- Email: [email protected]
Drug combination testing in acute lymphoblastic leukemia using the MTT assay
Leukemia Research Vol. 19, No. 3, pp. 175-181, 1995. Cowright 0 1995 Elsevier Science L.td Printed in &eat Britain. All rights reserved 0145-2126195 $9.50 + 0.W
Pergamon 0145-2126(94)00126-X
DRUG COMBINATION TESTING IN ACUTE LYMPHOBLASTIC LEUKEMIA USING THE MT-T ASSAY Gertjan J.L. Kaspers,* Anjo J.P. Veerman,*? Rob Pieters,* Ina Van Zantwijk,* and Elisabeth R. Van Wering?
Karel HBhlenQ
*Department of Pediatrics, Free University Hospital, De Boelelaan 1117, 1081 HV, Amsterdam; TDutch Childhood Leukemia Study Group, Dr. Van Welylaan 2, 2506 LP, The Hague; and ZSophia Children’s Hospital, Subdivision of Hemato-Oncology, On Molewaterplein 60, 3015 GJ Rotterdam, The Netherlands Abstract-Drug resistance assays may be useful to identify drug interactions. For this purpose, we studied three drug combinations, each at 8-12 concentrations, with the Ml7 assay in acute lymphoblastic leukemia (ALL) samples from 34 children obtained at initial diagnosis. This resulted in a total of 518 comparisons between expected and observed leukemic cell survivals. The combinations prednisolone (PRD) with vincristine (VCR), PRD with mafosfamide (MAF), and PRD with daunorubicin (DNR) were tested without technical difficulties, and without an increased assay variation as compared to single drugs. We observed a marked heterogeneity between combinations, and between different in drug interactions between patients, concentrations within one specific combination. Between PRD+VCR, synergism was found in 46%, antagonism in 18%, and additivity in 36% of the 228 observations. Between PRD+MAF, synergism was found in 51%, antagonism in 20%, and additivity in 29% of the 140 observations. Between PRD+DNR, synergism was found in 35%, antagonism in 31%, and additivity in 34% of the 150 observations. PRD+VCR and PRD+MAF showed more often synergism than PRD+DNR, while antagonism was observed more frequently between PRD+DNR (p
resistance,
drug
interactions,
In hematopoietic malignancies, a large number of different anticancer drugs are being used. With the in the treatment
assay,
leukemia,
synergism,
which are already being administered in combination clinically, or to identify interactions between drugs which might be candidates for combined administration. In vitro drug resistance assays can also be used to predict the clinical response to chemotherapy in leukemias. A large number of groups have reported good in vitro-in vivo correlations in childhood and adult leukemia, as reviewed by Veerman and Pieters [2] and as described in later reports. [3-81 However, in most studies single drugs were tested in vitro, while patients were treated with drug combinations. If relevant drug interactions do occur clinically, in vitro-in vivo correla-
Introduction
increased use of drug combinations
Ml7
of
leukemia, favorable and unfavorable drug interactions are more likely to occur. [l] However, interactions are difficult to identify clinically, which indicates the need for a reliable in vitro system. In vitro drug resistance assays can be used to study interactions between drugs Abbreviations: MTT, methyl-thiazol-tetrazolium; PRD, prednisolone; VCR, vincristine; DNR, daunorubicin; A4AF, mafosfamide; BM, bone marrow; PB, peripheral blood; ALL, acute lymphoblastic leukemia; TdT, terminal deoxynucleotidyltransferase; OD, optical density; KS, leukemic cell survival; SD, standard deviation. Correspondence too:Gertjan J. L. Raspers (Tel: +31 20 444 2420; Fax: +31 20 444 2422).
tions may improve if the clinically used drug combinations are tested in vitro. The aims of the present study were to determine the feasibility of testing drug combinations in vitro in acute
leukemia samples, and to investigate whether synergistic, antagonistic, or additive drug interactions could be 175
G. J. L. Kaspers et al.
176
detected between prednisolone (PRD) combined with either vincristine (VCR), mafosfamide (MAF, an active metabolite of cyclophosphamide), or daunorubicin (DNR). In vitro drug resistance was assessed using the methyl-thiazol-tetrazolium (MIT) assay. The combinations of PRD+VCR and PRD+DNR were chosen because of their clinical importance in the treatment of acute lymphoblastic leukemia (ALL), while PRD+MAF was studied becatise theoretically alkylating agents such as MAP might damage the glucocorticoid receptor. Such damage might decrease the receptor-mediated glucocorticoid-induced cell kill, resulting in antagonism.
blank the spectrophotometer. The drugs, either alone or in combination, did not influence the background color of the medium. Drug combinations and concentrations tested are shown in Table 1. Single drugs were tested in triplicate, drug combinations in duplicate. After 4 days of culture, MTI was added to all wells, and after shaking the plates until the cell pellet was resuspended, they were incubated for 6 h. The colored formazan crystals formed were dissolved with acidified isopropanol. The optical density (OD) of each well, which is linearily related to the viable cell number, [12] was measured with an EL-312 microplate spectrophotometer (Biotek Instruments Inc., Winooski, U.S.A.) at 562 nm. Leukemic cell survival (LCS) was calculated by the equation: LCS = (OD treated well/mean OD control wells) x 100%. The higher the cytotoxicity of a drug or drug combination, the lower the LCS.
Materials and Methods Patient samples Bone marrow (BM) or peripheral blood (PB) samples of 34 children with newly diagnosed ALL were successfully evaluated. The samples had been sent to the research laboratory for pediatric hemato-onto-immunology of the Free University Hospital in Amsterdam for cellular drug resistance testing, by the laboratory of the Dutch Childhood Leukemia Study Group in The Hague, the Sophia Children’s Hospital in Rotterdam, and by members of the German COALL-StudyGroup (Head: Professor Dr G. Janka-Schaub; Study center: Hamburg). Mononuclear cells were isolated by Ficoll density gradient centrifugation (Ficoll Paque; density 1.077 g/ml; Pharmacia, Sweden) for 15 min (room temperature, 1000 g). Part of the cell samples was tested after cryopreservation. ALL samples were further characterized by immunophenotyping as described previously. [9] Pro-B ALL was diagnosed in case of positivity for terminal deoxynucleotidyl transferase (TdT), HLA-DR and CD19 (n = 3), common-ALL (c-ALL) in case of positivity for TdT, HLA-DR, CD19, and CD10 (n = 15) pre-B ALL in case of positivity for TdT, HLA-DR, CD19, and cytoplasmic H chain (n = 8) and T-ALL in case of positivity for TdT, cytoplasmic CD3, and CD7 (n = 8).
