP DC
Bed tissue 7.5% human AB formation of lymphoid organs [137]
cells serum
[136]
Page 18 of 34
Table 9 Protocol features for stirred systems
Author Protocol Stirred system Expansion Static control Purity Functionality
Stimulation Medium Key Vessel Stir. Aerat. DO Seed Volume Fold Days Fold Days
strategy Supplements speed rate density [ml]
[rpm] [cells/
mL]
T cells
Ou −2019 IL-2;anti OpTmizer Serum-free 2L 70 0.01 70% 5 × 105 800 132– 4 – – > 99% CD4+ High CD3 expression (91.7–
[90] CD3/CD28 CTS Stirred vvm 1011 and CD8+ T 99.4%);
beads or vessel cells T cell co-stimulatory signal-
mABs ing receptor (ICOS/CD278)
elevated; upregulated ex-
pression of PD-1/CD279;
production of IFNγ
decreased by ~ 50%
Costariol – IL-2; anti RPMI 10% FBS; Spinner 35 – – 5 × 105 100 No – – – CD4:CD8 ratio –
2019 [89] CD3/CD28 1640 Glutamine growth decreased from
beads 4:1 to 1:1

ambr 250 100– 14.25 60% 5 × 105 250 23.2 ± 7 15.2 ± 3.1 7 Effector memory cells
200 mL/ 1.3 increased from 35.69 ±
min 10.98% to ~ 80%;
21% no significant difference in T
O2 cell subpopulation profiles
(2021) 15:13
between static and ambr®

250
Ramsborg – IL-2; anti AIM V 2% AP Spinner 60 – – 1.5 × 100 5.4 15 CD3 cells > –
2004 [142] CD3/CD28 105– (0.5– 90%;
5
mABs 3.0 × 10 87) CD8 cells
preferentially
expanded over
CD4 cells
Foster – 2004 PHA or RPMI 10% AB; Spinner 50 – – 5 × 105 100 > 10 7 > 10 7 NLV–tetramer + CD25 decreased
[143] OKT3; 50 IL2; 1640 Glutamine, CD8+ exponentially. This behavior
HEPES, phenotype > did not vary significantly
pyruvate 95% between suspension and
static cultures;CTL
maintained their specificity
toward CMVpp65
Bohnenkamp IL-2; pre anti aMEM 10% FBS Spinner – – – – – 394 9 – – > 94% CD3+ –
– 2002 [140] CD3/CD28 cells;
mABs 1 L stirred – – – 1.35 × – 44 10 Not – CD4:CD8 from IL-2R was downregulated
vessel 105 different 2.4:1 to 1:5 earlier than in the static T-
flask culture;
(stirred vessel),
IFNγ secretion assay against
1:2.7
(suspension a hCMV protein maintained
Stirred – – – 5 × 105 – 60 10 Not – bioreactor) and –
50–550 different 1:4.8 (T-flask)
mL
Page 19 of 34
Table 9 Protocol features for stirred systems (Continued)
Author Protocol Stirred system Expansion Static control Purity Functionality
Stimulation Medium Key Vessel Stir. Aerat. DO Seed Volume Fold Days Fold Days
strategy Supplements speed rate density [ml]
[rpm] [cells/
mL]
Hilbert −2001 IL-2;anti RPMI 10% FBS; 18 L – – – – 18,000 11 10 No > 90% T-cells –
[144] CD3/CD28 1640 OR Glutamine stirred difference
mABs AIMV vessel
550 mL – 20– 3.8 × 105 550 67 5 – – – –
cell 50%
retention
reactor
Carswell – IL-2, pre PHA RPMI 10% FBS; Spinner 60 – 5% – 100 1214 ± 16 – – – –
2000 [145] or antiCD3 1640 Glutamine; 374
pyruvate,
neAAs
NK cells
Pierson – IL-2; (DMEM)/ 10% FBS; Reactor 60 – 40% 106 530 352 33 – – > 90% NK-specific cytolytic function
1996 [81] pre anti- Ham’s F12 ascorbic acid; maintained
CD5/CD8 selenite
Spinner 60 – – 106 250 107 ± 28 43 ± 11 28 86 ± 9,5% 75% specific lysis of K562;
mABs
17 CD56+/CD3-; similar to static
(2021) 15:13
similar to static
Page 20 of 34
Productivity-wise, and probably because of the oper- Culturing platforms specialized on ACT
ational limitations to avoid any possible damage inflicted Besides traditional bioreactors, derived from long estab-
by impellers, T cells cultured in stirred reactors experi- lished bioprocess applications, some expansion technolo-
ence little [89, 140] to insignificant [61, 139, 144] boost- gies were developed to specifically address the
ing in their proliferation as compared to static systems. requirements of autologous and allogenic ACT. These
T cell differentiation have also been found not to be im- platforms aim to either provide a physiological-like en-
pacted by the agitation regime, with a similar phenotype vironment, or to efficiently integrate, from cell acquisi-
to static controls [89]. As with other expansion tech- tion to product formulation, the complex cell therapy
nologies, there is a high expansion variability for cultures workflow into a robust and GMP compliant automated
processed under the same conditions, likely due to raw system. Although these platforms have become available
material variability. Contrary to T cells, stirred bioreac- just in the last decade, they have been extensively and
tors have been found to increase the total NK cell pro- successfully tested. They are already implemented in
duction by 7 fold compared to static cultures [81], clinical practice and cell culture processes, that will be
however the application of this kind of reactor to NK discussed below, are evolved around the devices them-
cells has not been further explored and there is need on selves, hence the different processes are categorized by
additional comparable results to conclude on its poten- the culturing platforms they were performed on.
tial. Similarly, Peripheral Blood Mononuclear Cells
(PBMCs) cultured with 30 rpm stirring speed have Processes with the G-rex flask
shown comparable [147] or superior [148] expansion The G-rex flask is a cylindrical vessel, equipped with a
levels than in static systems. As a comparison with simi- silicon membrane for gas exchange, that enables the
lar cells, cord blood derived hematopoietic stem cells usage of a great amount of medium without requiring
(CB-HSC) have also been found to better expand in mechanical assistance for oxygen transfer [106]. Its
stirred systems than in static culture, when agitated be- geometry allows for a set of linearly scalable vessels with
tween 30 to 40 rpm [149, 150]. They also present a dif- a surface area from 5 to 500 cm2 [154], starting from
ferent expression of genes mainly responsible for permeable six well plates [155, 156], up to 4500 mL
chemotactic activity DNA repair and apoptosis [151]. flasks. The increased medium quantity, usually limited
Stirred reactors were also tested for ex-vivo expansion of by superficial gas diffusion to the cells, supplies nutrients
encapsulated primary human T lymphocytes, but growth and allows waste dilution into a greater volume, while
rates were lower in dynamic conditions [152]. enhancing close cell-to-cell contact [157], however, the
The possibility to develop robust control strategies in final cell density in a G-rex flask is mainly limited by gas
the ACT field would be one of the main advantages of exchange rather than by exhaustion of nutrients [158].
stirred bioreactors; however, little research has been The device includes an automated harvesting unit that
published in this regard. Pierson et al. [81] tested on-line allows to perform the expansion in a fully closed system
laser turbidity measurement that reportedly correlated [10]. The G-rex favors differential expansion of specific
well with cell counts. Recently [46], T-cells cultured in cell subsets, as it allows oxygen-demanding cells to bet-
stirred vessels fitted with Raman probes were used to de- ter survive and proliferate with a more oxidative pheno-
velop chemometric models for glucose (R = 0.987), lac- type and higher levels of mitochondrial activity [159].
tate (R = 0.986), ammonia (R = 0.936), glutamine (R = Furthermore, it could help to rescue certain lymphocyte
0.922), and glutamate (R = 0.829). Univariate Raman lines that can poorly grow in traditional culture devices
modeling for non-targeted analysis of the culture media [159, 160]. Because of the static culture environment, G-
was found useful to track the nutrient depletion (glucose rex bioreactors excel in protocols that use APCs such as
and glutamine) and metabolite production (glutamate TILs and antigen-specific T cells [11].
and lactate), with similar accuracy to the chemometric Compared to static systems, the G-rex decreases the
models. Despite of that, no further research on the amount of manual labor by approximately four times
application of advanced process analytical technology [158] and shortens manufacturing time of some T cell
for stirred cultures in ACT has been done. Manual protocols by half [161]. Even in protocols that couple
sampling, coupled with at-line and off-line measure- transfection and expansion, the materials’ cost is approxi-
ments is routinely performed to measure other mately 38% less than in a bag-based process [162]. As the
process parameters such as cell density, viability, and G-rex can be used for pre-REP operations, traditional
metabolites concentrations [11]. The lack of process flasks can be entirely eliminated from the manufacturing
understanding has prevented the development of suit- process [106]. On the other hand, cell expansion kinetics
able mechanistic models which still have major dis- is affected if the cells are disturbed, hindering process
crepancies between predictions and experimental data sampling [12]. The G-rex flasks cannot incorporate real-
[153]. time visualization of the cell culture because of their
standing nature [13]. Unfortunately, even when the G-rex after stationary intervals that promote cell contact and
process is scalable, there is still a need for using several clustering [188]. Its integrated enrichment and separ-
flasks to get an adequate number of cells for treatment, in- ation functionality has been successfully tested for
creasing cost and workload [33]. However, the versatility CD34+ hematopoietic stem cells [131, 189, 190], T cells
of this device has allowed its application to diverse ACT [191–194] and NK cells [195, 196], decreasing the risk of
fields: the manufacturing of multi-virus-specific [163– contamination and increasing process consistency, while
166], adenovirus-specific [167], cytomegalovirus-specific reducing personnel and processing time [187], behaving
[168] and EBV-specific [169] cytotoxic T-cells for infec- essentially as a “walk away” process for most of the cul-
tion control after hematopoietic cell transplantation. Fur- ture period. However, the Prodigy system was designed
thermore, the GMP manufacturing of NKG2D CAR-T to be fully closed and automated, rather than to
cells for acute myeloid leukemia and multiple myeloma maximize cell expansion. This characteristic is a limit to
[170], gene-edited NK-92 and YTS cell lines expansion use the system for protocols requiring large amounts of
[171] and the generation of TIL for ovarian epithelial can- cells [197]. Thus, if a vast amount of lymphocytes is re-
cer [172] were also made possible. quired for infusion, there will be a need to perform sev-
Most of the applications of the G-rex extensively use feeder eral expansions in multiple devices for a successful
cells (Table 10), profiting on the enhanced close cell-to-cell therapy [198]. It is expected that new tubing sets with
interaction. The platform allows expansion levels above 2000 enhanced culturing volume, or the integration of a
fold for TIL, T cell and NK cell within 2–3 culturing weeks higher scale bioreactor within the platform [198] could
using highly standardized protocols for GMP-grade cell enable a fully automated system that is able to scale
manufacturing. That includes peer-reviewed guides for CTL from an initial expansion phase to a late cultivation stage
from PBMC [182, 183] and UCB [184], NK cells [185], TILs at higher culture volumes.
[186] and CAR-T cells [157]. When comparing its yield, the Most of the protocols developed for the Prodigy sys-
expansion of T cells in the G-rex generates up to 100 [173] tem couple gene transfer and expansion (Table 11),
to 1000 [158] times more cells than classical static culturing reaching transduction efficiencies in the range of 20–
techniques, but there are lower yield exceptions [175, 177]. 30% [200, 201, 203], 50–60% [197, 202, 204] and 80%
In the same way, NK cells and TILs have also shown better [199]. Compared to rocking motion cultures, that re-
or similar expansion performance, but the difference is mod- quire a pre-cultivation phase to generate enough cells, a
erate [106, 159, 180]. Processes in the G-rex flask also lower cell amount is needed to inoculate an NK [198] or
achieve good results regarding cell purity and functionality. T cell culture [200, 201, 205]. The Prodigy-based lymph-
The CD4/CD8 ratio of T cells cultured in the G-rex flask oid cultures have shown similar growth kinetics to static
tend to be preserved [158, 160, 162], while the target cell systems, such as the G-rex [205], but final yield is usu-
purity is generally above 90% [158, 173, 174]. NK cells have ally lower [197, 198]. The maximum increase in cell
also been produced with above 95% purity, even in complex number is generally below 50 fold after 10–13 days of
protocols derived from Umbilical Cord Blood [179]. In TILs, culturing [187, 188, 197, 199, 202, 204] but there are
the Phenotype has been found to be similar to static systems some higher yield exceptions [200, 203]. Despite of this,
[180], with no evidence for cloning selection [159]. over short periods of time, the observed fold expansion
Functionality-wise, T cells grown in the G-rex preserve their is significantly higher than traditional methods, indicat-
cytolytic activity [158, 160, 176] and some research have ing a stable advantage of high yield in brief expansions
found better [177] to similar [162] cytokine expression profile [202]. NK culturing protocols have consistently
compared to static controls. The cytokine production of TIL reached target cell fractions around 99% [188, 198]
grown in G-rex flasks is similar to TIL produced with 24- and T cells have been produced at around 95% purity
well plates, T-175 flasks, and bags [180]. [187, 199, 203, 205]. The phenotype profiles of auto-
matically and manually expanded cells have been
Processes with the prodigy system found to be similar in NK [198] and T cells [197].
The Prodigy® system integrates cell washing, separation, Gene expression analysis has shown just slight diver-
enrichment and expansion into a fully automated GMP gences between NK cells expanded manually or
compliant workflow. Prodigy’s culturing function is exe- through automation and a similar IFN-γ expression in
cuted by a centrifugation chamber equipped with a automated and manual NK cultures have been re-
microscope camera [92]. The system is able to control ported [198]. However, IFN-γ has also been found to
temperature, DO, pCO2 and it can exchange media be decreased in Prodigy-based T cell cultures com-
while performing detailed stimulation protocols using a pared to static protocols [187, 197]. Despite this, the
single set of tubing [92, 187]. The expansion chamber cytotoxicity of products from a Prodigy system have
switches from static to dynamic culturing, applying short been generally found to be compliant for their clinical
centrifugation pulses in order to gently mix the cells application [187, 188, 197, 201–204].
Table 10 Protocol features for G-rex systems
Protocol features Starting material Expansion Comparator Purity Functionality changes
Author (ref) - IL-2 Stimulation Medium Suppl. Source Seed Vol. Culture Fold Days System Fold days Target Cytotoxicity
year [IU/ [cells [mL] system
mL] /mL]
T cell
Vera [158] - 50 Irr. EBV-LCLs RPMI; CM 10% FBS; glutamine T cells from 5× 30 G-rex40 1700– 23 plates 3–5 23 No change in Cytolytic activity
2010 PBMC 105 2200 phenotype; > maintained, killing vs
90% CD3+ cells EBV-LCL 62% ± 12 vs
(96.7 ± 1.7 vs 57% ± 8;
92.8 ± 5.6; G-rex G-rex vs 24-well plate
vs 24-well),
CD8+ (62.2% ±
38.3 vs 75% ±
21.7).
Chakraborty 50 Anti CD3/ RPMI 10% AB, glutamine nTregs from NP 400 G- > 600 21 plates 5–6 21 Percentage of FoxP3+ cells increased
[173] - 2013 CD28, BME PMBC rex100 CD4 + CD25+ from day 1 to day 21 of
temsirolimus, remained stable culture. No compromise

