The use of ATP bioluminescence assays in selecting a drug screen panel for chemosensitivity testing of uterine cancer cell lines

GYNECOLOGIC

ONCOLOGY

45, 185-191 (1992)

The Use of ATP Bioluminescence Assays in Selecting a Drug Screen Panel for Chemosensitivity Testing of Uterine Cancer Cell Lines HOA N. NGUYEN, M.D.,’ BERND-UWE SEVIN, M.D., PH.D., HERVY E. AVERETTE, M.D., JAMESPERRAS,PH.D., DANIEL DONATO, M.D., AND MANUEL PENALVER, M.D. Division

of GynecoIogic

Oncology,

Department of Obstetrics & Gynecology, University of Miami School of Medicine, P.O. Box 016940 (D-52), Miami, Florida 33101

Received July 26, 1991

The ATP bioluminescence assay has demonstrated a strong potential to become a clinical assay for chemosensitivity testing. Currently, chemotherapy of gynecologic cancers remains controversial and empirical. To optimize the patient’s chance of survival and to justify related toxicities, the chemoregimen should be individualized and basedon the patient’s chemosensitivity profiles. This study was performed to identify a pane1 of active drugs against uterine cancer cell lines for possible use in future chemosensitivity testing. We used the ATP chemosensitivity assays to screen 12 common cytotoxic agents against six uterine cancer cell lines. Drug concentrations required for a 50% surviving fraction were defined as ICSOs. When using an IC50 of 0.21 PPC (peak plasma concentration) as a cutoff value for sensitivity, the following 8 drugs were considered effective for uterine cancer cell lines: actinomycin D, Adriamycin, vinblastine, etoposide, S-fluorouracil, methotrexate, cytosine arabinoside, and mitomycin-C. Meanwhile, 4 drugs, cisplatin, 4OHCytoxan, bleomycin, and Alkeran with mean ICSOs of 2.1 + 0.7, 0.8 + 0.1, >S.O, and 0.75 + 0.36 PPC, respectively, were considered inactive or partially active with higher ICSOsthan peak plasma concentrations. In conclusion, the above pane1of promising drugs can be further tested in animal models or human cancer specimens for possible use in chemosensitivity testing of uterine cancer patients. D KSZ Academic

Press, Inc.

INTRODUCTION

Chemotherapy of gynecologic cancers remains a controversial and empirical treatment [1,2]. In current therapy, certain chemotherapeutic regimens are prescribed for certain types of cancer. The “established treatment” often does not take into consideration whether the pa’ American Cancer Society Clinical Oncology Fellow, recipient of American Cancer Society Institutional Grant for Young Investigators, and to whom all correspondence and reprint requests should be addressed.

tient’s cancer is sensitive to chemotherapy. Because most cytotoxic agents cause severe toxic reactions, it is important to make sure that the prescribed treatment offers the best chance of survival. Numerous scientific studies have demonstrated that different cancer patients have different responses to the same chemoregimen [3-121. One reason for the heterogeneous tumor response is the different chemosensitivity profile of the individual patient’s cancer. To better predict drug responses, many assays have been introduced [3]. However, only the human stem cell assay has been universally adopted for chemosensitivity testing for many years [4,.5]. Several retrospective studies have shown a good correlation between test prediction and clinical response [5,6]. In a study by Alberts et al., improved survival of recurrent ovarian cancer patients was reported when their treatments were based on the results of chemosensitivity testing [7]. However, colony-forming assays have several problems such as long turnover time, low applicability rate, etc., making them impractical for routine clinical use [8]. In addition, it has been argued that these assays measure chemosensitivity profiles of clonogenic stem cells and not those from the entire cell populations. The ATP chemosensitivity assay has several advantages. It is a highly quantitative assay which measures cellular response of the entire tumor population by cellular ATPs instead of counting colonies. In addition, the assay is reproducible and reliable and has a clinical applicability rate of more than 90%, making it well suited for clinical application [9-121. This feature is possible due to the extreme sensitivity of this assay, which can detect a difference of 50 cells [13]. In clinical medicine, bacterial culture and sensitivity were routinely performed to help select the right antibiotics for each type of infection. A panel of antibiotics known to be active against a particular infection is often

185 0090-82X492 $4.00 Copyright 0 1992 by Academic Press, Inc. All tights of reproduction in any form reserved.

