Leukemia (1999) 13, 918–925  1999 Stockton Press All rights reserved 0887-6924/99 $12.00 http://www.stockton-press.co.uk/leu Potentiation of antitumor effects of derivatives in B-chronic lymphocytic leukemia cells by 2-chloro-2′-deoxyadenosine E Van Den Neste1,2, F Bontemps2, A Delacauw2, S Cardoen2, I Louviaux1, JM Scheiff1, E Gillis1, P Leveugle1, V Deneys1, A Ferrant1 and G Van den Berghe2

1Services d’He´matologie et de Biologie He´matologique, Cliniques Universitaires Saint-Luc, Brussels; and 2Laboratory of Physiological Chemistry, International Institute of Cellular and Molecular Pathology; Brussels, Belgium

Because 2-chloro-2′-deoxyadenosine (CdA) is active in B- the most widely used anticancer agents and, moreover, active chronic lymphocytic leukemia (B-CLL), and may interfere with in CLL.6 Cross-linking of DNA strands is a major biochemical DNA repair, we investigated the potentiating effect of CdA on 7,8 the cytotoxicity induced in vitro in B-CLL lymphocytes by action accounting for cytotoxicity of CP and CP derivatives. cyclophosphamide (CP) derivatives, which induce DNA dam- To evaluate the potential benefits of the combination of CP age by DNA cross-linking. Exposure to CdA at clinically achiev- and CdA in B-CLL, the modulator effect of CdA on the cyto- able concentrations for 2 h, followed by mafosfamide (MAF) or toxicity of CP was examined in vitro. In these studies, CP was 4-hydroxycyclophosphamide (4HC) for 22 h, resulted in syner- replaced by 4-hydroxycyclophosphamide (4HC), its active gistic cytotoxicity in the majority of B-CLL samples tested. Syn- metabolite produced by microsomal oxidation in the liver, ergy between CdA and MAF was observed in cell samples and by mafosfamide (MAF), which yields 4HC by rapid of sensitive/untreated patients, as well as in cells of 7 resistant/pretreated patients, particularly at the highest con- hydrolysis in aqueous conditions. The results of these experi- centrations of MAF. In the cells treated with CdA and MAF, we ments provided a rationale for the clinical association of CdA observed loss in ATP and hallmarks of apoptosis, as evidenced and CP, a regimen barely investigated hitherto.9,10 by cellular morphology and high molecular weight DNA frag- mentation. The synergy could be explained neither by an influ- ence of MAF on the phosphorylation of CdA, nor by an increase in the incorporation of CdA into DNA in the presence of MAF. Patients and methods The in vitro synergy between CdA and CP derivatives provides a rationale for the use of this association in B-CLL patients. Patients Keywords: 2-chloro-2′-deoxyadenosine; ; cyclophos- phamide; chronic lymphocytic leukemia; mafosfamide; 4-hydroxy- cyclophosphamide Samples were freshly obtained from 25 patients with B-CLL. All patients fulfilled the National Cancer Institute criteria for the diagnosis of B-CLL. Clinical staging was based on the sys- 11 Introduction tem described by Binet. Patients were at all stages of disease and free of any anticancer treatment for at least 3 months. 2-Chloro-2′-deoxyadenosine (CdA) is an analogue of deoxya- Binet staging ranged from A to C. Fourteen patients had never denosine with major antitumor activity in indolent lymphoid received , while 11 had been treated with vari- = malignancies including chronic lymphocytic leukemia (CLL).1 ous regimens including (CLB) (n 8), a purine = = CdA is resistant to , an enzyme which analogue as single agent (CdA: n 6; fludarabine: n 1), = plays an important role in the degradation of purine nucleo- CHOP (CP, , , prednisone: n 1) and = sides. Intracellularly, CdA is phosphorylated initially by high-dose corticosteroids (n 1). The median number of prior deoxycytidine kinase into its 5′-monophosphate, CdAMP, and regimens was two (range 1–5). All the patients who had been thereafter converted successively into CdADP and CdATP. pretreated with a were in relapse or pro- The latter triphosphate is considered the active metabolite of gression. Eleven patients (eight pretreated, three untreated) CdA, and has been shown to interfere with DNA synthesis had a lymphocyte count doubling time of less than 12 months. through inhibition of DNA polymerase ␤ and ribonucleotide reductase.2 Since the latter enzymes are also involved in DNA repair, Materials it is conceivable that CdA may influence this process. Accord- ingly, CdA has been shown to block DNA repair synthesis in CdA (Ͼ99.9% purity) was synthesized and supplied by Pr L X-irradiated lymphocytes, a property which has been explored Ghosez, Laboratory of Organic Chemistry, Catholic University more extensively with fludarabine, another analogue of of Louvain, Belgium. [8-3H]CdA (24.2 Ci/mmol) was from deoxyadenosine, which inhibits DNA repair synthesis elicited Moravek Biochemicals (La Brea, CA, USA). 4HC was kindly by alkylating agents or UV-irradiation.3–5 provided by Dr S Ludeman, Duke University, Durham, NC, Interference of CdA with DNA repair raises the possibility USA. MAF was kindly provided by Dr J Pohl, Asta Medica, that CdA might produce synergistic antitumor effects when Germany. 3-[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazol- combined with agents that evoke a DNA repair response. ium bromide (MTT) was purchased from Sigma-Aldrich (St Cyclophosphamide (CP), a bifunctional DNA alkylating agent Louis, MO, USA). Fetal bovine serum (FBS) and belonging to the family, is one of the most penicillin/streptomycin were purchased from BioWhittaker active compounds in this respect, in addition to being one of Europe (Verviers, Belgium). DNA markers, agarose and RPMI 1640 were from Life-Technologies (Merelbeke, Belgium). Ficoll–Paque PLUS (density: 1.077 ± 0.001 g/ml) was from Correspondence: E Van Den Neste, Service d’He´matologie, Cliniques Universitaires Saint-Luc, 10, av. Hippocrate, B-1200 Brussels, Pharmacia Biotech (Roosendaal, The Netherlands). May– Belgium; Fax: 322 7648959 Gru¨nwald-Giemsa was from Merck (Darmstadt, Germany). Received 12 November 1998; accepted 1 March 1999 Fluorescein-conjugated human annexin V was from R&D Sys- CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 919 tems (Abingdon, Oxon, UK). All other chemicals, materials cytometer (San Jose, CA, USA), using Cell Quest software. and reagents were of the highest quality available. Drug toxicity was quantified by summing the proportion of single annexin V+ cells (purely apoptotic cells) and annexin V+/PI+ cells (necrotic and late-apoptotic cells). Isolation of lymphocytes and cell culture