Definitions of drug interaction On the assumption that each drug acts independently, the ‘multiplicative’ model according to Valeriote and Lin [13] predicts that the effect of a drug combination is the product of the effect of each single drug: expected LCS (A+B) = LCS (A) x LCS (B). The ‘maximum’ model according to Sondak et al. [14] predicts that the effect of a combination is similar to that of the most active single drug (Dmax): expected LCS (A+B) = LCS (A), if LCS (A) < LCS (B). These models were used to calculate the borderlines of synergistic, additive, and antagonistic drug interactions. The following definitions of interactions were used: -
observed LCS (A+B) < LCS (A) x LCS (B). -
Materials and drugs PRD, DNR, and acidified (0.04 N HCl) isopropanol were obtained from the hospital pharmacy. MAF (ASTA 27654) which is an active metabolite of cyclophosphamide, was initially a gift and later purchased from ASTA Medica B.O. (Diemen, NL). PRD was dissolved in saline, DNR in distilled water, and the lysine salt formulation of MAF was also dissolved in distilled water. RPM1 1640 (Dutch modification) was obtained from Gibco (Uxbridge, U.K.), fetal calf serum from Flow Laboratories (Irvine, U.K.), and MIT (3-[4,5dimethyl-thiazol-2-yl]-2,5-diphenyltetraolium-bromide) from Sigma Chemical Co. (St. Louis, MO). MTT assay The M’IT assay was essentially performed as described previously. [lo] Since BM and PB samples do not differ in drug resistance, these samples were evaluated together. [ll] Each sample contained 280% leukemic cells at the start of culture, and 2 70% leukemic cells in drug-free cultures after 4 days, which is required for reliable results. Leukemic cells were cultured in suspension (RPM1 1640 with fetal calf serum and other supplements) with or without drugs for 4 days in the wells of 96-well microculture plates, at 37°C in humidified air with 5% COa. Six wells contained culture medium only to
Synergism: the observed LCS for a drug combination is lower than the product of the effect of each single drug:
Additivity: the observed LCS for a drug combination is lower than the LCS for D,,,, but is higher than the product of the effect of each single drug: LCS (A) x LCS (B) < observed LCS (A+B) < LCS (Dmax).
Table 1. The drug combinations tested Drugs
Concentrations (in pgjml)
Prednisolone (PRD) VCR 0.16 + PRD 0.2 and 2 and VCR 0.63 + PRD 0.2, 2, 20 and 200 vincristine (VCR) VCR 2.5 + PRD 0.2, 2, 20 and 200 VCR 10 + PRD 0.2 and 2 Prednisolone MAF 1.7 + PRD 0.2, 2, ,20 and 200 and MAF 6.8 + PRD 0.2, 2, 20 and 200 mafosfamide WW Prednisolone DNR 0.008+ PRD 0.2 and 2 and DNR 0.03 + PRD 0.2 and 2 daunorubicin DNR 0.13 t PRD 0.2, 2, 20 and 200 @NW DNR 0.5 t PRD 0.2 and 2
Drug combination testing in leukemia using the Mm assay
-
177
considered to be clinically called antagonism.
Antagonism: the observed LCS for a drug combination is higher than the LCS for the most active drug, Dmax:
an unfavorable interaction, and
Statistics The Wilcoxon’s matched pair test and Spearman’s rank correlation test (parameter, Rho) were used at a level of significance of p = 0.05.
observed LCS (A+B) > LCS (Dmax). It was thus necessary to use two models in order to classify each observation for a drug combination as being either a synergistic, additive, or an antagonistic interaction. It is further reasoned that by combining two drugs, any cytotoxicity added to that of the most cytotoxic single agent is clinically a favorable interaction, and therefore can be defined as at least additive. Similarly, if the cytotoxicity of a combination is lower than that of the most cytotoxic single agent, this is
Results There were no technical problems associated with the combination of these drugs in our assay system. For the combinations, the coefficient of variation (standard
Table 2. In vitro synergistic (Sy), additive (Ad), and antagonistic (An) drug interactions in 34 acute lymphoblastic leukemia (ALL) samples. ’
Patient 1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Total Percentage of observations
Leukemia type pro-B ALL pro-B ALL pro-B ALL C-ALL c-ALL C-ALL c-ALL C-ALL C-ALL C-ALL C-ALL C-ALL C-ALL C-ALL C-ALL c-ALL C-ALL C-ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL T-ALL T-ALL T-ALL T-ALL T-ALL T-ALL T-ALL T-ALL
SY 4 3 7 0 5 5 4 6 1 9 10 4 9
PRD+VCR Ad
Drug combinations and interactions PRD+hJAF An
2 1 1 5 3 4 1 4 7 0 0 0 1
105
4 6 2 5 2 1 5 0 2 1 0 6 0 n.t. n.t. n.t. n.t. n.t. 4 8 10 4 7 6 n.t. n.t. 2 0 2 5 n.t. n.t. n.t. n.t. 82
41
46%
36%
18%
5 10 5 2
-
SY
-
Ad
71
2 5 6 1 0 0 4 0 3 n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. 1 2 0 0 2 n.t. 8 n.t. 2 5 0 n.t. n.t. n.t. n.t. n.t. 41
51%
29%
3 3 1 7 4 0 4 8 3
An
PRD+DNR SY 10 3
Ad
An
0
0
n.t. n.t. 3 n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. 2 3 2 2 4 3 3
4
n.t. n.t. n.t. n.t. n.t. 1 1
28
1 8 0 2 53
4 7 n.t. n.t. n.t. 7 1 4 6 51
2 1 6 2 46
20%
35%
34%
31%
5 2
Indicated are the number of combinations in which each specific interaction was detected. For PRD+VCR up to 10, for PRD+MAF up to 8, and for PRD+DNR 10 different concentrations were tested. Synergism: lethal effect of the combination greater than the product of the effects of each single drugs; Antagonism: lethal effect of the combination less than that of the most single active drug; Additivity: all other observations; n.t.=not tested.
178
G. J. L. Kaspers et
A
Prednisolone + vincristine Multiplicative model 100,
values according to the multiplicative model are given are useful to directly identify synergistic interactions,
a,
i.e. if the observed LCS is lower than the expected LCS.
Additive-antagonism 00
2 a g
Similarly, the figures in which the expected LCS values according to the maximum model are given, enable the identification of antagonistic interactions, i.e. if the observed LCS is higher than the expected LCS.
0
60 6too
0 0
a F2
Synergism
20
8 20
40
60
/
I
80
100
Expected LCS (%)
B
al.
Prednisolone+vincristine (Table 2, Fig. 1) This drug combination was tested in 23 samples at up to 10 different concentrations, which resulted in 228 comparisons between the expected and observed LCS values.
For all but one of these concentrations,
the
observed LCS values were significantly (p < 0.03) lower than the LCS values for the most active single agents. In
Prednisolone + vincristine Maximum model
other words, the drug combinations
100,
??
Antagonism
were significantly
more cytotoxic than the most active single drug. Drug interactions
in the individual
patients
are indicated
in
Table 2. A synergistic interaction was observed in 46%
80 ~
60~
A
0
Prednisolone t mafosfamide Multiplicative model
40 ~
Additive-antagonism
x
AdditiveI 0
I
20
I 80
100
Exifcled LCy(%) Fig. 1. Comparisons (n = 228) between observed leukemic cell survival (LCS) and expected LCS values for the combination prednisolone+vincristine, tested in 23 acute lymphoblastic leukemia samples. (A) Expected LCS according to the multiplicative model: LCS (A+B) = LCS (A) x LCS (B); (B) expected LCS according to the maximum model: LCS (A+B) = D,,,. The interrupted line is x = y.
