Journal of Immunological Methods, 127 (1990) 29-37
29
Elsevier JIM05456
Culture of human tumor infiltrating lymphocytes in hollow fiber bioreactors R i c h a r d A. K n a z e k
1 Yan-Wan
Wu
2, Paul
M. A e b e r s o l d 3 a n d S t e v e n A. R o s e n b e r g 3
1 Cellco Advanced Bioreactors, lnc., Kens,ington, MD, U.S.A., 2 Developmental Endocrinology Branch, National lnstitute of Child Health and Human Development, National lnstitutes of Health, Bethesda, MD, U.S.A., and 3 Surgery Branch, Division of Cancer Treatment, National Cancer Institute, National Institutes of Health, Bethesda, MD, U.S.A.
(Received22 September1989, accepted 11 October 1989)
Human Lumor infiltrating lymphocytes ( T I L ) f r o m metastatic melanoma of six patients were grown using a new hollow fiber bioreactor system. After inoculating 0.35-10 x 108 TIL into the extra-fiber space (EFS), each Cellmax bioreactor was perfused with AIM-V medium, supplemented with rlL-2. The cells subsequently expanded 124-1170-fold to yield 1.5-5.4 × 101° TIL over a 14-32 day period. TIL were flushed from the EFS using 200 ml medium and possessed an average viability = 91%. The phenotype and the autologous tumor cell lytic capacity of these TIL were similar to those of TIL grown in the currently used gas-permeable culture bags. Tissue culture media use averaged 4.3 liters/101° TIL harvested. The TIL of one patient were re-expanded twice from cells remaining within the same bioreactor after harvest suggesting that one bioreactor cartridge could be used for repetitive, periodic studies. An estimated 80% decrease in technical time expended and in incubator space requirements were realized using this methodology. Cell culture on hollow fibers appears to be a useful method for producing large quantities of primary human lymphocytes for experimental, and perhaps, therapeutic needs. Key words: Hollow fiber; Tumor infiltrating lymphocyte;Interleukin-2; Cell culture; Lymphocyte
Introduction Both animal studies and early clinical trials have demonstrated that adoptive immunotherapy holds promise as a potential mode of treatment for human malignancies. Initially, lymphokineactivated killer cells (LAK), obtained by incubating peripheral blood lymphocytes with interleukin-2 (IL-2) in vitro were shown to induce Correspondence to: R.A. Knazek, CellcoAdvanced Bioreactors, Inc., 5516 NicholsonLane, Kensington,MD 20895, U.S.A. Abbreviations: TIL, tumor infiltrating lymphocytes; rlL-2, recombinant interleukin-2; EFS, extra-fiber space; LAK, lymphokine-activatedkiller cells.
regression of certain types of animal and human tumors (Rosenberg, 1986, 1988; Rosenberg et al., 1987; Rosenberg, 1988). The low frequency of responses, while encouraging, prompted further investigations to find more specific and cytolytic subpopulations of lymphocytes. Subsequent animal studies showed that the lymphocytes infiltrating tumors (TIL) had, when expanded in vitro under the stimulus of IL-2, a 50-100-fold increase in their ability to induce tumor regression in mice when compared to LAK cells (Rosenberg et al., 1986). These T I L have been reported to induce tumor regression in some human patients, albeit not as dramatic as in the animal models (Rosenberg et al., 1988; Topalian et al., 1988).
0022-1759/90/$03.50 © 1990 ElsevierSciencePubfishers B.V. (BiomedicalDivision)
30 A protocol at the National Cancer Institute presently administers approximately 2 × 1011 autologous TIL to patients with metastatic cancer. Currently, this is accomplished by growing the initial TIL harvested from a metastatic lesion in plastic, gas permeable culture bags (Topalian et al., 1987). Each bag supports up to 3 × 109 TI1 in a 1.5 liter volume of tissue culture medium containing human serum albumin and recombinant human IL-2. The gradual expansion into 100 such bags over 4-6 weeks for each patient is currently a complex and costly procedure that limits the numbers of patients that can be treated. We now describe a new method for growing TIL to clinically useful quantities in vitro that significantly reduces cost, time, and space requirements. This involves the method of cell culture on hollow fibers (Knazek et al., 1972), wherein cells grew on tube-shaped, semi-permeable membranes to reach nearly solid tissue density. It is shown here, that TIL grown in hollow fiber cartridges expand well and are both phenotypically and functionally comparable to TIL grown in gas-permeable bags.
Materials and methods
Tumor infiltrating lymphocytes Tumors were excised from patients having metastatic melanoma and were transported sterilely from the surgical suite to a laminar flow hood. They were then minced into 1-2 mm pieces and suspended in RPMI 1640 tissue culture media (Biofluids, Rockville, MD) containing 10 mg collagenase/ml, 1 mg deoxyribonuclease/ml, and 2.5 U hyaluronidase/ml (Sigma). The suspension was stirred gently overnight at room temperature. It was then filtered through sterile Nitex mesh, washed twice, and suspended in either AIM-V (Gibco, Grand Island, NY) that contained 1000 U interleukin-2 (IL-2)/ml (kindly provided by Cetus, Emeryville, CA), 10 /~g gentamicin/ml (Gibco), 50 /~g streptomycin/ml (Gibco), 50 U penicillin/ml (Gibco), 1.25 /~g fungizone/ml (Flow Laboratories, MacLean, VA), 2.95 mg glucose/ml and 2 mM glutamine (Flow Laboratories), termed complete AIM-V, that was further supplemented with 20% autologous lymphokine -activated killer
cell (LAK) AIM-V supernatant as described previously (Muul et al., 1986) or a 50:50 mixture of the previous medium and RPMI 1640 having 10% heat-inactivated human serum and further supplemented with LAK supernatant containing 2% human serum. The suspension containing both TIL and tumor cells was plated in 6-well culture plates (Costar, Cambridge, MA) at densities of 5 x 105/ ml. After 1 week or when densities of 1-2 x 106/ ml were reached, the cells were replated in fresh medium at densities of 5 x 105 TIL/ml. Upon further expansion, the cell suspension was diluted to a density of 5 x 105/ml in 0.5-1.5 liters of complete AIM-V medium. This entire volume was then injected into a 1.5 liter polyolefin culture bag (PL-732 plastic, Fenwal Laboratories, Deerfield, IL) and incubated in a flat position on a perforated shelve without agitation a t 37®C for 3-4 days in a humidified 5% CO2/air incubator. Upon expansion, these cell suspensions were periodically diluted 1 / 3 - 4 in complete AIM-V medium in new bags.
Hollow fiber cell culture The hollow fiber cell culture system (Cellmax 100, Cellco Advanced Bioreactors, Kensington, MD) consists of a standard glass medium bottle which serves as the reservoir, a stainless steel/Ryton gear pump, an autoclavable hollow fiber cartridge in which cells are cultured, and medical grade silicone rubber tubing which serves as a gas exchanger to maintain the appropriate pH and pO 2 of the culture medium. All components are secured to a stainless steel tray of sufficiently small dimensions to enable four such systems to fit within a standard tissue culture incubator. The pump speed and automatically reversing flow direction are determined by an electronic control unit which sits outside of the incubator and is connected to the pump motor via a flat ribbon cable which passes through the gasket of the incubator door. The pump motor is magnetically coupled to the pump and is lifted from the system prior to steam autoclaving. Tissue culture medium, drawn from the reservoir, is pumped through the lumen of the hollow fibers. It then passes through the gas exchange tubing in which it is reoxygenated and its pH readjusted prior to returning to the reservoir for subsequent recirculation.
