Baker Company, Sanford Maine). The MDA-231 cell line was cultured in Roswell Park Memorial Institute. (RPMI) 1640 media (Invitrogen, Carlsbad, Cal) con-.
Springer 2006
Breast Cancer Research and Treatment (2006) 96: 217–225 DOI 10.1007/s10549-005-9004-z
Characterization of an in vitro cell culture bioreactor system to evaluate anti-neoplastic drug regimens Mark N. Kirstein1, Richard C. Brundage2, William F. Elmquist3, Rory P. Remmel4, Paul H. Marker5, Dan E. Guire2, and Douglas Yee6 1
Department of Experimental and Clinical Pharmacology, College of Pharmacy and Comprehensive Cancer Center; Department of Experimental and Clinical Pharmacology, College of Pharmacy; 3Department of Pharmaceutics, College of Pharmacy; 4Department of Medicinal Chemistry, College of Pharmacy; 5Department of Medicine and Stem Cell Institute; 6Department of Medicine, Department of Medicine and Cancer Center, University of Minnesota, 308 Harvard St SE, 55455, Minneapolis, MN, USA 2
Key words: bioreactor, Gemcitabine, in vitro, MDA-231 cells, pharmacokinetics, S-phase
Summary A dynamic 3-dimensional tissue culture system has been developed that will allow for control of gemcitabine exposure to mimic concentration-time profiles measured from biologic samples. Gemcitabine was infused into a central reservoir. Media is mixed and delivered through hollow fiber capillaries, where it diffuses into the extracapillary space containing anchorage-dependent MDA-231 cells. To test for control of gemcitabine concentration-time profiles, drug was first infused through bioreactors without cells, and gemcitabine concentrations were measured with HPLC. Concentrations could be controlled to simulate 30-min and 2.5 h infusions, and were similar in both the lumen and extracapillary space. MDA-231 cells were then seeded into control (n = 4) and gemcitabine treatment (n = 4) groups, and maintained in culture for 2 weeks. Gemcitabine (5.3 mg) was infused over 30 min to the treatment group, and blank media to the control group. Accuracy of measured gemcitabine maximum concentration (Cmax) was 83.4%, and area under the curve (AUC), 106.2%, relative to pre-experimental theoretical values. With cells present, gemcitabine AUC in the extracapillary space was 32% of the value in the lumen. For the control group, 21.2 million cells (94.3% viable) were recovered, and for the gemcitabine-treated group, 16.8 million cells (87.1 % viable). Flow cytometry showed that 13.3 % of cells in the control group were in S-phase and 34.3 % in the gemcitabine-treated group were in S-phase (p = 0.003). In conclusion, gemcitabine concentration-time profiles could be accurately controlled through dosage, infusion rate, and pump flow rate, and cells could be recovered afterward to evaluate drug treatment.
Introduction Chemotherapy dosing and schedule are often determined experimentally with tissue culture systems, animal models and in Phase I clinical trials. Therapy can be ineffective for a variety of reasons including differences in tumor sensitivity between patients, use of sub-optimal dosage regimens, incomplete understanding of optimal timing and scheduling of anticancer agent delivery, development of tumor resistance, and pharmacokinetic exposure of the drug to the tumor. The relation between exposure and antitumor activity may be dependent upon maximum drug concentrations (Cmax), area under the concentration-time curve (AUC), or steady state concentrations (Css) over a period of time. Pharmacokinetic-directed clinical trials that evaluated fractionated dosing and short versus prolonged infusion rates for
similar drug dosages, include trials of topotecan, paclitaxel, 5-fluorouracil, vincristine, etoposide, and gemcitabine [1–7]. Investigators often use static tissue cultures to evaluate minimum inhibitory concentrations, exposure-time relationships, and combination treatments (isobologram analysis) [8–10]. These model systems provide an important and facile methodology to estimate drug dosages that can be further tested in animal and clinical studies. These studies are limited, however, by their ability to simulate only constant concentrations of drug over time. Animal models (i.e., human xenograft tumors) are utilized to assess drug regimens, since tumors in mice mimic the microenvironment of human tumors [8]. Animal models allow for the pharmacokinetic assessment in a complex in vivo environment that includes the
218
MN Kirstein et al.
