Leukemia Research 24 (2000) 445–452

In vitro chemosensitivity testing in acute non lymphocyticleukemia using the bioluminescence ATP assay

Lars Mollgard a,*, Ulf Tidefelt b, Britt Sundman-Engberg a, Christina Lofgren a,Christer Paul a

a Department of Hematology, Huddinge Uni6ersity Hospital, S-141 86 Huddinge, Stockholm, Swedenb Department of Hematology, O8 rebro Medical Centre Hospital, O8 rebro and Karolinska Institute, Stockholm, Sweden

Received 8 July 1999; accepted 18 December 1999

Abstract

The ATP assay is a short term in vitro chemosensitivity assay where the amount of viable cells are determined by their contentof ATP. The aim of the study was to compare the in vitro results of six cytostatic drugs to the clinical outcome in 83 acutenon-lymphocytic leukemia (ANLL) patients. The secondary ANLL at diagnosis showed an in vitro resistance to daunorubicinthat was significantly higher compared to de novo ANLL at diagnosis (PB0.003). De novo ANLL at diagnosis that achievedcomplete remission (CR) were significantly more sensitive to daunorubicin compared to those who didn’t achieve CR (PB0.05).There was an vitro correlation between topoisomerase II active drugs but not between these drugs and ara-C. In vitro ara-Csensitivity (5 the median of the de novo ANLL at diagnosis) was correlated to poor overall survival (P=0.02). In vitrosensitivity to daunorubicin and mitoxantrone was associated with prolonged disease free survival (P=0.03 and P=0.04). Weconclude that despite significant correlation to clinical parameters for daunorubicin and mitoxantrone the predictive value of theATP assay in this material was insufficient for directing therapy. © 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Bioluminiscence ATP assay; Chemosensitivity; Cytotoxicity; Drug resistance; Myeloid leukemia

www.elsevier.com/locate/leukres

1. Introduction

Many attempts have been made to develop methodsfor in vitro chemosensitivity testing in different tumors.The aim has been to find a reliable test with highpredictive value that could be a useful instrument in thechoice of treatment both at diagnosis and in resistantdisease.

The clonogenic assays and assays based on incorpo-ration of DNA precursors have been used to predictresponse to chemotherapy but technical difficulties and

long culturing time have limited the application of themethods [1–3].

The differential staining cytotoxicity assay (DiSC)measures total cell kill by microscopic evaluation of dyeexclusion by viable cells and is capable of discriminat-ing between effects on tumor and normal cells [4,5].Studies have shown correlation, both to the initialresponse to chemotherapy and to the long term out-come [6–9]. The DiSC is a short time assay (96 h) butlabor-intensive and relies on the subjective assessmentby a skilled observer.

In the MTT-assay surviving cells convert MTT intoformazan which can be quantified by spectrophotome-try. The method have shown good correlation with theDiSC and correlation to clinical outcome in bothleukemias and solid tumors [10–15]. Alternatively, as inthe FMCA assay, fluorescein diacetate can be used as amarker of cell viability [16–18].

Another possibility is the bioluminescence ATP assaybased on metabolic activity measured as cellular ATPcontent [19,20]. The amount of ATP in a specific cell

Abbre6iations: ANLL, acute non lymphocytic leukemia; ara-C,cytarabine; ATP, adenosine 5%-triphophate; CR, complete remission;DiSC, differential staining cytotoxicity; FAB, French AmericanBritish; LC50, the drug concentration lethal to 50% of the leukemiccells; MTT, 3-[4,5-dimethylthiazol-2,5-diphenyl] tetrazolium bromide;Pgp, P-glycoprotein; TCA, trichloracetic acid.

* Corresponding author. Tel.: +46-8-58580000; fax: +46-8-58582525.

E-mail address: lars.mollgard@hematol.hs.sll.se (L. Mollgard)

0145-2126/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.PII: S 0 1 4 5 -2126 (00 )00003 -5

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452446

type is relatively constant [21,22]. ATP is rapidly de-graded by ATP-ases leading to prompt depletion if therespiratory cycle is disturbed in aerobic cells. Since theATP levels are constant in a given healthy cell it can beused as an indirect method for measuring cell growth ordeath. The bioluminescence ATP assay has shown goodcorrelation with the DiSC and clonogenic assays andhas been used in different tumors [2,23–26].

The objective of this study was to evaluate the feasi-bility and the predictive value of the bioluminescenceATP chemosensitivity assay in acute non-lymphocyticleukemia (ANLL).