Freshly obtained peripheral blood from B-CLL patients was Cellular morphology fractionated by Ficoll–Paque sedimentation. Mononuclear cells were washed with cold PBS and resuspended in RPMI At the indicated times, 5 × 105 cells were centrifuged on to 1640 supplemented with 10% FBS and 1% penicillin (10000 glass slides, fixed with methanol, and stained with May–Gru¨n- U/ml)–streptomycin (10 mg/ml). Cells were counted, diluted wald-Giemsa. Cell morphology was examined by light ° to indicated concentration in RPMI, and incubated at 37 Cin microscopy using a BH2 microscope (Olympus, Tokyo, 5% CO2. Japan). Apoptotic morphology was determined by standard criteria, including chromatin condensation, nuclear mar- gination and cellular shrinkage. Necrotic cells were defined Drug exposure, analysis of cytotoxicity and drug by the following characteristics: cytoplasmic vacuolization, interaction cell and nuclear swelling, rupture of nuclear and plasma membranes, appearance as faintly stained cells with nuclear × 6 B-CLL cells, resuspended at a concentration of 1.5 10 /ml, ghosts. At least 400 cells in a minimum of three fields of view were exposed to CdA or CP derivatives for 24 h. When CdA were counted from duplicate slides for each sample by and CP derivatives were combined, CdA was added 2 h different experienced observers. before CP derivatives. 4HC and MAF were freshly dissolved in sterile water prior to each experiment. Stock solution of CdA was prepared in ETOH/NaCl 0.9% (v/v). After drug Quantitation of adenylic nucleotides and intracellular exposure, 7 ml of cell suspension were washed with RPMI, metabolites of [8-3H]CdA resuspended in 200 ␮l RPMI and seeded in 96-well microtiter plates. Controls and each drug concentration were set up in CLL lymphocytes were incubated at a concentration of triplicate. Viability of cells after drug treatment was measured 10 × 106 cells/ml with 0.5 ␮M [8-3H]CdA. At the indicated 72 h after washing, using the MTT assay.12 The optical density time points, 3 ml of the cell suspension were washed twice in (OD) of each well was measured at 540 nm with a Multiwell 5 ml of cold PBS, before addition of 350 ␮lof1M HClO to Scanning Spectrophotometer (Molecular Devices, Sunnyvale, 4 the washed pellet. The supernatant obtained after centrifug- CA, USA). Leukemic cell survival (LCS) was calculated by the ation was neutralized with 3 M K CO . Nucleotides were sep- equation: (mean OD of treated well/mean OD of control 2 3 arated by HPLC on a 12.5-cm Partisphere 5 SAX column by wells) × 100%. Interaction between drugs, ie synergism, addi- the method of Hartwick and Brown.14 UV detection of nucleo- tivity or antagonism, was defined according to the multiplicat- tides was performed at 254 nm. Fractions corresponding to ive and maximum models as used by Kaspers et al.12 In the 1 min of elution were collected and their radioactivity was multiplicative model, the expected effect of a drug combi- determined by scintillation counting. Radioactivity incorpor- nation is the product of the effect of each single drug. In the ated in the acid-insoluble pellet was measured after washing maximum model, the expected effect of a combination is of the pellet followed by digestion with soluene. equal to that of the most active single drug. These models were combined to establish the interactions between drugs: synergism if observed LCS (drug A + drug B) Ͻ LCS (drug A) × LCS (drug B); additivity if LCS (drug A) × LCS (drug DNA fragmentation analysis Ͻ + Ͻ B) observed LCS (drug A drug B) LCS (Dmax), where High molecular weight DNA fragmentation (HMWF) was ana- Dmax is the effect of the most active single drug; and antagon- + Ͼ lyzed by pulsed field gel electrophoresis as previously ism if observed LCS (drug A drug B) LCS (Dmax). The LC50 15 5 was defined as the drug concentration that results in 50% LCS, described. Briefly, after drug treatment, 10 B-CLL cells were and was determined graphically as the point where the dose- embedded in 0.6% agarose plugs. After a 24-h incubation in response curve crosses the 50% LCS point. lysis buffer containing 1% Sarkosyl and proteinase K, the plugs were loaded into 1% agarose gels, and DNA was separated with a CHEF-DRII system (Bio-Rad Laboratories, Richmond, Annexin V binding CA, USA). Gels were stained with ethidium bromide and photographed. Apoptotic cells, with exposure of phosphatidylserine on the outer leaflet, and necrotic cells were quantified by flow cyto- metry after annexin V and propidium iodine (PI) double labe- Statistical analysis ling.13 Briefly, after 24 h incubation with or without drugs as indicated, cells were washed twice in cold PBS and then Differences in LCS between control and treated cells, or resuspended at a concentration of 1 × 106/ml in Hepes buff- between intracellular metabolites of [8-3H]CdA accumulated ered saline solution supplemented with 25 mM CaCl2. One with or without MAF, were analyzed for statistical significance hundred microliters of the cells were incubated for 15 min at by the two-tailed Student’s t-test for paired or unpaired room temperature in the presence of fluorescein-conjugated samples, at a level of significance of P р 0.05. The same annexin V (10 ␮lof10␮g/ml stock) and PI (10 ␮lof50␮g/ml method was applied to compare LCS from untreated and pre- stock). Following completion of cell staining, cells were viously treated patients. All mean values were calculated from immediately analyzed on a Becton Dickinson FACScan flow experiments performed in different patients. CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 920 Results