ability for metabolic bioactivation or detoxification that can affect antitumor activity. The in vivo hollow fiber assay is also used as a routine screening method for anticancer activity, since greater than 50 cell lines can be inserted into small fibers that are subsequently implanted into mice to allow for testing in a cost effective manner [11]. Studies that assess drug pharmacokinetics in small animal models are useful for determining minimum effective exposures required for antitumor activity, drug disposition in tissue, and drug–drug interactions [12]. Unfortunately, mouse models are expensive and labor intensive, and it is difficult to control drug pharmacokinetics, or obtain multiple sequential blood samples in individual animals. The ability to control such variables as maximum concentrations, steady state concentrations, and overall exposure could enhance the understanding of important relationships between these pharmacokinetic variables and antitumor activity. In the case of agents that are activated at the tumor site (e.g., 5¢-FU, gemcitabine), the ability to control drug concentrations through dosage adjustment, infusion rate, and clearance could optimize exposure, reduce toxicity, and enhance antitumor effect. Since tumor cells are difficult to harvest from animal models, other methods that allow for determination of these relationships are needed. A system that would permit control of drug disposition could be useful for optimizing drug delivery, particularly to the tumor site. A 3-dimensional system of cells grown (Figure 1) in a bioreactor containing hollow fibers and a flow path driven by a pump has several advantages [13–15]. Drug concentrations and thus
exposures can be controlled to simulate human pharmacokinetic drug profiles. Another advantage for this in vitro hollow fiber system is that when cells are grown on the outside of hollow fibers, the cells are accessible, and can be removed to determine cytotoxicity. This system allows one to evaluate cytotoxicity and when relevant, active intracellular metabolite exposure can be determined under dynamic flow conditions. Establishment of this system could then provide an important and novel methodology to model and optimize new drug combinations. Since we know of no other studies that utilize this system to evaluate antineoplastic agents, we developed an in vitro hollow fiber model system to evaluate gemcitabine treatment. Gemcitabine, a pyrimidine nucleoside, undergoes metabolism by plasma and liver cytidine deaminase to form 2¢, 2¢-difluorodeoxyuridine (dFdU), a compound with little antitumor activity. Overall, approximately 77% of administered gemcitabine is excreted either unchanged, or as the dFdU metabolite into the urine within 24 h. Gemcitabine also undergoes intracellular phosphorylation by deoxycytidine kinase at the tumor site to form difluoro-dCMP, and is phosphorylated further by other intracellular kinases to form difluorodCDP and difluoro-dCTP [16–18]. The diphosphate metabolite (dFdCDP) inhibits ribonucleotide reductase, an enzyme that catalyzes formation of deoxynucleosides required for DNA synthesis [19]. The triphosphate (dFdCTP) is incorporated into DNA, resulting in chain termination [20,21]. It is theorized that continuous infusions are superior to shorter infusions due to greater active metabolite accumulation. Plasma gemcitabine clearance varies 4–30 fold between patients receiving the same dose, and dFdU production also varies 2–11-fold between patients [22,23]. It is unclear how this variability affects intracellular metabolite production and antitumor activity. We used published pharmacokinetic studies [22–26] to guide our initial development of this in vitro system.
Materials and methods Cell culture
Figure 1. Cell culture perfusion system: Schematic of the bioreactor apparatus. Cells are seeded into the extracapillary space (ECS), and maintained as a sterile culture. Gemcitabine is delivered as continuous or short infusion by syringe pump to the central reservoir that also receives drug-free media from the diluent reservoir. The central reservoir serves as a mixing compartment, and drug is delivered by an adjustable flow-rate pump through the lumen, where it diffuses into the ECS. Media is subsequently delivered to the elimination reservoir.
All cell culture procedures were conducted in a sterile class II biological safety cabinet (Sterilgard III Advance, Baker Company, Sanford Maine). The MDA-231 cell line was cultured in Roswell Park Memorial Institute (RPMI) 1640 media (Invitrogen, Carlsbad, Cal) containing 5% fetal calf serum (Biosource, Rockville, MD), 2 mM glutamine, 50 units/ml penicillin, and 50 lg/ml streptomycin (Invitrogen, Carlsbad, Cal). Cells were grown in 75-cm2 flasks (Corning, NY) in a humidified incubator at 37 C with 5% CO2 (Forma Scientific, Marietta, OH). Cells are stained with trypan blue, and counted with a hemacytometer. This human breast tumor cell line was chosen on the basis of sensitivity toward gemcitabine in preliminary studies.
In vitro bioreactor system and gemcitabine Hollow Fiber preparation All hollow fiber procedures were developed according to manufacturerÕs specifications. Polysulfone Plus cartridges (FiberCell Systems, Frederick, MD) with a Mr (at 50%) 0.1 lm cutoff and 70 cm2 surface area were used. The bioreactor culturing system consists of hollow fiber capillaries made from semi-permeable polysulfone within a clear plastic chamber. The tumor cells are grown on the outside of these fibers (i.e., extracapillary space; ECS). Gas permeable tubing connects this chamber to a source of growth media. The semipermeable fibers permit constant perfusion of medium nutrients and oxygen for the cells and dialysis of metabolic waste products away from the cells. Tumor cells were cultured as described above, a confluent flask of cells attained and seeded as 5 106 cells into the sterile-enclosed extracapillary space surrounding the hollow fiber capillaries. The cartridge is mounted onto a duet pump that pumps sterile media from a reservoir bottle through the gas permeable tubing where it is oxygenated and saturated with CO2. Media is then perfused through the lumen of the capillaries, and returns to the reservoir bottle. The cells are continuously perfused with media as a flow rate of approximately 20 ml/min in a humidified 5% CO2 incubator kept at 37 C. Reservoir volume was 110 ml during the first week, and increased to 150 ml during the second week of incubation.
219
given as 10 mg/m2/min over 2.5 h to reach a Css of 10– 20 lM were simulated [26]. Flow rates for the pumps, drug concentrations, and the volume of media were calculated before each experiment according to the following equations: V ¼ Q=Kd
ð1Þ
0
V¼
K0 1 eKt 0 Kd Kd Cmax ðCmin eKt Þ
Css ¼ R0 =Q
ð2Þ
ð3Þ
whereV is volume in the central reservoir, Q is the pump flow rate, Kd is the elimination rate constant for gemcitabine, K0 is the rate of gemcitabine infusion, Cmax is the maximum desired gemcitabine concentration, Cmin is the minimum desired concentration (fixed to 0), tÕ is drug infusion time, Css is steady state concentration. The flow rates for the pumps were verified by measurement of the media volume, recovered at the end of the infusion. At the end of the experiment, cartridges were returned to the humidified incubator at 37 C with 5% CO2, placed on the duet pump, and incubated until cell harvest.
