Leukemia (1998) 12, 1230–1235  1998 Stockton Press All rights reserved 0887-6924/98 $12.00 http://www.stockton-press.co.uk/leu In vitro chemosensitivity of chronic lymphocytic leukaemia to purine analogues – correlation with clinical course TJ Bromidge1, DL Turner2, DJ Howe1, SA Johnson2 and SAJ Rule2

1Leukaemia Research Unit and 2Department of Haematology, Taunton and Somerset Hospital, Musgrove Park, Taunton, Somerset TA1 5DA, UK

Samples from 51 chronic lymphocytic leukaemia (CLL) patients analogues have produced inconsistent results with some (42 typical, nine atypical) were assessed for in vitro response reports of responses with one agent after failure with the to fludarabine and cladribine (2-CdA) using the flow cytometric other,11 while the experience of others is that clinical resist- terminal deoxynucleotidyl transferase (TdT) assay. No differ- 12–14 ence was demonstrated between the in vitro response of typi- ance extends to both compounds. In vitro, a correlation cal and atypical CLL and previous treatment did not result in of resistance/sensitivity between 2-CdA and fludarabine was a more apoptosis resistant phenotype. The assay could not dis- found in a proportion of CLL samples (54/78).15 This study tinguish those patients who required subsequent treatment also reported a correlation of drug resistance with expression from those whose disease remained stable, and universal of the FMC7 antigen. The association of this marker with non- cross-resistance/sensitivity to the two purine analogues was demonstrated. The assay’s potential for use in the rapid CLL malignant lymphoid disorders suggests that disease enti- assessment of in vivo response to purine analogue therapy in ties which are intrinsically less likely to be responsive to CLL was limited; correctly predicting the clinical outcome of purine analogue therapy may have been included in the 10/12 patients to treatment but failing to predict progression analyses. Consistent classification of lymphoproliferative dis- in two p53 deficient patients. The level of bcl-2 in the clonal orders is of great importance when correlating laboratory lymphocytes did not influence the in vitro, spontaneous or parameters with clinical course, as inclusion of a hetero- drug-induced, apoptosis. Keywords: chronic lymphocytic leukaemia; in vitro chemo- geneous group of patients will obscure emerging trends. In sensitivity; purine analogue therapy this report, strict morphological and immunophenotypic cri- teria have been used to define the patient population and inclusion has been restricted to CLL only.16 A recent report Introduction suggested that atypical CLL was more likely to progress clini- cally than typical CLL,17 implying that some fundamental dif- Drugs of the purine analogue group have considerable clinical ference exists between these two subtypes of disease. The efficacy in the treatment of low-grade lymphoproliferative dis- present belief that CLL arises through a defect in the apoptotic eases and there is extensive experience with the use of fluda- pathway raises the question as to whether typical and atypical rabine and cladribine to treat patients with chronic lympho- CLL show similar apoptotic responses, and led to the further cytic leukaemia (CLL).1 Both compounds are resistant to classification of patients in this report as typical and atypical deamination by adenosine deaminase within lymphoid cells CLL. The over-expression of the oncogene bcl-2, a well estab- resulting in their incorporation as abnormal nucleotides into 18 DNA2,3 and cell death by apoptosis.4 Various assays have lished characteristic of CLL cells arises from hypomethyl- ation of the bcl-2 gene and has been linked with the extended been applied to the assessment of the in vitro chemosensitivity 19 of CLL cells5–7 and the suggestion that the susceptibility of survival of CLL cells in culture. A study of CLL patients found cells to the induction of apoptosis may correlate with the clini- no correlation of bcl-2 expression and clinical features, how- 5 ever, higher bcl-2 levels appeared to be linked