Time course of CdA nucleotide concentrations in B-CLL cells incubated with [8-3H]CdA

B-CLL cells were incubated with 0.5 ␮M [8-3H]CdA. The kin- etics of accumulation of CdAMP, CdADP, and CdATP are shown in Figure 1. Intracellular phosphorylated metabolites of CdA were already detectable 15 min after the start of the incu- bation. Accumulation of CdAMP was predominant, followed by that of CdATP, whereas that of CdADP was negligible. The intracellular levels of CdATP (pmoles/106 cells, mean ± s.e.m.) reached 0.19 ± 0.05 after a 2-h, and 0.20 ± 0.07 after a 24-h incubation with CdA. Compared to the accumulation of its phosphorylated metabolites, incorporation of CdA into nucleic acids was limited: it reached 0.004 ± 0.001 and 0.045 ± 0.012 pmoles/106 cells after 2 and 24 h, respectively. This indicated that a 2-h preincubation with CdA before CP derivative addition was long enough to allow sufficient forma- tion of CdATP, the putative active metabolite of CdA.

In vitro cytotoxicity in B-CLL cells of CdA, MAF and 4HC Figure 2 Effect of CdA and CP derivatives on LCS. B-CLL cells were incubated with the indicated concentrations of CdA (᭺), MAF (ᮀ), or 4HC (̃), for 24 h. Cell survival was measured by MTT assay. As shown in Figure 2, a 24-h incubation of the cells with Results are means ± s.e.m. of four separate experiments. increasing concentrations of CdA, 4HC or MAF as single agents, resulted in a progressive decrease in LCS, as assessed ␮ ␮ by the MTT assay. LC50 were as follows: CdA, 0.36 M; MAF, respectively, upon preincubation of the cells with 0.05 M 3.33 ␮M; 4HC, 5.0 ␮M. CdA. Combination experiments were performed with two con- In order to determine the importance of drug scheduling, centrations of CdA, 0.05 and 0.1 ␮M, and with concentrations cells were treated either with 2-h CdA followed by 18-h MAF, of MAF and 4HC derived from the dose-response curves or with 2-h MAF followed by 18-h CdA. Although potentiation depicted in Figure 2. The B-CLL lymphocytes were incubated occurred in all conditions, the sequence of CdA followed by in the presence of CP derivatives during 22 h, after a 2-h pre- MAF was slightly, although non-significantly, more active incubation with or without CdA. As shown in Figure 3, the (data not shown). dose-response curve to MAF was markedly shifted to the left Effect of CdA on the toxicity of MAF was also analyzed upon preincubation with CdA. In the presence of 0.05 using annexin V/PI staining and two-parameter flow cytometer ␮ (Figure 3a) or 0.1 M (Figure 3b) CdA, the LC50 of MAF was analysis. Cytotoxicity of increasing concentrations of MAF was reduced by a factor of 8 and 23, respectively. Potentiation of augmented in the presence of 0.1 ␮M CdA. The percentage + + − MAF cytotoxicity by 0.05 ␮M CdA was obtained despite a low (mean ± s.e.m., for n = 4) of annexin V (and either PI or PI ) cytotoxicity of CdA as single agent at this concentration cells after MAF 0.2 ␮M,2␮M, and 5 ␮M was 45 ± 7, 48 ± 7, (median LCS: 76%, range 43–102% for n = 13). The greatest and 63 ± 8 without CdA, and 63 ± 8, 77 ± 6, and 83 ± 5in potentiating effect of CdA was recorded at 2 ␮M MAF. The the presence of 0.1 ␮M CdA, respectively. Potentiation of MAF cytotoxicity of 4HC was similarly increased by preincubation toxicity by CdA was found to be highly significant at all the of the cells with CdA. LCS (%, mean ± s.e.m. for n = 4), which concentrations of MAF tested (P р 0.02). Mean percentage + after incubation with 0.5 and 2 ␮M of 4HC alone was 89 ± 5 (± s.e.m., for n = 4) of annexin V cells in samples incubated and 69 ± 9, respectively, decreased to 46 ± 10 and 24 ± 10, during 24 h without drugs or with CdA 0.1 ␮M alone was 37 ± 2 and 58 ± 7, respectively.