Gemcitabine infusions
Pharmacokinetic sampling and analysis
On the day of gemcitabine infusion (i.e., day 14 of continuous culture), the bioreactor cartridges are removed from the duet pump, and placed in a class II biological safety cabinet. The experimental system is depicted in Figure 1. Control bioreactor cartridges are treated in a similar manner with drug-free media. Drug delivery is precisely controlled with a programmable syringe pump (Cole Parmer, Vernon Hills, IL). Drug is infused into the central reservoir, which also receives media through Masterflex silicone tubing (L/S 14; Cole Parmer) from the diluent reservoir. The media is mixed in the central reservoir, and is kept at a constant volume (80 ml). Media then travels the length of the fibers, delivering the drug where it diffuses to the ECS. Waste is collected in the elimination reservoir. Flow rates from the diluent reservoir and central reservoir are controlled with a Masterflex L/S digital drive peristaltic pump (Cole Parmer). Gemcitabine was infused to simulate 30min or continuous infusions in the hollow fiber system. Gemcitabine exposure was controlled through regulation of pharmacokinetic parameters (Cmax, clearance, volume). Published pharmacokinetic data for patients receiving gemcitabine as 800–1000 mg/m2 over 30 min were used to design the studies. From these studies, the Cmax ranged from 12 to 272 lM, AUC from 14 to 95 lM h, and most of the parent molecule was nonmeasurable 1 to 2 h after the end of infusion [22–25]. For development of continuous infusions, gemcitabine
Pharmacokinetic studies for gemcitabine were conducted during and after the end of the gemcitabine infusion. Media samples were collected from both the extracapillary space (port 3), and the inner lumen (port 4). For the 0.5 h infusion, samples were collected before and at 0.08, 0.25, and 0.47 h during the infusion, and at 0.08, 0.17, 0.33, 0.5, 0.83, 1, 1.25, and 1.5 h after the end of infusion. For the 2.5 h infusion, samples were collected before and at 0.08, 0.25, 0.5, 1, 2, 2.4, and 2.5 h during the infusion, and 0.08, 0.17, 0.33, 0.5, 0.83, 1, and 1.5 h after the end of infusion. At each time point, 0.25 ml of media was collected. These samples were then processed as described previously (Kirstein, et al. manuscript submitted). Briefly, 2¢-deoxycytidine was added as internal standard to make 100 lM final concentration. Samples were then acidified with 20 ll of 70% perchloric acid, and 25 ll of the resultant supernatant was injected onto an isocratic high-performance liquid chromatographic instrument with uv detection (267 nm). This method was determined to be accurate and precise (total assay %CV 3.7) at gemcitabine concentrations ranging from 2 to 200 lM. Structural pharmacokinetic model A two-compartment model was fit to the gemcitabine concentrations in the lumen and extracapillary space using maximum-likelihood estimation as implemented
220
MN Kirstein et al.
in ADAPT II [27]. Model parameters that were estimated included the elimination rate constant (Ke), volume of the central compartment (Vc), the intercompartmental rate constants (Kcp and Kpc). The volume of the E.C.S. was fixed to 2.5 ml as per manufacturer specification. Standard equations were used to calculate clearance and steady state concentrations from parameter estimates [28]. The model parameters for each experiment were used to simulate the concentration-time profile for both the lumen and extracapillary space, from which the area under the media concentration-time curve from time 0 to infinity (AUC0 fi ¥) was calculated by the log-linear trapezoidal method. Cell recovery and analyses Two days after gemcitabine treatment, MDA-231 cells were trypsinized and collected for analysis. Total cell recovery was determined with a hemacytometer as described previously. The cells were also prepared for flow cytometry [29]. Briefly, 3 105 cells were fixed in 1:1 (v:v) 400 ll 1 PBS:ethanol, and stored overnight at )20 C. Flow cytometry The MDA-231 cells were resuspended in 1 ml of 3.8 mM sodium citrate containing 50 lg/ml propidium iodide and 125 lg/ml RNAse A. Cells were analyzed on a Becton Dickinson FACSCalibur (Becton Dickinson, San Jose, CA) flow cytometer gated on forward light scatter pulse height, and side scatter pulse height for analysis of cell cycle fractions, and ungated mode for detection of cells with subG1 DNA content. Resulting histograms were evaluated with Flow Jo Watson Pragmatic v. 6 software (Tree Star, Stanford, CA). Comparison between control and gemcitabine-treated groups was evaluated with the paired students t-test. Lactate assay To measure cell lactate production, media samples were collected from the reservoir bottle daily and assayed for lactate concentration [30]. To account for changes in lactate concentrations due to media changes, samples were collected before and immediately afterwards. A cocktail, prepared at the time of assay as follows: 10 mg Nicotinamide adenine dinucleotide (Sigma, St. Louis, MO), 2 ml 0.6 mol/l glycine buffer, pH 9.2 (Sigma), 4 ml water, and 0.1 ml of 1,000 U/ml lactate dehydrogenase (Sigma). Media samples (167 ll) were subsequently added to 833 ll of the cocktail, followed by incubation at 37 C for 20 min. The spectrophotometric absorbance at 340 nm was read within 10 min for each sample. Lactate concentrations were determined with a lactate standard curve that is linear through a concentration range from 0.025 to 0.2 mg/ml (R2 > 99%). Briefly, 40 mg/dl lactate standard solution (Trinity Biotech, Wicklow, Ireland) was diluted serially, and then mixed
with cocktail to generate four calibrator solutions. A separate quality control lactate solution was prepared to assess for preparation accuracy (Range 90–110%). To determine lactate concentrations, sample absorption for the unknown samples were plotted. Lactate production rate was calculated according to the following equation:
Production Rate ¼
AF AI Elapsed Time
Where AF is the final lactate amount (mg), AI is the initial lactate amount measured on the previous day, and the elapsed time is the time span between the initial and final concentrations are measured, usually at 24 h intervals.