31 The flow rate was increased from 40 to 300 ml/min as the number of cells increased with time. The direction of perfusion through the hollow fiber lumen was reversed automatically every 10 rain. This feature served to provide a more uniform distribution of nutrient supply, waste product removal and cell distribution within the space surrounding the hollow fibers. The hollow fiber bioreactor (B3, Cellco Advanced Bioreactors, Kensington, MD) contains 8000 cellulosic hollow fibers, 12 inches in length, and provides a 1.1 mE surface area. The fiber'walls nominally restrict diffusion to substances having a molecular weight of less than 3000 and divide the cartridge into a 50 ml extra-fiber space (EFS) and the volume within the fiber lumina. The hollow fiber systems, excluding the pump motor, were steam autoclaved at 121°C for 20 rain. The system was then perfused with 1.3 liters deionized water overnight at 37 o C. The perfusion pathway and extra-fiber space of each system were drained and flushed with complete AIM-V medium before replacing the reservoir bottle with a fresh, warmed 1 liter bottle of complete medium. All operations were performed in a sterile, laminar flow hood. Cell suspensions taken from culture bags 17-33 days after excision of the tumor were centrifuged at 400 x g for 10 rain at room temperature. These cells were then injected into the EFS of the hollow fiber bioreactor through the 2 side-ports, each of which had been connected to a 100 ml glass side-port bottle in which TIL were suspended in complete AIM-V medium. The bottles were pressurized gently with a 20 ml plastic syringe to transfer the cell suspension into the extra-fiber space. The cells settled onto fibers from which they received nutrient support by diffusion from the perfusate. Simultaneously, low molecular weight metabolites diffused away from the cells and were diluted by the perfusate. The system was operated in a 37 ° C, humidified, 5 % CO2/air incubator. The perfusion medium was periodically replaced when its glucose concentration decreased to 1-1.5 g/1. To accomplish this, the hollow fiber system was taken from the incubator, placed in a laminar flow hood and a fresh bottle of pre-warmed, complete medium was substituted for the existing, spent bottle of
medium. This re-feeding procedure required approximately 5 rain to complete. The extra-fiber space was drained periodically to either harvest the culture medium, presumably enriched in non-diffusible products secreted by TIL, or to evaluate the number and functional characteristics of the cells. This was accomplished in a laminar flow hood by draining the extra-fiber fluid, by gravity, into one of the two loading side-port bottles. When performed gently, the medium contained only a small percentage of the total number of cells present. These were subsequently recovered during a 200 x g x 10 min centrifugation at room temperature, resuspended in fresh, complete AIM-V medium and then re-inoculated into the extra-fiber space. This procedure required 20-30 min to accomplish during which time, the cartridge was continuously perfused.
Viability Aliquots of cell suspensions were diluted in trypan blue/normal saline (Sigma) to a final concentration 0.04%. Viability was determined by dye exclusion.
Cytotoxicity The chromium release activity of TIL was tested against targets that consisted of autologous tumor cells, allogeneic tumor cells, the NK-sensitive K562 erythroleukemia cell line, and the NK-resistant Daudi B cell lymphoma line (Topalian et al., 1987). Approximately 108 target cells were labelled with 400 /~Ci sodium 51CRO4 (New England Nuclear, Boston, MA) in a 0.7 ml volume for 1 h, washed three times, incubated for another 30 rain at 37°C, and washed twice before use. Serial dilutions of effector cells were plated with 5 x 103 target cells in triplicate at effector:target cell ratios of 80:1, 20:1, and 5:1 in a total of 150 #1 culture medium in 96-well round-bottom microtiter plates (Costar). Supernatants were harvested with the Skatron-Titertek system (Skatron, Lier, Norway) after 4 h incubation and counted in a gamma counter (LKB Instruments, Gaithersburg, MD). Spontaneous target lysis was determined by incubation of tumor target cells in medium alone. Maximal lysis was achieved by target cell incuba-
32 tion with 2% SDS detergent. Specific lysis was determined as: % specificlysis = experimentalrelease- spontaneous release x 100 maximal release- spontaneous release
Phenotype analysis Lymphocytes were washed with cold staining medium (Hanks' buffered saline solution without phenol red containing 5% heat-inactivated fetal calf serum and 0.02% sodium azide) and resuspended at concentrations of 1 x 106 to 1 x 107 cells/ml. Undiluted fluorescein-conjugated monoclonal antibodies against human mononuclear cell antigens (Becton-Dickinson, Mountain View, CA) were added to 100/~1 volumes of Cell suspension at a concentration of 5% (v/v). The antibodies used were: anti-Leu-4 (against T cells), anti-Leu-3a (against T helper/inducer cells), anti-Leu-2b (against T cytotoxic/suppressor cells), anti-Leu-5b (against E rosette receptor-bearing cells), antiLeu-7 (against N K and some T cells), anti-Leu-lla ( N K cells and neutrophils), anti-Leu-14 and -16 (against B cells), anti-HLA-DR (against activated T cells, B cells, and monocytes/macrophages), anti-IL-2 receptor (against activated T cells, TAC), and anti-Leu-M3 (against m o n o c y t e s / m a c r o phages). The antibodies used also included the negative controls anti-Thy-l.2 (against murine T cells), and phycoerythroin. After staining for 30-60 min at 4°C, cells were washed w i t h staining medium, fixed with 1% paraformaldehyde, washed again and resuspended in staining medium. Labeled cells were stored at 4 ° C for 1-7 days. Fluorescence analysis was performed using a FACS 440 microfluorometer interfaced with a Consort 40 computer (Becton-Dickinson).
TIL harvest from Cellmax hollow fiber system The entire hollow fiber system was removed from the incubator and placed in a laminar flow hood. The electronic control unit was then reconnected to the pump motor and perfusion was continued at a rate of 40 m l / m i n to prevent the TIL from becoming anoxic during the harvest
procedure. Approximately one-third of the extrafiber medium was drained by gravity into a loading side-port bottle. The hollow fiber bioreactor was then shaken vigorously to detach TIL from the fibers. The remaining media and cells in suspension were drained into the side-port bottle. The procedure was then repeated × 3, each time with approximately 35 ml fresh, complete medium being placed in the empty side-port bottle and injected into the extra-fiber space prior to shaking the bioreactor. The last two flushes were accomplished by shaking the cartridge with a hand held vibrator (Oster, Milwaukee, WI) for 1 min to remove the more firmly attached cells. The entire procedure required 30 min and resulted in the removal of approximately 95% of the cells present in the bioreactor. After harvesting the TIL from the cartridge of patient G, the cartridge perfusion was re-instituted to determine if the residual cells within the EFS would continue to grow and eventually yield a clinically significant number of cells. The in vitro functional characteristics of cell suspensions harvested from the hollow fiber bioreactors were compared to those of cells harvested from bag cultures on the same day. After the final TIL harvest, the entire Cellmax system was steam-autoclaved at 121°C for 20 min. The system was then flushed with deionized water and the silicone rubber tubing and bioreactors then were discarded. Cleaning, reassembly and reautoclaving with fresh tubing and bioreactor in place was performed before the system was reused for a different patient.
Results TIL obtained from the excised tumors of eight patients with metastatic melanoma were first cultured in plates and then in 1.5 liter, gas permeable bags prior to being used to inoculate the hollow fiber culture system. After an initial lag phase, the cells entered the period of exponential growth within the bags. At that point, as illustrated in Figs. 1 and 2, a small volume of the cell suspension was withdrawn from the bag and injected into the EFS of the hollow fiber bioreactor cartridge. The inocula ranged from 0.35 to 4.3 x
33
?
~1011
d [ /
,o
/
i
/
u /
1010
_~
iI
/
1I
0 ,< a
,,=,
/ t
z
~:
~
,~ /
5
,I~
I
Q
I.-
t
'i ~'oZ
/
Z o 0.1
108
//
,'o
.o.
0.5
,~' ,'/ ~
~"
/ i
101o
/
/
=-
/I
i
I
_o
REPLACEMENTOF EXTRA*FIBER SPACE MEDIA
10 9
!
/
XOLCOWFIBEBS
a //
,,=,
//I
0,5
*
tt
I-"
iI
1011
/---O----CELL NUMBER, BAGS ~ - - - ~ - - - CELL NUMBER, HOLLOW FIBERS i .~.O.-,.--GLUCOSE CONSUMPTION,
---O---CELL NUMBER. BAGS ---~---CELL NUMBER, HOLLOW FIBERS GLUCOSE CONSUMPTION, HOLLOW FIBERS + REPLACEMENTOF EXTRA-FIBER SPACE MEDIA
, ' /,"
,/
/
l
J
¢~
o
z
0
;o TIME, D A Y S
0.1
~
/
~0.05 LJ
I
/
3
/
,,
o
_.~
I I
Fig. 1. Typical growth curve showing W-TIL increasing in cell number vs. time elapsed from date of excisional biopsy. TIL withdrawn from on bag on day 23 were inoculated into a Cellmax hollow fiber bioreactor and harvested on day 44. The inoculum of 4.3×108 TIL yielded a 5.4×101° TIL harvest. Glucose concentrations of bioreactor perfusate vs. duration of culture increased logarithmically with time.