2. Materials and methods

2.1. Patients

A total of 94 samples with ANLL cells were sepa-rated from bone marrow or peripheral blood from 86patients. Eighty-three of these samples, from 77 pa-tients, were technically successful and diagnosis andstage at inclusion are shown in Table 1. Seventy-four ofthe 77 patients were classified according to the FABcriteria’s [27]: 14 M1; 24 M2; 3 M3; 17 M4; 13 M5; 2M6 and 1 M7.

2.2. Sample collection

Peripheral blood and/or bone marrow was collectedin heparinized tubes before start of treatment. Theleukemic cells were separated by centrifugation (400×gfor 20 min) on metrizoate-dextran (Lymphoprep, Nye-gaard and Co., AS, Oslo, Norway) as previously de-scribed [28,29]. The blast cells (80–90% pure) were thenwashed twice in PBS (phosphate buffered 0.9% saline,pH 7.4).

2.3. Incubations and culturing

Only cells from fresh samples were used. The cells(1.0×105 cells/ml) were incubated in a medium consist-ing of 1.8 ml RPMI 1640 supplemented with 1% L-glu-tamine and 10% fetal calf serum and 0.2 ml of thecytostatic drug at final drug concentrations as follows:daunorubicin 0.2 mM for 1 h, Ara–C 0.5 mM continu-ously, mitoxantron 0.05 mM for 1 h, idarubicin 0.05 mMfor 1 h, amsacrine 1.0 mM continuously and etoposide20 mM for 1 h. All incubations were performed induplicate and with a drug-free control. After the shorttime incubations the cells were spun down (400×g for10 min) and the supernatant removed. A volume of 2ml of fresh medium as described above was added. Allthe samples were then cultured for 4 days in a humi-dified incubator (37°C, 5% CO2).

2.4. Extraction of ATP

Extraction of ATP in leukemic cells was performedby mixing equal volumes (100 ml) of cell-suspension and2.5% TCA (trichloracetic acid). The extracts were as-sayed immediately or stored in a freezer (−20°C) untilanalysis.

2.5. ATP assay

The bioluminescence assay was performed automati-cally in a Bio Orbit photometer (Turku, Finland) aspreviously described [23]. The ATP monitoring reagentand the ATP standard used were both supplied by BioOrbita (Turku, Finland). The ATP standard was recon-stituted in 10 ml distilled water giving a 10 mM solu-tion. The ATP monitoring reagent was reconstitutedwith 5 ml Tris-EDTA buffer at pH 7.75 (100 mM Trisand 2 mM EDTA, pH adjusted with acetic acid). Avolume of 20 ml of the sample was added to 900 mlTris-EDTA buffer. The cuvette was placed in the pho-tometer. Automatically 100 ml ATP monitoring reagentwas dispensed in a cuvette placed in the photometerand the resulting light emission was measured. TheATP standard was then automatically added (10 ml)and the emitted light remeasured. The amount of ATPwas calculated with correction for the blanks. With this

Table 1Diagnosis and stage at inclusion for all ANLL samples and clinicaloutcome in de novo ANLL

Diagnosis and stage at inclusion No. of samples

De no6o ANLLDiagnosis 46

CR after one to two courses 19No CR after two courses 10

4No CR after only onecourse

8Early deathsLow dose treatment 4

1Response to therapy not as-sessed

Relaps 11Resistant disease 9

Secondary ANLLDiagnosis 14

Myelodysplastic syndrome 82CML in blastcrisis2Other myeloproliferative dis-

ease2Other cytostatic treatment3Resistant disease

Methodological 11failure

94Total

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452 447

Fig. 1. In vitro results in de novo and secondary ANLL. The symbol indicates the median. The box encompasses 50% and the box-whiskers 90%of the observations. ** PB0.01 compared to de novo ANLL at diagnosis.

procedure the light emitted is proportional to theamount of ATP in the sample. The results were given asnmol ATP/sample. The percentage ATP in a samplewhen compared to the drug-free control was thencalculated.