Results Gemcitabine exposure without cells Before seeding the MDA-231 cells on the outside of the fibers (i.e., the E.C.S.), we determined gemcitabineÕs ability to pass through the lumen and to penetrate the semipermeable fibers into the ECS of a ‘‘blank’’ cartridge. We infused 2.4 mg gemcitabine over 30 min in water solution, the flow rate was set to 5 ml/min, and we sampled at various time points from sampling ports 3 (E.C.S.) and 4 (lumen). The AUCs in the lumen and E.C.S. were 21.4 and 21 lM h, respectively. Thus, gemcitabine readily penetrated the polysulfone fibers into the E.C.S., and non-specific binding of gemcitabine to the tubing or the cartridge was negligible. The media used for maintaining the MDA-231 cells was infused into a second cartridge, so that we could test the influence of the media on gemcitabineÕs passage into the extracapillary space of a ‘‘blank’’ cartridge. Furthermore, we began to test our measured gemcitabine concentrations, relative to the concentrations that were predicted from the pharmacokinetic equations described in the methods section. Shown in Figure 2 are representative concentration-time plots for gemcitabine after varying conditions of dosage, infusion rate, and flow rate. We infused 2.9 mg gemcitabine over 30 min, set the flow rate to 5 ml/min, and performed pharmacokinetic sampling. Based on our initial pharmacokinetic equations, our predicted Cmax and AUC were 58.3 lM and 32.2 lM h, respectively. Our measured values from the lumen were 48.8 lM (83.7% accuracy) and 29.4 lM h (accuracy 91.3%), respectively. The E.C.S. AUC was 30.3 lM h. We next infused 5.3 mg gemcitabine over 30 min, set the flow rate to 5 ml/min, and again sampled the media at various time points. Our measured values from the lumen for Cmax and AUC were 88 and 50.3 lM h, respectively (Figure 3). The E.C.S. AUC was 61.2 lM h. The AUC values were slightly higher in the E.C.S. than the lumen (112%). The Cmax concentrations in the lumen were 20% higher than in the E.C.S., and the time to maximum concentrations (tmax)
A
100
A
100
Gemcitabine Conc (µ M)
10
Gemcitabine Conc (µ M)
In vitro bioreactor system and gemcitabine
10
1 0 .0
0.5
1.0
1.5
2.0
1 0.0
2.5
0.5
Gemcitabine Conc (µ M)
10
0
1
2
3
4
10
10
0.5
1.0
1.5
2.0
2.5
1.0
1.5
2.0
2.5
Time (hr) Figure 3. Concentration-time plots for gemcitabine measured from the lumen (d) and E.C.S. (m). Plots depict gemcitabine infused as 5.3 mg over 0.5 h in bioreactors. Plot 3A is without cells and 3B is a bioreactor with MDA-231 cells maintained in culture for 2 weeks.
100
1 0.0
0.5
5
Time (hr)
Gemcitabine Conc (µ M)
100
1 0.0
1
C
B Gemcitabine Conc (µ M)
100
1.5
Time (hr)
Time (hr)
B
1.0
221
2.0
2.5
Time (hr) Figure 2. Concentration-time plots for gemcitabine measured from the lumen of the hollow fiber cartridge. Plot 2A depicts gemcitabine infused at either 2.9 mg (m) or 5.3 mg (d) over 0.5 h. Plot 2B depicts profiles for different infusion rates: 3.4 mg infused over 0.5 h (d) or 2.5 h (m). Plot 2C depicts different flow rates: 5.3 mg infused over 0.5 h and delivered at 5 ml/min (d) or 9 mL/min (m).
was 6 min earlier in the lumen. When the gemcitabine (5.3 mg) infusion was repeated over 0.5 h, but increased the flow rate to 9 mL/min, the calculated Cmax was 51 lM and the AUC was 26.9 lM h in the lumen, while the E.C.S. AUC was 30.7 lM h. To determine if the lower Cmax and later tmax values measured in the E.C.S. might be due to protein binding in the media, we assessed protein binding for gemcitabine in media containing 5% fetal calf serum. After incubating gemcitabine in media at 37 C for 15 min, quadruplicate samples were divided in half, and the first half was processed for total gemcitabine measurement, and the
other half were loaded onto micropartition concentrators (Amicon Inc., Beverly, MA), and centrifuged for 1 h at 2000 g. We observed 100% recovery of gemcitabine from the flow through, suggesting absence of protein binding in the media. We next tested our ability to control gemcitabine exposure after continuous infusion in a cartridge without cells. As described in the methods section, we targeted our steady state concentrations to range from 10 to 20 lM. We infused 2.7 mg over 2.5 h and our steady state concentration was 13 lM. We then infused 3.4 mg twice, and steady state concentrations reached 18.5 and 11 lM after 1 h of the infusion. MDA-231 cell culture To test the method for growing MDA-231 cells on the outsides of the fibers (i.e., E.C.S.), two bioreactors were seeded with cells. The total number of cells seeded per volume of media closely approximates the standard procedure for routine passaging of MDA-231 cells. A total of 5 106 cells were seeded into the E.C.S. and the total media volume was 110 ml. After 14 days of continuous culture on the fibers, 18 and 30 million cells were recovered from the two cartridges. To further characterize the MDA-231 cell growth, and to build a cohort from which to use as a control group (n = 4), 5 106 cells were seeded into the E.C.S.
222
MN Kirstein et al.