10 8 TIL. Cells from two of the eight cultures stopped growing simultaneously in both the bag and hollow fiber cultures for unknown reasons.
/ , ~
ool
1'o
;,o
3o
REGROWTH FROM RESIDUAL CELLS
4o
1o,
DURATION OF CULTURE, D A Y S
Fig. 2. Growth curve showing G-TIL increasing in cell number vs. time elapsed from date of exc~sional biopsy. TIL withdrawn from one bag on day 16 were inoculated into a Cellmax hollow fiber bioreactor and harvested on day 30. The inoculum of 1.0 × 10 s TIL yielded a 1.5 × 101° TIL harvest. Perfusion of the hollow fiber bioreactor was re-instituted and the residual TIL expanded again to 1.5 x 10 l° by the second harvest on day 52. Glucose concentrations of bioreactor perfusate vs. duration of culture increased logarithmically with time.
TABLE I TIL
HF inoculated (Days post bx)
HF culture duration (days)
Cell inoculum ( x l 0 -S)
Final harvest (×10 -1°)
B
33 31
21 32
58
29
24 24 17 31
26 22 14 21
0.5 0.35 1.3 10.0 0.4 1.8 4.3 1.0 Residual
4.5 85% 3.6 12 4.1 97% 3.5 25 No growth in bags or hollow fiber bioreactor (26 ml) 1.9 12 No growth in bags or hollow fiber bioreactor 2.8 91% 1.7 9 5.4 93% 2.4 12 1.5 87% 2.9 7 1.5 89% 3.3 8
M S K J H W G G
Harvest viability
Terminal glucose consumption (g/day)
Reservoir changes (n)
EFS changes (n)
Total media used (liter)
Media used per 10 l° (liter)
10 14
11.3 27
2.5 6.6
10
13
12 7 2 0
10 13 8 8.8
3.6 2.4 5.0 5.8
34
Fig. 3. a: scanning electron micrograph of W - T I L within the extra-fiber space of a Cellmax hollow fiber bioreactor. The space between the large, ovoid hollow fibers is filled with a nearly solid mass of TIL. The ovoid shape of the hollow fibers and the space between the TIL mass and the fiber surfaces are artifacts of histologic preparation, b: a higher magnification shows the individual cells near the outer surface of a single hollow fiber.
Since those bag and hollow fiber cultures were being maintained in separate incubators, it would seem that such a growth failure was due to an inherent biological characteristic of the cells as yet undefined. The remaining six TIL cultures grew well in both the hollow fiber system and bags. The hollow fiber cultures were maintained for periods ranging from 14 to 32 days after which time 1.5-5.4 × 101° TIL were harvested from single hollow fiber bioreactors (Table I). These represented 124-1170-fold expansions of the inocula that filled a significant percentage of the EFS and eventually formed a nearly solid mass of TIL (Fig. 3). The viability of these cells averaged 91%. One cell harvest was frozen as a 26 ml packed cell pellet before the cell number and viability had been determined. At the time of harvest from each hollow fiber cartridge, the TIL were flushed from the EFS by approximately 200 ml medium which was in turn centrifuged to yield the final, packed TIL pellet ready for subsequent use. Approximately 30 min were required to perform the harvest procedure. During the course of a culture, aliquots of perfusate were assayed to determine the glucose and lactate concentrations (Yellow Springs Instrument, Yellow Springs, OH) permitting the rates of
consumption and production, respectively, to be determined (Figs. 1 and 2). Such rates exhibited logarithmic increases over time, doubling every 1.5-3.6 days. These were commensurate with the 1.5-3.2 day doubling time of the cell number of the same TIL grown in bags. Presumably then, the glucose utilization reflects the number of metabolically active cells present and can be used as an index of the well-being of the culture. However, the rate of glucose consumption by TIL from different patients varied between 0.45 and 2.2 g/101° TIL per 24 h (Table I). The rate of lactic acid production approximated that of the glucose consumption rate for all patient TIL studied (data not shown). The spent reservoir bottle was replaced with another containing fresh medium when the glucose concentration had dropped into the range of 100-150 mg/dl. The amount of medium used, therefore, was dependent, in large part, on the rate of glucose consumed by each culture. This ranged from 8 to 27 liters per hollow fiber system: a utilization rate of 2.4-6.6 liters complete AIM-V medium per 101° TIL harvested (Table I). The time required to replace the reservoir bottle required less than 5 rain. Medium in the extra-fiber space (EFS) of the
35 TABLE II
first five successful h o l l o w fiber T I L cultures (B, M, K, H, W) was replaced, on average, every 2.2 days to decrease the p o t e n t i a l of a b u i l d u p of p o o r l y diffusable inhibitors of cell g r o w t h or function. T h e E F S o f the last culture (G), however, was replaced twice an d 0 times during the 14 a n d 21 day culture periods, respectively. T h e glucose c o n s u m p t i o n rate curves were n o t adversely effected by the decreased f r e q u e n c y of E F S changes. T h e a m o u n t of time r e q u i r e d for each E F S c h a n g e was 2 0 - 3 0 min. This i n c l u d e d a 10 m i n c e n t r i f u g a t i o n to recover a n d r e i n o c u l a t e T I L that h a d been flushed o u t of the E F S during the procedure. D e c r e a s i n g the n u m b e r of E F S changes in a culture, therefore, resulted in a r e d u c t i o n of the a m o u n t o f time e x p e n d e d in m a i n t a i n i n g that culture. T I L that r e m a i n e d within the h o l l o w fiber bior e a c t o r cartridge (patient G) after a harvest, regrew to clinically useful n u m b e r s w h e n the cartridge p er f u s i o n was c o n t i n u e d (Fig. 2). T h e
Phenotypic profile CD3 (Leu-4) CD4 (Leu-3) CD8 (Leu-2) CDllb (Leu-15) CD14 (Leu-M3) CD16 (Leu-ll) CD20 (Leu-16) CD56 (Leu-19) CD57 (Leu-7) HLA-DR Tac
K-HF
K-bag
G-HF
G-bag
94.1 1.1 95.9 0.3 -0.4 0.6 3.7 8.3 4.2 80.7 2.4
94.7 2.0 90.4 9.5 0.3 0.8 4.0 20.4 11.1 85.6 4.7
85.3 13.1 73.8 3.7 0.6 2.2 0.1 6.5 2.2 89.3 27.3
91.4 14.3 83.1 0.4 1.1 1.4 0.8 7.1 4.6 91.1 30.4
rates of increase in glucose c o n s u m p t i o n during the first a n d second g r o w t h p er i o d s were indistinguishable. E x t r a p o l a t i o n of the second h o l l o w fiber g r o w t h cu r v e b a c k to the date of the first harvest i n d i c a t e d that a p p r o x i m a t e l y 1 × 10 7 T I L h a d r e m a i n e d w i t h i n the h o l l o w fiber cartridge to
TABLE IIl CYTOLYTIC ACTIVITY (% LYSIS) Effector
Target cells
cell
K562
Daudi
Tumor Autologous
*
20:1
5:1
LAK M-TIL (HF) K-TIL (bag) K-TIL (HF)
44.4 6.6 58.7 21.6
12.9 5.3 33.0 11.2
6.9 0.8 12.5 7.4
28.4 2.5 19.1 4.8
9.1 1.1 11.5 5.0
5.7 4.1 3.7 4.5
LAK H-TIL (bag) H-TIL (HF) K-TIL (HF)
65.2 7.4 2.3 4.2
23.0 5.8 0.9 1.3
7.3 0.8 0.7 0.8
43.6 2.6 - 0.8 1.8
14.1 5.6 2.8 - 2.1
8.4 6.6 2.8 - 3.1
17.2 50.2 49.9
8.9 57.1 42.9 ND
4.3 33.7 39.6
4.6 6.6 0.8 3.6
1.1 0.1 5.0 4.9
3.6 - 7.3 - 0.6 0.1
LAK W-TIL (bag) W-TIL (HF)
81.0 30.5 22.5
44.0 24.5 11.2
18.0 22.7 12.5
38.8 64.0 55.1
18.6 53.2 45.0
6.0 29.2 19.8
5.1 30.6 19.4
0.9 14.9 11.9
2.0 3.2 4.4
9.1 4.8 2.0
6.6 6.8 4.1
0.9 4.5 5.8
LAK W-TIL (bag) W-TIL (HF)
86.8 15.0 8.6
75.6 6.9 2.5
27.4 0.9 0.3
69.8 72.2 55.9
45.8 53.1 41.3
17.2 21.7 17.0
34.3 15.4 11.5
13.6 7.9 0
4.1 2.8 2.0
26.0 5.5 3.9
11.7 12.7 5.4
4.5 2.5 5.0
LAK G-TIL (bag) G-TIL (HF)
74.3 13.5 11.1
56.1 8.5 10.1
49.9 5.2 6.1
59.7 21.3 3.9
45.9 4.9 9.6
40.0 2.6 7.6
69.9 51.1 50.4
41.3 32.0 40.4
18.1 22.3 38.2
37.1 3.9 3.4
43.1 2.9 3.2
17.7 2.5 3.6
a ND, not done. * Effector: target cell ratios
80:1
20:1
5:1
80:1
20:1
Allogeneic
80:1
5:1
80:1
ND ND ND ND
20:1
5:1
ND ND ND ND
36 re-initiate growth of the second TIL batch. This is based on the assumption that the TIL doubling time in the first and second period of growth were identical. Cells removed simultaneously from bag and hollow fiber cultures yielded similar surface antigen profiles (Table II). The slight variations between them were of the same magnitude as those seen between bags that had originated from a single bag culture (data not shown). The cytolytic capacities of three dilutions of TIL taken from simultaneous aliquots of bag and hollow fiber cultures are presented in Table III. TIL caused 51chromium to be released from the target cells in quantities that were similar whether they were cultured in bags or hollow fibers. The lytic capacity of TIL was most often greater in the autologous tumor system than in the allogeneic tumor system and the pattern of lysis of K562 and Daudi target cells showed no difference between bag-cultured and hollow fiber-cultured TIL.