with shorter cal response to purine analogue therapy has been made. As 20 cell death due to purine analogue treatment results from the survival. The belief that the level of bcl-2 can, by itself, pre- production of DNA strand breaks, the terminal deoxynucleo- dict the susceptibility of a sample to apoptosis is losing favour tidyl transferase (TdT) assay should be an appropriate method since the discovery of a bcl-2-related gene bax. The prop- osition that the ratio of bcl-2 to bax determines cell survival for assessing chemosensitivity in these patients. TdT is a DNA 21 polymerase which can be used to label DNA strand breaks rather than the absolute level of either has been made. We by the incorporation of a labelled nucleotide followed by a have applied the TdT assay to 51 patients with CLL and evalu- fluorescent detection step.8 ated the level of bcl-2 in their clonal lymphocytes. Correlation Exposure to therapeutic agents may result in changes in the of TdT assay results with the level of bcl-2 protein expression chemosensitivity of CLL cells. In vitro studies of previously was examined. treated CLL patients have reported the development of pleio- 9 tropic drug resistance and a difference in the spontaneous Materials and methods rate of apoptosis.5 Other investigators have demonstrated that in vitro resistance to chlorambucil increased following chlor- Patients ambucil therapy.10 However, in some cases, the increase was reversible and fluctuations in untreated patients could occur, Peripheral blood (PB) samples from 51 B-CLL patients were suggesting that factors other than therapy influence drug examined. All patients had a significant lymphocytosis; at the resistance. time of study the majority of the lymphocytes were malignant. Clinical evaluations of the comparative efficacy of purine Strict criteria were applied to define the diagnosis and patients were classified as typical CLL or atypical CLL according to surface immunophenotype and morphology16 as follows: Correspondence: TJ Bromidge; Fax: 441823271023 Received 12 February 1998; accepted 21 April 1998 Typical CLL (42 patients) – phenotype and morphology In vitro chemosensitivity of CLL to purine analogues TJ Bromidge et al 1231 characteristic of B-CLL. CD5+, SIg weak, CD23+, FMC7−, Bcl-2 staining typical morphology. Atypical CLL (nine patients) – either one atypical surface Separated mononuclear cells were fixed in 0.5% formal- marker or atypical morphology, did not include cases of dehyde (diluted directly before use from a 4% stock solution) CLL/PL. and stored at 4°C. Cells were centrifuged out of formaldehyde, permeablised by incubating at 37°C in PBS/0.2% Tween 20 Treatment details are shown in Table 1. Clinical staging was for 15 min, then dual stained with anti-CD19/PE and anti-Bcl- 22 according to the Binet system and response criteria were 2/FITC monoclonal antibodies (Dako, High Wycombe, UK) 23 defined by the NCI-sponsored working group. and analysed on a FACScan flow cytometer. A mouse IgG1- RD1/mouse IgG1-FITC isotype control was used (Coulter, Luton, UK). Drug sensitivity assay Ten thousand events were collected after gating on CD19- positive cells. Mean bcl-2 levels were expressed as a linear equivalent of channel number, in arbitrary units (AU), to Peripheral blood mononuclear cells were obtained by density enable direct comparison. gradient centrifugation on lymphocyte separation medium (Nycomed, Oslo, Norway). Aliquots of 106 cells were cultured in a total volume of 500 ␮l RPMI 1640, 25 mm Hepes with l- Statistical analysis glutamine (Life Technologies, Paisley, UK), 10% fetal calf ° serum (Life Technologies), in a 37 C, 5% CO2, humidified Patient groups were compared with the Mann–Whitney U test, incubator in the presence/absence of drug. Drug concen- the significance of response to drug analysed by Wilcoxon trations used were fludarabine (Schering AG, Berlin, signed rank test and the relationship between response to cla- Germany) 3 ␮m and 30 ␮m, cladribine (purchased from Drs dribine and fludarabine was compared using Spearman corre- Zygmunt