Synergism
cases of lower expected LCS values. Figures l-3 show the comparisons between the expected LCS values as predicted by the multiplicative model and the maximum model and the observed LCS values for the three drug
combinations. The figures in which the expected LCS
1
0
Exizcted LCF(%) B
deviation divided by the mean) of the LCS values was mean 7.7% (S.D. 7.4%), while for the single drugs this mean was 6.5% (S.D. 4.8%), a non-significant difference. In general, a marked heterogeneity in drug interactions was observed between patient samples, between drug combinations, and occasionally between different concentrations of the same drug combinations (Table 2). The type of drug interaction was not clearly related to the immunophenotype. In case of higher expected LCS values, synergism was slightly more likely to occur, while antagonism tended to occur more frequently in
/ 80
/ 20
0
Prednisolone t mafosfamide Maximum model 100,
2 2 +
"
Antagonism 8o
,
_
2 p: 60g31 ~ 0 40 I2 -0 2s
20-
"0
/ 0
/' 20
I 40
/ 60
I 80
100
Expected LCS (%) Fig. 2. Comparisons (n = 140) between observed leukemic cell survival (LCS) and expected LCS values for the combination prednisolonetmafosfamide, tested in 18 acute leukemia samples. (A) Expected LCS according to the multiplicative model; (B) expected LCS according to the maximum model.
Drug combination testing in leukemia using the MIT
A
drug combinations were significantly (p < 0.02) more cytotoxic than the most active single agent. These cytotoxicities did not differ for the combination of 2 l&ml PRD+MAF 1.7 &ml. Synergism was observed in 51% of observations, with a maximum absolute difference in LCS of 30%. Antagonism was observed in 20% of observations, with a maximum absolute difference in LCS of 22%. An additive interaction was found in the remaining 29% of observations.
Prednisolone + daunorubicin Multiplicative model 100
Additive-antagonism
Synergism
20
0
40
60
80
100
Expected LCS (%)
B
Prednisolone + daunorubicin Maximum model 100
Antagonism 52 e
O
,piD 0
SOQBJ
’ &
2
_
60-
g 3 B >
2
JO-
Additivesynergism
20-
6
,8%
/ 0
I
I
I
I
20
40
60
80
100
Expected LCS (%) Fig. 3. Comparisons
179
assay
(n = 150) between observed leukemic
cell
survival (LCS) and expected LCS values for the combination prednisolone+daunorubicin, tested in 15 acute leukemia samples. (A) Expected LCS according to the multiplicative model; (B) expected LCS according to the maximum model. of observations, with a maximum absolute difference in LCS between the product of the effect of each single drug and the observed LCS of 47%. Similarly, antagonism was seen in 18% of the observations, with a maximum absolute difference in LCS between the effect of the most active single drug and the observed LCS of 33%. Additivity was found for the remaining 36% of observations. Considering the whole group of samples, the concentrations of 0.16 pg/ml VCR+PRD 2 &ml (p = 0.02) and 0.16 pg/ml VCR+PRD 0.2 @ml (p = 0.009) were significantly synergistic. So, synergism was especially found with the lowest concentration of VCR tested. Prednisolone+mafosfamide (Table 2, Fig. 2) This drug combination was tested in 18 samples at up to 8 different concentrations, resulting in 140 comparisons between the expected and observed LCS values (Table 2). For 6 out of 8 of these concentrations, the
Prednisolone+daunorubicin (Table 2, Fig. 3) This combination was successfully tested at 10 concentrations in 15 samples, which resulted in 150 comparisons between the expected and observed LCS values. In contrast to the findings for PRD+VCR and PRDtMAF, the cytotoxicity of 7 out of the 10 different concentrations tested with PRDtDNR were not significantly different from that of the most active single drug. Synergism was seen in 35% of observations, with a maximum absolute difference in LCS of 20%. An antagonistic interaction was found in 31% of the observations, with a maximum absolute difference in LCS of 28%. Considering the whole group of samples, the concentration of 2 pg/ml PRDtDNR 0.008 &ml was significantly (p = 0.03) synergistic. Overall, there were no significantly antagonistic interactions in any of the drug combinations tested, if all samples are considered together. Synergism was observed between PRDtVCR at two different concentrations, and between PRDtDNR at one concentration. The frequency of synergistic, antagonistic, and additive interactions did not differ significantly between the combinations PRDtVCR and PRDtMAF. However, the latter combinations showed synergism more often than the combination PRDtDNR, while PRDtDNR was significantly more often antagonistic (p < 0.05). Discussion We studied interactions between PRD combined with either VCR, MAF or DNR in ALL samples, obtained from children at initial diagnosis, using the MTT assay. Testing drugs in combination gave no technical difficulties. The assay variation for the drug combinations was in the same range, mean 7-8%, as for single drugs. Drug interactions varied considerably between patients, between drug combinations, and even between different concentrations of the same drug combination. Synergistic interactions were more frequent between PRD+MAF and PRD+VCR, detected in about half of the observations, than between PRD+DNR, for which combination synergism was found in 35% of the observations. Antagonism, on the other hand, was seen more frequently between PRDtDNR than between the other two combinations. These differences were sig-
180
G. J. L. Kaspers et
nificantly different. However, the magnitude of antagonism did not differ much between the three drug combinations, nor was there a significant antagonistic interaction in any of the drug combinations tested, if all samples were considered together. We could not confirm our hypothesis that MAP might decrease the cytotoxicity of PRD by alkylating the glucocorticoid receptor, resulting in antagonism between these two drugs. In about one-third of the observations, an additive interaction was found. Additivity may be favorable clinically, although the extent of additive interaction is important to consider. We did not distinguish minor and major interactions in the present study. The frequent additive and synergistic interactions, observed in 70-80% of the observations for all 3 drug combinations, may help explain the known clinical effectiveness of these drug combinations in ALL. [15] The cause of the observed interactions is unknown. It is tempting to speculate that the drugs interact in the cascade of events which characterize drug-induced cell death by apoptosis. The ‘multiplicative model’, which predicts that the effect of a combination is the product of the effect of each single drug, was a better predictor of the observed in vitro cytotoxicity of the combinations of PRD+VCR and PRD+MAF than the ‘maximum’ model, which predicts that the effect of a combination is equal to the effect of the most active single drug. This is in agreement with the findings of Larsson et al. using the fluorometric microculture cytotoxicity assay in ANLL samples [16]. However, it is in contrast to the observations of Sondak et al. in solid tumor samples, who reported that the cytotoxicity of different drug combinations rarely exceeded the cytotoxicity of the most effective single drug. [14] Moreover, the latter authors did not detect synergistic drug interactions in any of 199 solid human tumor samples, while in human leukemia samples this type of interaction is relatively frequent as reported here and by others previously. [16, 171. The heterogeneity in drug interactions observed suggests that for an optimal prediction of the clinical response to combination chemotherapy, drugs should be tested in vitro in the same combinations and if possible in comparable concentrations. Theoretically, if only information about single drugs is used, antagonistic interactions would result in false sensitive predictions, while synergistic interactions result in false resistant predictions. However, the ability to predict clinical drug resistance did not show a major improvement using the in vitro results obtained with drug combinations as compared to single drugs in two studies [6, 181. We could not address this question in the present study, because of a heterogeneous treatment and a short follow-up. Aapro discussed several problems associated with
al.