Discussion
Tumor infiltrating lymphocytes hold promise as a means for the treatment of certain types of cancer (Rosenberg et al., 1988). The investigation and implementation of this form of adoptive immunotherapy, however, has been hindered by the cumbersome and expensive methods currently used to grow the large number of TIL required for each patient. The studies presented herein, demonstrate that large numbers of TIL can be generated by culture in hollow fiber bioreactors. The rates of growth of these cells in the gas permeable bags currently used and in the proposed hollow fiber device are indistinguishable. Likewise, the phenotypic and cytotoxic characteristics of TIL obtained from either culture method are not significantly different. Total amounts of medium required appear to vary between patients as a function of the metabolic activity of their TIL within the hollow fiber bioreactor. Medium thus expended can be minimized by monitoring the glucose concentration of the perfusate and replacing the reservoir bottle when their levels drop to 1-1.5 g/1. While cells
grown within bags routinely required 1 liter of medium for 1-2 x 109 T•L, the hollow fiber cultures required 1 liter of medium to yield 1.5-4.2 × 109 TIL. Although none of the TIL described in this paper were used for patient therapy, the numbers of cells obtained from a single hollow fiber bioreactor approximated clinically relevant amounts. More than one hollow fiber cartridge would be needed to provide 2 x 1011 TIL, but this would be accompanied by approximately an 80% reduction in the amounts of both technician time expended and incubator space required in comparison to bag cultures. Furthermore, multiple doses of TIL could be harvested from a single hollow fiber cartridge and thus be made available for periodic use, an approach to therapy that may prove to be advantageous. Absence of a requirement to periodically replace the extra-fiber space medium indicates that TIL do not produce growth inhibitors having molecular weights greater than 3000-4000. This is in contradistinction to the hybridomas which synthesize TGF-fll that causes an auto-inhibition of growth and antibody secretion (Kidwell, 1989). The first studies of mammalian cells grown within a hollow fiber cartridge demonstrated both that solid tissue densities could be reached in vitro (Knazek et al., 1972) and that such cultures secreted 3-13-fold more biologically active product per cell than did the same cells grown in monolayer culture (Knazek, 1974; Knazek et al., 1974). Such increased efficiency may be a result of improved nutrient support or metabolite removal, or be secondary to the marked intensity of cell-cell interaction and concomitant retention of poorly diffusible autocrine factors that stimulate a more in vivo-like behavior of the cells. Addition of various cytokines or irradiated autologous tumor cells (Kawakami et al., 1988; Stotter et al., 1989) into the extra-fiber space of the hollow fiber bioreactors may further increase specific cytotoxicity of TIL; such cells would have the dual advantage of access to a large nutrient pool and concurrent exposure to high concentrations of these poorly diffusible stimuli. It is apparent, that the method of cell culture on hollow fibers may be applicable to the expansion of TIL in vitro and thus ease the burdens of
37 such production within both clinical and research laboratories.
References Kawakami, Y., Rosenberg, S.A. and Lotze, M.T. (1988) Interleukin-4 promotes the growth of tumor-infiltrating lymphocytes cytotoxic for human autologous melanoma. J. Exp. Med. 168, 2183. Kidwell, W.R. (1989) Filtering out inhibition. Biotechnology 7, 462. Knazek, R.A. (1974) Solid tissue masses formed in vitro form cells cultured on artificial capillaries. Fed. Proc. 33, 1978. Knazek, R.A., Gullino, P.M., Kohler, P.O. and Dedrick, R.L. (1972) Cell culture on artificial capillaries: an approach to tissue growth in vitro. Science 178, 65. Knazek, R.A., Kohler, P.O. and Gullino, P.M. (1974) Hormone production by cells grown in vitro on artificial capillaries. Exp. Cell Res. 84, 251. Muul, L.M., Director, E.P., Hyatt, C.L. and Rosenberg, S.A. (1986) Large scale production of human lymphokineactivated killer cells for use in adoptive immunotherapy. J. Immunol. Methods 88, 265. Rosenberg, S.A. (1986) Adoptive irnmunotherapy of cancer using lymphokine activated killer cells and recombinant interleukin-2. In: V.T. DeVita, S. Hellman and S.A. Rosenberg (Eds.), Important Advances in Oncology. JBL Lippincott, Philadelphia, PA, pp. 55-91.
Rosenberg, S.A. (1988) The development of new immunotherapies for the treatment of cancer using interleukin-2. Ann. Surg. 208, 121. Rosenberg, S.A., Spiess, P. and Lafreniere, R. (1986) A new approach to the adoptive immunotherapy of cancer with tumor-infiltrating lymphocytes. Science 223, 1318. Rosenberg, S.A., Lotze, M.T., Muul, L.M., Chang, A.E., et al. (1987) A progress report on the treatment of 157 patients with advanced cancer using lymphokine-activated killer cells and interleukin-2 or high-dose interleukin-2 alone. New Engl. J. Med. 316, 889. Rosenberg, S.A., Packard, B.S., Aebersold, P.M., Solomon, D., et al. (1988) Use of tumor-infiltrating lymphocytes and interleukin-2 in the immunotherapy of patients with metastatic melanoma. New Engl. J. Med. 319, 1676. Stotter, H., Wiebke, E.A., Tomita, S., Belldegrun, A., et al. (1989) Cytokines alster target cell susceptibility to lysis. II. Evaluation of tumor infiltrating lymphocytes. J. Immunol. 142, 1767. Topalian, S.L., Muul, L.M., Solomon, D. and Rosenberg, S.A. (1987) Expansion of human tumor infiltrating lymphocytes for use in immunology trials. J. Immunol. Methods 102, 127. Topalian, S.L., Solomon, D., Avis, F.P., Chang, A.E., et al. (1988) Immunotherapy of patients with advanced cancer using tumor-infiltrating lymphocytes and recombinant interleukin-2: a pilot study. J. Clin. Oncol. 6, 839.