Characterization of in vitro drug interactions

To differentiate between synergism, additivity and antagon- ism, LCS were analyzed according to the multiplicative and maximum models as detailed in the Methods section. Figure 4 shows the comparison between the expected LCS values, as predicted by both the multiplicative and the maximum model, and the observed LCS values for each drug combination. The multiplicative model identifies synergistic interactions (when the observed LCS (drug A + drug B) is lower than the expected LCS (drug A + drug B)), whereas the maximum model enables Figure 1 Time course of the concentrations of CdA nucleotides in identification of antagonism (when the observed LCS (drug A 3 + B-CLL cells incubated with 0.5 ␮M [8- H]CdA. Points are drug B) is higher than the LCS of the most active single drug). means ± s.e.m. of three separate experiments. Analysis of 24 experiments at up to five different concen- CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 921

Figure 3 Effect of drug combinations on cell survival. B-CLL cells were incubated in the presence of the indicated concentrations of MAF for 22 h, after a 2-h preincubation without (ᮀ) or with (᭿) CdA 0.05 ␮M (a), or 0.1 ␮M (b). Cell survival was measured by the MTT assay. Points are means ± s.e.m. of three to 10 determinations; Student’s t-test on paired data; a, P Ͻ 0.005; b, P Ͻ 0.05. Figure 4 Comparisons (n = 43) between observed LCS and expected LCS values for the combination CdA + MAF in B-CLL trations, resulting in 43 individual comparisons between samples. (a) Expected LCS according to the multiplicative model observed and expected LCS, showed synergistic interaction in where LCS (drug A + drug B) = LCS (drug A) × LCS (drug B); (b) expected LCS according to the maximum model where LCS (drug 72%, additivity in 19%, and antagonism in 9% of the samples + = treated with CdA and MAF. Antagonism between CdA and A drug B) LCS of the most active single drug. The continuous line is x = y. MAF, which occurred most frequently at the lowest concen- tration of MAF, was found exclusively in previously treated patients (n = 4). When LCS observed for the combination of was found in 100% of the samples in which this combination CdA and MAF was compared to LCS for the most active single was tested (not illustrated). The observed LCS for the combi- drug in each individual experiment, a highly significant differ- nation of CdA and 4HC was significantly lower than the LCS ence was found (P Ͻ 0.01). This shows that, overall, the drug of the most active single drug when these values were com- combination was significantly more toxic than the most active pared in each individual experiment (P Ͻ 0.01). single drug. The combination of CdA and 4HC was tested in Accordingly, when the toxicity of MAF ± CdA, as assessed four different samples, at up to four different concentrations, by annexin V/PI double staining, was analyzed for drug inter- which resulted in eight comparisons between the expected action, the majority (11/12) of the pairs tested showed syner- and observed LCS values. Synergism between CdA and 4HC gistic interaction, and 1/12 showed additivity. CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 922 Characteristics of cell death following incubation with drugs

Cell death induced by CdA was associated with morphologi- cal changes of apoptosis such as chromatin condensation and cell shrinkage, in comparison with control cells (Figure 5a and b). Cells treated with MAF (Figure 5c) similarly displayed the hallmarks of apoptosis. In the samples treated with both CdA and MAF, evidence of cell death by necrosis occurred in association with apoptosis (Figure 5d). As calculated in three separate experiments, the percentage of apoptotic plus nec- rotic cells (mean ± s.e.m.) after a 24-h incubation without drugs, or with CdA 0.05 ␮M, MAF 5 ␮M, or CdA 0.05 ␮M + MAF 5 ␮M was 4.3 ± 2.9, 15.0 ± 12.6, 15.3 ± 8.9 and 37.3 ± 13.9, respectively. Counting of apoptotic/necrotic cells under the microscope was in keeping with cell viability as measured by the MTT assay. To confirm the nature of cell death, we analyzed whether incubation with drugs induced HMWF. Agarose gel electrophoresis showed fragmentation of genomic DNA into 50 kb pieces after incubation with MAF ± CdA, as shown in cells from two representative patients (Figure 6). Incubation with 4HC also induced HMWF in B-CLL cells (results not shown).

