Post-induction residual leukemia in childhood acute lymphoblastic leukemia quantified by PCR correlates with in vitro prednisolone resistance

Transcription

1 (2001) 15, Nature Publishing Group All rights reserved /01 $ Post-induction residual leukemia in childhood acute lymphoblastic leukemia quantified by PCR correlates with in vitro prednisolone resistance K Schmiegelow 1, C Nyvold 2, J Seyfarth 2, R Pieters 3, MMA Rottier 4, N Knabe 1, LP Ryder 2, HO Madsen 2, A Svejgaard 2 and GJL Kaspers 4 Departments of 1 Pediatrics and of 2 Clinical Immunology, The National University Hospital, Rigshospitalet, Copenhagen, Denmark; 3 Sophia Children s University Hospital, Rotterdam, and 4 University Hospital Vrije Universiteit, Amsterdam, The Netherlands Most prognostic factors in childhood acute lymphoblastic leukemia (ALL) are informative for groups of patients, whereas new approaches are needed to predict the efficacy of chemotherapy for the individual patient. The residual leukemia following 4 weeks of induction therapy with prednisolone, vincristine, doxorubicin and i.t. methotrexate and the in vitro resistance to prednisolone, vincristine, and doxorubicin were measured in 30 boys and 12 girls with B (n = 34) or T lineage (n = 8) ALL. The residual leukemia was quantified after 2 (MRD-D15, n = 29) and 4 weeks (MRD-PI, n = 42) of induction therapy with a precise and reproducible clone-specific PCR technique. The median MRD-D15 and MRD-PI were 0.50% (75% range %) and 0.014% (75% range %), respectively, and these levels correlated significantly (n = 29, r S = 0.75, P 0.001). Both the MRD-D15 and the MRD-PI were related to the age of the patient (MRD-D15: r S = 0.48, P = 0.009; MRD-PI: r S = 0.45, P = 0.003). Patients with T lineage ALL had higher MRD- PI than those with B lineage ALL (median MRD-PI: 0.5% vs 0.01%, P = 0.05). The median LC 50 (concentration lethal to 50% of cells) for prednisolone was 2.3 g/ml (75% range ). Both MRD-D15 and MRD-PI correlated significantly with the in vitro resistance to prednisolone (MRD-D15: r S = 0.41, P = 0.03; MRD-PI: r S = 0.39, P = 0.01); but not to in vitro vincristine or doxorubicin resistance. The correlations between MRD and in vitro prednisolone resistance were even more pronounced when B cell precursor and T cell leukemia were analyzed separately (B cell precursor ALL: MRD-PI vs prednisolone LC 50 : n = 33, r S = 0.47, P = 0.006; T cell ALL: MRD-PI vs prednisolone resistance: n = 8, r S = 0.84, P = 0.009). After a median followup of 5.0 years (75% range ) eight patients have relapsed. All of the 21 patients with a MRD-PI 0.5% and a prednisolone LC g/ml have remained in remission whereas the 7 year event-free survival for the remaining 20 patients was 0.45 ± 0.16 (P = 0.002) Prospective studies in childhood ALL are needed to clarify whether combined monitoring of in vitro drug resistance and residual leukemia early during chemotherapy could offer new ways to classify patients and stratify the intensity of therapy. (2001) 15, Keywords: child; drug resistance; leukemia, lymphocytic, acute; polymerase chain reaction methods; prednisolone; residual leukemia Introduction The early reduction of the tumor burden is known to be of prognostic significance for childhood acute lymphoblastic leukemia (ALL). This includes the blast count in peripheral blood following 1 week of a prednisolone prephase, 1 and the percentage of leukemic blasts in day 8, 15, or 29 bone marrow In addition, the in vitro resistance to anticancer agents given during induction therapy has been shown to correlate well both with the response to a steroid prephase 6 and with the long-term event-free survival (EFS) in childhood ALL. 7 9 Correspondence: K Schmiegelow, Section of Clinical Hematology and Oncology, The Juliane Marie Centre, The Pediatric Clinic II 4064, The National University Hospital, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen, Denmark; Fax: Received 27 June 2000; accepted 19 January 2001 However, it is not known whether the latter correlation reflects (1) an early reduction in the leukemic tumor burden second to the steroid exposure or (2) a general propensity for apoptosis when exposed to anti-cancer agents. Recently, novel methods for precise determination of the residual leukemia down to the level of have been developed. These include quantification of leukemic blasts by flow cytometry analysis using the immunophenotype as a marker or by PCR using the CDRIII regions of the rearranged immunoglobulin heavy chain (IgH) and T cell receptor (TCR) genes as clonal markers These gene rearrangements, which take place in normal as well as malignant lymphoblasts, can be detected in more than 95% of children with ALL by PCR amplification using consensus primers covering the junctional regions of the IgH and TCR genes followed by DNA sequencing. 