29
Elsevier JIM05456
Culture of human tumor infiltrating lymphocytes in hollow fiber bioreactors R i c h a r d A. K n a z e k
1 Yan-Wan
Wu
2, Paul
M. A e b e r s o l d 3 a n d S t e v e n A. R o s e n b e r g 3
1 Cellco Advanced Bioreactors, lnc., Kens,ington, MD, U.S.A., 2 Developmental Endocrinology Branch, National lnstitute of Child Health and Human Development, National lnstitutes of Health, Bethesda, MD, U.S.A., and 3 Surgery Branch, Division of Cancer Treatment, National Cancer Institute, National Institutes of Health, Bethesda, MD, U.S.A.
(Received22 September1989, accepted 11 October 1989)
Human Lumor infiltrating lymphocytes ( T I L ) f r o m metastatic melanoma of six patients were grown using a new hollow fiber bioreactor system. After inoculating 0.35-10 x 108 TIL into the extra-fiber space (EFS), each Cellmax bioreactor was perfused with AIM-V medium, supplemented with rlL-2. The cells subsequently expanded 124-1170-fold to yield 1.5-5.4 × 101° TIL over a 14-32 day period. TIL were flushed from the EFS using 200 ml medium and possessed an average viability = 91%. The phenotype and the autologous tumor cell lytic capacity of these TIL were similar to those of TIL grown in the currently used gas-permeable culture bags. Tissue culture media use averaged 4.3 liters/101° TIL harvested. The TIL of one patient were re-expanded twice from cells remaining within the same bioreactor after harvest suggesting that one bioreactor cartridge could be used for repetitive, periodic studies. An estimated 80% decrease in technical time expended and in incubator space requirements were realized using this methodology. Cell culture on hollow fibers appears to be a useful method for producing large quantities of primary human lymphocytes for experimental, and perhaps, therapeutic needs. Key words: Hollow fiber; Tumor infiltrating lymphocyte;Interleukin-2; Cell culture; Lymphocyte
Introduction Both animal studies and early clinical trials have demonstrated that adoptive immunotherapy holds promise as a potential mode of treatment for human malignancies. Initially, lymphokineactivated killer cells (LAK), obtained by incubating peripheral blood lymphocytes with interleukin-2 (IL-2) in vitro were shown to induce Correspondence to: R.A. Knazek, CellcoAdvanced Bioreactors, Inc., 5516 NicholsonLane, Kensington,MD 20895, U.S.A. Abbreviations: TIL, tumor infiltrating lymphocytes; rlL-2, recombinant interleukin-2; EFS, extra-fiber space; LAK, lymphokine-activatedkiller cells.
regression of certain types of animal and human tumors (Rosenberg, 1986, 1988; Rosenberg et al., 1987; Rosenberg, 1988). The low frequency of responses, while encouraging, prompted further investigations to find more specific and cytolytic subpopulations of lymphocytes. Subsequent animal studies showed that the lymphocytes infiltrating tumors (TIL) had, when expanded in vitro under the stimulus of IL-2, a 50-100-fold increase in their ability to induce tumor regression in mice when compared to LAK cells (Rosenberg et al., 1986). These T I L have been reported to induce tumor regression in some human patients, albeit not as dramatic as in the animal models (Rosenberg et al., 1988; Topalian et al., 1988).
0022-1759/90/$03.50 © 1990 ElsevierSciencePubfishers B.V. (BiomedicalDivision)
30 A protocol at the National Cancer Institute presently administers approximately 2 × 1011 autologous TIL to patients with metastatic cancer. Currently, this is accomplished by growing the initial TIL harvested from a metastatic lesion in plastic, gas permeable culture bags (Topalian et al., 1987). Each bag supports up to 3 × 109 TI1 in a 1.5 liter volume of tissue culture medium containing human serum albumin and recombinant human IL-2. The gradual expansion into 100 such bags over 4-6 weeks for each patient is currently a complex and costly procedure that limits the numbers of patients that can be treated. We now describe a new method for growing TIL to clinically useful quantities in vitro that significantly reduces cost, time, and space requirements. This involves the method of cell culture on hollow fibers (Knazek et al., 1972), wherein cells grew on tube-shaped, semi-permeable membranes to reach nearly solid tissue density. It is shown here, that TIL grown in hollow fiber cartridges expand well and are both phenotypically and functionally comparable to TIL grown in gas-permeable bags.
Materials and methods
Tumor infiltrating lymphocytes Tumors were excised from patients having metastatic melanoma and were transported sterilely from the surgical suite to a laminar flow hood. They were then minced into 1-2 mm pieces and suspended in RPMI 1640 tissue culture media (Biofluids, Rockville, MD) containing 10 mg collagenase/ml, 1 mg deoxyribonuclease/ml, and 2.5 U hyaluronidase/ml (Sigma). The suspension was stirred gently overnight at room temperature. It was then filtered through sterile Nitex mesh, washed twice, and suspended in either AIM-V (Gibco, Grand Island, NY) that contained 1000 U interleukin-2 (IL-2)/ml (kindly provided by Cetus, Emeryville, CA), 10 /~g gentamicin/ml (Gibco), 50 /~g streptomycin/ml (Gibco), 50 U penicillin/ml (Gibco), 1.25 /~g fungizone/ml (Flow Laboratories, MacLean, VA), 2.95 mg glucose/ml and 2 mM glutamine (Flow Laboratories), termed complete AIM-V, that was further supplemented with 20% autologous lymphokine -activated killer
cell (LAK) AIM-V supernatant as described previously (Muul et al., 1986) or a 50:50 mixture of the previous medium and RPMI 1640 having 10% heat-inactivated human serum and further supplemented with LAK supernatant containing 2% human serum. The suspension containing both TIL and tumor cells was plated in 6-well culture plates (Costar, Cambridge, MA) at densities of 5 x 105/ ml. After 1 week or when densities of 1-2 x 106/ ml were reached, the cells were replated in fresh medium at densities of 5 x 105 TIL/ml. Upon further expansion, the cell suspension was diluted to a density of 5 x 105/ml in 0.5-1.5 liters of complete AIM-V medium. This entire volume was then injected into a 1.5 liter polyolefin culture bag (PL-732 plastic, Fenwal Laboratories, Deerfield, IL) and incubated in a flat position on a perforated shelve without agitation a t 37®C for 3-4 days in a humidified 5% CO2/air incubator. Upon expansion, these cell suspensions were periodically diluted 1 / 3 - 4 in complete AIM-V medium in new bags.
Hollow fiber cell culture The hollow fiber cell culture system (Cellmax 100, Cellco Advanced Bioreactors, Kensington, MD) consists of a standard glass medium bottle which serves as the reservoir, a stainless steel/Ryton gear pump, an autoclavable hollow fiber cartridge in which cells are cultured, and medical grade silicone rubber tubing which serves as a gas exchanger to maintain the appropriate pH and pO 2 of the culture medium. All components are secured to a stainless steel tray of sufficiently small dimensions to enable four such systems to fit within a standard tissue culture incubator. The pump speed and automatically reversing flow direction are determined by an electronic control unit which sits outside of the incubator and is connected to the pump motor via a flat ribbon cable which passes through the gasket of the incubator door. The pump motor is magnetically coupled to the pump and is lifted from the system prior to steam autoclaving. Tissue culture medium, drawn from the reservoir, is pumped through the lumen of the hollow fibers. It then passes through the gas exchange tubing in which it is reoxygenated and its pH readjusted prior to returning to the reservoir for subsequent recirculation.