Irr. APCs at 92 ± 5% by in telomere length.
day 21
Gerdemann NP Irr. RPMI; CM 10% FBS; glutamine rCTLs from 2× 30 G-rex10 7–28 9–11 NP NP NP CD3+ T cells CTL lines specific for
[174] - 2013 nucleofected PBMC 107 (mean 98.6 ± EBV, CMV, and Adv
DCs, IL4 IL7 0.1%), CD8+ antigens but not
(2021) 15:13
(59.6 ± 2.7%) alloreactive.

CD4+ (34.1 ±
2.5%)
Ramanayake 50 Irr. AIMV 10% AB OR FBS T cells from 107 NP G-rex10 680.4– 22 plates 326.8– 22 57% CAR NP
[175] - 2015 Autologous PBMC 765.4 576.0 expression
PBMCs or fold fold (range 50–63%)
Nalm-6 cells; in G-rex10 vs.
IL15 66% (35–78%) in
plates; CD3+ in
G-rex10 88% vs.
96% in plates
Orio [160] - NP IL-4, IL-7, IL- RPMI; CM 10% AP, glutamine T cells from 2× NP G-rex10 9.2 12– NP NP NP Balanced CD4/ Antigen-specific, allo-
2015 21 PBMC 107 14 CD8 T-cell ratio, tolerant, anti-EBV T-cell
slight CD8 lines
predominance.
Jin [176] - 2018 3000 Irr. allogeneic AIMV 5% AB, glutamine T cells from 107 800– G- 1890 ± 21 NP NP NP CD4+ T-cells Transduced T-cells lyse
PBMCs; anti- PBMC 4500 rex500 138 were favored CASKi cells (48 ± 4%)
CD3 MCS transduction effi- and 293 cells expressing
ciency of 74 ± E6 (72 ± 1%). E6 and E7
10% and 93 ± TCR transduced T-cells
1.4% produced IFN-γ and
TNF-α.
Kuranda [177] - Used GM-CSF, IL-4, AIMV 10% HS Ag or AdV5- 107 NP G-rex10 11 fold 10 NP NP NP NP More CD8+ T cells
2019 IL-1b, Flt3L, reactive CD8+ vs. std. producing TNF-α, IL-2,
TLR8L, TNF-a, from PBMC or IFN-γ, and/or MIP-1b for
Page 23 of 34
Table 10 Protocol features for G-rex systems (Continued)
mL] /mL]
PGE2, IL-7, IL- CB G-rex cultures. IFN-γ,
2, IL-15, IL-7. and MIP-1b were un-
detectable in plates.
Gagliardi [162] Viral supern. TexMACS 10% FBS Transd. T cells NP 1000 G- Higher Bags No difference in Transduction in G-rex
- 2019 and GMP from PBMC rex100 than CD45RA and 55 ± 7%, vs. 73 ± 7% in
Vectofusin-1; MCS bags CD62 L cell bag. More viable and
IL7, IL15, expression or transgenic cells from G
antiCD3/ CD4:CD8 ratio rex. Secretion of cyto-
CD28 kines did not differ.
Xiao [178]- 300 Zometa; Irr AIMV 5% AB Vγ9Vδ2T cells NP NP G-rex10 10, 17 NP NP NP The Vγ9Vδ2T-cell G-rex promoted
2018 K562 Clone A from PBMC & 100 995 ± purity achieved phenotypic changes:
APCs; OKT3 3078 was 85.98% ± TEFF cell decreased
10.28%. from 66.6–51%; TEM cell