Table 1. Gemcitabine pharmacokinetic data for cartridges with MDA-231 cells Vc (ml)
Est. Cl
Measured Cl
Cmax Lumen
AUC Lumen
Cmax ECS
AUC ECS
(ml/min)
(ml/min)
(lM)
(lM h)
(lM)
(lM h)
54.6
17.2
17.3
Mean (n = 4)
87.0
4.81
4.72
88.4
D
15.2
0.79
0.07
13.6
CV%
17.5
16.4
1.50
15.4
Two weeks later, media without drug was infused to simulate drug treatment conditions. Two days after drug-free media was infused, 21.2 ± 7.4 million cells were recovered. As determined by trypan blue staining, 94.3% of the cells were viable, and the cells were analyzed by flow cytometry as described below. Gemcitabine exposure in the presence of cells A second cohort (n = 4) of cartridges were seeded with 5 106 cells. Two weeks later, 5.3 mg gemcitabine was infused over 30 min, and media samples were collected from the lumen and E.C.S. during and after the infusion. Pharmacokinetic results are shown in Table 1. Based upon the dose infused, predicted values for Cmax and AUC were 106 and 51.4 lM h. Accuracy of the calculated values versus predicted values for Cmax and AUC were 83.4% and 106.2%, respectively. In the presence of MDA-231 cells, the Cmax and AUC values for the E.C.S. were 19.4% and 31% of the values calculated for the lumen, indicating that the cells produced a diffusional barrier. The measured clearance is based upon the volume of media collected divided by the duration of the experiment, and was 94.4 % of the value that was programmed for the peristaltic pump (5 ml/ min), and 98.1 % of the clearance calculated from the pharmacokinetic equations.
8.38 15.3
7.02 40.9
6.51 37.5
(p = 0.45). The viability was also higher for the control group (94.3%), but was not significantly different (p = 0.2). Lactate measurement Since monitoring cell viability grown in a cartridge was not possible with a microscope, an assay to measure the lactate production rate was developed as a surrogate for viability during culture. Media samples were collected daily from 3 cartridges, and the lactate concentrations were determined. Shown in Figure 5 is a representative plot of lactate production rate versus time in culture. In general, it was observed that lactate production rates did not increase during the first week in culture, and most of the increases were observed during the second week of cell culture. It was not possible to collect daily cell counts in this system and thus, the daily rates of lactate production could not be normalized for cell number to
Flow cytometry and cell viability To determine viability and cell cycle analysis for cells grown on the fibers the MDA-231 cells collected from the cartridges for both the control and gemcitabine treated cohorts were analyzed by flow cytometry. Shown in Figure 4 are representative histograms for control and gemcitabine-treated cartridges, and data for both cohorts are summarized in Table 2. A significant increase in the fraction of cells during S-phase was observed after gemcitabine treatment (Figure 4b), compared with untreated cells (Figure 4a). Concomitantly, a significant decrease in the fraction of cells present in G1 and G2/M phases was observed for gemcitabine-treated versus untreated cells. The percentages of cells present in the subG1 fraction were similar between the 2 cohorts. As determined from trypan blue staining, 16.8 ± 4.5 106 cells were recovered from cartridges treated with gemcitabine (87.1% viable). This value was less than that for the control group (21.2 ± 7.4 106 cells), but was not significantly different
Figure 4. Representative flow histograms of MDA-231 cells recovered from separate bioreactors and treated with propidium iodide. Histogram 4A: MDA-231 cells treated with drug-free media as control and 4B: MDA-231 cells after 5.3 mg gemcitabine infused over 0.5 h.
In vitro bioreactor system and gemcitabine
223
Table 2. Cell cycle analysis for MDA-231 cells grown in the ECS with and without gemcitabine treatment
Controls (n = 4) Gemcitabine (n = 4)
G1(%)
S(%)
G2/M (%)
SubG1 (%)
Cell Recovery (million)
Viability (%)
Mean ± Std Dev Mean ± Std Dev
68.9 ± 1.83 42.4 ± 15.0
13.3 ± 0.78 34.3 ± 4.87
5.00 ± 0.76 2.90 ± 1.33
11.0 ± 3.12 15.9 ± 15.12
21.2 ± 7.35 16.8 ± 4.47
94.3 ± 1.28 87.1 ± 5.50
P
0.03
0.003
0.04
0.52
0.45
0.20
Lactate Prod. Rate (mg/day)
140 120 100 80 60 40 20 0 0
2
4
6
8
10
12
14
16
Time (days) Figure 5. Representative plot for lactate production rate versus time for MDA-231 cells maintained in bioreactor culture for 2 weeks.
determine whether the increased production rate over time was due to increased cell number or to cell lysis. A 10% decrease in the lactate production rate was observed for both unexposed and gemcitabine-treated cells on the day following treatment. However, recovery was observed on the day of cell harvest for the control cartridges, but not for the gemcitabine-treated bioreactor.