31 The flow rate was increased from 40 to 300 ml/min as the number of cells increased with time. The direction of perfusion through the hollow fiber lumen was reversed automatically every 10 rain. This feature served to provide a more uniform distribution of nutrient supply, waste product removal and cell distribution within the space surrounding the hollow fibers. The hollow fiber bioreactor (B3, Cellco Advanced Bioreactors, Kensington, MD) contains 8000 cellulosic hollow fibers, 12 inches in length, and provides a 1.1 mE surface area. The fiber'walls nominally restrict diffusion to substances having a molecular weight of less than 3000 and divide the cartridge into a 50 ml extra-fiber space (EFS) and the volume within the fiber lumina. The hollow fiber systems, excluding the pump motor, were steam autoclaved at 121°C for 20 rain. The system was then perfused with 1.3 liters deionized water overnight at 37 o C. The perfusion pathway and extra-fiber space of each system were drained and flushed with complete AIM-V medium before replacing the reservoir bottle with a fresh, warmed 1 liter bottle of complete medium. All operations were performed in a sterile, laminar flow hood. Cell suspensions taken from culture bags 17-33 days after excision of the tumor were centrifuged at 400 x g for 10 rain at room temperature. These cells were then injected into the EFS of the hollow fiber bioreactor through the 2 side-ports, each of which had been connected to a 100 ml glass side-port bottle in which TIL were suspended in complete AIM-V medium. The bottles were pressurized gently with a 20 ml plastic syringe to transfer the cell suspension into the extra-fiber space. The cells settled onto fibers from which they received nutrient support by diffusion from the perfusate. Simultaneously, low molecular weight metabolites diffused away from the cells and were diluted by the perfusate. The system was operated in a 37 ° C, humidified, 5 % CO2/air incubator. The perfusion medium was periodically replaced when its glucose concentration decreased to 1-1.5 g/1. To accomplish this, the hollow fiber system was taken from the incubator, placed in a laminar flow hood and a fresh bottle of pre-warmed, complete medium was substituted for the existing, spent bottle of
medium. This re-feeding procedure required approximately 5 rain to complete. The extra-fiber space was drained periodically to either harvest the culture medium, presumably enriched in non-diffusible products secreted by TIL, or to evaluate the number and functional characteristics of the cells. This was accomplished in a laminar flow hood by draining the extra-fiber fluid, by gravity, into one of the two loading side-port bottles. When performed gently, the medium contained only a small percentage of the total number of cells present. These were subsequently recovered during a 200 x g x 10 min centrifugation at room temperature, resuspended in fresh, complete AIM-V medium and then re-inoculated into the extra-fiber space. This procedure required 20-30 min to accomplish during which time, the cartridge was continuously perfused.
Viability Aliquots of cell suspensions were diluted in trypan blue/normal saline (Sigma) to a final concentration 0.04%. Viability was determined by dye exclusion.
Cytotoxicity The chromium release activity of TIL was tested against targets that consisted of autologous tumor cells, allogeneic tumor cells, the NK-sensitive K562 erythroleukemia cell line, and the NK-resistant Daudi B cell lymphoma line (Topalian et al., 1987). Approximately 108 target cells were labelled with 400 /~Ci sodium 51CRO4 (New England Nuclear, Boston, MA) in a 0.7 ml volume for 1 h, washed three times, incubated for another 30 rain at 37°C, and washed twice before use. Serial dilutions of effector cells were plated with 5 x 103 target cells in triplicate at effector:target cell ratios of 80:1, 20:1, and 5:1 in a total of 150 #1 culture medium in 96-well round-bottom microtiter plates (Costar). Supernatants were harvested with the Skatron-Titertek system (Skatron, Lier, Norway) after 4 h incubation and counted in a gamma counter (LKB Instruments, Gaithersburg, MD). Spontaneous target lysis was determined by incubation of tumor target cells in medium alone. Maximal lysis was achieved by target cell incuba-
32 tion with 2% SDS detergent. Specific lysis was determined as: % specificlysis = experimentalrelease- spontaneous release x 100 maximal release- spontaneous release
Phenotype analysis Lymphocytes were washed with cold staining medium (Hanks' buffered saline solution without phenol red containing 5% heat-inactivated fetal calf serum and 0.02% sodium azide) and resuspended at concentrations of 1 x 106 to 1 x 107 cells/ml. Undiluted fluorescein-conjugated monoclonal antibodies against human mononuclear cell antigens (Becton-Dickinson, Mountain View, CA) were added to 100/~1 volumes of Cell suspension at a concentration of 5% (v/v). The antibodies used were: anti-Leu-4 (against T cells), anti-Leu-3a (against T helper/inducer cells), anti-Leu-2b (against T cytotoxic/suppressor cells), anti-Leu-5b (against E rosette receptor-bearing cells), antiLeu-7 (against N K and some T cells), anti-Leu-lla ( N K cells and neutrophils), anti-Leu-14 and -16 (against B cells), anti-HLA-DR (against activated T cells, B cells, and monocytes/macrophages), anti-IL-2 receptor (against activated T cells, TAC), and anti-Leu-M3 (against m o n o c y t e s / m a c r o phages). The antibodies used also included the negative controls anti-Thy-l.2 (against murine T cells), and phycoerythroin. After staining for 30-60 min at 4°C, cells were washed w i t h staining medium, fixed with 1% paraformaldehyde, washed again and resuspended in staining medium. Labeled cells were stored at 4 ° C for 1-7 days. Fluorescence analysis was performed using a FACS 440 microfluorometer interfaced with a Consort 40 computer (Becton-Dickinson).
TIL harvest from Cellmax hollow fiber system The entire hollow fiber system was removed from the incubator and placed in a laminar flow hood. The electronic control unit was then reconnected to the pump motor and perfusion was continued at a rate of 40 m l / m i n to prevent the TIL from becoming anoxic during the harvest
procedure. Approximately one-third of the extrafiber medium was drained by gravity into a loading side-port bottle. The hollow fiber bioreactor was then shaken vigorously to detach TIL from the fibers. The remaining media and cells in suspension were drained into the side-port bottle. The procedure was then repeated × 3, each time with approximately 35 ml fresh, complete medium being placed in the empty side-port bottle and injected into the extra-fiber space prior to shaking the bioreactor. The last two flushes were accomplished by shaking the cartridge with a hand held vibrator (Oster, Milwaukee, WI) for 1 min to remove the more firmly attached cells. The entire procedure required 30 min and resulted in the removal of approximately 95% of the cells present in the bioreactor. After harvesting the TIL from the cartridge of patient G, the cartridge perfusion was re-instituted to determine if the residual cells within the EFS would continue to grow and eventually yield a clinically significant number of cells. The in vitro functional characteristics of cell suspensions harvested from the hollow fiber bioreactors were compared to those of cells harvested from bag cultures on the same day. After the final TIL harvest, the entire Cellmax system was steam-autoclaved at 121°C for 20 min. The system was then flushed with deionized water and the silicone rubber tubing and bioreactors then were discarded. Cleaning, reassembly and reautoclaving with fresh tubing and bioreactor in place was performed before the system was reused for a different patient.
Results TIL obtained from the excised tumors of eight patients with metastatic melanoma were first cultured in plates and then in 1.5 liter, gas permeable bags prior to being used to inoculate the hollow fiber culture system. After an initial lag phase, the cells entered the period of exponential growth within the bags. At that point, as illustrated in Figs. 1 and 2, a small volume of the cell suspension was withdrawn from the bag and injected into the EFS of the hollow fiber bioreactor cartridge. The inocula ranged from 0.35 to 4.3 x
33
?
~1011
d [ /
,o
/
i
/
u /
1010
_~
iI
/
1I
0 ,< a
,,=,
/ t
z
~:
~
,~ /
5
,I~
I
Q
I.-
t
'i ~'oZ
/
Z o 0.1
108
//
,'o
.o.
0.5
,~' ,'/ ~
~"
/ i
101o
/
/
=-
/I
i
I
_o
REPLACEMENTOF EXTRA*FIBER SPACE MEDIA
10 9
!
/
XOLCOWFIBEBS
a //
,,=,
//I
0,5
*
tt
I-"
iI
1011
/---O----CELL NUMBER, BAGS ~ - - - ~ - - - CELL NUMBER, HOLLOW FIBERS i .~.O.-,.--GLUCOSE CONSUMPTION,
---O---CELL NUMBER. BAGS ---~---CELL NUMBER, HOLLOW FIBERS GLUCOSE CONSUMPTION, HOLLOW FIBERS + REPLACEMENTOF EXTRA-FIBER SPACE MEDIA
, ' /,"
,/
/
l
J
¢~
o
z
0
;o TIME, D A Y S
0.1
~
/
~0.05 LJ
I
/
3
/
,,
o
_.~
I I
Fig. 1. Typical growth curve showing W-TIL increasing in cell number vs. time elapsed from date of excisional biopsy. TIL withdrawn from on bag on day 23 were inoculated into a Cellmax hollow fiber bioreactor and harvested on day 44. The inoculum of 4.3×108 TIL yielded a 5.4×101° TIL harvest. Glucose concentrations of bioreactor perfusate vs. duration of culture increased logarithmically with time.