The NK increased from 21 to
population 37.6%. around 40–80%
decreased from target cell killing.
20 to 4.6%
NK cell
(2021) 15:13
Lapteva [106] - 10 Irr. K562- SCGM 10% FBS; glucose CD56+ CD3- 2× 400 G- 442 ± 10 Bags 227 ± 10 Less than 35% NP
2012 mbIL15- NK cells from 106 rex100 29 91 proliferation of T
41BBL cells PBMC cells was
detected during
the NK cell
expansion.
Shah [179] - 100 Irr. APC cells RPMI; CM 10% AB NK from NP NP GP500 2389 14 NP NP NP 95% purity for Significant in vivo
2013 Cryopreserved NK cells (CD56+/ activity against MM in a
CB CD32) and less xenogeneic mouse
than 1% CD3+ model
cells.
TIL
Jin [180] - 2012 3000 Irr. allogeneic AIMV 5% AB Tumor 5× 400 G- 176 + 7 T175 167 ± 7 Phenotype IFN-γ similar to T-175
PBMC, anti- fragments or 106 rex100 136 37 similar to TIL flasks and bags (Table
CD3 tumor digest produced with 4).
cells 24-well plates, T-
175 flasks, and
bags.
Forget [181] - Used Anti CD3; Irr. RPMI 10% AB, Melanoma 5× NP G- > 2000 NP NP NP NP Mostly CD3+ Decreased CD28
2014 allogeneic glutaminepyruvate, tumors 106 rex100 cells for both expression in T-flask sys-
PBMC OR HEPES, BME M conditions tem was not observed
Aapc when using the G-rex
flasks
Forget [159] - Used anti-CD3, Irr. NP Melanoma 5× 400 G- 2653 14 Plate + 1210 14 G-rex did not No Va or Vb expression
Page 24 of 34
Table 10 Protocol features for G-rex systems (Continued)
mL] /mL]
2016 allogeneic tumors 106 rex100 bag favor clonal was lost with any
PBMC M selection system. TCR diversity
was not altered. The
OCR of the TIL in G-rex
almost tripled, demon-
strating an enhanced
mitochondrial capacity
(2021) 15:13
Page 25 of 34
Table 11 Protocol features for Prodigy systems
Authors Protocol features Starting material Culturing Expansion Purity Functionality
IL-2 Stimmul. Medium Serum Source Seed Vol. Tubing Shaking Fold Days Compared Comp.
[IU/ strategy [cell/ [mL] Set system Fold
mL] ml]
T cells
Mock [187] - Used CD3/CD28; TexMACS 3% HS T cells 7.45– 100 TS520 Shaker from 16.2 ± 8–10 G-rex 10 22.3 ± No difference with G-rex; Exhaustion profile did not
2016 Transd. GMP from 10 × day 4/5. More 7.9 12.2 98.0% max T cell purity differ. Prodigy cells
PBMC 108 vigorous produced less IFN-γ. Secre-
shaking tion of TNF-α and IL-2 was
depending on lower but not significant
cell density specific lysis to target
Priesner [199] - Not CD3/CD28; TexMACS 3% HS CD62L+ 3× NP TS520 Static culture 13–23 10– Not used 83.4–98.9% CD3 + CD45+ Transduction efficiency of
2016 used Transd.; IL7; GMP cells 109 until day 3; (cells); 13 T cells 83%
IL15 from total clusters were 28–42
PBMC dispersed (Tcells)
Lock [200] - Not CD3/ TexMACS 3% HS CD4 + 2– 250 TS520 Automated 65 ± 12 Not used 91.3–5.0% for Healthy Transduct.: 34.5–11.7% for
2017 used CD28(transact); GMP CD8 + 10 × media 36 donor and 88.3–7.1% for HD and 36.4–17.7% for PM.
Transd.; IL7; CD45+ 107 exchange Melanoma Patient Secretion of GM-CSF, IFN-γ,
IL15 from total every day sourced IL-2, and TNF-α, no IL-4, IL-
PBMC 5, or IL-10 detected.
Blaeschke [201] Not CD3/ TexMACS 3% HS CD4+/ 2.86 × NP TS520 NP 14.7– 12 Not used 50.3% CD4+ and 38.7% Transduction: 26.95%; No
- 2018 used CD28(transact); GMP CD8+ 107 102.4 CD8+ cells; 10.2% of the increased expression of
(2021) 15:13
Transd.; IL7; from total cells were NKT cells. exhaustion markers. CD19
IL15 PBMC CAR T cells killed 80% of
the target (5:1); cells were
able to secrete GM-CSF,
IFN-γ, IL-2, and TNF-α
Zhang [202] - 200 CD3/ TexMACS CD4+/ 108 250 TS520 Media 16 Not used CD4+ and CD8+ T cells Transduction 60% in CD4+
2018 CD28(transact); GMP CD8+ total exchanges (cells); changed from 45 and and 50% in CD8+ T cells.
Transd. from performed 20 (T 34% to 22 and 74%.The Cells produced significant
PBMC without cells) frequency of CD4+ Tregs amounts of Th1 cytokine,
disrupting the decreased IFN-γ and IL-2 after stimu-
cells lation. Cytotoxicity ~ 60%
at a 3:1 effector-to-target
ratio
Zhu [197] - 2018 Used CD3/ TexMACS 3% HS CD4+/ 108 NP TS520 As medium is 24.5– 13 G-rex100 Signific. CD4+ decreased; CD8+ Low levels of IFN-γ. IFN-γ;
CD28(transact); GMP CD8+ total added or 41.0 better increased. Treg all CAR-T-cell products
Transduct. from exchanged, it decreased. Little lysed Raji cells when
PBMC is difference between tested; transduction: 21 to
programmed Prodigy and G-rex in re- 56.6%. better transduction
to shake gard to phenotype in Prodigy.
Aleksandrova Not CD3/ TexMACS HS CD4+/ NR NP TS520 NP 46–81 12 Not used 97% T cells (range 96– Transduction: 22% on day
[203] - 2019 used CD28(transact); GMP CD8+ 99); some impurities of 5 (range 17–41%) and 23%
Transd.; IL7; from NKT and NK cells: median in the final product (range
IL15 PBMC 2.9 and 0.07%, 21–45%).; CAR T cells were
respectively. CD4/CD8 also cytotoxic against
ratio of 2.4 decreased to target cells
1.7 (range 1.1–2.3) in the
final product.
Page 26 of 34
Table 11 Protocol features for Prodigy systems (Continued)
Authors Protocol features Starting material Culturing Expansion Purity Functionality
IL-2 Stimmul. Medium Serum Source Seed Vol. Tubing Shaking Fold Days Compared Comp.
[IU/ strategy [cell/ [mL] Set system Fold
mL] ml]
Fernández [204] 100 CD3/ TexMACS CD45RA 1× NP TS520 NP 13.4– 10– Not used Enrichment in CD4+ vs. Cytotoxicity ≥20%. The
- 2019 CD28(transact); GMP − from 108 38.6 13 CD8+ T cells. Effector expression of NKG2D
Transd. PBMC total memory (TEM) increased during the
phenotype expansion of cells. NKG2D
CAR expression 55–60.6%
Marín [205] - 1000 T reg beads TexMACS 5%AB CD4 + Mean: 250 TS510 Shaked from Similar kinetics G-rex G-rex 95.6% (range 93.7– Suppressive capacity varied
2019 (aCD3/aCD8); GMP CD25− 56.9 × day 5; media 97.0; n = 4). Prodigy R between donors but
Rapamycin from 106 exchange was 92.8% (range 89.8–94.5%, appeared to be
PBMC conducted. n = 4; p = 0.08). independent of the
Agitation manufacturing regimen
paused to
facilitate bead
to cell contact
NK cells
Granzin [198] - 500 Irr. EBV-LCL TexMACS 5% AB CD56 + 2× 210 TS730 Shaking after 850 ± 14 T75 1344 ± > 99% CD3e/CD56þ,T or No differences in the
2015 GMP CD3- 106 day 7 509 1135 B cells could not be production of IFN-γ and
from detected. TNF-α and similar levels of
PBMC degranulation. No reduc-
tion in telomere length.
(2021) 15:13
T75 and Prodigy had a

comparable marker profile.
NKG2C, CX3CR1 and
KIR2DL2/DL3 higher in
Prodigy
Oberschmidt 500 IL-21, IL-15; NK NK MACS 5%AB CD56 + 2–4 × 250 TS310 0.17–0.3 g 11.31– 3–14 Not used 98.8–99.2% Total NK cells expressed
[188] - 2019 MACS CD3- 106 shaking mode 17.94 initially low to moderate
supplement from was activated levels of NKG2D, NKp30,
PBMC on day 7. NKp44, and NKp46. The
surface markers increased
during the 14-day expan-
sion. Death receptors and
activation markers in-
creased. NK cell cytotox-
icity increased
Page 27 of 34
Conclusions possibility of process intensification through quality

The goal of this review was to describe and compare dif- driven approaches and the capacity of harnessing incre-
ferent bioreactors that are available for lymphoid cell ex- mental knowledge from ACT manufacturing in a sys-
pansion, summarizing their design features and overall tematic way, compensating for source variability and cell
applicability to produce ACT products. Process yield, complexity. Furthermore, the bioreactors’ ability to pre-
purity and product functionality were compared to over- cisely control process parameters, mimicking in vivo
all expansion results in the context of expansion proto- conditions better than in static cultures, could be benefi-
col diversity across different devices. The dependency on cial for product quality.
a carefully tuned stimulation strategy, high sensitivity to From the user perspective, the main challenges are
initial conditions and process parameters, translate into found in two directions: augment ACT scale and improve
high unpredictability of the cultivation process. Reactors process economics. That means increasing predictability
are flexible enough to address various ACT challenges, of critical process stages such as stimulation schedule,
but also limited for some specific lymphocyte culture ap- feeding/splitting scheme, in-process testing or point of
plications (Fig. 3) Lymphocytes are still frequently ex- harvest. During the transition to phase II/III clinical trials
panded totally or partly in conventional, static culture and if cellular therapies for high impact cancers (e.g. lung
flasks or similar vessels. That is due to the fact that most and pancreatic cancer) prove to be successful, a large pro-
immune cell types can be grown using this simple and duction scale has to be considered, needing thousands of
cheap approach without special equipment, but also be- cell therapeutic doses per year. To get a stable manufac-
cause those methods fulfill their purpose and, in some turing pipeline for the large-scale needs, tight process con-
instances, have been shown to be even more successful trol through continuous monitoring in bioreactors must
than current bioreactors. The key difference between be already extensively used in routine ACT expansion,
static and dynamic bioreactor cultures resides in the which is yet to be achieved. This will be instrumental in
Fig. 3 Summary of the features of the compared bioreactors. * compared to static cultures in bags or flasks
parameter targeting for optimization, contributing to- Authors’ contributions

wards process efficiency, increasing the accessibility of OG screened the literature about the usage of bioreactors for cell therapy
expansion and prepared the draft of the manuscript. CH provided
therapies to patients. supervision and scope related guidance. BK was a major contributor in
From the research perspective, the efforts in the devel- formulating and structuring the document. All authors read and approved
opment of ACT have been mostly product-oriented, with- the final manuscript.
out thoroughly considering the importance of the Funding

production process itself [14]. Bioreactors with associated The authors declare that they have not received any specific funding for the
computational modeling and process control will be of manuscript.
great benefit for understanding the mechanisms in which
Availability of data and materials
process parameters interact with raw material attributes All data generated or analyzed during this study are included in this
and the selected stimulation strategy. In that way, cellular published article.
metabolic profiles would also provide an additional
Declarations
phenotypic information that can be used to guide cell fate
decisions directing the expansion of preferred cell subpop- Ethics approval and consent to participate
ulations in an automated fashion. This aspect of process Not Applicable.
automation can be used to remove sampling requirements Consent for publication