Kazimierczuk and Pawel Grieb, The Foundation for lation. All analysis was performed using SPSS PC + software. the Development of Diagnostics and Therapy, Warsaw, Poland) 1.7 ␮m and 17 ␮m. After 22 h culture, cells were cen- trifuged out of media at 200 g, fixed in 1% formaldehyde Results and discussion (made fresh from a 4% stock solution) for 15 min on ice, washed in PBS (pH 7.4) and resuspended in 70% ethanol. The Drug sensitivity assay samples were kept at −20°C for 1–3 days before performing the TdT assay. The DNA strand breaks were labelled with Previous investigators have demonstrated that a substantial digoxigenin-ddUTP (Boehringer Mannheim, Lewes, UK) and proportion of PB B-CLL lymphocytes die spontaneously by detected using FITC conjugated anti-digoxigenin (Boehringer apoptosis upon culture,25 that this could be increased by the Mannheim).24 Samples were analysed on a FACScan flow addition of purine analogue to the culture, and that inter- cytometer (Becton Dickinson, Oxford, UK) using a 488 nm patient variation occurred.5 Using the TdT assay to assess in beam from an argon ion laser. Green fluorescence was col- vitro response of CLL cells to culture and exposure to purine lected after a 530 nm (band width 30 nm) filter and measured analogues, cladribine and fludarabine, we report that this using logarithmic amplification. Calibration and sensitivity cohort of CLL patients exhibited widely varying responses. were checked using fluorescent-labelled beads (Calibrite Addition of cladribine or fludarabine resulted in a dose- Beads; Becton Dickinson). Debris was gated out on the for- dependent increase in apoptosis. Statistical analysis showed ward scatter vs side scatter dot plot and 10 000 events col- the response to addition of drug to be significant for both con- lected. The data were analysed using LYSIS II software centrations of fludarabine and cladribine (Wilcoxon signed (Becton Dickinson). rank test P = 0.000 in all cases). This analysis was repeated Background fluorescence was determined using a patient- on all subsets of patients subsequently analysed and the sig- specific sample fixed pre-culture and a region corresponding nificance remained. Although the higher concentrations of to increased green fluoresence set. The percentage of cells drug used (fludarabine 30 ␮m, cladribine 17 ␮m) are higher responding was taken as those showing increased green than that therapeutically achievable, a complete range of fluorescence. response was seen between the patients (30 ␮m fludarabine Duplicate wells were analysed in several patients, the vari- range 14–91%, 17 ␮m cladribine range 13–94%) with both ation in percentage of cells responding was such that in the extremes being rare. As our aim was to identify comparative remaining patients only a single assay was performed. differences between patients and the higher drug concen- trations exaggerated the effect rather than turning comparative non-responders into responders, we restricted our analyses to the high concentrations. Table 1 Patient details Clinically, fludarabine is administered as fludarabine mono- phosphate which is rapidly dephosphorylated to the nucleo- Sub-type No. of Previously Previous Subsequent side F-ara-A, by 5Ј-nucleotidase, before transportation into the patients treated purine purine cell and rephosphorylation to its tri-phosphate form. A con- analogue analogue cern with using the monophosphate form (fludarabine) in an in vitro system is that transport of the drug into the cells will Typical CLL 42a 12 4 10 be impaired. However, in our system the drug had an effect Atypical CLL 9 2 0 3 on lymphocytes and the effect was dose dependent indicating Total 51 14 4 13 that the drug had access to the cells. For some patients whole blood cultures were set up in parallel with cultures of separ- aClinical data were not available for two patients in this group. ated lymphocytes and no difference in the percentage of cells In vitro chemosensitivity of CLL to purine analogues TJ Bromidge et al 1232 Table 2 Comparison of in vitro response of typical and atypical CLL