2.6. Drug therapy

Thirty-seven de novo ANLL and 11 secondaryANLL all at diagnosis received one or two intensiveinduction courses according to different protocols.Most of the patients received ara-C (n=47) and oneanthraquinone, e.g. mitoxantrone (n=23) anddaunorubicin (n=12). Twenty-six patients also re-ceived etoposide and 11 thioguanine. Four patientsreceived the resistance modifying cyclosporine analoguePSC 833. One patient with ANLL M3 also receivedall-trans retinoic acid (ATRA). Two patients underwentan autologous bone marrow transplantation and threepatients underwent an allogeneic bone marrowtransplantation

2.7. Clinical e6aluation

Complete remission (CR) was defined as =5% blastcells for M1 and M5a and for the other FAB-groupsalso =10% leukemic cells (blastcells, promyelocytesand promonocytes), absence of Auer rods and absenceof leukemic cell clusters in a bone marrow aspirate. Invivo sensitivity was defined as CR after one to twoinduction courses and in vivo resistance as no CR aftertwo induction courses.

2.8. In 6itro– in 6i6o comparison and sur6i6al analysis

In the patients with ANLL at diagnosis who achievedinduction therapy the result of the single in vitro mostactive drug that the patient received was compared tothe clinical outcome. In the survival analysis the medianfor each drug from the 46 de novo ANLL at diagnosiswas used as a cut off level to separate in vitro sensitivityfrom in vitro resistance (Fig. 1.). In the overall survivalanalysis the ANLL patients at diagnosis were includedon an intention to treat basis. Disease free survival wasdefined as the time from complete remission to the dateof relapse, death or last follow up.

2.9. Statistical analysis

The differences in cytotoxic effect in vitro betweendifferent groups of patients were evaluated with t-testfor independent samples. The Kaplan–Meier methodand the log-rank test were used to estimate differencesin survival.

3. Results

3.1. Methodological considerations

Assays from 11/94 patients could not be evaluated,10 due to failure in the control (six samples had too lowlevels of ATP in the drug free control (B20 nM) and infour samples only one of the duplicates in the control

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452448

was successful) and one due to extremely high levels ofATP compared to the drug free control. In 33 cases onesingle drug sample had to be excluded mainly becauseof failure in one of the duplicates. Totally 443 out of542 individual drug samples (82%) were successful.

3.2. Clinical outcome

The clinical outcome for the 33 evaluable de novoANLL patients at diagnosis who achieved inductiontreatment is shown in Table 1. The eight early deathswere due to infections or haemorrhages. Four of thesepatients died before treatment had started and fourdied after the first course but before the response couldbe evaluated.

3.3. In 6itro– in 6i6o comparison in different groups ofpatients

In Fig. 1 the in vitro results from the de novo ANLLat diagnosis, the de novo ANLL at relapse/resistantdisease and secondary ANLL at diagnosis are shown.There was a wide distribution of the results for eachdrug and an obvious overlapping when comparing theseparate drugs in the different groups. Although themean values were higher in the de novo ANLL atrelapse/resistant disease group compared to the de novoANLL at diagnosis group for daunorubicin (48 and43%), mitoxantrone (50 and 46%) and idarubicin (46and 43%) these differences were not significant. The in

vitro effect of daunorubicin was significantly higher atdiagnosis in the de novo ANLL group compared to thesecondary ANLL (43 and 66%, P=0.003). The meaneffect of mitoxantrone, idarubicin, amsacrine andetoposide was also higher at diagnosis in the de novoANLL group compared to the secondary ANLL groupbut the differences were not significant.

The effect of daunorubicin was significantly lower inthe group of patients who did not achieve CR after twocourses compared to those who achieved CR after oneto two courses (56 and 38%, P50.05), (Fig. 2). Themean values for ara-C (60 and 48%), mitoxantrone (55and 40%) idarubicin (44 and 38%) and amsacrine (40and 35%) showed the same tendency but the differenceswere not significant. In the patients with ANLL atdiagnosis who achieved induction therapy and whowere evaluable for clinical response, the single in vitromost active drug of the drugs that the patient actuallyreceived could not predict the short term clinicaloutcome.

3.4. In 6itro correlation between different drugs

In vitro cross resistance can be indicated by thecorrelation between different drugs. Results from allANLL samples which were included in this study areshown in Table 2. There was a clear correlation be-tween the in vitro effect of daunorubicin, idarubicin,mitoxantrone, etoposide and amsacrine but not be-tween these drugs and ara-C.

Fig. 2. In vitro results and clinical outcome in previously untreated de novo ANLL. The symbol indicates the median. The box encompasses 50%and the box-whiskers 90% of the observations. * P50.05 compared to the CR group.