and operator input on run conditions, thereby producing Not Applicable.
a more consistent, metabolically driven control scheme.
Competing interests
The years to come will be framed by the transition to-
The authors declare that they have no competing interests.
wards a process oriented and data intensive ACT para-
digm, that should translate current biological Received: 16 December 2020 Accepted: 29 March 2021
understanding into a digitally driven predictive manufac-
ture approach, regardless of the design characteristics of References
the culturing platform. Every bioreactor layout will still 1. Hegde PS, Chen DS. Top 10 challenges in Cancer immunotherapy.
define a specific niche within the immune cell therapeu- Immunity. 2020;52(1):17–35.
2. Yang F, Jin H, Wang J, Sun Q, Yan C, Wei F, et al. Adoptive cellular therapy
tics, but only those that able to utilize batch-to-batch (ACT) for cancer treatment. Adv Exp Med Biol. 2016;909:169–239.
process knowledge will remain competitive, as ACT is 3. Klaver Y, Kunert A, Sleijfer S, Debets R, Lamers CHJ. Adoptive T-cell therapy:
made accessible to a wider range of patients demanding a need for standard immune monitoring. Immunotherapy. 2015;7(5):513–33.
4. Rosenberg SA, Restifo NP. Adoptive cell transfer as personalized
higher flexibility and treatment opportunity. immunotherapy for human cancer. Science. 2015;348(6230):62–8.
5. Rohaan MW, Wilgenhof S, Haanen JBAG. Adoptive cellular therapies: the
current landscape. Virchows Arch Int J Pathol. 2019;474(4):449–61.
Abbreviations 6. da Silva JL, Dos Santos ALS, Nunes NCC, de Moraes Lino da Silva F, Ferreira
AB: Pooled human AB Serum; ACT: Adoptive Cell Therapy; Ag- CGM, de Melo AC. Cancer immunotherapy: the art of targeting the tumor
mDCs: Autologous LNCaP-loaded mature DCs; AP: Autologous Plasma; immune microenvironment. Cancer Chemother Pharmacol. 2019;84(2):227–40.
APCs: Antigen Presenting Cells; ATMPs: Advanced Therapy Medicinal 7. Dai X, Mei Y, Nie J, Bai Z. Scaling up the manufacturing process of adoptive
Products; BME: ß-mercaptoethanol; CAR: Chimeric Antigen Receptor; T cell immunotherapy. Biotechnol J. 2019;14(4):e1800239.
CIK: Cytokine-Induced Killer; CMV: Cytomegalovirus; CQAs: Critical Quality 8. Brenner MK. Adoptive cell therapy: ACT-up or ACT-out? Mol Ther J Am Soc
Attributes; CTLs: Cytotoxic T Lymphocytes; DC: Dendritic Cells; Gene Ther. 2019;27(4):693–4.
DMEM: Dulbecco’s Modified Eagle’s Medium; DO: Dissolved Oxygen; 9. Abbasalizadeh S, Pakzad M, Cabral JMS, Baharvand H. Allogeneic cell
EBV: Epstein-Barr virus; EC: Extra-Capillary; FBS: Fetal Bovine Serum; GC– therapy manufacturing: process development technologies and facility
MS: Gas chromatography–mass spectrometry; G-CSF: Granulocyte colony- design options. Expert Opin Biol Ther. 2017;17(10):1201–19.
stimulating factor; GLN: Glutamine; GM-CSF: Granulocyte-macrophage 10. Highfill SL, Stroncek DF. Overcoming challenges in process development of
Colony-stimulating Factor; GMP: Good Manufacturing Practices; HLA- cellular therapies. Curr Hematol Malig Rep. 2019;14(4):269–77.
DR: Human Leukocyte Antigen – DR isotype complex; HS: Human Serum; 11. Iyer RK, Bowles PA, Kim H, Dulgar-Tulloch A. Industrializing autologous
HSC: Hematopoietic Stem Cells; HSSE: Headspace Sorptive Extraction; adoptive immunotherapies: manufacturing advances and challenges. Front
IC: Intra-Capillary; IFN-γ: Interferon Gamma; IMDM: Iscove’s Modified Med (Lausanne). 2018;5:150
Dulbecco’s Medium; Irr.: Irradiated; LAK: Lymphokine-Activated Killer; 12. Wang X, Rivière I. Clinical manufacturing of CAR T cells: foundation of a
LCL: Lymphoblastoid Cell Lines; mAb: Monoclonal Antibody; NCRs: Natural promising therapy. Mol Ther - Oncolytics. 2016;3:16015.
Cytotoxic Receptors; NeAAs: Non-Essential Amino Acids; NK: Natural Killer; 13. Dwarshuis NJ, Parratt K, Santiago-Miranda A, Roy K. Cells as advanced
NMR: Nuclear Magnetic Resonance; OCR: Oxygen Consumption Rate; therapeutics: state-of-the-art, challenges, and opportunities in large scale
PAT: Process Analytical Technology; PBL: Peripheral Blood Lymphocytes; biomanufacturing of high-quality cells for adoptive immunotherapies. Adv
PBMCs: Peripheral Blood Mononuclear Cells; PCA: Principal Component Drug Deliv Rev. 2017;114:222–39.
Analysis; PHA: Phytohaemagglutinin; QbD: Quality by Design; REP: Rapid 14. Scibona E, Morbidelli M. Expansion processes for cell-based therapies.
Expansion Protocol; RPMI: Roswell Park Memorial Institute Medium; Biotechnol Adv. 2019;37(8):107455.
SBSE: Stirbar Sorptive Extraction; SCGM: Stem Cell Growth Medium; TCR: T- 15. Geukes Foppen MH, Donia M, Svane IM, Haanen JB. a. G. tumor-infiltrating
Cell Receptor; TEFF: Effector T Cells; TILs: Tumor-Infiltrating Lymphocytes; lymphocytes for the treatment of metastatic cancer. Mol Oncol. 2015;9(10):
TNFa: Tumor Necrosis Factor Alpha; Treg: Regulatory T-Cell; UCB: Umbilical 1918–35.
Cord Blood; VCD: Viable Cell Density; VOC: Volatile Organic Compound 16. Brindley DA, French AL, Baptista R, Timmins N, Adams T, Wall I, et al. Cell
therapy bioprocessing technologies and indicators of technological
convergence. BioProcess Int. 2014;12(3 SUPPL):14–21.
Acknowledgements 17. Naing MW, Williams DJ. Three-dimensional culture and bioreactors for
Not applicable. cellular therapies. Cytotherapy. 2011;13(4):391–9.
18. Fischbach MA, Bluestone JA, Lim WA. Cell-based therapeutics: the next pillar 43. Kirouac DC, Zandstra PW. The systematic production of cells for cell
of medicine. Sci Transl Med. 2013;5(179):179ps7. therapies. Cell Stem Cell. 2008;3(4):369–81.
19. Pörtner R, Sebald C, Parida SK, Hoffmeister H. Single-use bioreactors for 44. Pollard D, Kistler C. Disposable bioreactors. In: Current developments in
manufacturing of immune cell therapeutics. In: Single-use Technology in biotechnology and bioengineering: bioprocesses, bioreactors and controls;
Biopharmaceutical Manufacture: Wiley; 2019. p. 327–34. Available from: 2017. p. 353–79.
https://onlinelibrary.wiley.com/doi/abs/10.1002/9781119477891.ch30. Cited 45. Cierpka K, Elseberg CL, Niss K, Kassem M, Salzig D, Czermak P. hMSC
2019 Nov 26. production in disposable bioreactors with regards to GMP and PAT. Chem
20. MacPherson S, Kilgour M, Lum JJ. Understanding lymphocyte metabolism Ing Tech. 2013;85(1–2):67–75.
for use in cancer immunotherapy. FEBS J. 2018;285(14):2567–78. 46. Baradez M-O, Biziato D, Hassan E, Marshall D. Application of Raman
21. Morrow D, Ussi A, Migliaccio G. Addressing pressing needs in the spectroscopy and Univariate Modelling as a process analytical Technology
development of advanced therapies. Front Bioeng Biotechnol. 2017;5:55. for Cell Therapy Bioprocessing. Front Med. 2018;5:47.
22. de Almeida FM, de Matos Branco AD, Fernandes-Platzgummer A, da Silva 47. Chen M, McReynolds N, Campbell EC, Mazilu M, Barbosa J, Dholakia K, et al.
CL, Cabral JMS. Addressing the manufacturing challenges of cell-based The use of wavelength modulated Raman spectroscopy in label-free
therapies. Adv Biochem Eng Biotechnol. 2020;171:225–78. identification of T lymphocyte subsets, natural killer cells and dendritic cells.
23. Vormittag P, Gunn R, Ghorashian S, Veraitch FS. A guide to manufacturing PLoS One. 2015;10(5):e0125158.
CAR T cell therapies. Curr Opin Biotechnol. 2018;53:164–81. 48. Hobro AJ, Kumagai Y, Akira S, Smith NI. Raman spectroscopy as a tool for
24. Roh K-H, Nerem RM, Roy K. Biomanufacturing of therapeutic cells: state of label-free lymphocyte cell line discrimination. Analyst. 2016;141(12):3756–64.
the art, current challenges, and future perspectives. Annu Rev Chem Biomol 49. Alhusban AA, Breadmore MC, Gueven N, Guijt RM. Capillary electrophoresis
Eng. 2016;7:455–78. for automated on-line monitoring of suspension cultures: correlating cell
25. Eaker S, Abraham E, Allickson J, Brieva TA, Baksh D, Heathman TRJ, et al. density, nutrients and metabolites in near real-time. Anal Chim Acta. 2016;
Bioreactors for cell therapies: current status and future advances. 920:94–101.
Cytotherapy. 2017;19(1):9–18. 50. Jung B, Lee S, Yang IH, Good T, Cote GL. Automated on-line noninvasive
26. Kuznetsov SA, Mankani MH, Robey PG. Effect of serum on human bone optical glucose monitoring in a cell culture system. Appl Spectrosc. 2002;
marrow stromal cells: ex vivo expansion and in vivo bone formation. 56(1):51–7.
Transplantation. 2000;70(12):1780–7. 51. Cerignoli F, Abassi YA, Lamarche BJ, Guenther G, Ana DS, Guimet D, et al. In
27. Selvaggi TA, Walker RE, Fleisher TA. Development of antibodies to fetal calf vitro immunotherapy potency assays using real-time cell analysis. PLoS One.
serum with arthus-like reactions in human immunodeficiency virus-infected 2018;13(3):e0193498
patients given syngeneic lymphocyte infusions. Blood. 1997;89(3):776–9. 52. Oberg H-H, Peters C, Kabelitz D, Wesch D. Real-time cell analysis (RTCA) to
28. Stock S, Schmitt M, Sellner L. Optimizing manufacturing protocols of measure killer cell activity against adherent tumor cells in vitro. Methods
chimeric antigen receptor T cells for improved anticancer immunotherapy. Enzymol. 2020;631:429–41.
Int J Mol Sci. 2019;20(24):6223. 53. Peper JK, Schuster H, Löffler MW, Schmid-Horch B, Rammensee H-G,
29. Preethy S, Dedeepiya VD, Senthilkumar R, Rajmohan M, Karthick R, Stevanović S. An impedance-based cytotoxicity assay for real-time and
Terunuma H, et al. Natural killer cells as a promising tool to tackle label-free assessment of T-cell-mediated killing of adherent cells. J Immunol
cancer—a review of sources, methodologies, and potentials. Int Rev Methods. 2014;405:192–8.
Immunol. 2017;36(4):220–32. 54. Crowe SM, Kintzios S, Kaltsas G, Palmer CS. A bioelectronic system to
30. Lee DA. Cellular therapy: adoptive immunotherapy with expanded natural measure the glycolytic metabolism of activated CD4+ T cells. Biosensors.
killer cells. Immunol Rev. 2019;290(1):85–99. 2019;9(1):10.
31. Poznanski SM, Ashkar AA. What defines NK cell functional fate: phenotype 55. Singh M, Truong J, Reeves WB, Hahm J-I. Emerging cytokine biosensors with
or metabolism? Front Immunol. 2019;10:1414. optical detection modalities and nanomaterial-enabled signal enhancement.
32. Piscopo NJ, Mueller KP, Das A, Hematti P, Murphy WL, Palecek SP, et al. Sensors. 2017;17(2):428.
Bioengineering solutions for manufacturing challenges in CAR T cells. 56. Revzin A, Maverakis E, Chang H-C. Biosensors for immune cell analysis—a
Biotechnol J. 2018;13(2). perspective. Biomicrofluidics. 2012;6(2):021301.
33. Granzin M, Wagner J, Köhl U, Cerwenka A, Huppert V, Ullrich E. Shaping of 57. McCartney MM, Yamaguchi MS, Bowles PA, Gratch YS, Iyer RK, Linderholm
natural killer cell antitumor activity by ex vivo cultivation. Front Immunol. AL, et al. Volatile organic compound (VOC) emissions of CHO and T cells
2017;8:458. correlate to their expansion in bioreactors. J Breath Res. 2019;14(1):016002.
34. Pörtner R, Parida SK, Schaffer C, Hoffmeister H. Landscape of manufacturing 58. Martin I, Simmons PJ, Williams DF. Manufacturing challenges in regenerative
process of ATMP cell therapy products for unmet clinical needs. Stem Cells medicine. Sci Transl Med. 2014;6(232):232fs16.
Clin Pract Tissue Eng. 2017; Available from: https://www.intechopen.com/ 59. Henstock JR, El Haj AJ. Bioreactors. In: Mechanobiology: Exploitation for
books/stem-cells-in-clinical-practice-and-tissue-engineering/landscape-of-ma Medical Benefit; 2016. p. 275–96.
nufacturing-process-of-atmp-cell-therapy-products-for-unmet-clinical-needs. 60. Jones M, Nankervis B, Santos-Roballo K, Pham H, Bushman J, Coeshott C. A
Cited 2019 Nov 26. comparison of automated perfusion- and manual diffusion-based human
35. Abraham E, Ahmadian BB, Holderness K, Levinson Y, McAfee E. Platforms for regulatory T cell expansion and functionality using a soluble activator
manufacturing allogeneic, autologous and iPSC cell therapy products: an complex. Cell Transplant. 2020;29:1–15.
industry perspective. Adv Biochem Eng Biotechnol. 2018;165:323–50. 61. Carswell KS, Papoutsakis ET. Culture of human T cells in stirred bioreactors
36. Kaiser AD, Assenmacher M, Schröder B, Meyer M, Orentas R, Bethke U, et al. for cellular immunotherapy applications: shear, proliferation, and the IL-2
Towards a commercial process for the manufacture of genetically modified receptor. Biotechnol Bioeng. 2000;68(3):328–38.
T cells for therapy. Cancer Gene Ther. 2015;22(2):72–8. 62. Larbi A, Cabreiro F, Zelba H, Marthandan S, Combet E, Friguet B, et al.
37. Marshall D, Ward S, Baradez M-O. Requirement for smart in-process control Reduced oxygen tension results in reduced human T cell proliferation and
systems to deliver cell therapy processes fit for the 21st century. Cell Gene increased intracellular oxidative damage and susceptibility to apoptosis
Therapy Insights. 2016;2(6):683–9. upon activation. Free Radic Biol Med. 2010;48(1):26–34.
38. Lipsitz YY, Timmins NE, Zandstra PW. Quality cell therapy manufacturing by 63. Nielsen LK. Bioreactors for hematopoietic cell culture. Annu Rev Biomed
design. Nat Biotechnol. 2016;34(4):393–400. Eng. 1999;1:129–52.
39. Stephenson M, Grayson W. Recent advances in bioreactors for cell-based 64. Somerville RPT, Devillier L, Parkhurst MR, Rosenberg SA, Dudley ME. Clinical
therapies. F1000Res. 2018;7:F1000 Faculty Rev-517. scale rapid expansion of lymphocytes for adoptive cell transfer therapy in
40. Levinson Y, Beri RG, Holderness K, Ben-Nun IF, Shi Y, Abraham E. Bespoke the WAVE ® bioreactor. J Transl Med. 2012;10:69.
cell therapy manufacturing platforms. Biochem Eng J. 2018;132:262–9. 65. Lapteva N, Szmania SM, van Rhee F, Rooney CM. Clinical grade purification
41. Pigeau GM, Csaszar E, Dulgar-Tulloch A. Commercial scale manufacturing of and expansion of natural killer cells. Crit Rev Oncog. 2014;19(1–2):121–32.