Sub-type No. of No drug range 30 ␮M fludarabine 17 ␮M cladribine patients (median)a range (median)a range (median)a

Typical 42 4–66 (22) 14–88 (58) 20–86 (65) Atypical 9 5–83 (21) 12–91 (51) 13–94 (63) Mann–Whitney U test P 0.780 0.971 0.932

aExpressed as percentage of responding cells.

responding to fludarabine was observed, indicating that 5Ј- response (Table 3B); demonstrating that this in vitro assay is nucleotidase activity on erythrocytes did not increase the not capable of distinguishing the subset of CLL patients who availability of fludarabine to cells in this assay. Due to the will require future treatment from those whose disease relative insolubility of F-ara-A and the results of these prelimi- remains stable. This is an interesting observation, implying nary experiments fludarabine was used in this assay. that the susceptibility of CLL cells to apoptosis in vitro bears In this cohort of strictly classified CLL patients the rates of little relation to their accumulative potential in vivo. spontaneous apoptosis and response to fludarabine and cladri- The correlation between the in vitro response of patients to bine showed wide variation and no significant difference was cladribine and fludarabine was striking (Figure 1), only 2/51 apparent between patients with typical or atypical CLL (Table patients varied in their response to the two drugs (P = 0.000 2). Atypical CLL is clinically more aggressive than typical by Spearman correlation analysis). Previous investigators have CLL,17 however, the results presented here imply that atypical documented a general concordance of sensitivity to 2-CdA CLL lymphocytes are not more resistant to the in vitro induc- and fludarabine but did find a percentage of cases where 15 tion of apoptosis. When the effect of prior treatment was cross-resistance was not evident. The use of cut off LD50 examined, 13 patients who had completed therapy more that values for placing patients into sensitive/resistant categories in 6 months before assessment (but with significant disease present) were compared with 35 previously untreated patients (one patient receiving treatment immediately prior to the assay was excluded from analyses); analyses showed no statistically significant difference (Mann–Whitney U test) between the two groups for either spontaneous or drug-induced response (Table 3A). It appears from these results that chemotherapy does not necessarily result in the irreversible selection of a lymphocyte subpopulation more resistant to apoptotic cell death. This is in conflict with previous reports5,9,10 although the latter investigators found that in some cases the increase was reversible and that fluctuations in untreated patients could also occur. Our group of previously treated patients had all received treatment over 6 months before the assay was performed; this may explain the discrepancy between findings and also supports the suggestion that factors other than ther- apy cause changes in drug resistance. Comparison of previously untreated patients who sub- sequently required treatment (n = 10) with those who have never needed treatment (n = 25) showed no statistically sig- Figure 1 Correlation of in vitro response to 30 ␮m fludarabine and nificant differences (Mann–Whitney U test) in their in vitro 17 ␮m cladribine.

Table 3 Correlation of clinical course with in vitro response

Treatment No. of No drug P 30 ␮M P17␮M P patients range fludarabine cladribine (median)a range range (median)a (median)a

(A) Untreated before 35 5–66 (20) 12–88 (60) 13–87 (68) TdT assay 0.871 0.296 0.057 Treated Ͼ6 months 13 4–83 (31) 31–91 (44) 33–94 (55) before TdT assay (B) Subsequently treated 10 7–61 (17.5) 12–88 (55) 13–85 (60) 0.627 0.928 0.198 Never treated 25 5–66 (26) 14–83 (60) 20–87 (68)

aExpressed as percentage of responding cells. Statistical analysis performed using Mann–Whitney U test. In vitro chemosensitivity of CLL to purine analogues TJ Bromidge et al 1233 Table 4 Clinical details of purine analogue treated patients

Patient Sub-type Stage Pre-treatment Time of No. cycles Response Follow-up test

1 atypical C none pre CdA 2 progression Deceased 3 m post treatment. Pneumonia 2 atypical A none pre CdA 9 PR 17 m post treatment, stable PR 3 typical C none pre CdA 3 PR 1 m post treatment progressed, died gastrointestinal bleed 4 typical C CAP, CBP, F pre CdA 3 PR Deceased 1 m post treatment. Pneumonia 5 typical C none pre CdA 6 CR 26 m post treatment, CR 6 typical A none pre CdA 8 PR 5 m post treatment, stable PR 7 typical B CdA pre F 6 PR 36 m post treatment, progressed 8 typical B CdA pre DCF/F 4 + 2 progression Deceased 1 m post treatment. Pneumonia 9 atypical B CBP pre CdA 6 CR 6 m post treatment progressed. Deceased 10 m. Pneumonia 10 typical B none pre CdA 8 PR 2 m post treatment, stable PR 11 typical C none pre CdA 10 PR 24 m post treatment, progressed 12 typical B CBP pre F 2 progressed Withdrawn due to concurrent medical problems 13 typical C CVP, CBP, P pre F 2 responding Continuing treatment