16 In the present study we have explored the correlation of the day 15 and the post-induction residual leukemia (MRD-D15, MRD-PI) with the in vitro resistance to prednisolone, vincristine, and doxorubicin. Patients and methods Patients Patients were eligible for the study if they (1) were diagnosed with non-b cell ALL at The National University Hospital, Rigshospitalet, Copenhagen, between June and March ; (2) were between 1 and 15 years of age at diagnosis; and (3) received induction therapy according to the common Nordic NOPHO ALL-92 protocol. 17 Of the 83 patients who fulfilled these criteria, bone marrow DNA from the time of diagnosis was not available for one patient, three patients did not have detectable clonal IgH or TCR rearrangements with the set of primers applied, 18 two children died during induction therapy, and for 21 patients bone marrow DNA was not available for quantification of MRD after 4 weeks of induction chemotherapy. Thus, 56 patients had MRD quantified following induction therapy. In vitro resistance testing was performed successfully in 42 of the 56 children for whom MRD-PI was quantified. The reasons for the remaining 14 patients not being included were lack of sufficient amount of leukemic cells (eight patients) or technical failures (six patients). Risk classification was determined by age and WBC at diagnosis (standard risk, SR, years and WBC /l; intermediate risk (IR) 2.0 years or 10 years and/or WBC /l; high risk, HR, WBC /l), and the presence of CNS or testicular leukemia, T cell disease, a mediastinal mass, lymphomatous leukemia, t(9;22), t(4;11), or a day 15 M3 or a day 29 M2 bone marrow (all HR criteria).

2 Patient characteristics The 30 boys and 12 girls included in the study did not differ significantly from the remaining 41 patients with respect to sex (girls/boys: 12/30 vs 16/25), WBC at diagnosis (median: 10 vs /l; P = 0.12 (Mann Whitney)), risk grouping (SR/IR/HR: 11/17/14 vs 12/15/14), or phenotype (B/T-lineage: 34/8 vs 33/8); but they were younger (median: (3.8 vs 6.5 years; P = 0.01 (Mann Whitney)). Chemotherapy All patients were treated according to the NOPHO ALL-92 protocol. 17 The first 4 weeks of induction therapy consisted of prednisolone (60 mg/m 2 /day), weekly vincristine (2.0 mg/m 2 ), doxorubicin (40 mg/m 2 days 1 and 22; + day 8 for HR-ALL), and i.t. methotrexate (days 1, 8, 15 and 29). In addition, the 14 patients with HR-ALL received a prednisolone prephase of 1 7 days (median: 4 days) prior to the co-administration of vincristine and doxorubicin. To evaluate the morphologic response to induction therapy bone marrow aspirates were to be taken 2 and 4 weeks from the first administration of doxorubicin ( day 15 and day 29 bone marrow). When calculated from the first administration of prednisolone, the day 15 bone marrow aspirates were taken after days (median: 14 days) and the day 29 bone marrow aspirates were taken after days (median: 28 days) of prednisolone therapy. All but two of the patients (one with an M1 bone marrow aspirate at day 15 ) had less than 5% lymphoblasts in the day 29 bone marrow aspirate. In vitro resistance The in vitro resistance assays were performed as previously described at the Research Laboratory of Pediatric Oncology of the University Hospital Vrije Universiteit, Amsterdam. 19 Briefly, mononuclear cells were seperated from freshly obtained bone marrow aspirates by density gradient centrifugation and washed twice. Eighty l l of /ml leukemic cell suspension were added to the wells of microculture plates containing 20 l of drug solutions. 19 Untreated control leukemic cells were set up in 8-fold. After 4 days of incubation, dimethylthiazol tetrazoliumbromide solution (MTT, Sigma, St Louis, MO, USA) was added to each well, and the purple blue formazan crystals formed by the reduction of MTT (by living cells only) were determined by automated reading of the optical density for each well. The optical density is linearly related to the number of viable cells. 