allogeneic cell therapy. Front Med (Lausanne). 2018;5:233. 66. Tran C-A, Burton L, Russom D, Wagner JR, Jensen MC, Forman SJ, et al.
42. Moutsatsou P, Ochs J, Schmitt RH, Hewitt CJ, Hanga MP. Automation in cell Manufacturing of large numbers of patient-specific T cells for adoptive
and gene therapy manufacturing: from past to future. Biotechnol Lett. 2019; immunotherapy: an approach to improving product safety, composition,
41(11):1245–53. and production capacity. J Immunother. 2007;30(6):644–54.
67. Chrobok M, Dahlberg CIM, Sayitoglu EC, Beljanski V, Nahi H, Gilljam M, et al. 90. Ou J, Si Y, Tang Y, Salzer GE, Lu Y, Kim S, et al. Novel biomanufacturing
Functional assessment for clinical use of serum-free adapted NK-92 cells. platform for large-scale and high-quality human T cells production. J Biol
Cancers. 2019;11(1):69. Eng. 2019;13(1):34.
68. Rosenberg SA. Adoptive immunotherapy of cancer using lymphokine 91. Bröker K, Sinelnikov E, Gustavus D, Schumacher U, Pörtner R, Hoffmeister H,
activated killer cells and recombinant interleukin-2. Important Adv Oncol. et al. Mass production of highly active nk cells for cancer immunotherapy in
1986:55–91. a gmp conform perfusion bioreactor. Front Bioeng Biotechnol. 2019;7:194.
69. Rosenberg SA, Spiess P, Lafreniere R. A new approach to the adoptive 92. Apel M, Brüning M, Granzin M, Essl M, Stuth J, Blaschke J, et al. Integrated
immunotherapy of cancer with tumor-infiltrating lymphocytes. Science. clinical scale manufacturing system for cellular products derived by
1986;233(4770):1318–21. magnetic cell separation, centrifugation and cell culture. Chem Ing Tech.
70. Rosenberg SA, Packard BS, Aebersold PM, Solomon D, Topalian SL, Toy ST, 2013;85(1–2):103–10.
et al. Use of tumor-infiltrating lymphocytes and interleukin-2 in the 93. Singh V. Disposable bioreactor for cell culture using wave-induced agitation.
immunotherapy of patients with metastatic melanoma. A preliminary Cytotechnology. 1999;30(1–3):149–58.
report. N Engl J Med. 1988;319(25):1676–80. 94. Sutlu T, Stellan B, Gilljam M, Quezada HC, Nahi H, Gahrton G, et al. Clinical-
71. Rosenberg SA, Lotze MT, Muul LM, Chang AE, Avis FP, Leitman S, et al. A grade, large-scale, feeder-free expansion of highly active human natural
progress report on the treatment of 157 patients with advanced cancer killer cells for adoptive immunotherapy using an automated bioreactor.
using lymphokine-activated killer cells and interleukin-2 or high-dose Cytotherapy. 2010;12(8):1044–55.
interleukin-2 alone. N Engl J Med. 1987;316(15):889–97. 95. Eibl R, Eibl D. Application of disposable bag bioreactors in tissue
72. Yannelli JR. The preparation of effector cells for use in the adoptive cellular engineering and for the production of therapeutic agents. Adv Biochem
immunotherapy of human cancer. J Immunol Methods. 1991;139(1):1–16. Eng Biotechnol. 2009;112:183–207.
73. Bödeker BG, Lehmann J, Mühlradt PF. Lymphokine (interleukin-2) 96. EIBL R, EIBL D. Design and use of the wave bioreactor for plant cell culture.
production by mitogen-stimulated human lymphocytes in small reactors. In: Gupta SD, Ibaraki Y, editors. Plan tissue culture engineering. Dordrecht:
Dev Biol Stand. 1981;50:193–200. Springer Netherlands; 2006. p. 203–27. (Focus on Biotechnology). Available
74. Bödekeri BGD, Lehmann J, Van Damme J, Kappmeyer H, Gassel W-D, from: https://doi.org/10.1007/978-1-4020-3694-1_12. Cited 2019 Nov 10.
Havemann K, et al. Production of five human Lymphokines 97. Meng Y, Sun J, Hu T, Ma Y, Du T, Kong C, et al. Rapid expansion in the
(GranulocyteMacrophage Colony stimulating factor, interferon-γ, interleukin WAVE bioreactor of clinical scale cells for tumor immunotherapy. Hum
2, macrophage cytotoxicity factor and macrophage migration inhibitory Vaccines Immunother. 2018;14(10):2516–26.
factor) from con a stimulated lymphocyte cultures in bioreactors. 98. Hami LS, Green C, Leshinsky N, Markham E, Miller K, Craig S. GMP
Immunobiology. 1984;166(1):12–23. production and testing of Xcellerated T cells™ for the treatment of patients
75. Pauly JL, Russell CW, Planinsek JA, Minowada J. Studies of cultured human T with CLL. Cytotherapy. 2004;6(6):554–62.
lymphocytes. I. Production of the T cell growth-promoting lymphokine 99. Hollyman D, Stefanski J, Przybylowski M, Bartido S, Borquez-Ojeda O, Taylor
interleukin-2. J Immunol Methods. 1982;50(2):173–86. C, et al. Manufacturing validation of biologically functional T cells targeted
76. Braude IA. A simple and efficient method for the production of human to CD19 antigen for autologous adoptive cell therapy. J Immunother. 2009;
gamma interferon. J Immunol Methods. 1983;63(2):237–46. 32(2):169–80.
77. Grote W, Klaar J, Mühlradt PF, Monner DA. Large scale production and 100. Janas M, Nunes C, Marenghi A, Sauvage V, Davis B, Bajaj A, et al. Perfusion’s
purification of human IL-2 from buffy coat lymphocytes stimulated with 12- role in maintenance of high-density T-cell cultures. BioProcess Int. 2015;
O-tetradecanoylphorbol 13-acetate and calcium ionohore A23187. J 13(1):18–26.
Immunol Methods. 1987;103(1):15–25. 101. Vavrova K, Vrabcova P, Filipp D, Bartunkova J, Horvath R. Generation of T
78. Knazek RA, Wu Y-W, Aebersold PM, Rosenberg SA. Culture of human tumor cell effectors using tumor cell-loaded dendritic cells for adoptive T cell
infiltrating lymphocytes in hollow fiber bioreactors. J Immunol Methods. therapy. Med Oncol. 2016;33(12):136.
1990;127(1):29–37. 102. O’Hanlon CF, Fedczyna T, Eaker S, Shingleton WD, Helfer BM. Integrating a
79. Alter BJ, Ochoa AC, Gruenberg ML, Keyport GM, Bach FH. The growth of 19F MRI tracer agent into the clinical scale manufacturing of a T-cell
cells with LAK activity in an automated tissue culture system (Acusyst P). immunotherapy. Contrast Media Mol Imaging. 2017;2017:9548478.
Prog Clin Biol Res. 1987;244:301–11. 103. Smith TA. CAR-T cell expansion in a Xuri cell expansion system W25.
80. Tanji Y, Tanaka T, Kimoto Y, Fujiwara A, Fujita M, Taguchi T. Mass culture of Methods Mol Biol Clifton NJ. 2020;2086:151–63.
LAK cells by hollow-fiber bioreactor system. Jpn J Cancer Chemother. 1989; 104. Spanholtz J, Preijers F, Tordoir M, Trilsbeek C, Paardekooper J, de Witte T,
16(4 II-3):1888–92. et al. Clinical-grade generation of active NK cells from cord blood
81. Pierson BA, Europa AF, Hu W-S, Miller JS. Production of human natural killer hematopoietic progenitor cells for immunotherapy using a closed-system
cells for adoptive immunotherapy using a computer-controlled stirred-tank culture process. PLoS One. 2011;6(6):e20740 Kaufman D, editor.
bioreactor. J Hematother. 1996;5(5):475–83. 105. Rujkijyanont P, Chan WK, Eldridge PW, Lockey T, Holladay M, Rooney B,
82. Murata M, Yano T, Togami M, Yasumoto K, Sugimachi K, Kimura G, et al. et al. Ex vivo activation of CD56+ immune cells that eradicate
Development of a new culture system for human lymphokine-activated neuroblastoma. Cancer Res. 2013;73(8):2608–18.
killer cells. J Immunol Methods. 1990;129(1):89–95. 106. Lapteva N, Durett AG, Sun J, Rollins LA, Huye LL, Fang J, et al. Large-scale
83. Walther I, Pippia P, Meloni MA, Turrini F, Mannu F, Cogoli A. Simulated ex vivo expansion and characterization of natural killer cells for clinical
microgravity inhibits the genetic expression of interleukin-2 and its receptor applications. Cytotherapy. 2012;14(9):1131–43.
in mitogen-activated T lymphocytes. FEBS Lett. 1998;436(1):115–8. 107. Sadeghi A, Pauler L, Annerén C, Friberg A, Brandhorst D, Korsgren O, et al.
84. Bakos A, Varkonyi A, Minarovits J, Batkai L. Effect of simulated microgravity on Large-scale bioreactor expansion of tumor-infiltrating lymphocytes. J
human lymphocytes. J Gravit Physiol J Int Soc Gravit Physiol. 2001;8(1):69–70. Immunol Methods. 2011;364(1–2):94–100.
85. Risin D, Pellis NR. Modeled microgravity inhibits apoptosis in peripheral 108. Donia M, Larsen SM, Met T, Svane IM. Simplified protocol for clinical-grade
blood lymphocytes. In Vitro Cell Dev Biol Anim. 2001;37(2):66–72. tumor-infiltrating lymphocyte manufacturing with use of the wave
86. Ward NE, Pellis NR, Risin SA, Risin D. Gene expression alterations in activated bioreactor. Cytotherapy. 2014;16(8):1117–20.
human T-cells induced by modeled microgravity. J Cell Biochem. 2006;99(4): 109. Andersen R, Donia M, Ellebaek E, Borch TH, Kongsted P, Iversen TZ, et al.
1187–202. Long-lasting complete responses in patients with metastatic melanoma
87. Li Q, Mei Q, Huyan T, Xie L, Che S, Yang H, et al. Effects of simulated after adoptive cell therapy with tumor-infiltrating lymphocytes and an
microgravity on primary human NK cells. Astrobiology. 2013;13(8):703–14. attenuated il2 regimen. Clin Cancer Res. 2016;22(15):3734–45.
88. Bradley JH, Stein R, Randolph B, Molina E, Arnold JP, Gregg RK. T cell 110. Bjoern J, Lyngaa R, Andersen R, Rosenkrantz LH, Hadrup SR, Donia M, et al.
resistance to activation by dendritic cells requires long-term culture in Influence of ipilimumab on expanded tumour derived T cells from patients
simulated microgravity. Life Sci Space Res. 2017;15:55–61. with metastatic melanoma. Oncotarget. 2017;8(16):27062–74.
89. Costariol E, Rotondi M, Amini A, Hewitt CJ, Nienow AW, Heathman TRJ, et al. 111. Nankervis B, Jones M, Vang B, Brent Rice R Jr, Coeshott C, Beltzer J.
Establishing the scalable manufacture of primary human T-cells in an Optimizing T cell expansion in a hollow-Fiber bioreactor. Curr Stem Cell
automated stirred-tank bioreactor. Biotechnol Bioeng. 2019;116(10):2488–502. Rep. 2018;4(1):46–51.
112. Coeshott C, Vang B, Jones M, Nankervis B. Large-scale expansion and 132. Klapper JA, Thomasian AA, Smith DM, Gorgas GC, Wunderlich JR, Smith FO,
characterization of CD3+ T-cells in the quantum® cell expansion system. J et al. Single-pass, closed-system rapid expansion of lymphocyte cultures for
Transl Med. 2019;17(1):258. adoptive cell therapy. J Immunol Methods. 2009;345(1–2):90–9.
113. Cadwell JJS. New developments in hollow-fiber cell culture. Am Biotechnol 133. Goltry KL, Hampson BS, Venturi NA, Bartel RL. Large-scale production of
Lab. 2004;22(8):14. adult stem cells for clinical use. In: Emerging technology platforms for stem
114. De Bartolo L, Piscioneri A, Cotroneo G, Salerno S, Tasselli F, Campana C, cells: Wiley; 2009. p. 153–68. Available from: https://onlinelibrary.wiley.com/
et al. Human lymphocyte PEEK-WC hollow fiber membrane bioreactor. J doi/abs/10.1002/9780470454923.ch9. Cited 2020 Mar 30.
Biotechnol. 2007;132(1):65–74. 134. Guardino AE, Rajapaksa R, Ong KH, Sheehan K, Levy R. Production of
115. Hillman GG, Wolf ML, Montecillo E, Younes E, Ali E, Pontes JE, et al. myeloid dendritic cells (DC) pulsed with tumor-specific idiotype protein for
Expansion of activated lymphocytes obtained from renal cell carcinoma in vaccination of patients with multiple myeloma. Cytotherapy. 2006;8(3):277–
an automated hollow fiber bioreactor. Cell Transplant. 1994;3(4):263–71. 89.
116. Yannelli JR, Hyatt C, McConnell S, Hines K, Jacknin L, Parker L, et al. Growth 135. Schlegel P, Lang A-M, Matela M, Horrer A, Schilling A, Jöchner A, et al. Ex
of tumor-infiltrating lymphocytes from human solid cancers: summary of a vivo expansion of autologous, donor-derived NK-, γδT-, and cytokine
5-year experience. Int J Cancer. 1996;65(4):413–21. induced killer (CIK) cells post haploidentical hematopoietic stem cell
117. Lewko WM, Hall PB, Oldham RK. Growth of tumor-derived activated T cells transplantation results in increased antitumor activity. Bone Marrow
for the treatment of advanced cancer. Cancer Biother Radiopharm. 2000; Transplant. 2019;54:727–32.
15(4):357–66. 136. Kuzin I, Sun H, Moshkani S, Feng C, Mantalaris A, Wu JD, et al. Long-term
118. Malone CC, Schiltz PM, Mackintosh AD, Beutel LD, Heinemann FS, Dillman immunologically competent human peripheral lymphoid tissue cultures in a
RO. Characterization of human tumor-infiltrating lymphocytes expanded in 3D bioreactor. Biotechnol Bioeng. 2011;108(6):1430–40.
hollow-fiber bioreactors for immunotherapy of cancer. Cancer Biother 137. Giese C, Demmler CD, Ammer R, Hartmann S, Lubitz A, Miller L, et al. A
Radiopharm. 2001;16(5):381–90. human lymph node in vitro--challenges and progress. Artif Organs. 2006;
119. Freedman RS, Tomasovic B, Templin S, Atkinson EN, Kudelka A, Edwards CL, 30(10):803–8.
et al. Large-scale expansion in interleukin-2 of tumor-infiltrating 138. Gilbertson JA, Sen A, Behie LA, Kallos MS. Scaled-up production of
lymphocytes from patients with ovarian carcinoma for adoptive mammalian neural precursor cell aggregates in computer-controlled
immunotherapy. J Immunol Methods. 1994;167(1–2):145–60. suspension bioreactors. Biotechnol Bioeng. 2006;94(4):783–92.
120. Pan D, Shankar R, Stroncek DF, Whitley CB. Combined ultrafiltration- 139. Foster AE, Forrester K, Gottlieb DJ, Barton GW, Romagnoli JA, Bradstock KF.
transduction in a hollow-fiber bioreactor facilitates retrovirus-mediated gene Large-scale expansion of Cytomegalovirus-specific cytotoxic T cells in
transfer into peripheral blood lymphocytes from patients with suspension culture. Biotechnol Bioeng. 2004;85(2):138–46.
mucopolysaccharidosis type II. Hum Gene Ther. 1999;10(17):2799–810. 140. Bohnenkamp H, Hilbert U, Noll T. Bioprocess development for the
121. Shankar R, Whitley CB, Pan D, Burger S, McCullough J, Stroncek D. Retroviral cultivation of human T-lymphocytes in a clinical scale. Cytotechnology.
transduction of peripheral blood leukocytes in a hollow- fiber bioreactor. 2002;38(1–3):135–45.
Transfusion (Paris). 1997;37(7):685–90. 141. Hu W, Berdugo C, Chalmers JJ. The potential of hydrodynamic damage to
122. Stroncek DF, Hubel A, Shankar RA, Burger SR, Pan D, McCullough J, et al. animal cells of industrial relevance: current understanding. Cytotechnology.
Retroviral transduction and expansion of peripheral blood lymphocytes for 2011;63(5):445–60.