drug combination testing in clonal assays, which also apply to short-term non-clonogenic assays such as the MTT assay [19]. Drug doses and schedules may not represent the clinical situation. We did not attempt to investigate different time schedules. In addition, chemical interactions between the drugs may occur in vitro, and substances present in the culture medium may influence the drug interaction. Despite these cautionary notes, testing drug combinations in vitro may still be a valuable tool to identify drug interactions. Promising new drug combinations might be selected for clinical application based on the interactions observed in vitro. It might even be possible to use the in vitro results of drug combinations tested in resistance assays to tailor currently used chemotherapeutic drug combinations in individual patients. In conclusion, the MIT assay can be used to study drug interactions in vitro in ALL samples. The combinations PRDtVCR and PRD+MAF generally showed additive and even synergistic interactions. The cytotoxicity of PRDtDNR was generally not markedly higher than that of the most active single drug. Acknowledgement-This study was financially supported by the Dutch Cancer Society (IKA 89-06) and by the project VONK (VU Onderzoek Naar Kinderkanker). Computer equipment was provided by Olivetti Nederland BV. The Dutch Childhood Leukemia Study Group (DCLSG) provided most of the patient samples. Board members of the DCLSG are H. Van Den Berg, M. V. A. Bruin, J. P. M. Bokkerink, P. J. Van Dijken, K. Hahlen, W. A. Kamps, F. A. E. Nabben, A. Postma, J. A. Rammeloo, I. M. Risseeuw-Appel, A. Y. N. SchoutenVan Meeteren, G. A. M. De Vaan, E. Th. Van ‘t Veerkorthof, A. J. P. Veerman, M. Van Weel-Sipman and R. S. Weening.
References 1. Pinkel D. (1987) Curing children of leukemia. Cancer 59, 1683. 2. Veerman A. J. P. & Pieters R. (1990) Drug sensitivity assays in leukaemia and lymphoma. Br. J. Huemat. 74, 381. 3. Bosanquet A. G. (1991) Correlations between therapeutic response of leukaemias and in-vitro drug-sensitivity assay. Lancet 337, 711. 4. Hongo T. & Fujii Y. (1991) In vitro chemosensitivity of lymphoblasts at relapse in childhood leukemia using the M’IT assay. ht. J. Hemat. 54, 219. 5. Pieters R., Huismans D. R., Loonen A. H., Hahlen K., Van Der Does-Van Den Berg A., Van Wering E. R. & Veerman A.J.P (1991) Relation of cellular drug resistance to longterm clinical outcome in childhood acute lymphoblastic leukaemia. Lancet 338, 399. 6. Larsson R., Kristensen J., Sandberg C. & Nygren P. (1992) Laboratory determination of chemotherapeutic drug resistance in tumor cells from patients with leukemia, using a fluorimetric microculture cytotoxicity assay (FMCA). Int.
J. Cancer 50, 177. 7. Kaspers G. J. L., Pieters R., Van Zantwijk C. H., De Waal F. C., Van Wering E. R. & Veerman A. J. P. (1993) Is resistance to prednisolone in vitro related to the response to
Drug combination testing in leukemia using the MTI assay prednisolone at initial diagnosis in childhood acute lymphoblastic leukemia? A preliminary analysis. In Drug Resistance in Leukemia and Lymphoma. The Clinical Value ofLaboratory Assays (Kaspers G. J. L., Pieters R., Twentyman P. R., Weisenthal L. M. & Veerman A. J. P., Eds), pp. 321-328. Chur, Harwood Academic Publ. 8. Kaspers G. J. L., Pieters R., Van Zantwijk C. H., Van Wering E. R., Van Der Does-Van Den Berg A. & Veerman A. J. P. (1993) Resistance to prednisolone (PRD) in vitro: a new prognostic factor in childhood acute lymphoblastic leukemia (ALL) at initial diagnosis. Proceedings of the American Society of Clinical Oncology 12, 320 (abstract). 9. Van Wering E. R., Brederoo P., Van Staalduinen G. J., Van Der Meulen J., Van Der Linden-Schrever B. E. M. & Van Dongen J. J. M. (1993) Contribution of electron microscopy to the classification of minimally differentiated acute leukemias in children. Recent Results Cancer Res. 131,77. 10. Pieters R., Loonen A. H., Huismans D. R., Broekema G. J., Dirven M. W. J., Heyenbrok M. W., Hahlen K. & Veerman A. J. P. (1990) In vitro sensitivity of cells from children with leukemia using the MTT assay with improved culture conditions. Blood 76, 2327. 11. Kaspers G. J. L., Pieters R., Van Zantwijk C. H., De Laat P. A. J. M., De Waal F. C., Van Wering E. R. & Veerman A. J. P. (1991) In vitro drug sensitivity of normal peripheral blood lymphocytes and childhood leukaemic cells from bone marrow and peripheral blood. Br. J. Cancer 64, 469. 12. Pieters R., Huismans D. R., Lcyva A. & Veerman A. J. P.
181
(1988) Adaptation of the rapid automated tetrazolium dye based M’IT assay for chemosensitivity testing in childhood leukemia. Cancer Lett. 41, 323. 13. Valeriote F. & Lin H.-S. (1975) Synergistic interaction of anticancer agents: a cellular perspective. Cancer Chemother. Rep. 59, 89.5. 14. Sondak 0. K., Korn E. L. & Kern D. H. (1988) In vitro testing of chemotherapeutic combinations in a rapid thymidine incorporation assay. Znt. J. Cell Cloning 6, 378. 15. Henderson E. S. (1969) Treatment of acute leukemia. Semin Hemat. 6, 271. 16. Larsson R., Fridborg H., Kristensen J., Sundstriim C. & Nygren P. (1993) In vitro testing of chemotherapeutic drug combinations in acute myelocytic leukaemia using the fluorimetric microculture cytotoxicity assay (FMCA). Br. J. Cancer 67, 969. 17. Sargent J., Taylor C. & Elgie A (1993) Study of drug combinations in acute myeloid leukemia (AML) using the MIT assay. In Drug Resistance in Leukemia and Lymphoma. The Clinical Value of Laboratory Assays (Kaspers G. J. L., Pieters R., Twentyman P. R., Weisenthal L. M. and Veerman A. J. P., Eds), pp. 293-297. Chur, Harwood Academic Publ. 18. Viano I., Silvestro L., Infelise O., Falda M., Lovisone E., Compagnoni G. & Genazzani E. (1986) A short-term chemosensitivity test with different labeled precursors of DNA or protein synthesis: correlation with clinical response. Tumori 72, 357. 19. Aapro M. S. (1984) Drug combination testing with in vitro clonal cultures. Recent Results Cancer Res. 94, 224.