186

NGUYEN ET AL.

used to test for drug sensitivity. Because it is impractical to test all available cytotoxic agents, it makes good sense to select a panel of drugs with known activity against uterine cancers for chemosensitivity testing. Embracing the same concept, this study was designed to evaluate a panel of 12 common cytotoxic agents and to determine the patterns of drug response in uterine cancer cell lines. This drug screen panel might then be further developed in animal models or human tumor systems prior to routine ATP chemosensitivity testing of uterine cancer patients. MATERIALS

AND METHODS

Cell culture. Six human uterine cancer cell lines were used: endometrial cell lines AE7 and ECCl were obtained from Dr. Satyaswaroop (Hershey, PA). They were cultured from primary and untreated well-differentiated adenocarcinoma of the endometrium. The other cell lines AN3, HEClA, HEClB, and SKUTlB were obtained from American Type Culture Collection. Both HEClA and HEClB were derived from untreated moderately differentiated adenocarcinoma of the endometrium at 122nd and 128th passages, respectively. Cell line AN3 was derived from a metastatic para-aortic lymph node of a poorly differentiated adenocarcinoma of the endometrium of a patient who was previously treated with hormones. SKUTlB was isolated from a patient with poorly differentiated leiomyosarcoma who was previously treated with irradiation. All cell lines were grown in Eagle’s modified essential medium. The media was prepared with 10% fetal bovine serum, 100 U/ml penicillin, 100 pg/ml streptomycin, and 2.5 pg/ml amphotericin B. Cells were incubated at 37°C with 5% carbon dioxide. Medium was replaced every 3 days and cells were subcultured weekly following detachment with 0.25% trypsin/0.02% EDTA. Drugs. The following 12 drugs have been tested using the reported peak plasma concentrations (PPC) as reference values [14]: actinomycin D (ACT D), 0.075 vg/ml; cytosine arabinoside (ARA C), 10 pg/ml; melphalan (ALK), 3.4 pg/ml; bleomycin (BLEO), 3 @g/ml; cisplatin (DDP), 2.5 pg/ml; Adriamycin (DXR), 0.5 pg/ml; 5fluorouracil (5FU), 50 pg/ml; methotrexate (METHO), 2.75 hg/ml; mitomycin-C (MIT0 C), 0.5 pg/ml, etoposide (VP16), 30 pg/ml; and vinblastine (VIN), 0.78 pg/ml. Because cyclophosphamide is not active without hepatic conversion, we used its active metabolite 4-hydroperoxy cyclophosphamide (40H-CYT), which was kindly provided by Dr. M. Colvin, Johns Hopkins Oncology Center. This active metabolite blood level is 20% that of cyclophosphamide, which has a PPC of 30 pg/ml [14,15]. Thus, a PPC of 6 pug/ml was chosen for 40HCYT. A TP chemosemitivity assays. Since the amount of cellular ATP corresponds to cell number and cell mass, the

ATP bioluminescence assayshave been used successfully to measure drug response. Briefly, a suspension of 20,000 cells/ml was used to plate a 24-well tissue culture flask in triplicates. Drug exposure was performed for 90 min on the next day. For each drug, besides the control wells, the following concentrations were used: 0.1, 0.2, 0.5, 1, 2, and 5 PPC. Dose-response curves were obtained on Day 7 by extracting ATP from the cells in situ with an equal volume of 2% trichloroacetic acid. ATP bioluminescence was determined as previously described [ll]. Criteria of response. Dose-response curves of the above six uterine cancer cell lines were determined for each of 12 drugs. For ease of description, a drug-induced reduction of control ATP level of at least 50% at 0.2 PPC was considered a sensitive response, while those with less than 50% reduction of cellular ATP were considered resistant. Data analysis. For purpose of comparison, ICSOswere measured in PPCs by the median effect plot of log(F,/SF) versus log C (SF, surviving fraction; F, , fraction affected = 1 - SF; C, concentration) [16]. Experiments were repeated from two to four times for each cell line with each drug. Assay coefficient of variation of triplicates ranged from 0.5 to 39% with a median of less than 10%. RESULTS