the treatment of mucopolysaccharidosis type II, Hunter’s syndrome. 142. Ramsborg CG, Windgassen D, Fallon JK, Paredes CJ, Papoutsakis ET.
Transfusion (Paris). 1999;39(4):343–50. Molecular insights into the pleiotropic effects of plasma on ex vivo-
123. Wolf ML, Hirschel MD. Growing TIL and LAK cells in hollow fibre bioreactors. expanded T cells using DNA-microarray analysis. Exp Hematol. 2004;32(10):
In: Animal Cell Biotechnology; 1994. p. 237–58. 970–90.
124. Lamers CHJ, van de Griend RJ, Gratama JW, RLH B. Activation and 143. Foster AE, Forrester K, Li Y-C, Gottlieb DJ. Ex-vivo uses and applications of
expansion of Hunan cytotoxic T lymphocytes in hollow fiber bioreactors. In: cytokines for adoptive immunotherapy. Curr Pharm Des. 2004;10(11):1207–20.
Spier RE, Griffiths JB, Berthold W, editors. Animal cell technology: 144. Hilbert U, Jelinek N, Schmidt S, Biselli M. Bioprocess development for the
Butterworth-Heinemann; 1994. p. 735–7. Available from: http://www. cultivation of human T-lymphocytes. Eng Life Sci. 2001;1(1):20–3.
sciencedirect.com/science/article/pii/B9780750618458501629. Cited 2020 145. Carswell KS, Weiss JW, Papoutsakis ET. Low oxygen tension enhances the
Mar 24. stimulation and proliferation of human T lymphocytes in the presence of IL-
125. Lamers CHJ, Gratama JW, Luider-Vrieling B, Bolhuis RLH, Bast EJEG. Large- 2. Cytotherapy. 2000;2(1):25–37.
scale production of natural cytokines during activation and expansion of 146. Krieger JA, Landsiedel JC, Lawrence DA. Differential in vitro effects of
human T lymphocytes in hollow fiber bioreactor cultures. J Immunother. physiological and atmospheric oxygen tension on normal human peripheral
1999;22(4):299–307. blood mononuclear cell proliferation, cytokine and immunoglobulin
126. Liu Z, Wong JT. Proliferative and regenerative capacities of CD4+ T cells production. Int J Immunopharmacol. 1996;18(10):545–52.
upon TCR stimulation. Clin Immunol. 1999;93(1):16–23. 147. Collins PC, Nielsen LK, Patel SD, Papoutsakis ET, Miller WM. Characterization of
127. Trickett AE, Kwan YL, Cameron B, Dwyer JM. Ex vivo expansion of functional hematopoietic cell expansion, oxygen uptake, and glycolysis in a controlled,
T lymphocytes from HIV-infected individuals. J Immunol Methods. 2002; stirred-tank bioreactor system. Biotechnol Prog. 1998;14(3):466–72.
262(1–2):71–83. 148. Collins PC, Miller WM, Papoutsakis ET. Stirred culture of peripheral and cord
128. Curcio E, Piscioneri A, Salerno S, Tasselli F, Morelli S, Drioli E, et al. Human blood hematopoietic cells offers advantages over traditional static systems
lymphocytes cultured in 3-D bioreactors: influence of configuration on for clinically relevant applications. Biotechnol Bioeng. 1998;59(5):534–43.
metabolite transport and reactions. Biomaterials. 2012;33(33):8296–303. 149. Hosseinizand H, Ebrahimi M, Abdekhodaie MJ. Agitation increases
129. Lewko WM, Good RW, Bowman D, Smith TL, Oldham RK. Growth of tumor expansion of cord blood hematopoietic cells and promotes their
derived activated T-cells for the treatment of cancer. Cancer Biother. 1994; differentiation into myeloid lineage. Cytotechnology. 2016;68(4):969–78.
9(3):211–24. 150. De león A, Mayani H, Ramírez OT. Design, characterization and application
130. Freedman RS, Kudelka AP, Kavanagh JJ, Verschraegen C, Edwards CL, Nash of a minibioreactor for the culture of human hematopoietic cells under
M, et al. Clinical and biological effects of intraperitoneal injections of controlled conditions. Cytotechnology. 1998;28(1–3):127–38.
recombinant interferon-gamma and recombinant interleukin 2 with or 151. Li Q, Liu Q, Cai H, Tan W-S. A comparative gene-expression analysis of
without tumor-infiltrating lymphocytes in patients with ovarian or CD34+ hematopoietic stem and progenitor cells grown in static and stirred
peritoneal carcinoma. Clin Cancer Res Off J Am Assoc Cancer Res. 2000;6(6): culture systems. Cell Mol Biol Lett. 2006;11(4):475–87.
2268–78. 152. Kaiser P, Werner M, Jérôme V, Freitag R. Scale-up of the ex vivo expansion
131. Stroncek DF, Tran M, Frodigh SE, David-Ocampo V, Ren J, Larochelle A, et al. of encapsulated primary human T lymphocytes. Biotechnol Bioeng. 2018;
Preliminary evaluation of a highly automated instrument for the selection of 115(10):2632–42.
CD34+ cells from mobilized peripheral blood stem cell concentrates. 153. Barton G, Forrester K, Bradstock K, Gottlieb D, Foster A. Optimizing
Transfusion (Paris). 2016;56(2):511–7. bioreactor productivity for therapeutic immune cells. Food Bioprod Process.
2005;83(2 C):158–63.
154. Bajgain P, Mucharla R, Wilson J, Welch D, Anurathapan U, Liang B, et al. 174. Gerdemann U, Katari UL, Papadopoulou A, Keirnan JM, Craddock JA, Liu H,
Optimizing the production of suspension cells using the G-Rex “M” series. et al. Safety and clinical efficacy of rapidly-generated trivirus-directed T cells
Mol Ther - Methods Clin Dev. 2014;1:14015. as treatment for adenovirus, EBV, and CMV infections after allogeneic
155. Bajgain P, Tawinwung S, D’Elia L, Sukumaran S, Watanabe N, Hoyos V, et al. hematopoietic stem cell transplant. Mol Ther. 2013;21(11):2113–21.
CAR T cell therapy for breast cancer: harnessing the tumor milieu to drive T 175. Ramanayake S, Bilmon I, Bishop D, Dubosq M-C, Blyth E, Clancy L, et al.
cell activation. J Immunother Cancer. 2018;6(1):34. Low-cost generation of Good manufacturing practice-grade CD19-specific
156. Pampusch MS, Haran KP, Hart GT, Rakasz EG, Rendahl AK, Berger EA, et al. chimeric antigen receptor-expressing T cells using piggyBac gene transfer
Rapid transduction and expansion of transduced T cells with maintenance and patient-derived materials. Cytotherapy. 2015;17(9):1251–67.
of central memory populations. Mol Ther Methods Clin Dev. 2020;16:1–10. 176. Jin J, Gkitsas N, Fellowes VS, Ren J, Feldman SA, Hinrichs CS, et al. Enhanced
157. Ludwig J, Hirschel M. Methods and process optimization for large-scale CAR clinical scale manufacturing of TCR transduced T-cells using closed culture
T expansion using the G-Rex cell culture platform. Methods Mol Biol Clifton system modules. J Transl Med. 2018;16(1):13.
NJ. 2020;2086:165–77. 177. Kuranda K, Caillat-Zucman S, You S, Mallone R. In vitro expansion of anti-
158. Vera JF, Brenner LJ, Gerdemann U, Ngo MC, Sili U, Liu H, et al. Accelerated viral T cells from cord blood by accelerated co-cultured dendritic cells. Mol
production of antigen-specific T cells for preclinical and clinical applications Ther - Methods Clin Dev. 2019;13:112–20.
using gas-permeable rapid expansion cultureware (G-Rex). J Immunother. 178. Xiao L, Chen C, Li Z, Zhu S, Tay JC, Zhang X, et al. Large-scale expansion of
2010;33(3):305–15. Vγ9Vδ2 T cells with engineered K562 feeder cells in G-Rex vessels and their
159. Forget M-A, Haymaker C, Dennison JB, Toth C, Maiti S, Fulbright OJ, et al. use as chimeric antigen receptor–modified effector cells. Cytotherapy. 2018;
The beneficial effects of a gas-permeable flask for expansion of 20(3):420–35.
tumorinfiltrating lymphocytes as reflected in their mitochondrial function 179. Shah N, Martin-Antonio B, Yang H, Ku S, Lee DA, Cooper LJN, et al. Antigen
and respiration capacity. Oncoimmunology. 2016;5(2):e1057386. presenting cell-mediated expansion of human umbilical cord blood yields
160. Orio J, Carli C, Janelle V, Giroux M, Taillefer J, Goupil M, et al. Early exposure log-scale expansion of natural killer cells with anti-myeloma activity. PLoS
to interleukin-21 limits rapidly generated anti-Epstein-Barr virus T-cell line One. 2013;8(10):e76781.
differentiation. Cytotherapy. 2015;17(4):496–508. 180. Jin J, Sabatino M, Somerville R, Wilson JR, Dudley ME, Stroncek DF, et al.
161. Rooney CM, Leen AM, Vera JF, Heslop HE. T lymphocytes targeting native Simplified method of the growth of human tumor infiltrating lymphocytes
receptors. Immunol Rev. 2014;257(1) Available from: https://www.ncbi.nlm. in gas-permeable flasks to numbers needed for patient treatment. J
nih.gov/pmc/articles/PMC3869095/. Cited 2019 Dec 7. Immunother. 2012;35(3):283–92.
162. Gagliardi C, Khalil M, Foster AE. Streamlined production of genetically 181. Forget M-A, Malu S, Liu H, Toth C, Maiti S, Kale C, et al. Activation and
modified T cells with activation, transduction and expansion in closed- propagation of tumor infiltrating lymphocytes on clinical-grade designer
system G-Rex bioreactors. Cytotherapy. 2019;21(12):1246–57. artificial antigen presenting cells for adoptive immunotherapy of melanoma.
163. Spielmann G, Bollard CM, Kunz H, Hanley PJ, Simpson RJ. A single exercise J Immunother Hagerstown Md 1997. 2014;37(9):448–60.
bout enhances the manufacture of viral-specific T-cells from healthy donors: 182. Bollard CM, Gottschalk S, Huls MH, Leen AM, Gee AP, Rooney CM.
implications for allogeneic adoptive transfer immunotherapy. Sci Rep. 2016; Manufacture of GMP-grade cytotoxic T lymphocytes specific for LMP1 and
6:25852. LMP2 for patients with EBV-associated lymphoma. Cytotherapy. 2011;13(5):
164. Leen AM, Bollard CM, Mendizabal AM, Shpall EJ, Szabolcs P, Antin JH, et al. 518–22.
Multicenter study of banked third-party virus-specific T cells to treat severe 183. Sili U, Leen AM, Vera JF, Gee AP, Huls H, Heslop HE, et al. Production of
viral infections after hematopoietic stem cell transplantation. Blood. 2013; good manufacturing practice-grade cytotoxic T lymphocytes specific for
121(26):5113–23. Epstein–Barr virus, cytomegalovirus and adenovirus to prevent or treat viral
165. Dave H, Luo M, Blaney JW, Patel S, Barese C, Cruz CR, et al. Toward a rapid infections post-allogeneic hematopoietic stem cell transplant. Cytotherapy.
production of multivirus-specific T cells targeting BKV, adenovirus, CMV, and 2012;14(1):7–11.
EBV from umbilical cord blood. Mol Ther Methods Clin Dev. 2017;5:13–21. 184. Hanley PJ, Lam S, Shpall EJ, Bollard CM. Expanding cytotoxic T lymphocytes
166. Gerdemann U, Vera JF, Rooney CM, Leen AM. Generation of multivirus- from umbilical cord blood that target cytomegalovirus, Epstein-Barr virus,
specific T cells to prevent/treat viral infections after allogeneic and adenovirus. J Vis Exp. 2012;63:e3627.
hematopoietic stem cell transplant. J Vis Exp. 2011;51:e2736. 185. Lapteva N, Parihar R, Rollins LA, Gee AP, Rooney CM. Large-scale culture and
167. Horlock C, Skulte A, Mitra A, Stansfield A, Bhandari S, Ip W, et al. genetic modification of human natural killer cells for cellular therapy.
Manufacture of GMP-compliant functional adenovirus-specific T-cell therapy Methods Mol Biol. 2016;1441:195–202.
for treatment of post-transplant infectious complications. Cytotherapy. 2016; 186. Wickström S, Lövgren T. Expansion of tumor-infiltrating lymphocytes from
18(9):1209–18. melanoma tumors. Methods Mol Biol. 2019;1913:105–18.
168. Luah YH, Sundar Raj K, Koh MBC, Linn YC. A novel simplified method of 187. Mock U, Nickolay L, Philip B, Cheung GW-K, Zhan H, Johnston ICD, et al.
generating cytomegalovirus-specific cytokine-induced killer cells of high Automated manufacturing of chimeric antigen receptor T cells for adoptive
specificity and superior potency with GMP compliance. Clin Immunol. 2019; immunotherapy using CliniMACS prodigy. Cytotherapy. 2016;18(8):1002–11.
205:83–92. 188. Oberschmidt O, Morgan M, Huppert V, Kessler J, Gardlowski T, Matthies N,
169. Nakazawa Y, Huye LE, Salsman VS, Leen AM, Ahmed N, Rollins L, et al. et al. Development of automated separation, expansion, and quality control
PiggyBac-mediated Cancer immunotherapy using EBV-specific cytotoxic T- protocols for clinical-scale manufacturing of primary human NK cells and
cells expressing HER2-specific chimeric antigen receptor. Mol Ther. 2011; Alpharetroviral chimeric antigen receptor engineering. Hum Gene Ther
19(12):2133–43. Methods. 2019;30(3):102–20.
170. Murad JM, Baumeister SH, Werner L, Daley H, Trébéden-Negre H, Reder J, 189. Spohn G, Wiercinska E, Karpova D, Bunos M, Hümmer C, Wingenfeld E, et al.
et al. Manufacturing development and clinical production of NKG2D Automated CD34+ cell isolation of peripheral blood stem cell apheresis
chimeric antigen receptor–expressing T cells for autologous adoptive cell product. Cytotherapy. 2015;17(10):1465–71.
therapy. Cytotherapy. 2018;20(7):952–63. 190. Hümmer C, Poppe C, Bunos M, Stock B, Wingenfeld E, Huppert V, et al.
171. Parlar A, Sayitoglu EC, Ozkazanc D, Georgoudaki A-M, Pamukcu C, Aras M, Automation of cellular therapy product manufacturing: results of a split
et al. Engineering antigen-specific NK cell lines against the melanoma- validation comparing CD34 selection of peripheral blood stem cell
associated antigen tyrosinase via TCR gene transfer. Eur J Immunol. 2019; apheresis product with a semi-manual vs. an automatic procedure. J Transl
49(8):1278–90. Med. 2016;14(1):76.
172. Sakellariou-Thompson D, Forget M-A, Hinchcliff E, Celestino J, Hwu P, Jazaeri 191. Bunos M, Hümmer C, Wingenfeld E, Sorg N, Pfirrmann V, Bader P, et al.
AA, et al. Potential clinical application of tumor-infiltrating lymphocyte Automated isolation of primary antigen-specific T cells from donor
therapy for ovarian epithelial cancer prior or post-resistance to lymphocyte concentrates: results of a feasibility exercise. Vox Sang. 2015;
chemotherapy. Cancer Immunol Immunother. 2019;68(11):1747–57. 109(4):387–93.
173. Chakraborty R, Mahendravada A, Perna SK, Rooney CM, Heslop HE, Vera JF, 192. Kumaresan P, Figliola M, Moyes JS, Huls MH, Tewari P, Shpall EJ, et al.
et al. Robust and cost effective expansion of human regulatory T cells Automated cell enrichment of cytomegalovirus-specific T cells for clinical
highly functional in a xenograft model of graft-versus-host disease. applications using the cytokine-capture system. J Vis Exp. 2015;2015(104):
Haematologica. 2013;98(4):533–7. 52808.
193. Priesner C, Esser R, Tischer S, Marburger M, Aleksandrova K, Maecker-Kolhoff