P, prednisolone; CAP, cyclophosphamide, adriamycin, prednisolone; CBP, chlorambucil, prednisolone; CVP, chlorambucil, vincristine, prednisolone; F, fludarabine; DCF, 2-deoxycoformycin; m, months. that report, however, may result in slight variations leading to forecast the clinical response in 10/12 patients. The assay misleading classification. The results presented here suggest failed in two patients (patients 8 and 12), predicting a that the two purine analogues can be used interchangeably response, when in both patients the disease actually pro- and that clinical considerations should carry more weight than gressed while on therapy, a major drawback as these are the preferential sensitivity. patients it would be most useful to identify. Patient 12, how- Study of the 13 CLL patients who received purine analogue ever, only received two courses of fludarabine before having therapy (patient details Table 4) following the in vitro assay to be withdrawn from treatment due to concurrent medical showed the assay to be a general guideline to clinical problems and so may have had an inadequate trial of response. Patient 13 had not completed treatment at the time treatment. of writing and hence is excluded from further consideration. These failings highlight the importance of monitoring The clinical and in vitro response of the remaining twelve defects in apoptosis controlling genes alongside the in vitro patients was compared (Figure 2). Culturing in the absence of assay. Mutations of p53 have been reported to be associated drug held no predictive value, however, when the responses with a poor response to purine analogue therapy in CLL.26 to addition of fludarabine and cladribine were considered and Western blot analyses suggested that patient 8 could not a cut off set at 20% of cells responding, the assay correctly induce expression of p53 on exposure to drug and that patient 12 expressed mutant p53 (data not shown). The TdT assay indicates the level of DNA strand breaks within a sample, if p53 is non-functional it is possible that cells could accumulate DNA strand breaks without being able to enter apoptosis27 and this would appear to explain the anomalous results in these two patients. The assay correctly predicted one patient resistant to therapy (patient 1) with less than 20% of cells showing a response to cladribine or fludarabine. Although the patient group is too small to draw conclusions from, it may be that the use of this assay alongside p53 assays could provide a helpful guideline.

