Synergistic Interaction between Cisplatin and Gemcitabine in Vitro1

2’,2’-Difluorodeoxycytidine (dFdC; gemcitabine) is a new antineoplastic agent that is active against ovarian carcinoma, non-small-cell lung carcinoma, and head and neck squamous cell carcinoma. cis-diamminedichioropiatinum (CDDP; cisplatin) is used commonly for the treatment of these tumors. Because the two drugs have mechanisms of action that might be complementary, we investigated a possible synergism between dFdC and CDDP on growth inhibition. The combination was tested in the human ovarian carcinoma cell line A2780, its CDDP-resistant variant ADDP and its dFdC-resistant variant AG6000, the human head and neck squamous cell carcinoma cell line UMSCC-22B, and the murine colon carcinoma cell line C26-10. The cells were exposed to dFdC and CDDP as single agents and to combinations in a molar ratio of 1:500 for 1, 4, 24, and 72 h with a total culture time of 72 h. Synergy was evaluated using the multiple drug effect analysis. In A2780 and ADDP cells, simultaneous exposure to the drugs for 24 and 72 h resulted in synergism, but shorter exposure times were antagonistic. No synergism was found in the UMSCC-22B and C26-10 cell lines at prolonged simultaneous exposure. However, a preincubation with CDDP for 4 h followed by a dFdC incubation for 1, 4, 24, and 72 h was synergistic in all cell lines except C26-10 cells. A 4-h preincubation with dFdC followed by an incubation with the combination for 20 and 68 h was synergistic in all cell lines. Initial studies of the mechanism of interaction concentrated on the effect of CDDP on dFdCTP accumulation and DNA strand break formation. In all cell lines, CDDP failed to increase dFdCTP accumulation at 4or 24-h exposure to dFdC; in two cell lines, CDDP even tended to decrease dFdCTP accumulation. Neither dFdC nor CDDP caused more than 25% double strand break formation, whereas in the combination, CDDP even tended to decrease this type of DNA damage. The synergistic interaction between the two drugs is possibly the result of dFdC incorporation into DNA and/or CDDP-DNA adduct formation, which may be affected by each other. INTRODUCTION dFdC3 is a deoxycytidine analogue with two fluorine atoms substituted for the two hydrogens at the 2’ position of the ribose ring. The compound has established activity against solid tumors, as described for several solid tumor models, including ovarian and head and neck cancer (1-3). In the clinical setting, dFdC has proven to be effective against ovarian cancer and non-small cell lung cancer and moderately effective against head and neck cancers (4-6). After entering the cell, the drug has to be phosphorylated by dCK to dFdCMP and subsequently to dFdCTP (7), which can be incorporated into both DNA and RNA (8, 9). After incorporation of dFdCTP, one more deoxynucleotide can be incorporated, after which DNA polymerization stops (9). Unlike ara-C, exonuclease activity is unable to excise dFdCMP from the DNA (9). dFdC can be inactivated by the action of deoxycytidine deaminase to 2’,2’-difluorodeoxyuridine (7). For the treatment of the above-mentioned tumors, CDDP, a square planar coordination compound, is an important agent (10). The antitumor activity of CDDP is the result of the formation of adducts within the DNA (1 1-17). The binding of the CDDP moiety with two adjacent guanines on the same DNA strand is the most abundant lesion and is held responsible for the antitumor activity (12, 13). This lesion is thought to introduce a distortion in the DNA that is large enough to stop the division of the cells without being recognized rapidly and, thus, removed efficiently by repair enzymes (18). There is a possible relation between CDDP-DNA adduct levels and cell growth inhibition in cultured cells (14, 15). Synergistic interactions of CDDP with ara-C and DAC, two other deoxycytidine analogues, have been described previously. Studies on the mechanism of the synergism suggested differences in the interactions between ara-C and CDDP and DAC and CDDP (19, 20). CDDP-DNA adduct formation was enhanced by DAC incorporation, but ara-C did not affect the binding of CDDP to DNA. However, ara-C delayed the recovery of DNA synthesis inhibited