10 8 TIL. Cells from two of the eight cultures stopped growing simultaneously in both the bag and hollow fiber cultures for unknown reasons.
/ , ~
ool
1'o
;,o
3o
REGROWTH FROM RESIDUAL CELLS
4o
1o,
DURATION OF CULTURE, D A Y S
Fig. 2. Growth curve showing G-TIL increasing in cell number vs. time elapsed from date of exc~sional biopsy. TIL withdrawn from one bag on day 16 were inoculated into a Cellmax hollow fiber bioreactor and harvested on day 30. The inoculum of 1.0 × 10 s TIL yielded a 1.5 × 101° TIL harvest. Perfusion of the hollow fiber bioreactor was re-instituted and the residual TIL expanded again to 1.5 x 10 l° by the second harvest on day 52. Glucose concentrations of bioreactor perfusate vs. duration of culture increased logarithmically with time.
TABLE I TIL
HF inoculated (Days post bx)
HF culture duration (days)
Cell inoculum ( x l 0 -S)
Final harvest (×10 -1°)
B
33 31
21 32
58
29
24 24 17 31
26 22 14 21
0.5 0.35 1.3 10.0 0.4 1.8 4.3 1.0 Residual
4.5 85% 3.6 12 4.1 97% 3.5 25 No growth in bags or hollow fiber bioreactor (26 ml) 1.9 12 No growth in bags or hollow fiber bioreactor 2.8 91% 1.7 9 5.4 93% 2.4 12 1.5 87% 2.9 7 1.5 89% 3.3 8
M S K J H W G G
Harvest viability
Terminal glucose consumption (g/day)
Reservoir changes (n)
EFS changes (n)
Total media used (liter)
Media used per 10 l° (liter)
10 14
11.3 27
2.5 6.6
10
13
12 7 2 0
10 13 8 8.8
3.6 2.4 5.0 5.8
34
Fig. 3. a: scanning electron micrograph of W - T I L within the extra-fiber space of a Cellmax hollow fiber bioreactor. The space between the large, ovoid hollow fibers is filled with a nearly solid mass of TIL. The ovoid shape of the hollow fibers and the space between the TIL mass and the fiber surfaces are artifacts of histologic preparation, b: a higher magnification shows the individual cells near the outer surface of a single hollow fiber.
Since those bag and hollow fiber cultures were being maintained in separate incubators, it would seem that such a growth failure was due to an inherent biological characteristic of the cells as yet undefined. The remaining six TIL cultures grew well in both the hollow fiber system and bags. The hollow fiber cultures were maintained for periods ranging from 14 to 32 days after which time 1.5-5.4 × 101° TIL were harvested from single hollow fiber bioreactors (Table I). These represented 124-1170-fold expansions of the inocula that filled a significant percentage of the EFS and eventually formed a nearly solid mass of TIL (Fig. 3). The viability of these cells averaged 91%. One cell harvest was frozen as a 26 ml packed cell pellet before the cell number and viability had been determined. At the time of harvest from each hollow fiber cartridge, the TIL were flushed from the EFS by approximately 200 ml medium which was in turn centrifuged to yield the final, packed TIL pellet ready for subsequent use. Approximately 30 min were required to perform the harvest procedure. During the course of a culture, aliquots of perfusate were assayed to determine the glucose and lactate concentrations (Yellow Springs Instrument, Yellow Springs, OH) permitting the rates of
consumption and production, respectively, to be determined (Figs. 1 and 2). Such rates exhibited logarithmic increases over time, doubling every 1.5-3.6 days. These were commensurate with the 1.5-3.2 day doubling time of the cell number of the same TIL grown in bags. Presumably then, the glucose utilization reflects the number of metabolically active cells present and can be used as an index of the well-being of the culture. However, the rate of glucose consumption by TIL from different patients varied between 0.45 and 2.2 g/101° TIL per 24 h (Table I). The rate of lactic acid production approximated that of the glucose consumption rate for all patient TIL studied (data not shown). The spent reservoir bottle was replaced with another containing fresh medium when the glucose concentration had dropped into the range of 100-150 mg/dl. The amount of medium used, therefore, was dependent, in large part, on the rate of glucose consumed by each culture. This ranged from 8 to 27 liters per hollow fiber system: a utilization rate of 2.4-6.6 liters complete AIM-V medium per 101° TIL harvested (Table I). The time required to replace the reservoir bottle required less than 5 rain. Medium in the extra-fiber space (EFS) of the
35 TABLE II
first five successful h o l l o w fiber T I L cultures (B, M, K, H, W) was replaced, on average, every 2.2 days to decrease the p o t e n t i a l of a b u i l d u p of p o o r l y diffusable inhibitors of cell g r o w t h or function. T h e E F S o f the last culture (G), however, was replaced twice an d 0 times during the 14 a n d 21 day culture periods, respectively. T h e glucose c o n s u m p t i o n rate curves were n o t adversely effected by the decreased f r e q u e n c y of E F S changes. T h e a m o u n t of time r e q u i r e d for each E F S c h a n g e was 2 0 - 3 0 min. This i n c l u d e d a 10 m i n c e n t r i f u g a t i o n to recover a n d r e i n o c u l a t e T I L that h a d been flushed o u t of the E F S during the procedure. D e c r e a s i n g the n u m b e r of E F S changes in a culture, therefore, resulted in a r e d u c t i o n of the a m o u n t o f time e x p e n d e d in m a i n t a i n i n g that culture. T I L that r e m a i n e d within the h o l l o w fiber bior e a c t o r cartridge (patient G) after a harvest, regrew to clinically useful n u m b e r s w h e n the cartridge p er f u s i o n was c o n t i n u e d (Fig. 2). T h e
Phenotypic profile CD3 (Leu-4) CD4 (Leu-3) CD8 (Leu-2) CDllb (Leu-15) CD14 (Leu-M3) CD16 (Leu-ll) CD20 (Leu-16) CD56 (Leu-19) CD57 (Leu-7) HLA-DR Tac
K-HF
K-bag
G-HF
G-bag
94.1 1.1 95.9 0.3 -0.4 0.6 3.7 8.3 4.2 80.7 2.4
94.7 2.0 90.4 9.5 0.3 0.8 4.0 20.4 11.1 85.6 4.7
85.3 13.1 73.8 3.7 0.6 2.2 0.1 6.5 2.2 89.3 27.3
91.4 14.3 83.1 0.4 1.1 1.4 0.8 7.1 4.6 91.1 30.4
rates of increase in glucose c o n s u m p t i o n during the first a n d second g r o w t h p er i o d s were indistinguishable. E x t r a p o l a t i o n of the second h o l l o w fiber g r o w t h cu r v e b a c k to the date of the first harvest i n d i c a t e d that a p p r o x i m a t e l y 1 × 10 7 T I L h a d r e m a i n e d w i t h i n the h o l l o w fiber cartridge to
TABLE IIl CYTOLYTIC ACTIVITY (% LYSIS) Effector
Target cells
cell
K562
Daudi
Tumor Autologous
*
20:1
5:1
LAK M-TIL (HF) K-TIL (bag) K-TIL (HF)
44.4 6.6 58.7 21.6
12.9 5.3 33.0 11.2
6.9 0.8 12.5 7.4
28.4 2.5 19.1 4.8
9.1 1.1 11.5 5.0
5.7 4.1 3.7 4.5
LAK H-TIL (bag) H-TIL (HF) K-TIL (HF)
65.2 7.4 2.3 4.2
23.0 5.8 0.9 1.3
7.3 0.8 0.7 0.8
43.6 2.6 - 0.8 1.8
14.1 5.6 2.8 - 2.1
8.4 6.6 2.8 - 3.1
17.2 50.2 49.9
8.9 57.1 42.9 ND
4.3 33.7 39.6
4.6 6.6 0.8 3.6
1.1 0.1 5.0 4.9
3.6 - 7.3 - 0.6 0.1
LAK W-TIL (bag) W-TIL (HF)
81.0 30.5 22.5
44.0 24.5 11.2
18.0 22.7 12.5
38.8 64.0 55.1
18.6 53.2 45.0
6.0 29.2 19.8
5.1 30.6 19.4
0.9 14.9 11.9
2.0 3.2 4.4
9.1 4.8 2.0
6.6 6.8 4.1
0.9 4.5 5.8
LAK W-TIL (bag) W-TIL (HF)
86.8 15.0 8.6
75.6 6.9 2.5
27.4 0.9 0.3
69.8 72.2 55.9
45.8 53.1 41.3
17.2 21.7 17.0
34.3 15.4 11.5
13.6 7.9 0
4.1 2.8 2.0
26.0 5.5 3.9
11.7 12.7 5.4
4.5 2.5 5.0
LAK G-TIL (bag) G-TIL (HF)
74.3 13.5 11.1
56.1 8.5 10.1
49.9 5.2 6.1
59.7 21.3 3.9
45.9 4.9 9.6
40.0 2.6 7.6
69.9 51.1 50.4
41.3 32.0 40.4
18.1 22.3 38.2
37.1 3.9 3.4
43.1 2.9 3.2
17.7 2.5 3.6
a ND, not done. * Effector: target cell ratios
80:1
20:1
5:1
80:1
20:1
Allogeneic
80:1
5:1
80:1
ND ND ND ND
20:1
5:1
ND ND ND ND
36 re-initiate growth of the second TIL batch. This is based on the assumption that the TIL doubling time in the first and second period of growth were identical. Cells removed simultaneously from bag and hollow fiber cultures yielded similar surface antigen profiles (Table II). The slight variations between them were of the same magnitude as those seen between bags that had originated from a single bag culture (data not shown). The cytolytic capacities of three dilutions of TIL taken from simultaneous aliquots of bag and hollow fiber cultures are presented in Table III. TIL caused 51chromium to be released from the target cells in quantities that were similar whether they were cultured in bags or hollow fibers. The lytic capacity of TIL was most often greater in the autologous tumor system than in the allogeneic tumor system and the pattern of lysis of K562 and Daudi target cells showed no difference between bag-cultured and hollow fiber-cultured TIL.