B, et al. Comparative analysis of clinical-scale IFN-γ-positive T-cell
enrichment using partially and fully integrated platforms. Front Immunol.
2016;7:393.
194. Kállay K, Kassa C, Réti M, Karászi É, Sinkó J, Goda V, et al. Early experience
with CliniMACS prodigy CCS (IFN-gamma) system in selection of virus-
specific T cells from third-party donors for pediatric patients with severe
viral infections after hematopoietic stem cell transplantation. J Immunother.
2018;41(3):158–63.
195. Kim E-K, Ahn Y-O, Kim S, Kim TM, Keam B, Heo DS. Ex vivo activation and
expansion of natural killer cells from patients with advanced cancer with
feeder cells from healthy volunteers. Cytotherapy. 2013;15(2):231–241.e1.
196. Klöß S, Oberschmidt O, Morgan M, Dahlke J, Arseniev L, Huppert V, et al.
Optimization of human NK cell manufacturing: fully automated separation,
improved ex vivo expansion using IL-21 with autologous feeder cells, and
generation of anti-CD123-CAR-expressing effector cells. Hum Gene Ther.
2017;28(10):897–913.
197. Zhu F, Shah N, Xu H, Schneider D, Orentas R, Dropulic B, et al. Closed-
system manufacturing of CD19 and dual-targeted CD20/19 chimeric
antigen receptor T cells using the CliniMACS prodigy device at an academic
medical center. Cytotherapy. 2018;20(3):394–406.
198. Granzin M, Soltenborn S, Müller S, Kollet J, Berg M, Cerwenka A, et al. Fully
automated expansion and activation of clinical-grade natural killer cells for
adoptive immunotherapy. Cytotherapy. 2015;17(5):621–32.
199. Priesner C, Aleksandrova K, Esser R, Mockel-Tenbrinck N, Leise J, Drechsel K,
et al. Automated enrichment, transduction, and expansion of clinical-scale
CD62L + T cells for manufacturing of gene therapy medicinal products.
Hum Gene Ther. 2016;27(10):860–9.
200. Lock D, Mockel-Tenbrinck N, Drechsel K, Barth C, Mauer D, Schaser T, et al.
Automated manufacturing of potent CD20-directed chimeric antigen
receptor T cells for clinical use. Hum Gene Ther. 2017;28(10):914–25.
201. Blaeschke F, Stenger D, Kaeuferle T, Willier S, Lotfi R, Kaiser AD, et al.
Induction of a central memory and stem cell memory phenotype in
functionally active CD4+ and CD8+ CAR T cells produced in an automated
good manufacturing practice system for the treatment of CD19+ acute
lymphoblastic leukemia. Cancer Immunol Immunother. 2018;67(7):1053–66.
202. Zhang W, Jordan KR, Schulte B, Purev E. Characterization of clinical grade
CD19 chimeric antigen receptor T cells produced using automated
CliniMACS prodigy system. Drug Des Devel Ther. 2018;12:3343–56.
203. Aleksandrova K, Leise J, Priesner C, Melk A, Kubaink F, Abken H, et al.
Functionality and cell senescence of CD4/ CD8-selected CD20 CAR T cells
manufactured using the automated CliniMACS prodigy® platform. Transfus
Med Hemother. 2019;46(1):47–54.
204. Fernández L, Fernández A, Mirones I, Escudero A, Cardoso L, Vela M, et al.
GMP-compliant manufacturing of NKG2D CAR memory T cells using
CliniMACS prodigy. Front Immunol. 2019;10 Available from: https://www.
ncbi.nlm.nih.gov/pmc/articles/PMC6795760/. Cited 2019 Dec 5.
205. Marín Morales JM, Münch N, Peter K, Freund D, Oelschlägel U, Hölig K, et al.
Automated clinical grade expansion of regulatory T cells in a fully closed
system. Front Immunol. 2019;10:38.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.