Pergamon 0145-2126(94)00126-X
DRUG COMBINATION TESTING IN ACUTE LYMPHOBLASTIC LEUKEMIA USING THE MT-T ASSAY Gertjan J.L. Kaspers,* Anjo J.P. Veerman,*? Rob Pieters,* Ina Van Zantwijk,* and Elisabeth R. Van Wering?
Karel HBhlenQ
*Department of Pediatrics, Free University Hospital, De Boelelaan 1117, 1081 HV, Amsterdam; TDutch Childhood Leukemia Study Group, Dr. Van Welylaan 2, 2506 LP, The Hague; and ZSophia Children’s Hospital, Subdivision of Hemato-Oncology, On Molewaterplein 60, 3015 GJ Rotterdam, The Netherlands Abstract-Drug resistance assays may be useful to identify drug interactions. For this purpose, we studied three drug combinations, each at 8-12 concentrations, with the Ml7 assay in acute lymphoblastic leukemia (ALL) samples from 34 children obtained at initial diagnosis. This resulted in a total of 518 comparisons between expected and observed leukemic cell survivals. The combinations prednisolone (PRD) with vincristine (VCR), PRD with mafosfamide (MAF), and PRD with daunorubicin (DNR) were tested without technical difficulties, and without an increased assay variation as compared to single drugs. We observed a marked heterogeneity between combinations, and between different in drug interactions between patients, concentrations within one specific combination. Between PRD+VCR, synergism was found in 46%, antagonism in 18%, and additivity in 36% of the 228 observations. Between PRD+MAF, synergism was found in 51%, antagonism in 20%, and additivity in 29% of the 140 observations. Between PRD+DNR, synergism was found in 35%, antagonism in 31%, and additivity in 34% of the 150 observations. PRD+VCR and PRD+MAF showed more often synergism than PRD+DNR, while antagonism was observed more frequently between PRD+DNR (p
resistance,
drug
interactions,
In hematopoietic malignancies, a large number of different anticancer drugs are being used. With the in the treatment
assay,
leukemia,
synergism,
which are already being administered in combination clinically, or to identify interactions between drugs which might be candidates for combined administration. In vitro drug resistance assays can also be used to predict the clinical response to chemotherapy in leukemias. A large number of groups have reported good in vitro-in vivo correlations in childhood and adult leukemia, as reviewed by Veerman and Pieters [2] and as described in later reports. [3-81 However, in most studies single drugs were tested in vitro, while patients were treated with drug combinations. If relevant drug interactions do occur clinically, in vitro-in vivo correla-
Introduction
increased use of drug combinations
Ml7
of
leukemia, favorable and unfavorable drug interactions are more likely to occur. [l] However, interactions are difficult to identify clinically, which indicates the need for a reliable in vitro system. In vitro drug resistance assays can be used to study interactions between drugs Abbreviations: MTT, methyl-thiazol-tetrazolium; PRD, prednisolone; VCR, vincristine; DNR, daunorubicin; A4AF, mafosfamide; BM, bone marrow; PB, peripheral blood; ALL, acute lymphoblastic leukemia; TdT, terminal deoxynucleotidyltransferase; OD, optical density; KS, leukemic cell survival; SD, standard deviation. Correspondence too:Gertjan J. L. Raspers (Tel: +31 20 444 2420; Fax: +31 20 444 2422).
tions may improve if the clinically used drug combinations are tested in vitro. The aims of the present study were to determine the feasibility of testing drug combinations in vitro in acute
leukemia samples, and to investigate whether synergistic, antagonistic, or additive drug interactions could be 175
G. J. L. Kaspers et al.
176
detected between prednisolone (PRD) combined with either vincristine (VCR), mafosfamide (MAF, an active metabolite of cyclophosphamide), or daunorubicin (DNR). In vitro drug resistance was assessed using the methyl-thiazol-tetrazolium (MIT) assay. The combinations of PRD+VCR and PRD+DNR were chosen because of their clinical importance in the treatment of acute lymphoblastic leukemia (ALL), while PRD+MAF was studied becatise theoretically alkylating agents such as MAP might damage the glucocorticoid receptor. Such damage might decrease the receptor-mediated glucocorticoid-induced cell kill, resulting in antagonism.
blank the spectrophotometer. The drugs, either alone or in combination, did not influence the background color of the medium. Drug combinations and concentrations tested are shown in Table 1. Single drugs were tested in triplicate, drug combinations in duplicate. After 4 days of culture, MTI was added to all wells, and after shaking the plates until the cell pellet was resuspended, they were incubated for 6 h. The colored formazan crystals formed were dissolved with acidified isopropanol. The optical density (OD) of each well, which is linearily related to the viable cell number, [12] was measured with an EL-312 microplate spectrophotometer (Biotek Instruments Inc., Winooski, U.S.A.) at 562 nm. Leukemic cell survival (LCS) was calculated by the equation: LCS = (OD treated well/mean OD control wells) x 100%. The higher the cytotoxicity of a drug or drug combination, the lower the LCS.
Materials and Methods Patient samples Bone marrow (BM) or peripheral blood (PB) samples of 34 children with newly diagnosed ALL were successfully evaluated. The samples had been sent to the research laboratory for pediatric hemato-onto-immunology of the Free University Hospital in Amsterdam for cellular drug resistance testing, by the laboratory of the Dutch Childhood Leukemia Study Group in The Hague, the Sophia Children’s Hospital in Rotterdam, and by members of the German COALL-StudyGroup (Head: Professor Dr G. Janka-Schaub; Study center: Hamburg). Mononuclear cells were isolated by Ficoll density gradient centrifugation (Ficoll Paque; density 1.077 g/ml; Pharmacia, Sweden) for 15 min (room temperature, 1000 g). Part of the cell samples was tested after cryopreservation. ALL samples were further characterized by immunophenotyping as described previously. [9] Pro-B ALL was diagnosed in case of positivity for terminal deoxynucleotidyl transferase (TdT), HLA-DR and CD19 (n = 3), common-ALL (c-ALL) in case of positivity for TdT, HLA-DR, CD19, and CD10 (n = 15) pre-B ALL in case of positivity for TdT, HLA-DR, CD19, and cytoplasmic H chain (n = 8) and T-ALL in case of positivity for TdT, cytoplasmic CD3, and CD7 (n = 8).