by CDDP markedly. Because dFdC is related structurally to ara-C and DAC, it seemed appropriate to investigate the combination of CDDP with this compound too. One possible mechanism for synergy could be the inhibitory effect of CDDP on ribonucleotide reductase (21), possibly causing a depletion of dCTP pools, a potent feedback inhibitor of dCK. Subsequently, this could lead to increased dFdC phosphorylation. Alternatively, DNA might become more accessible to DNA adduct formation by distortions caused by Received 6/26/95; revised 1 1/15/95; accepted 12/5/95. I This work was supported by grants from the Dutch Cancer Society by Grants IKA-VU 90-19 and VU 94-753, and by a grant from Eli Lilly, Inc. (Nieuwegein, the Netherlands). 2To whom requests for reprints should be addressed. Phone: + 31-204442633; Fax: +31-20-4443844. 3 The abbreviations used are: dFdC, 2’,2’-difluorodeoxycytidine (gemcitabine); dCK, deoxycytidine kinase; ara-C, i3-D-arabino furanosyl cytosine; CDDP, cis-diamminedichloroplatinum (cisplatin); DAC, 2’-deoxy-5-azacytidine; 22B, UMSCC-22B; TCA, trichloroacetic acid; SRB, sulphorhodamine B; IC50, 50% inhibitory concentration; CI, combination index; FA, fraction affected; Dm, dose required to produce a 50% growth inhibition; FADU, fluorometric analysis of DNA unwinding. Research. on December 31, 2017. © 1996 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 522 Synergistic Interaction between Cisplatin and Gemcitabine dFdCTP incorporation. Furthermore, because both drugs have different, nonoveriapping side effects, synergistic or even additive antitumor effects of the combination may be clinically worthwhile. In addition, when using combinations, it is less likely that resistance will develop. Here, we describe the in vitro synergism between dFdC and CDDP. Three human ovarian carcinoma cell lines, the dFdCand CDDP-sensitive A2780 cell line and its CDDPand dFdCresistant variants ADDP and AG6000, respectively, the human head and neck squamous cell carcinoma cell line 22B, and the dFdC-insensitive murine colon tumor cell line C26-lO were exposed to both drugs in different concentrations and schedules. To investigate possible mechanisms of synergism, several parameters known to be important in the action of dFdC were studied, and the effect of CDDP on the dFdCTP accumulation and the number of DNA strand breaks induced by dFdC were measured to clarify the mechanism of the synergistic interaction. MATERIALS AND METHODS Chemicals and Reagents. DMEM and RPMI 1640 were purchased from Flow Laboratories (Irvine, United Kingdom); FCS was from GIBCO (Grand Island NY), TCA, glutamine, and gentamicin were from Merck (Darmstadt, Germany), trypsin, and SRB was from Sigma Chemical Co. (St. Louis, MO). dFdC was kindly supplied by Eli Lilly Research Labs (Indianapolis, USA). CDDP was purchased from Bristol-Myers Squibb (Weesp, the Netherlands). All other chemicals were of analytical grade and commercially available. Cell Culture. The in vitro experiments were performed with five different cell lines: A2780, a human ovarian cancer cell line (22, 23), and two variants, ADDP, with a 50-fold induced resistance to CDDP (kindly provided by Dr. K. J. Scanlon City of Hope National Medical Center, Duarte, CA; Ref. 22), and AG6000, with a 150,000-fold induced resistance to dFdC (23); 22B, a human head and neck squamous cell carcinoma cell line (24); and the murine colon carcinoma cell line C26-lO (25). Doubling times of the cell lines were 21, 32, 37, 40, and 14 h, respectively. Cells were grown in monolayers in DMEM supplemented with 5% heat-inactivated FCS, 1 mM L-glutamine, and 250 ng/ml gentamicin, except for ADDP, which was cultured in RPMI 1640 medium supplemented with 10% heat-inactivated FCS, 1 mM L-giutamine, and 250 ng/ml gentamicin at 37#{176}Cat 5% CO2. Chemosensitivity Testing. The determination of the IC50 was performed using the SRB assay (26, 27). The assay was performed using the National Cancer Institute protocol with some small modifications (27). Culture conditions were optimized for each cell line. The five cell lines were exposed to dFdC and CDDP as separate agents and to a combination of both. At day 1 , the cells were plated in 96-well plates