Effect of MAF on the metabolism of [8-3H]CdA

To detect an influence of MAF on the phosphorylation of CdA, B-CLL cells were incubated in the presence of 0.5 ␮M [8- 3H]CdA followed or not by 2 ␮M MAF. After a 24-h incu- bation, the concentrations of the phosphorylated metabolites of [8-3H]CdA (pmoles/106 cells, mean ± s.e.m. for n = 3) were not modified in the presence of MAF. In particular, the con- centration of [8-3H]CdATP was not significantly different, without (0.26 ± 0.21), or with (0.18 ± 0.15) MAF (P Ͼ 0.1). Likewise, incorporation of [8-3H]CdA into nucleic acids was similar (0.025 ± 0.010 pmoles/106 cells, mean ± s.e.m.) with or without MAF. Cellular ATP concentrations (pmoles/106 cells, mean ± s.e.m. for n = 3) tended to be lower in samples treated with CdA and MAF (0.13 ± 0.05) than in samples treated with CdA alone (0.22 ± 0.08, P = 0.12).

Correlations between sensitivity of B-CLL cells to CdA ± MAF and clinical status

B-CLL cells from different patients showed a wide range of sensitivities to CdA, to MAF and to 4HC. Sensitivity varied up to 2.3-fold for CdA 0.05 ␮M and up to 2.8-fold for CdA 0.1 ␮M. Sensitivity to MAF 0.2, 2 and 5 ␮M varied up to 2.1-, 4.6- and 46.7-fold, respectively. LCS (%, mean ± s.e.m.) after 2 ␮M MAF was significantly lower (P = 0.02) in previously untreated patients (55 ± 8 for n = 13) than in pretreated Figure 5 Induction of cell death by CdA ± MAF in B-CLL cells in patients (78 ± 6 for n = 9). After 5 ␮M MAF, LCS in samples vitro. (a) Morphology of control cells 24 h after incubation in drug- from chemotherapy naive patients (10 ± 2 for n = 7) was sig- = ± free medium. Morphological changes of cells after 24-h incubation nificantly lower (P 0.01) than in pretreated patients (43 10 with (b) 0.05 ␮M CdA, (c) 5 ␮M MAF, or (d) 0.05 ␮M CdA + 5 ␮M MAF. for n = 8). Similarly, cells from untreated patients were sig- Cells were stained with May–Gru¨nwald-Giemsa and observed by light nificantly more sensitive to CdA 0.05 ␮M (P Ͻ 0.05) than cells microscopy (original magnification ×165). from pretreated patients. Although the combination of CdA 0.5 ␮M with MAF 0.2 ␮M caused significantly (P Ͻ 0.01) more cytotoxicity in untreated patients than in pretreated patients, no differences in LCS were found according to the clinical status with the other combinations of CdA and MAF. CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 923

Figure 6 High molecular weight DNA fragmentation in B-CLL cells from two representative patients after 24-h incubation. Patient 1: lanes 1–4, patient 2: lanes 5–8. Lane 1, control cells; lane 2, cells incubated with 0.05 ␮M CdA; lane 3, cells incubated with 5 ␮M MAF; lane 4, cells incubated with 0.05 ␮M CdA and 5 ␮M MAF; lane 5, control cells; lane 6, cells incubated with 0.1 ␮M CdA; lane 7, cells incubated with 2 ␮M MAF; lane 8, cells incubated with 0.1 ␮M CdA and 2 ␮M MAF; lane 9, molecular weight markers.