Definitions of drug interaction On the assumption that each drug acts independently, the ‘multiplicative’ model according to Valeriote and Lin [13] predicts that the effect of a drug combination is the product of the effect of each single drug: expected LCS (A+B) = LCS (A) x LCS (B). The ‘maximum’ model according to Sondak et al. [14] predicts that the effect of a combination is similar to that of the most active single drug (Dmax): expected LCS (A+B) = LCS (A), if LCS (A) < LCS (B). These models were used to calculate the borderlines of synergistic, additive, and antagonistic drug interactions. The following definitions of interactions were used: -
observed LCS (A+B) < LCS (A) x LCS (B). -
Materials and drugs PRD, DNR, and acidified (0.04 N HCl) isopropanol were obtained from the hospital pharmacy. MAF (ASTA 27654) which is an active metabolite of cyclophosphamide, was initially a gift and later purchased from ASTA Medica B.O. (Diemen, NL). PRD was dissolved in saline, DNR in distilled water, and the lysine salt formulation of MAF was also dissolved in distilled water. RPM1 1640 (Dutch modification) was obtained from Gibco (Uxbridge, U.K.), fetal calf serum from Flow Laboratories (Irvine, U.K.), and MIT (3-[4,5dimethyl-thiazol-2-yl]-2,5-diphenyltetraolium-bromide) from Sigma Chemical Co. (St. Louis, MO). MTT assay The M’IT assay was essentially performed as described previously. [lo] Since BM and PB samples do not differ in drug resistance, these samples were evaluated together. [ll] Each sample contained 280% leukemic cells at the start of culture, and 2 70% leukemic cells in drug-free cultures after 4 days, which is required for reliable results. Leukemic cells were cultured in suspension (RPM1 1640 with fetal calf serum and other supplements) with or without drugs for 4 days in the wells of 96-well microculture plates, at 37°C in humidified air with 5% COa. Six wells contained culture medium only to
Synergism: the observed LCS for a drug combination is lower than the product of the effect of each single drug:
Additivity: the observed LCS for a drug combination is lower than the LCS for D,,,, but is higher than the product of the effect of each single drug: LCS (A) x LCS (B) < observed LCS (A+B) < LCS (Dmax).
Table 1. The drug combinations tested Drugs
Concentrations (in pgjml)
Prednisolone (PRD) VCR 0.16 + PRD 0.2 and 2 and VCR 0.63 + PRD 0.2, 2, 20 and 200 vincristine (VCR) VCR 2.5 + PRD 0.2, 2, 20 and 200 VCR 10 + PRD 0.2 and 2 Prednisolone MAF 1.7 + PRD 0.2, 2, ,20 and 200 and MAF 6.8 + PRD 0.2, 2, 20 and 200 mafosfamide WW Prednisolone DNR 0.008+ PRD 0.2 and 2 and DNR 0.03 + PRD 0.2 and 2 daunorubicin DNR 0.13 t PRD 0.2, 2, 20 and 200 @NW DNR 0.5 t PRD 0.2 and 2
Drug combination testing in leukemia using the Mm assay
-
177
considered to be clinically called antagonism.
Antagonism: the observed LCS for a drug combination is higher than the LCS for the most active drug, Dmax:
an unfavorable interaction, and
Statistics The Wilcoxon’s matched pair test and Spearman’s rank correlation test (parameter, Rho) were used at a level of significance of p = 0.05.
observed LCS (A+B) > LCS (Dmax). It was thus necessary to use two models in order to classify each observation for a drug combination as being either a synergistic, additive, or an antagonistic interaction. It is further reasoned that by combining two drugs, any cytotoxicity added to that of the most cytotoxic single agent is clinically a favorable interaction, and therefore can be defined as at least additive. Similarly, if the cytotoxicity of a combination is lower than that of the most cytotoxic single agent, this is
Results There were no technical problems associated with the combination of these drugs in our assay system. For the combinations, the coefficient of variation (standard
Table 2. In vitro synergistic (Sy), additive (Ad), and antagonistic (An) drug interactions in 34 acute lymphoblastic leukemia (ALL) samples. ’
Patient 1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Total Percentage of observations
Leukemia type pro-B ALL pro-B ALL pro-B ALL C-ALL c-ALL C-ALL c-ALL C-ALL C-ALL C-ALL C-ALL C-ALL C-ALL C-ALL C-ALL c-ALL C-ALL C-ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL pre-B ALL T-ALL T-ALL T-ALL T-ALL T-ALL T-ALL T-ALL T-ALL
SY 4 3 7 0 5 5 4 6 1 9 10 4 9
PRD+VCR Ad
Drug combinations and interactions PRD+hJAF An
2 1 1 5 3 4 1 4 7 0 0 0 1
105
4 6 2 5 2 1 5 0 2 1 0 6 0 n.t. n.t. n.t. n.t. n.t. 4 8 10 4 7 6 n.t. n.t. 2 0 2 5 n.t. n.t. n.t. n.t. 82
41
46%
36%
18%
5 10 5 2
-
SY
-
Ad
71
2 5 6 1 0 0 4 0 3 n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. 1 2 0 0 2 n.t. 8 n.t. 2 5 0 n.t. n.t. n.t. n.t. n.t. 41
51%
29%
3 3 1 7 4 0 4 8 3
An
PRD+DNR SY 10 3
Ad
An
0
0
n.t. n.t. 3 n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. n.t. 2 3 2 2 4 3 3
4
n.t. n.t. n.t. n.t. n.t. 1 1
28
1 8 0 2 53
4 7 n.t. n.t. n.t. 7 1 4 6 51
2 1 6 2 46
20%
35%
34%
31%
5 2
Indicated are the number of combinations in which each specific interaction was detected. For PRD+VCR up to 10, for PRD+MAF up to 8, and for PRD+DNR 10 different concentrations were tested. Synergism: lethal effect of the combination greater than the product of the effects of each single drugs; Antagonism: lethal effect of the combination less than that of the most single active drug; Additivity: all other observations; n.t.=not tested.
178
G. J. L. Kaspers et
A
Prednisolone + vincristine Multiplicative model 100,
values according to the multiplicative model are given are useful to directly identify synergistic interactions,
a,
i.e. if the observed LCS is lower than the expected LCS.
Additive-antagonism 00
2 a g
Similarly, the figures in which the expected LCS values according to the maximum model are given, enable the identification of antagonistic interactions, i.e. if the observed LCS is higher than the expected LCS.
0
60 6too
0 0
a F2
Synergism
20
8 20
40
60
/
I
80
100
Expected LCS (%)
B
al.
Prednisolone+vincristine (Table 2, Fig. 1) This drug combination was tested in 23 samples at up to 10 different concentrations, which resulted in 228 comparisons between the expected and observed LCS values.
For all but one of these concentrations,
the
observed LCS values were significantly (p < 0.03) lower than the LCS values for the most active single agents. In
Prednisolone + vincristine Maximum model
other words, the drug combinations
100,
??
Antagonism
were significantly
more cytotoxic than the most active single drug. Drug interactions
in the individual
patients
are indicated
in
Table 2. A synergistic interaction was observed in 46%
80 ~
60~
A
0
Prednisolone t mafosfamide Multiplicative model
40 ~
Additive-antagonism
x
AdditiveI 0
I
20
I 80
100
Exifcled LCy(%) Fig. 1. Comparisons (n = 228) between observed leukemic cell survival (LCS) and expected LCS values for the combination prednisolone+vincristine, tested in 23 acute lymphoblastic leukemia samples. (A) Expected LCS according to the multiplicative model: LCS (A+B) = LCS (A) x LCS (B); (B) expected LCS according to the maximum model: LCS (A+B) = D,,,. The interrupted line is x = y.