Figure 1 shows dose-response curves of the above cell lines against DDP, 40H-CYT, MIT0 C, and ALK. By using the above criteria, all cell lines were sensitive to mitomycin-C. Two cell lines, AE7 and ECClB, were sensitive to Alkeran but AN3, SKUTlB, HEClA, and HEClB were resistant to this drug. For 4OH-CYT, no cell line exhibited sensitivity to this drug at 0.2 PPC. However, all cell lines responded to this drug at 1.0 PPC. Similarly, when our strict criteria for drug sensitivity was used, all cell lines were considered resistant to cisplatin. In Fig. 2, no cell line had a sensitive response to bleomycin. Meanwhile, all cell lines were considered sensitive to the antimetabolites 5FU, METHO, and ARA C. The majority of cell kill occurred at 0.1 and 0.2-PPC doses. As the concentration increased to 5 PPC, the surviving fractions remained the same as those at 0.1 or 0.2 PPC. This resulted in plateau dose-response curves at high doses, indicating a lack of further drug response. Figure 3 shows the survival curves of vinca alkaloids and intercalating agents. All cell lines except HEClB were sensitive to VIN. Similar to the antimetabolites, dose-response curves of vinblastine did not exhibit further cytotoxicity as drug concentrations were raised from 0.2 to 5 PPC. All six cell lines were sensitive to VP16. By dose comparison, etoposide exposure led to more cell death than vinblastine in every cell line. Thus, etoposide appeared to be a more potent vinca alkaloid than vin-

187

USE OF ATP ASSAY TO SELECT A DRUG SCREEN PANEL CISPIATIN

fk 100 4 $

80

z g lb

60

0 5

40

6 g n

20

AN3 AE7 Kc HEGA Eta SKUT

OI 0.0

2.5

5.0 7.5 DOSE, ug/ml

10.0

d

12.5

s

ALKERAN

d 4

MlToMYCIN

6 E 0’

80 60

E

0

4

8 12 DOSE, uglml

12 1s DOSE, uglml

16

2’0

2h

3.0

2.0

2.5

c

80 60

0.0

0.5

1.0 1.5 DOSE, ug/ml

FIG. 1. Dose-response curves of AN3, AE7, ECCl, HECIA, HEClB, and SKUTlB cell lines when treated with cisplatin, Alkeran, cyclophosphamide, aid mitomycin-C.

40%

blastine. Adriamycin and actinomycin D were also very 0.7; and BLEO, >5. Eight drugs with mean ICSOs less effective drugs. Four out of six cell lines were sensitive than 0.21 PPC were considered potent drugs for the above to Adriamycin, while all six cell lines were sensitive to uterine cancer cell lines. These were VP& ACT D, ARA actinomycin D. By direct comparison of ATP surviving C, VIN, METHO, MIT0 C, 5FU, and DXR. The other fractions at each concentration, actinomycin D appeared four drugs, 40H-CYT, ALK, DDP, and BLEO, were more potent than Adriamycin. As drug concentrations of donsidered suboptimal drugs by the above cell lines. VP16, ACT D, and DXR increased, the surviving fractions continued to decrease proportionally. There were no “plateau effects” as seen with VIN, METHO, 5FU, DISCUSSION and ARA C. The above uterine cancer cell lines were derived from IC5Os were calculated from the above dose-response curves (Table 1). For ease of comparison, IC50 was mea- cancer patients who were not previously treated with any sured in peak plasma concentration. The actual ICSOcon- cytotoxic agents. These cell lines represent a spectrum of centrations could be obtained by multiplying IC50 values uterine cancers including the well-differentiated, modwith peak plasma concentrations as listed above. For erately differentiated, and poorly differentiated adenosome drugs, the dose-response curves did not include a carcinomas of the endometrium. Four cell lines were culdrop-off area within the range of concentrations used. tured from primary sites and one came from a metastatic This indicated drug resistance even at 5 PPC. The esti- lymph node. In addition, SKUTlB is a leiomyosarcoma. mated IC5Os would be subjected to errors and were thus Thus, this panel of representative uterine cancer cell lines listed as greater than 5 PPC. Mean IC5Os of 12 cytotoxic can serve as a model to evaluate new chemotherapeutic agents are listed in order of decreasing potency: VP16, agents for uterine cancers. 0.06 + 0.01; ACT D, 0.07 & 0.02; ARA C, 0.08 tIn general, tumor response to a chemotherapeutic agent 0.02; VIN, 0.09 z!z0.8; METHO, 0.09 ? 0.02; MIT0 is characterized by its heterogeneity [17,18]. An individual C, 0.09 f 0.02; 5FU, 0.13 -c 0.03; DXR, 0.21 r 0.10; patient’s tumor can be expected to mount different reALK, 0.75 -+ 0.36; BOH-CYT, 0.8 + 0.1; DDP, 2.1 -+ sponses to the same chemotherapeutic agents. Besides