in different densities, depending on their doubling times. The optimal plating number was the highest number of cells possible to enable log linear growth for 72 h (A2780, 5,000 cells/well; ADDP, 12,000 cells/well; AG6000, 18,000 cells/well; C26-lO, 1,500 cells/well; and 22B, 15,000 cells/well) in a volume of 100 i.l/well. Of the 96-well plates, the upper and lower rows (rows A and H) and the last two columns at the right side (columns 1 1 and I 2) were not used for cells, because we observed evaporation in these wells earlier. Thus, the first column contained only medium, the second column contained cells not exposed to drugs, and eight columns contained cells exposed to increasing concentrations of drugs. For each drug or drug combination, three wells were used. On day 2, dFdC and CDDP were added in a volume of 100 pal, resulting in series of final concentrations of 2.10 ‘ I_1.106 M dFdC and l.10_8_5,10_4 M of CDDP. The concentrations in the combination had the same range as the separate agents, with a dFdC:CDDP ratio of 1:500, which was based on the separate IC50 values in the various cell lines. Similar to the National Cancer Institute protocol, we chose the same culture time for all cell lines. The cells were exposed for 1 , 4, 24, and 72 h. After the exposure, the medium in the control and the drug-containing wells was removed and replaced by fresh, drug-free medium after a washing step with 50 il PBS (pH 7.4), and the cells were cultured until 72 h after the initial drug addition. Care was taken to ensure that cell loss was minimal in the procedure; furthermore, control wells were also washed. Final A values in washed and nonwashed wells were hardly different. At the end of the culture, the cells were precipitated with 50 pi ice-cold 50% (w/v) TCA and fixed for 60 mm, after which the SRB assay was performed as described (27). Only proteins of intact cells are stained, because proteins from killed cells would be degraded to amino acids by endogenous proteases after lysis (27). Growth inhibition curves were made relative to control As of every SRB assay. Two control values were always included, that of the A of cells at the day of drug administration (day 0, set at 0%) and that of cells after 3 days (set at 100%). This enabled us to calculate IC50 values (relative A at 50%), total growth inhibition (A of drug-treated cells similar to the initial value, 0%) and cell killing (A of drug treated cells lower than the initial value 0%). The points were connected by straight lines, and the IC50 values were determined from the interpolated graph (28). Also, the method of Chou and Talalay (29) was used; for these cell lines similar values (Dm) were achieved. The cells were not only exposed to the combination simultaneously, but two sequential schedules were also used: (a) 4-h preincubation with CDDP, after which the cells were exposed to dFdC for 1, 4, 24, or 72 h; and (b) 4-h preincubation with dFdC, followed by an incubation with both dFdC and CDDP for 20 or 68 h (Fig. 1). Dose-response interactions (antagonism, additivity, and synergism) between dFdC and CDDP were expressed as a nonexclusive case CI for every FA, using the method of Chou and Talalay (29), processed by a computer program developed by Chou and Chou (Biosoft, Cambridge, United Kingdom). This program was chosen because it provided one of the few objective computerized evaluation procedures (30). The data are supported by classic isobolograms made by the same computer program, which were based on the original method of Steel and Peckham (3 1 ). Other observations in favor for the choice of this program were the sigmoidal forms of the growth curves, ideal for median effect analysis. Thus, IC50 values (by interpolation) and Dm values (by the computer program) were usually similar. Measurable IC50 values for each drug separately are not required for computerized evaluation; Dm values would be calculated then by the program by extrapolation. For the separate drugs, we had to introduce their respective growth inhibition parameters, expressed as FA (e.g. , a FA of 0.25 is a growth Research. on December 31, 2017. © 1996 American Association for Cancer clincancerres.aacrjournals.org Downloaded from