Synergism
cases of lower expected LCS values. Figures l-3 show the comparisons between the expected LCS values as predicted by the multiplicative model and the maximum model and the observed LCS values for the three drug
combinations. The figures in which the expected LCS
1
0
Exizcted LCF(%) B
deviation divided by the mean) of the LCS values was mean 7.7% (S.D. 7.4%), while for the single drugs this mean was 6.5% (S.D. 4.8%), a non-significant difference. In general, a marked heterogeneity in drug interactions was observed between patient samples, between drug combinations, and occasionally between different concentrations of the same drug combinations (Table 2). The type of drug interaction was not clearly related to the immunophenotype. In case of higher expected LCS values, synergism was slightly more likely to occur, while antagonism tended to occur more frequently in
/ 80
/ 20
0
Prednisolone t mafosfamide Maximum model 100,
2 2 +
"
Antagonism 8o
,
_
2 p: 60g31 ~ 0 40 I2 -0 2s
20-
"0
/ 0
/' 20
I 40
/ 60
I 80
100
Expected LCS (%) Fig. 2. Comparisons (n = 140) between observed leukemic cell survival (LCS) and expected LCS values for the combination prednisolonetmafosfamide, tested in 18 acute leukemia samples. (A) Expected LCS according to the multiplicative model; (B) expected LCS according to the maximum model.
Drug combination testing in leukemia using the MIT
A
drug combinations were significantly (p < 0.02) more cytotoxic than the most active single agent. These cytotoxicities did not differ for the combination of 2 l&ml PRD+MAF 1.7 &ml. Synergism was observed in 51% of observations, with a maximum absolute difference in LCS of 30%. Antagonism was observed in 20% of observations, with a maximum absolute difference in LCS of 22%. An additive interaction was found in the remaining 29% of observations.
Prednisolone + daunorubicin Multiplicative model 100
Additive-antagonism
Synergism
20
0
40
60
80
100
Expected LCS (%)
B
Prednisolone + daunorubicin Maximum model 100
Antagonism 52 e
O
,piD 0
SOQBJ
’ &
2
_
60-
g 3 B >
2
JO-
Additivesynergism
20-
6
,8%
/ 0
I
I
I
I
20
40
60
80
100
Expected LCS (%) Fig. 3. Comparisons
179
assay
(n = 150) between observed leukemic
cell
survival (LCS) and expected LCS values for the combination prednisolone+daunorubicin, tested in 15 acute leukemia samples. (A) Expected LCS according to the multiplicative model; (B) expected LCS according to the maximum model. of observations, with a maximum absolute difference in LCS between the product of the effect of each single drug and the observed LCS of 47%. Similarly, antagonism was seen in 18% of the observations, with a maximum absolute difference in LCS between the effect of the most active single drug and the observed LCS of 33%. Additivity was found for the remaining 36% of observations. Considering the whole group of samples, the concentrations of 0.16 pg/ml VCR+PRD 2 &ml (p = 0.02) and 0.16 pg/ml VCR+PRD 0.2 @ml (p = 0.009) were significantly synergistic. So, synergism was especially found with the lowest concentration of VCR tested. Prednisolone+mafosfamide (Table 2, Fig. 2) This drug combination was tested in 18 samples at up to 8 different concentrations, resulting in 140 comparisons between the expected and observed LCS values (Table 2). For 6 out of 8 of these concentrations, the
Prednisolone+daunorubicin (Table 2, Fig. 3) This combination was successfully tested at 10 concentrations in 15 samples, which resulted in 150 comparisons between the expected and observed LCS values. In contrast to the findings for PRD+VCR and PRDtMAF, the cytotoxicity of 7 out of the 10 different concentrations tested with PRDtDNR were not significantly different from that of the most active single drug. Synergism was seen in 35% of observations, with a maximum absolute difference in LCS of 20%. An antagonistic interaction was found in 31% of the observations, with a maximum absolute difference in LCS of 28%. Considering the whole group of samples, the concentration of 2 pg/ml PRDtDNR 0.008 &ml was significantly (p = 0.03) synergistic. Overall, there were no significantly antagonistic interactions in any of the drug combinations tested, if all samples are considered together. Synergism was observed between PRDtVCR at two different concentrations, and between PRDtDNR at one concentration. The frequency of synergistic, antagonistic, and additive interactions did not differ significantly between the combinations PRDtVCR and PRDtMAF. However, the latter combinations showed synergism more often than the combination PRDtDNR, while PRDtDNR was significantly more often antagonistic (p < 0.05). Discussion We studied interactions between PRD combined with either VCR, MAF or DNR in ALL samples, obtained from children at initial diagnosis, using the MTT assay. Testing drugs in combination gave no technical difficulties. The assay variation for the drug combinations was in the same range, mean 7-8%, as for single drugs. Drug interactions varied considerably between patients, between drug combinations, and even between different concentrations of the same drug combination. Synergistic interactions were more frequent between PRD+MAF and PRD+VCR, detected in about half of the observations, than between PRD+DNR, for which combination synergism was found in 35% of the observations. Antagonism, on the other hand, was seen more frequently between PRDtDNR than between the other two combinations. These differences were sig-
180
G. J. L. Kaspers et
nificantly different. However, the magnitude of antagonism did not differ much between the three drug combinations, nor was there a significant antagonistic interaction in any of the drug combinations tested, if all samples were considered together. We could not confirm our hypothesis that MAP might decrease the cytotoxicity of PRD by alkylating the glucocorticoid receptor, resulting in antagonism between these two drugs. In about one-third of the observations, an additive interaction was found. Additivity may be favorable clinically, although the extent of additive interaction is important to consider. We did not distinguish minor and major interactions in the present study. The frequent additive and synergistic interactions, observed in 70-80% of the observations for all 3 drug combinations, may help explain the known clinical effectiveness of these drug combinations in ALL. [15] The cause of the observed interactions is unknown. It is tempting to speculate that the drugs interact in the cascade of events which characterize drug-induced cell death by apoptosis. The ‘multiplicative model’, which predicts that the effect of a combination is the product of the effect of each single drug, was a better predictor of the observed in vitro cytotoxicity of the combinations of PRD+VCR and PRD+MAF than the ‘maximum’ model, which predicts that the effect of a combination is equal to the effect of the most active single drug. This is in agreement with the findings of Larsson et al. using the fluorometric microculture cytotoxicity assay in ANLL samples [16]. However, it is in contrast to the observations of Sondak et al. in solid tumor samples, who reported that the cytotoxicity of different drug combinations rarely exceeded the cytotoxicity of the most effective single drug. [14] Moreover, the latter authors did not detect synergistic drug interactions in any of 199 solid human tumor samples, while in human leukemia samples this type of interaction is relatively frequent as reported here and by others previously. [16, 171. The heterogeneity in drug interactions observed suggests that for an optimal prediction of the clinical response to combination chemotherapy, drugs should be tested in vitro in the same combinations and if possible in comparable concentrations. Theoretically, if only information about single drugs is used, antagonistic interactions would result in false sensitive predictions, while synergistic interactions result in false resistant predictions. However, the ability to predict clinical drug resistance did not show a major improvement using the in vitro results obtained with drug combinations as compared to single drugs in two studies [6, 181. We could not address this question in the present study, because of a heterogeneous treatment and a short follow-up. Aapro discussed several problems associated with
al.