Discussion the clinical situation where rapid intracellular accumulation of CdA nucleotides has been shown after 2-h i.v. or oral Purine analogues such as CdA and fludarabine demonstrate CdA.24 A similar level of synergy was observed between CdA marked antitumor activity in terms of frequency and duration and 4HC, although studies with the latter were more limited. of responses in B-CLL.1,16 However, despite these results, fol- The mechanisms underlying the synergism between CdA low-up of CLL patients treated with CdA or fludarabine in first and CP derivatives remain speculative. This interaction was line does not clearly show a survival advantage over conven- due neither to an enhancement in the phosphorylation of CdA tional therapies of CLL.2,17,18 Theoretically, a possible way to to its active metabolites by MAF, nor to a greater incorpor- achieve better results in CLL is to combine purine analogues ation of CdA into nucleic acids in the presence of MAF. In with other agents. Indeed, CdA has been shown to interfere keeping with previous reports, the morphological and bio- with various enzymes associated with DNA repair.2,19 DNA chemical characteristics of cell death by apoptosis were repair inhibition by CdA may hence increase the accumu- observed in samples treated with CdA and/or MAF.26,27 The lation and slow the removal of DNA damage induced by morphological data also suggests that the gain in cell death drugs such as alkylating agents, leading to enhanced cytotox- achieved by adding CdA to MAF may be partially due to an icity of the latter. This strategy could also offer an opportunity increased incidence of necrosis in addition to cell death by to circumvent cellular resistance due to enhanced DNA repair apoptosis. Compared to the cells treated with CdA alone, the capability, a phenomenon which may partially play a role in decline in cellular ATP in the cells treated with CdA and MAF the resistance to alkylating agents in CLL.20–23 accords with a greater cellular dysfunction with the latter In the present report, we have investigated the effect of the combination. Although not demonstrated in this study, the combination of CdA with CP derivatives on B-CLL cells in synergism between CdA and CP derivatives may proceed from vitro. A 2-h preincubation with CdA before MAF or 4HC inhibition of DNA repair by CdA, which thereby enhances the addition was judged to be adequate given the rapid intracellu- toxicity of MAF or 4HC. CdA has indeed been shown to lar accumulation of CdATP. In the majority of samples tested, inhibit the polymerization step of DNA repair in X-irradiated we observed synergistic or additive cytotoxicity between CdA normal lymphocytes, and to interfere with the repair of DNA and CP derivatives. Antagonism between CdA and MAF, double-strand breaks in X-irradiated Chinese hamster V79 although rare, occurred more frequently in pretreated/resistant cells, thereby enhancing the lethal effects of X-rays.3,28 Fur- patients and at the lowest concentration of MAF. However, thermore, therapeutic synergy between CdA and CP was by raising the concentration of MAF, antagonism with CdA observed in mice bearing leukemia P388 and L1210.29 Syner- was no longer seen in these patients and all had enhancement gistic antitumor activity was also shown in CLL cells between of the cellular toxicity of MAF by CdA. In two patients relaps- CLB and CdA.30 However, this synergy was only observed if ing after CdA, and clinically refractory to CLB, who displayed the cells were treated with CLB prior to CdA and no effect of in vitro resistance to MAF, CdA increased the toxicity of MAF. CdA on the repair of CLB-induced cross-links was shown. This These two patients further proved to be clinically sensitive to last observation, as mentioned by the authors, could be the combination of CdA with CP. These data suggest that, by explained by the higher concentrations of CLB required to adding CdA, B-CLL cells isolated from patients pretreated detect DNA cross-links in the alkaline elution assay, as com- and/or refractory to alkylating agents were brought into a more pared to the cytotoxicity assay. The concentrations of CdA sensitive range at the in vitro level and that this favorable allowing synergy with CLB were also higher (10-fold) than in interaction is susceptible to hold true in vivo. Although in vitro our study and superior to those observed in the plasma of studies are not always translatable to the clinic, the concen- patients receiving the standard dose of CdA. Other purine ana- trations of CdA and 4HC chosen were in the range of what logues, including fludarabine and deoxycoformycin, have can be achieved in the plasma of patients treated with CdA been clearly shown to restrain DNA repair elicited by UV- or CP.24,25 Furthermore, our in vitro schedule was relevant to irradiation, X-irradiation, or DNA cross-linking.4,5,31,32 CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 924 Several clinical combination regimens including fludarab- 7 Niemeyer U, Engel J, Hilgard P, Peukert M, Pohl J, Sindermann ine or CdA and a DNA damaging agent have been reported H. Mafosfamide – a derivative of 4-hydroxycyclophosphamide. in B-CLL. Although proven efficacious, administration of Prog Clin Biochem Med 1989; 9: 35–60. 