188

NGUYEN ET AL. 5-FLUOROUFWCIL

P 100 4’ g E ;

80

CYTOSINE AFIABINOSIDE

p 2

100

<

80

t

c”

60

$

60

b

40

-

AN3 AE7

-RX *

HECA liza3 SKUT

s

k? 20 % 0 0

IdO DOSE,

p

2

2do

1;o

2;o

0

METHOTREXATE

100

10

uglml

E

20

30

DOSE,

uglml

40

50

BLEOMYCIN

100

u

t

d F

80

;

60

b

40 k E 20 w" P 0 0.00

0

2.75

5.50 DOSE,

a.25

11 .oo

13.75

I

I

I

I

I

I

0

3

6

9

12

15

uglml

FIG. 2. Dose-response curves of AN3, AE7, ECCl, HEClA, arabinoside, methotrexate, and bleomycin.

DOSE,

HEClB, and SKUTlB

external influences such as the immune system and nutritional status, a patient’s drug response depends heavily on whether the tumor is sensitive to cytotoxic treatment. Numerous studies over the years have demonstrated the need for chemosensitivity testing [3-121. The colonyforming assay has shown excellent negative predictive value. However, its positive predictive value was only about 50%. Moreover, in a prospective study by Von Hoff et al., the practical applicability rate was only 45% for human stem cell assay [5]. This low applicability rate was mostly due to poor tumor growth and the large number of cells required for colony-forming assay [6]. As a consequence, current chemotherapy remains empirical. Treatments were guided by established protocols [1,2]. This “blanket” approach unwillingly includes a large number of cancer patients whose tumors are not sensitive to the prescribed treatments. Except for a few exceptions, current chemotherapy has not improved the survival of recurrent or advanced gynecologic cancer patients over several decades [ 11.Thus, it seemsimportant to determine the individual patient’s chemosensitivity profile to help select the right cytotoxic agents. In a survey of literature in 1974, Donovan reported up to 16 different cytotoxic agents used to treat uterine cancers [19]. Since then, the number has grown as new cy-

uglml

cell lines when treated

with

.5-fluorouracil,

cytosine,

totoxic agents become available. Because of the large number of cytotoxic agents, it is necessary to select a panel of active drugs which are likely to be sensitive to uterine cancers to be used as a drug screen panel for chemosensitivity testing. The ATP chemosensitivity assays can help define a panel of active drugs and select effective agents for combination chemotherapy. To devise an effective combination regimen, each drug component should demonstrate activity against uterine cancers [20,21]. A regimen which employed an active drug and an inactive drug probably would not improve response rate. In a prospective and randomized study by the Gynecologic Oncology Group (GOG), a combination of Adriamycin (a sensitive drug with an IC50 of 0.21 2 0.10 PPC) and Cytoxan (an inactive drug with an IC50 of 0.80 + 0.10 PPC) did not perform any better than Adriamycin alone [22,23]. Since cisplatin is not an active drug for uterine cancers with an IC50 of 2.1 + 0.7 PPC, the result of this study predicts that the combination of cisplatin and Adriamycin might not prove better than Adriamycin alone. It will be interesting to wait for the results of GOG protocol 107 comparing the activity of Adriamycin alone and in combination with cisplatin in advanced and recurrent endometrial cancer. As drug concentrations were increased from 0.2 to 5

189

USE OF ATP ASSAY TO SELECT A DRUG SCREEN PANEL

E

100

u A 0 F

60

;

60

b z

40

yi

ETOKXIDE

ADWYCIN

100

g

60

2 ; b

60 40

5 ti

20

-

AN3 AE7

--BX -

HE’3 Hxa SKUT

20

i 0 0

60

30

DOSE,

2 u -J i?