Comparison of Bcl-2 protein level and in vitro response to drug

The level of bcl-2 protein expression did not vary significantly between the two subtypes of CLL (typical CLL range 117–382, Figure 2 Comparison of in vitro response and clinical response to median 226; atypical CLL range 131–400, median 248.5). The purine analogue therapy. For each patient first bar represents response level of bcl-2 in lymphocytes did not correlate with the in to culture in the absence of drug, second bar response to addition of vitro, spontaneous or drug-induced apoptosis in this group of 30 ␮m fludarabine and third bar response to addition of 17 ␮m cladri- bine. Clinical responses are coded as follows: clear, complete patients (Figure 3). remission; hatched bars, partial remission; black bars, progression of The lack of influence of bcl-2 levels on the in vitro response disease. Patient 13 was continuing therapy at the time of writing. of lymphocytes in this study supports the belief that the level In vitro chemosensitivity of CLL to purine analogues TJ Bromidge et al 1234 References 1 Johnson SA. Purine analogues in the management of lympholpro- liferative diseases. Clin Oncol 1996; 8: 289–296. 2 Gandhi V, Huang P, Plunkett W. Fludarabine inhibits DNA repli- cation: a rationale for its use in the treatment of acute leukemias. Leuk Lymphoma 1994; 14: 3–9. 3 Seto S, Carrera CJ, Kubota M, Wasson DB, Carson DA. Mechanism of deoxyadenosine and 2-chlorodeoxyadenosine toxicity to non- dividing human lymphocytes. J Clin Invest 1985; 75: 377–383. 4 Sandoval A, Consoli U, Plunkett W. Fludarabine-mediated inhi- bition of nucleotide excision repair induces apoptosis in quiescent human lymphocytes. Clin Cancer Res 1996; 2: 1731–1741. 5 Robertson LE, Plunkett W. Apoptotic cell death in chronic lympho- cytic leukemia. Leuk Lymphoma 1993; 11: 71–74. 6 Raimondo FD, Palumbo GA, Romeo MA, Cacciola E, Milone G, Impera S, Giustolisi R, Cacciola E. In vitro sensitivity of B-CLL cells to fludarabine and interferons. Leuk Lymphoma 1995; 17: 449–453. 7 Li X, Gong J, Feldman E, Seiter K, Traganos F, Darzynkiewicz Z. Apoptotic cell death during treatment of leukemias. Leuk Lym- phoma 1994; 13: 65–70. 8 Gorczyca W, Gong J, Darzynkiewicz Z. Detection of DNA strand breaks in individual apoptotic cells by the in situ terminal deoxyn- ucleotidyl transferase and nick translation assays. Cancer Res 1993; 53: 1945–1951. 9 Bosanquet AG, Bell PB. Novel ex vivo analysis of nonclassical, pleiotropic drug resistance and collateral sensitivity induced by therapy provides a rationale for treatment strategies in chronic lymphocytic leukemia. Blood 1996; 86: 1962–1971. 10 Silber R, Degar B, Costin D, Newcombe EW, Mani M, Rosenberg CR, Morse L, Drygas JC, Canellakis ZN, Potmesil M. Chemosensi- tivity of lymphocytes from patients with B-cell chronic lympho- cytic leukemia to chlorambucil, fludarabine, and camptothecin analogs. Blood 1994; 84: 3440–3446. 11 Juliusson G, Elmhorn-Rosenborg A, Liiemark J. Response to 2- chlorodeoxyadenosine in patients with B-cell chronic lymphocytic leukemia resistant to fludarabine. New Engl J Med 1992; 327: 1056–1061. 12 Delannoy A, Hanique G, Ferrant A. 2-Chlorodeoxyadenosine for patients with B-cell chronic lymphocytic leukemia resistant to fludarabine. New Engl J Med 1993; 328: 812. 13 Saven A, Lemon RH, Piro LD. 2-Chlorodeoxyadenosine for Figure 3 Correlation of level of bcl-2 in clonal lymphocytes with patients with B-cell chronic lymphocytic leukemia resistant to the in vitro (a) spontaneous apoptosis and (b) response to addition of fludarabine. New Engl J Med 1993; 328: 812–813. 30 ␮m fludarabine to the culture. 