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452 449

Table 2Relationship between the in vitro effect of different drugs in all ANLL patients expressed as correlation coefficients

MitoxantronIdarubicin Ara-C Amsacrine Etopside

0.74 0.20Daunorubicin 0.770.74 0.65–Idarubicin 0.84 0.31 0.80 0.77

– 0.32– 0.81Mitoxantrone 0.74– –Ara-C 0.19– 0.22– –– –Amsacrine 0.64

3.5. In 6itro results and sur6i6al analysis

Patients who were in vitro sensitive to daunorubicinshowed a tendency towards better over all survival(P=0.14). In vitro sensitivity to the other drugs wasnot associated with better overall survival. Patients whowere in vitro sensitive to ara-C had a poor over allsurvival (P=0.02). Patients who were in vitro sensitiveto daunorubicin (Fig. 3a) or mitoxantrone showed aprolonged disease free survival that was significant(P=0.03 and P=0.04). Idarubicin showed the sametendency (P=0.06). In vitro sensitivity to amsacrine oretoposide could not predict the length of disease freesurvival. Patients that were in vitro sensitive to ara-Cdid not differ from in vitro resistant patients in diseasefree survival (P=0.99, Fig. 3b).

4. Discussion

In this study with samples from patients with denovo ANLL we showed that in vitro sensitivity todaunorubicin, using the short term bioluminescenceATP-assay, was associated with response to inductiontherapy in previously untreated patients. In vitro sensi-tivity to daunorubicin and mitoxantrone was associatedwith prolonged disease free survival but not overallsurvival. As expected, in vitro resistance to daunoru-bicin was more common in the group of secondaryANLL at diagnosis. In the analysis of different drugs’correlation in vitro, the results corresponded well towhat is known about these drugs mechanism of actionand cross resistance.

The significantly higher in vitro effect of daunoru-bicin in patients who entered a CR compared to pa-tients who did not, is in accordance with other studies[8,15]. Ara-C is another important drug in the treat-ment of ANLL, but here we found no significant invitro differences between responders and non respon-ders. One reason could be that ara-C has anothermechanism of action affecting the cell in the S-phase[30]. In the short term incubations, contrary to clono-genic assays, the cells are just kept alive and furthergrowth and cell-divisions are not required. Anotherreason could be that the in vitro continuous ara-C

incubation corresponds better to in vivo continuos infu-sion of the drug than higher doses administered asshort infusion which most of the patients in our studyreceived. The ara-C results are in contrast to a previousstudy with the DiSC assay [8] but in accordance withanother study using the MTT assay where non-respon-ders were in vitro more resistant to daunorubicin butnot to ara-C compared to responders [15].

Fig. 3. (a) Disease free survival in de novo and secondary ANLL. Themedian for daunorubicin is used to distinguish in vitro sensitivityfrom in vitro resistance. (b) Disease free survival in de novo andsecondary ANLL. The median for ara-C is used to distinguish invitro sensitivity from in vitro resistance.

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452450

In the survival analysis sensitivity to both daunoru-bicin and mitoxantrone was associated with prolongeddisease free survival. There is no obvious explanation ofthe shorter overall survival in patients who were invitro sensitive to ara-C. The specific problem withara-C incubations has already been discussed and inaddition there are reports showing that a high numberof S-phase cells in ANLL is associated with a poorsurvival [31]. Thus one could theoretically assume thata high proportion of cells in S-phase makes the cellsmore sensitive to ara-C in the in vitro incubation. Inthe in vivo situation a high proportion of cells inS-phase instead indicates a highly proliferative diseasewith a high potential of regrowth and consequently apoor survival. This can illustrate the difficulties intranslating in vitro results to the in vivo situation butdue to the limited material for the survival analysis inour study and a high proportion of early deaths in theara-C sensitive group no further conclusions can bedrawn. In a previous study, using the DiSC assay, weshowed that in vitro sensitivity to anthracyclines and/orara-C predicted overall survival (P50.01) [8]. Anotherstudy that used the MTT assay showed that in vitrosensitivity to both daunorubicin and ara-C predicteddisease free survival (P=0.02) and that in vitro sensi-tivity to ara-C could predict continuous complete re-mission (P=0.02) [15]. Interestingly, six out of sevenpatients who were in vitro sensitive to daunorubicinand resistant to ara-C entered a complete remission.