20 The leukemic cell survival (resistance) in each well was calculated as the relation between the optical density in the drug-exposed well in relation to the drug-free (control) wells. 19 The LC 50, ie the drug concentration lethal to 50% of the leukemic cells, was calculated from the dose-response curves and used as a parameter for in vitro drug resistance. The results of the assay were regarded as reliable when at least 70% of the leukemic cells were present in the control wells after 4 days of incubation, and when the control optical density was higher than ,22 If too few cells were available the priority of the drugs to be tested was prednisolone, vincristine, and doxorubicin. In vitro resistance results were available for all three drugs for 36 patients, for prednisolone and vincristine in three patients, for prednisolone and doxorubicin in one patient, and for prednisolone only or doxorubicin only in the last two patients. Thus, in vitro resistance to prednisolone was tested successfully in 41 of the 42 patients. Six patients had drug resistance testing performed on stored, frozen bone marrow samples, which does not influence MTT assay results as reported earlier. 23 Clonal Ig and TCR gene rearrangements and quantification of residual leukemia Bone-marrow samples were obtained at diagnosis and after 2 (n = 29) and 4 (n = 42) weeks of induction therapy. Mononuclear cells were isolated by Ficoll density centrifugation (Lymphoprep; Pharmacia, Uppsala, Sweden). Quantification of residual leukemia was performed as described elsewhere. 18 Briefly, the rearranged genes of the immunoglobulin heavy chain (IgH), the T cell receptor gamma (TCR- ), and the incomplete V2D3 rearrangement of T cell receptor delta (TCR- ) were used as clonal markers. At diagnosis, PCRs were performed on genomic DNA using primers located in conserved domains on each side of the rearranged IgH, TCR-, and TCR- sites. The PCR products were sequenced and for 24 of the 42 patients it was possible to identify more than one clonal marker. The marker TCR- was preferentially used, then TCR-, and finally IgH for construction of a clone-specific primer (DNA Technology, Aarhus, Denmark). For each patient a clone-specific competitor was constructed by introducing the restriction site XhoI in a PCR product that was identical to parts of the highly specific rearranged IgH, TCR-, or TCR- genes of the malignant clone (Figure 1). Using primers located externally to the XhoI site the competitor and the DNA from the malignant clone will be amplified under identical conditions. To quantify the amount of MRD in a bone marrow sample a semi-nested PCR was performed where the clonespecific primer was reused and the consensus primer moved. After cleavage with XhoI, the PCR products originating from the competitor and the malignant clone could be separated by gel electrophoresis and the amount of MRD could be determined. The method has a detection limit of down to one malignant cell in 10 6 normal cells depending on the number of mononuclear cells analyzed. In patients for whom several clonal immune gene rearrangements could be identified the amount of MRD detected by the different clonal markers varied only two-fold or less. 18 Statistics The Mann Whitney U test, Kruskal Wallis test, chi-square test, Fischer s exact test, and Spearman s rank order correlation analyses were applied to compare distribution of parameters between subgroups and correlation between parameters (r S = correlation coefficient). 24 Backward linear regression analyses were performed to explore the best-fit model to predict the MRD-PI. Cox regression analysis was done to explore the best-fit model to predict the risk of relapse. 25 Parameters were exluded from the regression models at P values Logarithmic transformation of the MRD- PI and the prednisolone LC 50 values were performed prior to the multivariate linear regression analyses to obtain approximation to a normal distribution. The Kaplan Meier method was applied for estimation of remission duration and for the generation of survival curves. 26 Subgroups were compared with the log-rank test. 27 In all analyses two-sided P values 1067