Discussion An in vitro hollow fiber system for controlling gemcitabine pharmacokinetic exposure in bioreactors containing the anchorage-dependent MDA-231 breast cancer cell line was characterized. Gemcitabine exposure was controlled through dosage adjustment, infusion rate adjustment, or flow rate of the media, and mimicked gemcitabine concentration-time profiles measured from biological samples (i.e., patients). Treated cells can be recovered to analyze pharmacodynamic endpoints, including total cell recovery, viability, cell cycle disruption, and potentially intracellular gemcitabine triphosphates. Since gemcitabine is metabolized within tumor cells to form active phosphorylated metabolites, this system could test the relationship between infusion rate and metabolite production. Tempero and colleagues [26] evaluated patients with pancreatic carcinoma, and they report a median survival for patients receiving prolonged gemcitabine infusion (i.e., 2.5 h infusion) of 8 months, whereas for the standard 0.5 h infusion, median survival was 5 months. The authors demonstrated a two-fold increase in intracellular gemcitabine triphosphate concentrations in the fixed dose rate arm, by using peripheral blood mononuclear cells as surro-
gate. However, metabolite accumulation is difficult to evaluate at the tumor site, and this may differ from mononuclear cells. Since gemcitabine pharmacokinetic variability can affect intracellular metabolite accumulation, we plan to use this system to control gemcitabine exposure more precisely, and to evaluate metabolite accumulation. Several other aspects make gemcitabine a suitable drug for testing in this in vitro system. In patients, the half life for the parent molecule is 10–15 min, meaning that gemcitabine concentrations fall below detectable limits in plasma 1–2 h after infusion. This allows for us to perform infusions much more quickly than would be possible for other agents with much longer half-lives. Gemcitabine has relatively low binding affinity for circulating plasma proteins [31], and protein binding was also negligible in tissue culture media. For highly protein-bound drugs (e.g. cisplatin, docetaxel), the system could be modified to exclude fetal calf serum during the infusion, and permit greater access of the drug to the extracapillary space (E.C.S.). This underscores the importance of measuring drug concentrations, since protein binding and other unexpected non-specific binding to the bioreactor may alter the drug disposition. Whenever cells were present, lower gemcitabine concentrations (in the E.C.S.) were observed, suggesting that the cells created a diffusional barrier for drug penetration. To assess gemcitabine penetration intracellularly, future studies will be directed towards measurement of intracellular gemcitabine as well as active phosphorylated metabolite production. The seeding density for the bioreactors, relative to media volume was similar to seeding densities used for passaging the cells in static culture. The doubling times for MDA-231 cells grown in static culture are approximately 24 h when seeded in this manner. The doubling time observed for the same cells grown in the bioreactors was approximately 5–7 days. We attempted to increase the seeding density (30 million); however, this did not improve the doubling time (data not shown). Even though doubling times were slower in this culture system, cell cycle effects after gemcitabine treatment were still observed. Furthermore, others have reported that median potential doubling times for human breast tumors range from 8 to 13 days [32], suggesting that growth rates that we observed may be more representative of human tumor growth. Flow cytometry results demonstrated that 11% of the cells recovered from the bioreactor (control) cartridges were classified as subG1, which is higher than values reported for MDA-231 cells
224
MN Kirstein et al.
grown in static culture (1.3%) [33], as well as our own unpublished values (2.2%). As a consequence, we began measurement of lactate production rate as an indicator of cell viability. Initial results suggest that lactate production increases during the second week, and suggests that cells treated with cell cycle-dependent cytotoxic chemotherapy may be more sensitive to drug exposure during the second week of continuous cell culture than during the first week. To evaluate effects after treatment, flow cytometry was completed 2 days after gemcitabine treatment or in unexposed controls. A significant decrease in G1 and G2/M phases and an increase in S-phase fraction was found after treatment with gemcitabine. This demonstrates our ability to successfully harvest cells from the bioreactors, and to evaluate the effects of drug treatment. Accumulation of cells in S-phase has been found for various cell lines treated with gemcitabine concentrations ranging from 1 nM to 100 lM [34–38]. This concentration range is similar to the gemcitabine concentrations that were measured in the bioreactor during the gemcitabine infusion. These results are also consistent with gemcitabineÕs mechanism of action where the activated molecules interfere with DNA synthesis. Others report an accumulation of cells in subG1 and apoptotic fractions [39–41], whereas neither subG1 fractions nor total cell recovery were different between untreated and treated cells in the hollow fiber system. It is possible that prolonging the interval between drug treatment and cell recovery beyond 2 days may result in a higher fraction of cells that have undergone apoptosis, especially since the cells grown in the bioreactor are observed to have a slower dividing time than cells grown under static conditions. It is also possible that the growth conditions on the fibers are not identical to monolayer growth, and this may affect the response of the cells to drug treatment. In summary, this study has characterized the use of the in vitro hollow fiber system to evaluate gemcitabine treatment of the breast cancer cell line, MDA-231. With published gemcitabine pharmacokinetic data as a target guide, gemcitabine exposure was controlled through dosage, infusion rate, and flow rate. Cells can be recovered from the system, and evaluated for the effects of gemcitabine treatment. Future directions for this system include determining the accumulation of active intracellular gemcitabine phosphorylated metabolites and markers of cell death following different gemcitabine treatment regimens. After work is completed with single agent treatment, work can be extended to evaluate the addition of adjunct agents to either short or prolonged infusion gemcitabine treatment. Acknowledgements This work was supported in part by the Cancer Center Translational Breast Cancer Award to M.N.K.