The study of in vitro correlation between differentdrugs illustrates cross resistance and how differentmechanisms are involved in the drug resistance. ThePgp mediated multidrug resistance (MDR) is associatedwith resistance mainly to anthracyclines, vinca alkaloidsand epipodophyllotoxins [32]. Topoisomerase II medi-ated drug resistance affects important drugs used fortreatment of ANLL, e.g. anthracyclines, anthracen-dions, acridines and epipodophyllotoxins. In our studythese mechanisms were illustrated by the strong correla-tion between these drugs. The nucleoside analogue ara-C is not affected by these two mechanisms and therewas no cross resistance between ara-C and the otherdrugs in our study.

In the survival analysis the definition of ‘in vitrosensitive’ had to be stated. One way is just to arbitrarilydefine a cut-off level (often 30%) between in vitrosensitive and resistant [33]. Others used each drug inseveral concentrations and defined sensitive/resistantusing the median LC50 (the drug concentration lethal to50% of the leukemic cells) as cut-off point [15]. Anotherway is to find the cut-off level that best separates invivo sensitive and resistant patients [8]. In one studyKristenssen et al. divided the results for each drug,from all patients, into quartiles considering the lowerquartile as in vitro sensitive. This made it possible tofind an individual cut-off level for each drug [17]. We

chose this last method in our survival analysis butinstead of the lower quartile we used the median whichseparated the material in groups where the number ofpatients were equal. For most of the cytostatic drugsthe level for in vitro sensitivity was about 40% com-pared to the arbitrarily 30% limit (see Fig. 1). Com-pared to Tidefelt et al our ara-C level was higher, 49%compared to 35%, and the daunorubicin level lower, 39compared to 60% [8].

In our study, patients with de novo ANLL at diagno-sis who achieved induction therapy, the single in vitromost active drug of the drugs that the patient actuallyreceived could not predict the short term clinical out-come. The majority of the patients in our study re-ceived ara-C in their induction treatment but the otherdrugs varied. Some of the patients were included inprotocols where the induction course had a more or lessnovel design and in some cases the bioluminescenceATP-assay did not include all drugs the patient re-ceived. On the other hand all patients got an anthracy-cline or anthracendion derivate in addition to ara-Cand the correlation study suggested that in vitro sensi-tivity to daunorubicin corresponded to the in vitrosensitivity to other anthracyclines and anthracendions.A possible explanation to the discrepancy between ourin vitro and in vivo data could be, as mentioned above,that short term assays may not be appropriate in the invitro testing of ara-C. We have not estimated the blastpercentage at the end of the test and there is a possibil-ity that non malignant cells present at this stage mayhave decreased the predictive power of the test.

The short term assays which are used today are inmany aspects comparable, e.g. the culture procedure.The main difference is the various techniques that areused to estimate viable cells after the incubation. Thebioluminescence ATP assay has been in use for a longtime within the fields of biochemistry and microbiology,and a previous study has shown that it correlatessatisfactorily to the DiSC assay (r=0.8) [21–23]. Theuse of drug concentrations mimicking in vivo condi-tions, in combination with the DiSC assay, have showngood correlations to clinical outcome [8]. We have onlyused fresh samples, in contrast to other studies, wherecryopreserved samples also have been analyzed [15,17].One crucial point is the amount of leukemic cells afterthe density gradient centrifugation. In this aspect theDiSC assay has an advantage in the possibility tomorphologically distinguish leukemic cells from othercells. Even if that procedure is very time consuming itmay contribute to the good correlation between in vitroand in vivo data [8]. The success rate in our study was82%. Four samples were excluded because of failure inone of the duplicates in the drug free control. If thatcould have been avoided by using three instead of twodrug free controls the success rate would have been86%.

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452 451

Even if our study, as well as other previous studies,has shown correlation’s to the clinical outcome inANLL, more convincing results are needed before theseshort term assays will be accepted for tailoringchemotherapy treatment in clinical practice [8,14,15,17].Another possible application of the in vitro chemosensi-tivity data is in the risk group stratification of ANLLpatients, but also in this aspect further prospectivestudies are necessary to confirm that, e.g. in vitroresistance to daunorubicin is an independent risk factorin ANLL. In all different total cell kill assays there isalways the risk of contamination of non malignantcells. Even if the proportion of blasts so far generallyexceeded 80–90%, the remaining normal cells may af-fect the result. In acute lymphocytic leukemia therehave been attempts to overcome this problem by theuse of flow cytometry were the blast population can beseparated from non malignant cells [34]. We have estab-lished a similar method for ANLL that currently isevaluated.