8 Crook TR, Souhami RL, McLean AEM. Cytotoxicity, DNA cross- adequate doses of CdA or fludarabine in association with CLB linking, and single strand breaks induced by activated cyclophos- has been limited by the occurrence of protracted pancytop- phamide and in human leukemia cells. Cancer Res 1986; enia following these combinations.17,33,34,35 Yet, adminis- 46: 5029–5034. tration of oral CdA with CP in patients with Waldenstro¨m’s 9 Weber DM, Delasalle K, Gavino M, Wood A, Cabanillas F, Alex- disease was reported to be active and safe.9 The combination anian R. Primary treatment of Waldenstrom’s macroglobulinemia of CdA, CP and prednisone was feasible and active in pre- with subcutaneous 2-chlorodeoxyadenosine and oral cyclophos- 10 phamide. Proc Am Soc Hematol 1997, 90: 1592. viously untreated patients with CLL or low grade NHL. That 10 Zulian GB, Guetty-Alberto M, Iten P-A, Cerny T, Alberto P. CCP of fludarabine and CP was shown to be the most active sal- Cladribine with cyclophosphamide and prednisone as first-line vage regimen in patients in relapse after, or refractory to, flu- treatment of low grade B-lymphoproliferative disorders. Ann darabine as single agent. Other combination therapies includ- Oncol 1996; 7 (Suppl. 3): 553. ing purine analogues are active in CLL, although there is no 11 Binet JL, Auquier A, Dighiero G, Chastang C, Piguet A, Goasguen clear evidence yet for a superiority over fludarabine or CdA J, Vaugier G, Potron G, Colona P, Oberling F, Thomas M, Tchierna in this disease.2,16 G, Jacquillat C, Boivin P, Lesty C, Duault MT, Monconduit M, Belabbes S, Gremy F. A new prognostic classification of chronic The assumption that CP, compared to CLB, has a favorable lymphocytic leukemia derived from a multivariate survival analy- profile of clinical toxicity in combination with purine ana- sis. Cancer 1981; 48: 198–206. logues, together with the results of this study, led us to design 12 Kaspers GJL, Veerman AJP, Pieters R, Van Zantwijk I, Ha¨hlen K, an ongoing phase I trial of CdA in combination with CP in Van Wering ER. Drug combination testing in acute lymphoblastic patients with CLL. Further studies are also in progress to deter- leukemia using the MTT assay. Leukemia Res 1995; 19: 175–181. mine whether in vitro sensitivity to CdA and MAF may predict 13 Koopman G, Reutelingsperger CPM, Kujiten GAM, Keehnen RMJ, Pals ST, van Oers MHJ. Annexin V for flow cytometric detection clinical response to this combination, and, hence, help to of phosphatidylserine expression on B cells undergoing apoptosis. tailor chemotherapy on a more individual basis. Blood 1994; 84: 1415–1420. In conclusion, the results of the present study, demonstrat- 14 Hartwick RA, Brown PR. The performance of microparticle chemi- ing synergistic cytotoxicity in vitro between CdA and CP cally-bonded anion-exchange resins in the analysis of nucleotides. derivatives, further support the rationale of combining purine J Chromatogr 1975; 112: 650–662. analogues with DNA-damaging agents in the clinical manage- 15 Huang P, Robertson LE, Wright S, Plunkett W. High molecular ment of patients with B-CLL. As suggested by our data, this weight DNA fragmentation: a critical event in nucleoside ana- logue-induced apoptosis in leukemia cells. Clin Cancer Res 1995; strategy should not be limited to previously untreated or 1: 1005–1013. sensitive patients. 16 Keating MJ, O’Brien S, Lerner S, Koller C, Beran M, Robertson LE, Freidreich EJ, Estey E, Kantarjian H. Long-term follow-up of patients with chronic lymphocytic leukemia receiving fludarabine Acknowledgements regimens as initial therapy. Blood 1998; 4: 1165–1171. 17 Rai KR, Peterson B, Elias L, Shepherd L, Hines J, Nelson D, Cheson B, Kolitz J, Schiffer CA. A randomized comparison of fludarabine We thank P van der Bruggen and V Gre´goire for allowing and chlorambucil for patients with previously untreated chronic us to use the multiwell scanning spectrophotometer and the lymphocytic leukemia. A CALGB, SWOG, CTG/NCI-C and ECOG electrophoresis system, respectively. We are grateful to V Gre´- inter-group study. Blood 1996; 88: 141a. goire and A Delannoy for critical review of the manuscript. 