References 1. Tubergen DG, Stewart CF, Pratt CB, Zamboni WC, Winick N, Santana VM, Dryer ZA, Kurtzberg J, Bell B, Grier H, Vietti TJ: Phase I trial and pharmacokinetic (PK) and pharmacodynamics (PD) study of topotecan using a five-day course in children with refractory solid tumors: a pediatric oncology group study. J Pediatr Hematol Oncol 18: 352–361, 1996 2. Stewart CF, Iacono LC, Chintagumpala M, Kellie SJ, Ashley D, Zamboni WC, Kirstein MN, Fouladi M, Seele LG, Wallace D, Houghton PJ, Gajjar A: Results of a phase II upfront window of pharmacokinetically guided topotecan in high-risk medulloblastoma and supratentorial primitive neuroectodermal tumor. J Clin Oncol 22: 3357–3365, 2004 3. Markman M, Rose PG, Jones E, Horowitz IR, Kennedy A, Webster K, Belinson J, Fusco N, Fluellen L, Kulp B, Peterson G, McGuire WP: Ninety-six-hour infusional paclitaxel as salvage therapy of ovarian cancer patients previously failing treatment with 3-hour or 24-hour paclitaxel infusion regimens. J Clin Oncol 16: 1849–1851, 1998 4. Brenner B, Ilson DH, Minsky BD, Bains MS, Tong W, Gonen M, Kelsen DP: Phase I trial of combined-modality therapy for localized esophageal cancer: escalating doses of continuous-infusion paclitaxel with cisplatin and concurrent radiation therapy. J Clin Oncol 22: 45–52, 2004 5. de Gramont A, Bosset JF, Milan C, Rougier P, Bouche O, Etienne PL, Morvan F, Louvet C, Guillot T, Francois E, Bedenne L: Randomized trial comparing monthly low-dose leucovorin and fluorouracil bolus with bimonthly high-dose leucovorin and fluorouracil bolus plus continuous infusion for advanced colorectal cancer: a French intergroup study. J Clin Oncol 15: 808–815, 1997 6. Kellie SJ, Koopmans P, Earl J, Nath C, Roebuck D, Uges DR, De Graaf SS: Increasing the dosage of vincristine: a clinical and pharmacokinetic study of continuous-infusion vincristine in children with central nervous system tumors. Cancer 100: 2637–2643, 2004 7. Clark PI, Slevin ML, Joel SP, Osborne RJ, Talbot DI, Johnson PW, Reznek R, Masud T, Gregory W, Wrigley PF: A randomized trial of two etoposide schedules in small-cell lung cancer: the influence of pharmacokinetics on efficacy and toxicity. J Clin Oncol 12: 1427–1435, 1994 8. Suggitt M, Bibby MC: 50 years of preclinical anticancer drug screening: empirical to target-driven approaches. Clin Cancer Res 11: 971–981, 2005 9. Chou TC, Motzer RJ, Tong Y, Bosl GJ: Computerized quantitation of synergism and antagonism of taxol, topotecan, and cisplatin against human teratocarcinoma cell growth: a rational approach to clinical protocol design. J Natl Cancer Inst 86: 1517– 1524, 1994 10. Chou TC, Talalay P: Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul 22: 27–55, 1984 11. Suggitt M, Swaine DJ, Pettit GR, Bibby MC: Characterization of the hollow fiber assay for the determination of microtubule disruption in vivo. Clin Cancer Res 10: 6677–6685, 2004 12. Kirstein MN, Houghton PJ, Cheshire PJ, Richmond LB, Smith AK, Hanna SK, Stewart CF: Relation between 9-aminocamptothecin systemic exposure and tumor response in human solid tumor xenografts. Clin Cancer Res 7: 358–366, 2001 13. Stanness KA, Westrum LE, Fornaciari E, Mascagni P, Nelson JA, Stenglein SG, Myers T, Janigro D: Morphological and functional characterization of an in vitro blood–brain barrier model. Brain Res 771: 329–342, 1997 14. Drusano GL, Bilello PA, Symonds WT, Stein DS, McDowell J, Bye A, Bilello JA: Pharmacodynamics of abacavir in an in vitro hollow-fiber model system. Antimicrob Agents Chemother 46: 464–470, 2002 15. Bilello JA, Bauer G, Dudley MN, Cole GA, Drusano GL: Effect of 2¢,3¢-didehydro-3¢-deoxythymidine in an in vitro hollow-fiber pharmacodynamic model system correlates with results of dose-
In vitro bioreactor system and gemcitabine
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
ranging clinical studies. Antimicrob Agents Chemother 38: 1386– 1391, 1994 Krishnan P, Fu Q, Lam W, Liou JY, Dutschman G, Cheng YC: Phosphorylation of pyrimidine deoxynucleoside analog diphosphates: selective phosphorylation of L-nucleoside analog diphosphates by 3-phosphoglycerate kinase. J Biol Chem 277: 5453–5459, 2002 Bouffard DY, Laliberte J, Momparler RL: Kinetic studies on 2¢,2¢difluorodeoxycytidine (Gemcitabine) with purified human deoxycytidine kinase and cytidine deaminase. Biochem Pharmacol 45: 1857–1861, 1993 Liou JY, Dutschman GE, Lam W, Jiang Z, Cheng YC: Characterization of human UMP/CMP kinase and its phosphorylation of D- and L-form deoxycytidine analogue monophosphates. Cancer Res 62: 1624–1631, 2002 Heinemann V, Xu YZ, Chubb S, Sen A, Hertel LW, Grindey GB, Plunkett W: Inhibition of ribonucleotide reduction in CCRFCEM cells by 2¢,2¢-difluorodeoxycytidine. Mol Pharmacol 38: 567– 572, 1990 Ross Cuddy DD. DP: Molecular effects of 