Acknowledgements

This work was supported by grants from the SwedishCancer Society. The authors thank Sofia Bengtsson,Ulrika Broberg and Malin Prenkert for their technicalassistance.

References

[1] Hamburger AW, Salmon SE. Primary bioassay of human stemcells. Science 1977;197:461.

[2] Cree IA, Pazzagli M, Mini E, Mazzei T, Hunter EM, SutherlandLA, Pinzani P, Gerli A, Andreotti PE. Methotrexate chemosensi-tivity by ATP luminiscence in human leukemia cell lines and inbreast cancer primary cultures: comparison of the TCA-100assay with a clonogenic assay. Anticancer Drugs 1995;6:398.

[3] Delmer A, Marie JP, Thevenin D, Suberville AM, Zittoun R.Treatment of relapsing and refractory adult acute myeloidleukemia according to in vitro clonogenic leukemic cell drugsensitivity. Leuk Lymphoma 1993;10:67.

[4] Weisenthal LM, Marsden JA, Dill PL, Macaluso CK. A noveldye exclusion method for testing in vitro chemosensitivity ofhuman tumors. Cancer Res 1983;43:749.

[5] Bird MC, Bosanquet AG, Forskitt S, Gillby ED. Semi-microadaption of a 4-day differential staining cytotoxicity (DiSC)assay for determining the in-vitro chemosensitivity of haemtolog-ical malignancies. Leuk Res 1986;10:445.

[6] Wilbur DW, Camacho ES, Hilliard DA, Dill PL, WeisenthalLM. Chemotherapy of non-small cell lung carcinoma guided byan in vitro drug resistance assay measuring total tumour cell kill.Br J Cancer 1992;65:27.

[7] Bosanquet AG, Copplestone JA, Johnson SA, Smith AG, PoveySJ, Orchard JA, Oscier DG. Response to cladribine in previoslytreated patients with chronic lymphocytic leukemia identified byex vivo assessment of drug sensitivity by DiSC assay. Br JHaematol 1999;106:474.

[8] Tidefelt U, Sundman-Engberg B, Rhedin A-S, Paul C. In vitrodrug testing in patients with acute leukemia with incubationsmimicking in vivo intracellular drug concentrations. Eur JHaematol 1989;43:374.

[9] Staib P, Lathan B, Schinkothe T, Wiedenmann S, Pantke B,Dimski T, Voliotis V, Diehl V. Prognosis in adult AML isprecisely predicted by the DiSC-assay using the chemosensitivity-index Ci. Adv Exp Med Biol 1999;457:437.

[10] Pieters R, Huismans DR, Leyva A, Veerman AJ. Comparison ofa rapid automated MTT assay with a dye-exclusion assay forchemosensitivity testing of childhood leukemia. Br J Cancer1989;59:217.

[11] Kaspers GJ, Veerman AJ, Pieters R, Van Zantwijk CH, SmetsLA, Van Wering ER, Van Der Does-Van Den Berg A. In vitrocellular drug resistance and prognosis in newly diagnosed child-hood acute lymphoblastic leukemia. Blood 1997;90:2723.

[12] Norgaard JM, Langkjer ST, Palshof T, Clausen N, Pedersen B,Hokland P. Relation of blast cell survival and proliferation tochemotherapy resistance in AML. Br J Haematol 1996;93:888.

[13] Norgaard JM, Olesen G, Kristensen JS, Pedersen B, Hokland P.Leukemia cell drug resistance and prognostic factors in AML.Eur J Haematol 1999;63:219.

[14] Stute N, Kohler T, Lehmann L, Wetzstein W, Ehninger G. Drugresistance testing of acute myeloid leukemia in adults using theMTT assay. Adv Exp Med Biol 1999;457:445.

[15] Klumper E, Ossenkoppele GJ, Pieters R, Huismans R, LoonenAH, Rottier A, Westra G, Veerman AJ. In vitro resistance tocytosin arabinoside, not to daunorubicin, is associated with therisk of relapse in de novo acute myeloid leukaemia. Br J Haema-tol 1996;93:903.

[16] Larsson R, Kristensen J, Sanberg C, Nygren P. Laboratorydetermination of chemotherapeutic drug resistance in tumor cellsfrom patients with leukemia using a fluorometric microculturecytotoxicity assay (FMCA). Int J Cancer 1992;50:177.