18 French cooperative group on CLL. Comparison of fludarabine This work was supported in part by grants from the Fonds (FDB), CAP and CHOP in previously untreated stage B and C National de la Recherche Scientifique (grant Te´le´vie No. chronic lymphocytic leukemia (CLL). First interim result of a ran- domized trial in 247 patients. Blood 1994; 83: 461a. 7.4541.97), the Salus Sanguinis Foundation, and the Fond- 19 Chunduru SK, Appleman JR, Blakley RL. Activity of human DNA ation pour la Leuce´mie. polymerases ␣ and ␤ with 2-chloro-2′-deoxyadenosine 5′-tri- phosphate as a substrate and quantitative effects of incorporation on chain extension. Arch Biochem Biophys 1993; 302: 19–30. References 20 Geleziunas R, McQuillan A, Malapetsa A, Hutchinson M, Kopriva D, Wainberg MA, Hiscott J, Bramson J, Panasci L. Increased DNA 1 Delannoy A. 2-Chloro-2′-deoxyadenosine: clinical application in synthesis and repair-enzyme expression in lymphocytes from hematology. Blood Rev 1996; 10: 148–166. patients with chronic lymphocytic leukemia resistant to nitrogen 2 Plunkett W, Gandhi V. Nucleoside analogs: cellular pharma- mustards. J Natl Cancer Inst 1991; 83: 557–564. cology, mechanisms of action, and strategies for combination ther- 21 Torres-Garcia SJ, Cousineau L, Caplan S, Panasci L. Correlation apies. In: Cheson BD, Keating MJ, Plunkett W (eds). Nucleoside of resistance to nitrogen mustards in chronic lymphocytic leuke- Analogs in Cancer Therapy. Marcel Dekker: New York, 1997, mia with enhanced removal of -induced DNA cross- pp 1–35. links. Biochem Pharmacol 1989, 38: 3122–3123. 3 Seto S, Carrera CJ, Wasson DB, Carson DA. Inhibition of DNA 22 Mu¨ller MR, Seiler F, Thomale J, Buschfort C, Rajewski MF, Seeber repair by deoxyadenosine in resting human lymphocytes. J Immu- S. Capacity of individual chronic lymphocytic leukemia lympho- nol 1986; 8: 2839–2843. cytes and leukemic blasts for repair of O6-ethylguanine in DNA: 4 Sandoval A, Consoli U, Plunkett W. -mediated inhi- relation to chemosensitivity in vitro and treatment outcome. Can- bition of nucleotide excision repair induces apoptosis in quiescent cer Res 1994; 54: 4524–4531. human lymphocytes. Clin Cancer Res 1996; 2: 1731–1741. 23 Muller C, Christodoulopoulos G, Salles B, Panasci L. DNA-depen- 5 Koehl U, Li L, Nowak B, Ruiz van Haperen V, Kornhuber B, dent protein kinase activity correlates with clinical and in vitro Schwabe D, O’Brien S, Keating M, Plunkett W, Yang L-Y. Fludara- sensitivity of chronic lymphocytic leukemia lymphocytes to nitro- bine and cyclophosphamide: synergistic cytotoxicity associated gen mustards. Blood 1998; 92: 2213–2219. with inhibition of interstrand cross-link removal. Proc Am Assoc 24 Liliemark J. The clinical pharmacokinetics of cladribine. Clin Phar- Cancer Res 1997; 38: 10. macokinet 1997; 32: 120–131. 6 Fayad L, O’Brien S. Chronic lymphocytic leukemia and associated 25 Sladek NE, Doeden D, Powers JF, Krivit W. Plasma concentrations disorders. In: Pazdur R (ed). Medical Oncology – A Comprehen- of 4-hydroxycyclophosphamide and phosphoramide mustards in sive Review. PRR: Huntington, NY, 1995, pp 37–56. patients repeatedly given high doses of cyclophosphamide in CdA ± cyclophosphamide in B-CLL cells E Van Den Neste et al 925 preparation for bone marrow transplantation. Cancer Treat Rep and inhibition of DNA damage repair in irradiated murine L5178Y 1984; 68: 1247–1254. lymphoblasts and human chronic lymphocytic leukemia cells 26 Davidoff AN, Mendelow BV. Cell-cycle disruptions and apoptosis treated with 2′-deoxycoformycin and deoxyadenosine in vitro. induced by the cyclophosphamide derivative mafosfamide. Exp Cancer Res 1988; 48: 3981–3986. Hematol 1993; 21: 922–927. 32 Li L, Liu X-M, Glassman AB, Keating MJ, Stros M, Plunkett W, 27 Robertson LE, Chubb S, Meyn RE, Story M, Ford R, Hittelman W, Yang L-Y. Fludarabine triphosphate inhibits nucleotide excision Plunkett W. Induction of apoptotic cell death in chronic lympho- repair of -induced DNA adducts in vitro. Cancer Res cytic leukemia by 2-chloro-2′-deoxyadenosine and 9-␤-D-arabino- 1997; 57: 1487–1494. syl-2-fluoroadenine. Blood 1993; 81: 143–150. 33 Tefferi A, Witzig TE, Reid JM, Li C-Y, Ames M. Phase I Study of 28 Kuwabara M, Tanabe K, Hiraoka W, Tamura Y, Sato F, Matsuda combined 2-chlorodeoxyadenosine and chlorambucil in chronic A, Ueda T. 2-chlorodeoxyadenosine inhibits the repair of DNA lymphoid leukemia and low-grade lymphoma. J Clin Oncol 1994; double-strand breaks in X-irradiated Chinese hamster V79 cells. 12: 569–574. Chem–Biol Interactions 1991; 79: 349–358. 34 Weiss M, Spiess T, Berman E, Kempin S. Concomitant adminis- 29 Gora-Tybor J, Robak T. Synergistic action of 2-chlorodeoxyadeno- tration of chlorambucil limits dose intensity of fludarabine in pre- sine and cyclophosphamide on murine leukemias L1210 and viously treated patients with chronic lymphocytic leukemia. Leu- P388. Acta Haematol Pol 1993; 24: 177–182. kemia 1994; 8: 1290–1293. 30 Begleiter A, Wang H, Verburg L, Lee K, Israels LG, Mowat MRA, 35 Elias L, Stock-Novack D, Head D, Grever M, Weick JK, Chapman Johnston JB. In vitro cytotoxicity of 2-chlorodeoxyadenosine and RA, Godwin JE, Metz EN, Appelbaum FR. A phase I trial of combi- chlorambucil in chronic lymphocytic leukemia. Leukemia 1996; nation fludarabine monophosphate and chlorambucil in chronic 10: 1959–1965. lymphocytic leukemia: a Southwest oncology group study. Leuke- 31 Begleiter A, Pugh L, Israels LG, Johnston JB. Enhanced cytotoxicity mia 1993; 7: 361–365.