2¢,2¢-difluorodeoxycytidine (Gemcitabine) on DNA replication in intact HL-60 cells. Biochem Pharmacol 48: 1619–1630, 1994 Huang P, Chubb S, Hertel LW, Grindey GB, Plunkett W: Action of 2¢,2¢-difluorodeoxycytidine on DNA synthesis. Cancer Res 51: 6110–6117, 1991 Abbruzzese JL, Grunewald R, Weeks EA, Gravel D, Adams T, Nowak B, Mineishi S, Tarassoff P, Satterlee W, Raber MN, : A phase I clinical, plasma, and cellular pharmacology study of gemcitabine. J Clin Oncol 9: 491–498, 1991 Venook AP, Egorin MJ, Rosner GL, Hollis D, Mani S, Hawkins M, Byrd J, Hohl R, Budman D, Meropol NJ, Ratain MJ: Phase I and pharmacokinetic trial of gemcitabine in patients with hepatic or renal dysfunction: cancer and Leukemia Group B 9565. J Clin Oncol 18: 2780–2787, 2000 Dumez H, Louwerens M, Pawinsky A, Planting AS, de Jonge MJ, Van Oosterom AT, Highley M, Guetens G, Mantel M, de Boeck G, de Bruijn E, Verweij J: The impact of drug administration sequence and pharmacokinetic interaction in a phase I study of the combination of docetaxel and gemcitabine in patients with advanced solid tumors. Anticancer Drugs 13: 583–593, 2002 Kroep JR, Giaccone G, Voorn DA, Smit EF, Beijnen JH, Rosing H, van Moorsel CJ, van Groeningen CJ, Postmus PE, Pinedo HM, Peters GJ: Gemcitabine and paclitaxel: pharmacokinetic and pharmacodynamic interactions in patients with non-small-cell lung cancer. J Clin Oncol 17: 2190–2197, 1999 Tempero M, Plunkett W, Van Ruiz Haperen V, Hainsworth J, Hochster H, Lenzi R, Abbruzzese J: Randomized phase II comparison of dose-intense gemcitabine: thirty-minute infusion and fixed dose rate infusion in patients with pancreatic adenocarcinoma. J Clin Oncol 21: 3402–3408, 2003 DÕArgenio DZ, AADAPT II UserÕs Guide: Pharmacokinetic/ Pharmacodynamic Systems Analysis Software. Biomedical Simulations Resource Los Angeles, 1997 Gibaldi M, Koup JR: Pharmacokinetic concepts – drug binding, apparent volume of distribution and clearance. Eur J Clin Pharmacol 20: 299–305, 1981
225
29. Gooch JL, Van Den Berg CL, Yee D: Insulin-like growth factor (IGF)-I rescues breast cancer cells from chemotherapy-induced cell death–proliferative and anti-apoptotic effects. Breast Cancer Res Treat 56: 1–10, 1999 30. Marbach Weil EP. MH: Rapid enzymatic measurement of blood lactate and pyruvate. Use and significance of metaphosphoric acid as a common precipitant. Clin Chem 13: 314–325, 1967 31. Company EL, Package Insert 32. Rew DA, Wilson GD: Cell production rates in human tissues and tumours and their significance. Part II: clinical data. Eur J Surg Oncol 26: 405–417, 2000 33. Yamamoto D, Tanaka K, Nakai K, Baden T, Inoue K, Yamamoto C, Takemoto H, Kamato K, Hirata H, Morikawa S, Inubushi T, Hioki K: Synergistic effects induced by cycloprodigiosin hydrochloride and epirubicin on human breast cancer cells. Breast Cancer Res Treat 72: 1–10, 2002 34. Tolis C, Peters GJ, Ferreira CG, Pinedo HM, Giaccone G: Cell cycle disturbances and apoptosis induced by topotecan and gemcitabine on human lung cancer cell lines. Eur J Cancer 35: 796– 807, 1999 35. Cappella P, Tomasoni D, Faretta M, Lupi M, Montalenti F, Viale F, Banzato F, DÕIncalci M, Ubezio P: Cell cycle effects of gemcitabine. Int J Cancer 93: 401–408, 2001 36. Ng SSW, Tsao MS, Chow S, Hedley DW: Inhibition of phosphatidylinositide 3-kinase enhances gemcitabine-induced apoptosis in human pancreatic cancer cells. Cancer Res 60: 5451–5455, 2000 37. Pauwels B, Korst AE, Pattyn GG, Lambrechts HA, Van Bockstaele DR, Vermeulen K, Lenjou M, de Pooter CM, Vermorken JB, Lardon F: Cell cycle effect of gemcitabine and its role in the radiosensitizing mechanism in vitro. Int J Radiat Oncol Biol Phys 57: 1075–1083, 2003 38. Perabo FG, Lindner H, Schmidt D, Huebner D, Blatter J, Fimmers R, Muller SC, Albers P: Preclinical evaluation of gemcitabine/paclitaxel-interactions in human bladder cancer lines. Anticancer Res 23: 4805–4814, 2003 39. Ali S, El-Rayes BF, Aranha O, Sarkar FH, Philip PA: Sequence dependent potentiation of gemcitabine by flavopiridol in human breast cancer cells. Breast Cancer Res Treat 90: 25–31, 2005 40. Chandler NM, Canete JJ, Callery MP: Caspase-3 drives apoptosis in pancreatic cancer cells after treatment with gemcitabine. J Gastrointest Surg 8: 1072–1078, 2004 41. Serrano MJ, Sanchez-Rovira P, Algarra I, Jaen A, Lozano A, Gaforio JJ: Evaluation of a gemcitabine-doxorubicin-paclitaxel combination schedule through flow cytometry assessment of apoptosis extent induced in human breast cancer cell lines. Jpn J Cancer Res 93: 559–566, 2002 Address for offprints and correspondence: Mark N. Kirstein, Department of Experimental and Clinical Pharmacology, College of Pharmacy and Comprehensive Cancer Center, University of Minnesota, 308 Harvard St SE, Minneapolis, MN 55455, USA; Tel.: +1-612-6245689; Fax: +1-612-625-3927