Lymphocyte Expansion in Bioreactors: Upgrading Adoptive Cell Therapy

Uploaded by

Copyright:

Available Formats

Lymphocyte Expansion in Bioreactors: Upgrading Adoptive Cell Therapy

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lymphocyte Expansion in Bioreactors: Upgrading Adoptive Cell Therapy

Uploaded by

Copyright:

Available Formats

Garcia-Aponte et al.

Journal of Biological Engineering (2021) 15:13

REVIEW Open Access

Lymphocyte expansion in bioreactors:

Background characterized as a “living” treatment that can be en-

of the authors, and based on previous frameworks of re- GMP compliance

Table 1 Advanced process monitoring tools for ACT

Gas transfer Different reactor configurations may fulfill these re-

- 2009 500 CD28 beads from Bioreactor periods (200–1600 mL/day)

[64] - 2012 allogenic pluronic bioreactor 2/10 concentration at ~ 170 mg/dL.

Table 4 Further applications of the rocking motion bioreactor

[117] - LAK supern. between 1.0–1.5

[118] - samples specified 150 below 1000

Table 6 Performance differences between static and perfusion reactor cultures

Table 7 Performance of non-comparative cultures in hollow fiber reactors

beads 4:1 to 1:1

between static and ambr®

temsirolimus, remained stable culture. No compromise

(59.6 ± 2.7%) alloreactive.

10.28%. from 66.6–51%; TEM cell

T75 and Prodigy had a

Conclusions possibility of process intensification through quality

parameter targeting for optimization, contributing to- Authors’ contributions

out thoroughly considering the importance of the Funding

automation can be used to remove sampling requirements Consent for publication

193. Priesner C, Esser R, Tischer S, Marburger M, Aleksandrova K, Maecker-Kolhoff

You might also like