drug combination testing in clonal assays, which also apply to short-term non-clonogenic assays such as the MTT assay [19]. Drug doses and schedules may not represent the clinical situation. We did not attempt to investigate different time schedules. In addition, chemical interactions between the drugs may occur in vitro, and substances present in the culture medium may influence the drug interaction. Despite these cautionary notes, testing drug combinations in vitro may still be a valuable tool to identify drug interactions. Promising new drug combinations might be selected for clinical application based on the interactions observed in vitro. It might even be possible to use the in vitro results of drug combinations tested in resistance assays to tailor currently used chemotherapeutic drug combinations in individual patients. In conclusion, the MIT assay can be used to study drug interactions in vitro in ALL samples. The combinations PRDtVCR and PRD+MAF generally showed additive and even synergistic interactions. The cytotoxicity of PRDtDNR was generally not markedly higher than that of the most active single drug. Acknowledgement-This study was financially supported by the Dutch Cancer Society (IKA 89-06) and by the project VONK (VU Onderzoek Naar Kinderkanker). Computer equipment was provided by Olivetti Nederland BV. The Dutch Childhood Leukemia Study Group (DCLSG) provided most of the patient samples. Board members of the DCLSG are H. Van Den Berg, M. V. A. Bruin, J. P. M. Bokkerink, P. J. Van Dijken, K. Hahlen, W. A. Kamps, F. A. E. Nabben, A. Postma, J. A. Rammeloo, I. M. Risseeuw-Appel, A. Y. N. SchoutenVan Meeteren, G. A. M. De Vaan, E. Th. Van ‘t Veerkorthof, A. J. P. Veerman, M. Van Weel-Sipman and R. S. Weening.
References 1. Pinkel D. (1987) Curing children of leukemia. Cancer 59, 1683. 2. Veerman A. J. P. & Pieters R. (1990) Drug sensitivity assays in leukaemia and lymphoma. Br. J. Huemat. 74, 381. 3. Bosanquet A. G. (1991) Correlations between therapeutic response of leukaemias and in-vitro drug-sensitivity assay. Lancet 337, 711. 4. Hongo T. & Fujii Y. (1991) In vitro chemosensitivity of lymphoblasts at relapse in childhood leukemia using the M’IT assay. ht. J. Hemat. 54, 219. 5. Pieters R., Huismans D. R., Loonen A. H., Hahlen K., Van Der Does-Van Den Berg A., Van Wering E. R. & Veerman A.J.P (1991) Relation of cellular drug resistance to longterm clinical outcome in childhood acute lymphoblastic leukaemia. Lancet 338, 399. 6. Larsson R., Kristensen J., Sandberg C. & Nygren P. (1992) Laboratory determination of chemotherapeutic drug resistance in tumor cells from patients with leukemia, using a fluorimetric microculture cytotoxicity assay (FMCA). Int.
J. Cancer 50, 177. 7. Kaspers G. J. L., Pieters R., Van Zantwijk C. H., De Waal F. C., Van Wering E. R. & Veerman A. J. P. (1993) Is resistance to prednisolone in vitro related to the response to
Drug combination testing in leukemia using the MTI assay prednisolone at initial diagnosis in childhood acute lymphoblastic leukemia? A preliminary analysis. In Drug Resistance in Leukemia and Lymphoma. The Clinical Value ofLaboratory Assays (Kaspers G. J. L., Pieters R., Twentyman P. R., Weisenthal L. M. & Veerman A. J. P., Eds), pp. 321-328. Chur, Harwood Academic Publ. 8. Kaspers G. J. L., Pieters R., Van Zantwijk C. H., Van Wering E. R., Van Der Does-Van Den Berg A. & Veerman A. J. P. (1993) Resistance to prednisolone (PRD) in vitro: a new prognostic factor in childhood acute lymphoblastic leukemia (ALL) at initial diagnosis. Proceedings of the American Society of Clinical Oncology 12, 320 (abstract). 9. Van Wering E. R., Brederoo P., Van Staalduinen G. J., Van Der Meulen J., Van Der Linden-Schrever B. E. M. & Van Dongen J. J. M. (1993) Contribution of electron microscopy to the classification of minimally differentiated acute leukemias in children. Recent Results Cancer Res. 131,77. 10. Pieters R., Loonen A. H., Huismans D. R., Broekema G. J., Dirven M. W. J., Heyenbrok M. W., Hahlen K. & Veerman A. J. P. (1990) In vitro sensitivity of cells from children with leukemia using the MTT assay with improved culture conditions. Blood 76, 2327. 11. Kaspers G. J. L., Pieters R., Van Zantwijk C. H., De Laat P. A. J. M., De Waal F. C., Van Wering E. R. & Veerman A. J. P. (1991) In vitro drug sensitivity of normal peripheral blood lymphocytes and childhood leukaemic cells from bone marrow and peripheral blood. Br. J. Cancer 64, 469. 12. Pieters R., Huismans D. R., Lcyva A. & Veerman A. J. P.
181
(1988) Adaptation of the rapid automated tetrazolium dye based M’IT assay for chemosensitivity testing in childhood leukemia. Cancer Lett. 41, 323. 13. Valeriote F. & Lin H.-S. (1975) Synergistic interaction of anticancer agents: a cellular perspective. Cancer Chemother. Rep. 59, 89.5. 14. Sondak 0. K., Korn E. L. & Kern D. H. (1988) In vitro testing of chemotherapeutic combinations in a rapid thymidine incorporation assay. Znt. J. Cell Cloning 6, 378. 15. Henderson E. S. (1969) Treatment of acute leukemia. Semin Hemat. 6, 271. 16. Larsson R., Fridborg H., Kristensen J., Sundstriim C. & Nygren P. (1993) In vitro testing of chemotherapeutic drug combinations in acute myelocytic leukaemia using the fluorimetric microculture cytotoxicity assay (FMCA). Br. J. Cancer 67, 969. 17. Sargent J., Taylor C. & Elgie A (1993) Study of drug combinations in acute myeloid leukemia (AML) using the MIT assay. In Drug Resistance in Leukemia and Lymphoma. The Clinical Value of Laboratory Assays (Kaspers G. J. L., Pieters R., Twentyman P. R., Weisenthal L. M. and Veerman A. J. P., Eds), pp. 293-297. Chur, Harwood Academic Publ. 18. Viano I., Silvestro L., Infelise O., Falda M., Lovisone E., Compagnoni G. & Genazzani E. (1986) A short-term chemosensitivity test with different labeled precursors of DNA or protein synthesis: correlation with clinical response. Tumori 72, 357. 19. Aapro M. S. (1984) Drug combination testing with in vitro clonal cultures. Recent Results Cancer Res. 94, 224.