3 were regarded as being significant. Data analyses were done with the SPSS statistical software. 28 Results Residual leukemia The median MRD-D15 was 0.50% (n = 29, 75% range: %) and the median MRD-PI was 0.014% (n = 42, 75% range: %) and these were significantly correlated (n = 29, r S = 0.75 P 0.001) (Figure 1), being somewhat stronger for T cell ALL (n = 5, r S = 0.90, P = 0.04) than for B cell precursor ALL (n = 24, r S = 0.63, P = 0.001). The MRD- D15 and MRD-PI were also correlated, if only those with detectable leukemia were included (r S = 0.73, P 0.001). Neither for B lineage nor for T cell ALL was the MRD inversely related to the interval between the start of therapy (first day of doxorubicin/vincristine) and the day of MRD-PI measurement. In fact for both lineages, patients with longer intervals weakly tended to have higher MRD levels (B lineage: r S = 0.14; T lineage: r S = 0.27). If a logarithmic fall in the residual leukemia was assumed and the expected MRD at day 29 was calculated from the start of doxorubicin and vincristine adjusting for the differences in day 29 bone marrow sampling (since not all samples were taken exactly on day 29 from the first day of doxorubicin/vincristine), the adjusted residual leukemia and the unadjusted MRD-PI at day 29 were highly correlated (r S = 0.97, P 0.001). Neither the MRD-D15 nor the MRD-PI were significantly related to gender (male/female: median MRD-PI = 0.014/0.015), WBC at diagnosis (r S = 0.17, P = 0.29), or risk group (median MRD-PI; SR: 0.02%, IR: 0.01%, B lineage HR- ALL: 0.05%; P = 0.72). Patients with T cell ALL had higher MRD-PI than did patients with B lineage ALL (median MRD- PI: 0.5% vs 0.01%, P = 0.05). Both the MRD-D15 and the MRD-PI were significantly related to the age of the patient (MRD-D15: r S = 0.48, P = 0.009; MRD-PI: r S = 0.45, P = 0.003). This was also the case if B lineage ALL was analyzed seperately (MRD-D15: r S = 0.57, P = 0.004; MRD-PI: r S = 0.48, P = 0.004). Figure 2 Post-induction residual leukemia in relation to the in vitro resistance to prednisolone (LC 50 ). B lineage:, T lineage:. Open markers indicate that no leukemic cells could be detected. The level was set to be equal to the reciprocal value of the number of cells analyzed. In vitro resistance The median LC 50 for prednisolone was 2.3 g/ml (75% range ). The in vitro resistance to prednisolone was related to the resistance to vincristine (n = 39, r S = 0.29, P = 0.08) and to doxorubicin (n = 37, r S = 0.50, P = 0.002). In addition, the vincristine and doxorubicin resistance were correlated (n = 36, r S = 0.51, P = 0.001). Both MRD-D15 and MRD-PI were related to in vitro resistance to prednisolone (MRD-D15: r S = 0.41, P = 0.03; MRD-PI: r S = 0.39, P = 0.01). The correlations between MRD-PI and in vitro prednisolone resistance were even more pronounced when B cell precursor and T cell leukemia were analyzed separately: B cell precursor ALL: MRD- D15 vs prednisolone resistance: n = 23, r S = 0.47, P = 0.02; MRD-PI vs prednisolone resistance: n = 33, r S = 0.47, P = 0.006; T cell ALL: MRD-D15 vs prednisolone resistance: n = 5, r S = 0.87, P = 0.05; MRD-PI vs prednisolone resistance: n = 8, r S = 0.84, P = (Figure 2). In contrast, neither the day 15 nor the day 29 bone marrow MRD were significantly related to the in vitro vincristine or doxorubicin resistance (Table 1). In a backward multivariate regression analysis estimating the impact of age and white cell count at diagnosis, B lineage (=1) vs T lineage disease (=2), and prednisolone LC 50 Table 1 Correlations between residual leukemia and in vitro drug resistance Day 15 residual leukemia Day 29 residual leukemia n r s n r s Figure 1 The residual leukemia (measured by PCR) in bone marrow aspirates taken at treatment days 15 and 29. If no residual leukemia (RL) could be detected the level was set to be equal to the reciprocal value of the number of cells analyzed. : RL detectable at both time points; : RL not detectable at day 15 ; : RL neither detectable at day 15 nor at day 29. Prednisolone * *** B lineage ** *** T lineage * *** Vincristine Doxorubicin Correlations between residual leukemia measured by PCR at day 15 or day 29 and in vitro resistance to prednisolone, vincristine, and doxorubicin. P values 0.10, 0.05, and 0.01 are marked as *, ** and ***.