90

120

150

0.0

60

f d

60

60 40

8

60

F ;

b 5

40

8

5

1.5

2.0

2.5

uglml

ACTINOMYCIN

100

P-

E

1.0 DOSE,

VINBLASTINE

100

0.5

uglml

D

20 0 C

I

0.76

1.56 DOSE,

2.34

3.12

3.90

0.000

0.075

uglml

0.150 DOSE,

FIG. 3. Dose-response curves of AN3, AE7, ECCl, HEClA, Adriamycin, and actinomycin D.

HEClB,

PPC, the cytotoxic response of METHO, 5FU, ARA C, VIN, and MIT0 C failed to exhibit further cytotoxicity. Fractional cellular ATP remained more or less constant, indicating that surviving cells were able to tolerate a further increase of drug concentration. These “plateauing effects” are interesting and may explain the wide range

0.225

0.300

0.375

uglml

and SKUTlB cell lines when treated with etoposide, vinblastine,

of reportedly effective doses for METHO, 5FU, VIN, and ARA C by Thigpen [l]. If these findings are confirmed in animal models or gynecologic tumor specimens, these drugs should be administered at the lowest effective doses and preferably in combinations with other active drugs such as DXR, ACT D, or VP16.

TABLE 1 Mean IC5Os of Six Uterine Cancer Cell Lines when Treated with 12 Common Cytotoxic Drugs

ACID ALK ARA C BLEO 40H-CYT DDP DXR 5FU METHO MIT0 C VIN VP16

AN3

AE7

ECCl

0.03 0.43 0.07 0.54 0.88 0.54 0.06 0.14 0.12 0.06 0.06 0.03

0.04 0.13 0.05 >5.0 0.24 1.58 0.12 0.09 0.08 0.07 0.03 0.07

0.07 0.19 0.05 B5.0 0.89 1.22 0.67 0.21

0.10 0.07 0.05 0.09

HEClA

HECIB

0.08

0.14

0.04

2.40 0.18 0.67 0.94 3.39 0.27 0.18 0.07 0.17 >5.0 0.10

0.25 0.02 5.61 0.89 0.84 0.02 0.04 0.03 0.02

1.11 0.10 >5.0 0.96 >5.0 0.14 0.09 0.12 0.14 0.09 0.02

SKUTlB

0.11 0.04

Note. ICSOswere measured in PPCs for ease of comparison. When surviving fractions at 5 PPC were still greater than 50% of control ATP, determination of these ICSOsbecame inaccurate. Thus, they were listed as >5.0 PPC for these drugs.

190

NGUYEN ET AL.