Discussion
Tumor infiltrating lymphocytes hold promise as a means for the treatment of certain types of cancer (Rosenberg et al., 1988). The investigation and implementation of this form of adoptive immunotherapy, however, has been hindered by the cumbersome and expensive methods currently used to grow the large number of TIL required for each patient. The studies presented herein, demonstrate that large numbers of TIL can be generated by culture in hollow fiber bioreactors. The rates of growth of these cells in the gas permeable bags currently used and in the proposed hollow fiber device are indistinguishable. Likewise, the phenotypic and cytotoxic characteristics of TIL obtained from either culture method are not significantly different. Total amounts of medium required appear to vary between patients as a function of the metabolic activity of their TIL within the hollow fiber bioreactor. Medium thus expended can be minimized by monitoring the glucose concentration of the perfusate and replacing the reservoir bottle when their levels drop to 1-1.5 g/1. While cells
grown within bags routinely required 1 liter of medium for 1-2 x 109 T•L, the hollow fiber cultures required 1 liter of medium to yield 1.5-4.2 × 109 TIL. Although none of the TIL described in this paper were used for patient therapy, the numbers of cells obtained from a single hollow fiber bioreactor approximated clinically relevant amounts. More than one hollow fiber cartridge would be needed to provide 2 x 1011 TIL, but this would be accompanied by approximately an 80% reduction in the amounts of both technician time expended and incubator space required in comparison to bag cultures. Furthermore, multiple doses of TIL could be harvested from a single hollow fiber cartridge and thus be made available for periodic use, an approach to therapy that may prove to be advantageous. Absence of a requirement to periodically replace the extra-fiber space medium indicates that TIL do not produce growth inhibitors having molecular weights greater than 3000-4000. This is in contradistinction to the hybridomas which synthesize TGF-fll that causes an auto-inhibition of growth and antibody secretion (Kidwell, 1989). The first studies of mammalian cells grown within a hollow fiber cartridge demonstrated both that solid tissue densities could be reached in vitro (Knazek et al., 1972) and that such cultures secreted 3-13-fold more biologically active product per cell than did the same cells grown in monolayer culture (Knazek, 1974; Knazek et al., 1974). Such increased efficiency may be a result of improved nutrient support or metabolite removal, or be secondary to the marked intensity of cell-cell interaction and concomitant retention of poorly diffusible autocrine factors that stimulate a more in vivo-like behavior of the cells. Addition of various cytokines or irradiated autologous tumor cells (Kawakami et al., 1988; Stotter et al., 1989) into the extra-fiber space of the hollow fiber bioreactors may further increase specific cytotoxicity of TIL; such cells would have the dual advantage of access to a large nutrient pool and concurrent exposure to high concentrations of these poorly diffusible stimuli. It is apparent, that the method of cell culture on hollow fibers may be applicable to the expansion of TIL in vitro and thus ease the burdens of
37 such production within both clinical and research laboratories.
References Kawakami, Y., Rosenberg, S.A. and Lotze, M.T. (1988) Interleukin-4 promotes the growth of tumor-infiltrating lymphocytes cytotoxic for human autologous melanoma. J. Exp. Med. 168, 2183. Kidwell, W.R. (1989) Filtering out inhibition. Biotechnology 7, 462. Knazek, R.A. (1974) Solid tissue masses formed in vitro form cells cultured on artificial capillaries. Fed. Proc. 33, 1978. Knazek, R.A., Gullino, P.M., Kohler, P.O. and Dedrick, R.L. (1972) Cell culture on artificial capillaries: an approach to tissue growth in vitro. Science 178, 65. Knazek, R.A., Kohler, P.O. and Gullino, P.M. (1974) Hormone production by cells grown in vitro on artificial capillaries. Exp. Cell Res. 84, 251. Muul, L.M., Director, E.P., Hyatt, C.L. and Rosenberg, S.A. (1986) Large scale production of human lymphokineactivated killer cells for use in adoptive immunotherapy. J. Immunol. Methods 88, 265. Rosenberg, S.A. (1986) Adoptive irnmunotherapy of cancer using lymphokine activated killer cells and recombinant interleukin-2. In: V.T. DeVita, S. Hellman and S.A. Rosenberg (Eds.), Important Advances in Oncology. JBL Lippincott, Philadelphia, PA, pp. 55-91.
Rosenberg, S.A. (1988) The development of new immunotherapies for the treatment of cancer using interleukin-2. Ann. Surg. 208, 121. Rosenberg, S.A., Spiess, P. and Lafreniere, R. (1986) A new approach to the adoptive immunotherapy of cancer with tumor-infiltrating lymphocytes. Science 223, 1318. Rosenberg, S.A., Lotze, M.T., Muul, L.M., Chang, A.E., et al. (1987) A progress report on the treatment of 157 patients with advanced cancer using lymphokine-activated killer cells and interleukin-2 or high-dose interleukin-2 alone. New Engl. J. Med. 316, 889. Rosenberg, S.A., Packard, B.S., Aebersold, P.M., Solomon, D., et al. (1988) Use of tumor-infiltrating lymphocytes and interleukin-2 in the immunotherapy of patients with metastatic melanoma. New Engl. J. Med. 319, 1676. Stotter, H., Wiebke, E.A., Tomita, S., Belldegrun, A., et al. (1989) Cytokines alster target cell susceptibility to lysis. II. Evaluation of tumor infiltrating lymphocytes. J. Immunol. 142, 1767. Topalian, S.L., Muul, L.M., Solomon, D. and Rosenberg, S.A. (1987) Expansion of human tumor infiltrating lymphocytes for use in immunology trials. J. Immunol. Methods 102, 127. Topalian, S.L., Solomon, D., Avis, F.P., Chang, A.E., et al. (1988) Immunotherapy of patients with advanced cancer using tumor-infiltrating lymphocytes and recombinant interleukin-2: a pilot study. J. Clin. Oncol. 6, 839.