14 O’Brien S, Kantarjian H, Estey E, Koller C, Robertson B, Beran M, Andreef M, Pierce S, Keating M. Lack of effect of 2-chlorodeoxya- denosine therapy in patients with chronic lymphocytic leukemia of bcl-2 expression by itself is not a good predictive factor of refractory to fludarabine therapy. New Engl J Med 1994; 330: susceptibility to apoptosis and that level of bcl-2 should be 319–322. considered alongside those of bax. Although levels of bcl-2 15 Morabito F, Stelitano C, Callea I, Filangeri M, Oliva B, Sculli G, Callea V, Nobile F, Brugiatelli M. In vitro sensitivity of chronic alone have been reported to be inversely related with sur- lymphocytic leukemia B-cells to fludarabine, 2-chlorodeoxyad- 20 vival, drug resistance in B-CLL seems more dependent on enosine and chlorambucil: correlation with clinico-hematological the ratio of bcl-2:bax.28 Unintentional inclusion of NHL and immunophenotypic features. Haematologica 1996; 81: would bias results such as these. This study, as stated earlier, 224–231. includes only strictly classified CLL. We are now beginning 16 Howe DJ, Bromidge TJ, Stack-Dunne M, Davies SV, Paxton A, an investigation of bcl2:bax ratios in our current CLL patient Phillips MJ, Rule SJ, Johnson SA. Trisomy 12 is a rare event in cases of CLL with typical immunophenotype and morphology. population. Hematology 1997; 2: 373–378. In conclusion, although this in vitro TdT assay appears to 17 Oscier DG, Matutes E, Copplestone A, Pickering RM, Chapman give a guide to response to purine analogue therapy in most R, Gillingham R, Catovsky D, Hamblin TJ. Atypical lymphocyte patients it can be misleading in those with defects in p53 and morphology: an adverse prognostic factor for disease progression cannot identify at diagnosis those patients who will require in stage A CLL independent of trisomy 12. Br J Haematol 1997; treatment at a later date. These results highlight the impor- 98: 934–949. 18 Schena M, Larsson LG, Gottardi D, Gaidano GN, Carlsson M, tance of monitoring apoptosis controlling genes and the Nilsson K, Caligaris-Cappio P. Growth and differentiation-associa- search for new genetic markers of poor prognosis in CLL. ted expression of bcl-2 in B-chronic lymphocytic leukemia cells. Blood 1992; 79: 2981–2989. 19 Hanada H, Delia D, Aiello A, Stadtmauer E, Reed JC. Bcl-2 gene Acknowledgements hypomethylation and high-level expression in B-cell chronic lym- phocytic leukemia. Blood 1993; 82: 1820–1828. 20 Robertson LE, Plunkett W, McConnell K, Keating MJ, McDonnell We thank Dr Paul Ewings for help with statistical analyses. TJ. Bcl-2 expression in chronic lymphocytic leukemia and its This work was supported by the Musgrove Leukaemic correlation with the induction of apoptosis and clinical outcome. Group Somerset. Leukemia 1996; 10: 456–459. In vitro chemosensitivity of CLL to purine analogues TJ Bromidge et al 1235 21 Oltvai ZN, Milliman CL, Korsmeyer SJ. Bcl-2 heterodimerises in JF. Spontaneous programmed death (apoptosis) of B-chronic lym- vitro with a conserved homolog, bax, that accelerates programed phocytic leukaemia cells following their culture in vitro. Br J cell death. Cell 1993; 74: 609–619. Haematol 1989; 71: 343–350. 22 Binet JL, Auquier A, Dighiero G, Chastang C, Piguet H, Goasguen 26 Dohner H, Fischer K, Bentz M, Hansen K, Benner A, Cabot G, J, Vaugier G, Potron G, Colona P, Oberling F, Thomas M, Tchernia Diehl D, Schlenk R, Coy J, Stilgenbauer S, Volkman M, Galle PR, G, Jacquillat C, Boivin P, Lesty C, Dualt MT, Monconduit M, Poustka A, Hunstein W, Lichter P. p 53 gene deletion predicts for Belabbes S, Gremy F. A new prognostic classification of chronic poor survival and non-response to therapy with purine analogs in lymphocytic leukemia derived from a multi-variate survival analy- chronic B-cell leukemias. Blood 1995; 85: 1580–1589. sis. Cancer 1981; 48: 198–206. 27 Lowe SW, Ruley HE, Jacks T, Housman DE. p 53-dependent 23 Cheson BD, Bennett JM, Grever M, Kay N, Keating MJ, O’Brien apoptosis modulates the cytotoxicity of anticancer agents. Cell S, Rai KR. National Cancer Institute-sponsored working group 1993; 74: 957–967. guidelines for chronic lymphocytic leukemia: revised guidelines 28 Thomas A, El Rouby S, Reed JC, Krajewski S, Silber R, Potmesil for diagnosis and treatment. Blood 1996; 87: 4990–4997. M, Newcombe EW. Drug-induced apoptosis in B-cell chronic 24 Bromidge TJ, Howe DJ, Johnson SA, Phillips MJ. Adaptation of the lymphocytic leukemia: relationship between p 53 gene mutation TdT assay for semi-quantitative flow cytometric detection of DNA and bcl-2/bax proteins in drug resistance. Oncogene 1996; 12: strand breaks. Cytometry 1995; 20: 257–260. 1055–1062. 25 Collins RJ, Verschuer LA, Harmon BV, Prentice RL, Pope JH, Kerr