The above observation also supported the concept of tumor population heterogeneity [17,18]. As drug concentrations increased to 5 PPC, resistant tumor subpopulations survived and showed no further reduction of ATP surviving fractions. An alternative explanation for the plateauing effects was the development of a new drug resistance. Mechanisms of resistance could range from gene amplification, as has been shown for methotrexate, or increased drug efflux from mdr glycoprotein [24-261. However, this was an unlikely possibility because of the short-term drug exposure (90 min) and the fact that the above cell lines were not exposed to drugs previously. Among the above six cell lines, HEClB demonstrated a broad pattern of drug resistance to cisplatin, Cytoxan, vinblastine, bleomycin, and Alkeran. Thus, HEClB cell line might possesssome pleiotropic mechanisms of drug resistance and would be an ideal candidate for studying the problem of multidrug resistance [27]. This study demonstrated the heterogeneous response of each cell line to cytotoxic agents. These results reaffirmed the need to perform chemosensitivity testing on gynecologic cancer patients and individualize their chemotherapy accordingly. From the mean ICSOs, we identified eight active drugs which could be used as a possible drug screen panel for uterine cancers. These are ACT D, ARA C, VIN, 5FU, METHO, MIT0 C, DXR, and VP16. All of these drugs have demonstrated efficacy as single agents with mean IC5Os of less than 0.21 PPC. At the clinically recommended dose, peak plasma concentrations of the above drugs will be several fold higher than IC50 values, thus ensuring an optimal response. The other four drugs including DDP, 40H-CYT, BLEO, and ALK were not chosen because of inferior performance with high ICSOs: cisplatin, 2.1 ? 0.7; 40H-Cytoxan, 0.8 + 0.1; bleomycin, >5; and Alkeran, 0.75 + 0.36 PPC. Because of the high ICSOs required for 50% cell kill, the peak plasma concentrations at the recommended dose of these drugs would be below the IC50 levels. This meant some tumor subpopulations would respond and the rest would not. This finding was somewhat surprising because cisplatin, 40H-Cytoxan, and Alkeran were among the most commonly used drugs in uterine cancers. In our previous publications, cisplatin and 40H-Cytoxan did not perform as well as other drugs [11,12]. One possible explanation was that the experimental conditions in this study did not mimic in vivo situations, which led to poor predictability of certain drugs. In addition, the acute 90-min drug exposure did not reflect cycle-specific drugs such as 5-fluorouracil which were often given by continuous infusion. Because of these discrepancies, further studies in animal models and human cancer specimens will be necessary to determine the clinical value of the above drug screen panel.

CONCLUSION This study revealed several important findings. Both cisplatin and Cytoxan were not as active as other cytotoxic agents directly compared in in vitro testing. Evaluation of 12 common cytotoxic agents yielded the following 8 promising drugs: ACT D, ARA C, DXR, 5FU, METHO, MIT0 C, VIN, and VP16. This preliminary drug screen panel should be studied further in animal models or human cancer specimens to establish its clinical application. ACKNOWLEDGMENTS This study was supported in part by the American Cancer Society Institutional Grant for Young Investigators and the American Cancer Society Clinical Oncology Fellowship program.

REFERENCES 1. Thigpen, J. T. Chemotherapy, in Synopsis of gynecologic oncology (C. P. Morrow and D. Townsend, Eds.), Wiley, New York, 3rd ed., pp. 412-458 (1987). 2. Deppe, G. Chemotherapy for endometrial cancer, in Chemotherapy of gynecologic cancer (G. Deppe, Ed.), Liss, Inc., New York, 2nd ed., pp. 155-174 (1990). 3. Hoffman, R. M. In vitro sensitivity assays in cancer: A review, analysis, and prognosis, J. Clin. Lab. Anal. 5, 133-143 (1991). 4. Weisenthal, L. M., and Lippman, M. E. Clonogenic and nonclonogenic in vitro chemosensitivity assays, Cancer Treat. Rep. 69(6), 615-632 (1985). 5. Von Hoff, D. D., Casper, J., and Bradley, E. Association between human tumor colony forming assay results and response of an individual patient’s tumor to chemotherapy, Am. J. Med. 70, 10271032 (1981). 6. Bertelsen, C. A., Sondak, V. K., Mann, B. D., Korn, E. L., and Kern, D. H. Chemosensitivity testing of human solid tumors. A review of 1582assayswith 258 clinical correlations, Cancer 53,12401245 (1984). 7. Alberts, D. S., Chang, S., Chen, H., Gross, J., Walson, P., Moon, J., and Salmon, S. E. Improved survival for relapsing ovarian cancer patients using the human tumor stem cell assay to select chemotherapy, Proc. Am. Sot. Clin. Oncol. 22, 462 (1981). 8. Selby, P., Buick, R., and Tannock, I. A critical appraisal of the human tumor stem cell assay, N. Engl. J. Med. 308,129-134 (1983). 9. Garewal, H., Ahmann, F., Schifman, R., and Celniker, A. ATP assay: Ability to distinguish cytostatic from cytocidal anticancer drug effects, .I. Natl. Cancer Inst. 77, 1039-1045 (1986). 10. Kuzmits, R., Rumpold, H., Mueller, M., and Schoof, G. The use of bioluminescence to evaluate the infhrence of chemotherapeutic drugs on ATP-levels of malignant cell lines, J. Clin. Chem. Clin. Biochem. 24, 293-298 (1986). 11. Sevin, B. U., Peng, Z. L., Perras, J. P., Ganjei, P., Penalver, M., and Averette, H. E. Application of an ATP-bioluminescence assay in human tumor chemosensitivity testing, Gynecol. Oncol. 31, 191204 (1988). 12. Petru, E., Sevin, B. U., Perras, J., Boike, G., Ramos, R., Nguyen, H., and Averette, H. E. Comparative chemosensitivity profiles in four human ovarian carcinoma cell lines measuring ATP bioluminescence, Gynecol. Oncol. 38, 155-160 (1990).