[17] Kristensen J, Jonsson B, Sundstrom C, Nygren P, Larsson R. Invitro analysis of drug resistance in tumor cells from patients withacute myelocytic leukemia. Med Oncol Tumor Pharmacother1992;9:65.

[18] Lonnerholm G, Frost BM, Larsson R, Liliemark E, Nygren P,Peterson C. In vitro cytotoxic drug activity and in vivo pharma-cokinetics in childhood acute myeloid leukemia. Adv Exp MedBiol 1999;457:429.

[19] Kuzmits R, Aiginger P, Muller MM, Steurer G, Linkesch W.Assessment of sensitivity of leukemic cells to cytotoxic drugs bybioluminescence measurement of ATP in cultured cells. Clin Sci1986;71:81.

[20] Kangas L, Gronroos M, Nieminen AL. Bioluminiscence ofcelullar ATP: a new method for evaluating cytotoxic drugs invitro. Med Biol 1984;62:338.

[21] Thore A, A, nsehn S, Lundin A, Bergman S. Detection of bacteri-uria by luciferase assay of adenosine triphosphate. J Clin Micro-biol 1975;1:1.

[22] Lundin A, Hasenson M, Persson J, Pousette A, . Estimation ofbiomass in growing cell lines by adenosine triphosphate assay.Methods Enzymol 1986;133:27.

[23] Rhedin AS, Tidefelt U, Jonsson K, Lundin A, Paul C. Compari-sion of a bioluminescence assay with differential staining cyto-toxicity for cytostatic drug testing in vitro in human leukemiccells. Leuk Res 1993;17:271.

[24] Andreotti PE, Cree IA, Kurbacher CM, Hartmann DM, LinderD, Harel G, Gleiberman I, Caruso PA, Ricks SH, Untch M.Chemosensitivity testing of human tumors using a microplateadenosine triphosphate luminiscence assay: clinical correlationfor cisplatin resistance of ovarian carcinoma. Cancer Res1995;55:5276.

[25] Kurbacher CM, Bruckner HW, Cree IA, Kurbacher JA, Wil-helm L, Poch G, Indefrei D, Mallman P, Andreotti PE. Mitox-

L. Mollgard et al. / Leukemia Research 24 (2000) 445–452452

antrone combined with paclitaxel as salvage therapy for plat-inum-refractory ovarian cancer: laboratory study and clinicalpilot trial. Clin Cancer Res 1997;3:1527.

[26] Cree IA, Neale MH, Myatt NE, de Takats PG, Hall P, Grant J,Kurbacher CM, Reinhold U, Neuber K, MacKie RM, Chana J,Weaver PC, Khoury GG, Sartori C, Andreotti. Heterogeneity ofchemosensitivity of metastatic cutaneous melanoma, AnticancerDrugs 1999;10:437.

[27] Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA,Gralnick HR, Sultan C. Proposed revised criteria for the clas-sification of acute myeloid leukemia: a report of the French–American–British Cooperative Group. Ann Intern Med1985;103:620.

[28] Boyum A. Isolation of leucocytes from human blood. Furtherobservations. Methylcellulose, dextran and ficoll as erythrocyteaggregating agents. Scand J Clin Lab Invest 1968;21:31.

[29] Paul C, Peterson C, Gahrton G, Lockner D. Uptake of free and

DNA-bound daunorubicin and doxorubicin in human leukemiccells. Cancer Chemother Pharmacol 1979;2:49.

[30] Pallavicini MG. Cytosine arabinoside: molecular, pharmacoki-netic and cytokinetic considerations. Pharmacol Ther 1984;25:207.

[31] Vidriales MB, Orfao A, Lopez-Berges MC, Gonzalez M, Lopez-Macedo A, Ciudad J, Lopez A, Garcıa MA, Hernandez J,Borrego D, San Miguel JF. Prognostic value of S-phase cells inAML patients. Br J Haematol 1995;89:342.

[32] McKenna SL, Padau RA. Multidrug resistance in leukemia(review). Br J Haematol 1997;96:659.

[33] Bird MC, Bosanquet AG, Forskitt S, Gillby ED. Longtermcomparison of results of a drug-sensitivity assay in vitro withpatient response in lymphatic neoplasms. Cancer 1988;61:1104.

[34] Campana D, Manabe A, Evans WE. Stroma-Supported Immun-cytometric Assay (SIA): a novel method for testing the sensitiv-ity of acute lymphoblastic leukemia cells to cytotoxic drugs.Leukemia 1993;7:482.

.