USE OF ATP ASSAY TO SELECI A DRUG SCREEN PANEL 13. Moyer, J. D., and Henderson, J. F. Ultrasensitive assay for ribonucleoside triphosphates in 50-1000 cells. Biochem. Pharmacol. 32, 3831-3834 (1983). 14. Alberts, D. S., and Chen, H. S. Tabular summary of pharmacokinetic parameters relevant to in vitro drug testing, in Cloning of human tumor stem cells (S. Salmon, Ed.), Liss, New York, pp. 3.5-359 (1980). 15. Chabner, B. A., and Myers, C. E. Clinical pharmacology of cancer chemotherapy, in Cancer Principles and Practice of Oncology, (V. T. Devita, S. Hellman, and S. A. Rosenberg, Eds.), Lippincott, Philadelphia, 3rd ed., pp. 369-371 (1989). 16. Chou, T., and Talaly, P. Quantitative analysis of dose-effect relationships: The combined effects of multiple drugs or enzyme inhibitors, Adv. Enzyme Regul. 22, 27-55 (1984). 17. Kusumoto, H., Maehara, Y., Kusumoto, T., Anai, H., et al. Chemosensitivity differences between primary and metastatic lesions of clinical gastric cancer, Eur. J. Surg. Oncol. 14(6), 685-689 (1988). 18. Zaffaroni, N., Silvestrini, R., Sanfilippo, O., Daidone, M., Bolis, G., et al. Drug sensitivity of different tumor lesions from the same patients evaluated by a short-term assay, Tumori 74(2), 137-144 (1988). 19. Donovan, J. F. Nonhormonal chemotherapy of endometrial adenocarcinoma: A review, Cancer 34, 1587-1592 (1974). 20. De Vita, V. T., Wasserman, T. H., Young, R. C., and Carter, S. K.

21.

22.

23.

24. 25. 26.

27.

191

Perspectives and research in gynecologic oncology. Treatment protocols, Cancer 38, 509 (1976). De Vita, V. T., and Schein, P. The use of drugs in combination for the treatment of cancer: Rationale and results, N. Engl. J. Med. 288, 998 (1973). Thigpen, J. T., Buchsbaum, H. J., and Blessing, J. A. Phase II trial of adriamycin in the treatment of advanced and recurrent endometrial carcinoma: A Gynecologic Oncology Group study, Cancer Treat. Rep. 63(l), 21-27 (1979). Thigpen, J. T., Blessing, J., and DiSaia, P. A randomized comparison of adriamycin with or without cyclophosphamide in the treatment of advanced or recurrent endometrial cancer, Proc. Am. Sot. Clin. Oncol. 4, 115 (1985). Schimke, R. T. Gene amplification, drug resistance, and cancer, Cancer Res. 44, 1735-1742 (1984). Bertino, J. R. The multidrug resistance phenotype, J. Clin. Oncol. 3, 293 (1985). Benard, J., Bourhis, J., and Riou, G. Clinical significance of multiple drug resistance in human cancers, Anticancer Res. 10, 12971302 (1990). Myers, C., Cowan, K., Sinha, B., Chabner, B. The phenomenon of pleiotropic drug resistance, in Important advances in oncology, 1987 (J. T. DeVita, Jr., S. Hellman, and S. A. Rosenberg, Eds.), Lippincott, Philadelphia, pp. 27-38 (1987).