4 (logarithmically transformed to approximate a normal distribution) on the MRD-PI (logarithmically transformed), only the immunophenotype (P = 0.001) and the prednisolone LC 50 (P = 0.003) were included in the final model. If B lineage disease was analyzed separately only the prednisolone LC 50 was included in the final model (P = 0.03). Relapse For patients still in remission the median length of follow-up from diagnosis was 5.0 years (75% range: y) at the end of follow-up (31 August 2000). The overall EFS was 0.80 ± 0.06 at 5 years and 0.70 ± 0.11 at 7 years, which is comparable to the event-free survival for all children years with ALL in the Nordic countries (0.77 at 8 years). 17 Eight patients developed a relapse, and these all had a MRD-PI 0.014% (median: 0.45%) and a prednisolone LC g/ml (median: 24.1 g/ml) (Figure 2). In addition, all eight patients who relapsed had either a MRD-PI 0.5% or a prednisolone LC g/ml. Half of the study population had a MRD-PI 0.5% and a prednisolone LC g/ml and these 21 patients had a 7 year event-free survival of 1.0 compared to only 0.45 ± 0.16 for the remaining 20 patients (P = 0.002) (Figure 3). In a backward multivariate Cox regression analysis estimating the impact of age and white cell count at diagnosis, B lineage (=1) vs T lineage disease (=2), prednisolone LC 50 (logarithmically transformed), and the MRD-PI (logarithmically transformed) on the risk of relapse, only MRD-PI (P = 0.03) and the white cell count at diagnosis (P = 0.07) were included in the final model. The prednisolone LC 50 was excluded from the model by a P value of Discussion Since the mid 1970s it has been known that the response to the first 4 weeks of induction therapy is of long-term prognostic value in childhood ALL. However, although cytomorphological examination of a bone marrow aspirate by day 15 or 29 reflects the treatment response, it is of minor clinical importance in the prediction of the majority of relapses, since only a few of the patients who will experience recurrence of Figure 3 The event-free survival curves from the time of diagnosis for the 20 patients with residual leukemia 0.5% or a prednisolone LC g/ml (lower curve) vs the remaining 21 patients (upper curve) (P = 0.002). Each patient at risk is marked by vertical lines. their disease will be identified. Thus, new approaches are needed to identify the children who may do equally well with less intensive post-induction chemotherapy and the children who need more intensive or different treatment. In this context the present study although being rather small carries several important findings. First, although all but one of the patients were in morphologic remission at day 29 there was a more than 1000-fold variation in the residual burden of the leukemic clone. This is in accordance with previous findings and emphasizes, that the term remission should be interpreted in the context of the sensitivity of the techniques applied It is noteworthy that almost half of the patients had a MRD-PI below 0.01% and one fifth of the children 0.001% after just 4 weeks of chemotherapy and that none of these patients relapsed. It is possible that these very rapid responders may include the subset of patients who participated in trials in the 1960s and early 1970s and were cured by treatments that were shorter and less intensive compared to most contemporary protocols. 29,30 These results are in accordance with those of a recent International BFM study which demonstrated that patients without detectable disease by PCR after 4 weeks of therapy (which equals leukemic cells) have a 5-year relapse rate of only 2%. 14 Similarly, the EORTC and the St Jude Childrens Research Hospital study groups have found that patients who after induction therapy are without detectable disease by PCR or by flow cytometry have an excellent outcome. 13,15 Since a level of residual leukemia at 10 4 is both the median of all patients and is the upper cut-off limit for patients with an excellent prognosis, stratification of the post-remission therapy by this level of MRD would require a very reproducible and precise PCR method to correctly classify the patients. The competitive PCR-technique applied in the present study could be one approach to achieve this. 18 Second, the significant correlation between the MRD-D15 and the MRD-PI (r S = 0.75) in the present study could indicate that the disappearance of the leukemic clone follows a kinetic pattern characteristic for the individual clone. At day 15 most patients have only a few percent or less blasts in the bone marrow and at day 29 half of all patients have 1 leukemic blast in mononuclear bone marrow cells. Whether early morphological classification of the response to therapy such as a day 8 morphological blast count will identify the same good and poor responders as a post-induction quantification with the competitive PCR is explored in an on-going Nordic multicenter study which compares the results on bone marrow samples of (1) histomorphology days 8, 15 and 29, (2) determination of residual leukemia at days 29 and 106 by flow cytometric quantification of the leukemic phenotype (when feasible), or by (3) the competitive PCR of the clonespecific immune gene rearrangements. Such data would be very important in finding the clinically most informative and practical approach to early identification of those patients who could do with less intensive chemotherapy as well as those who would need to be switched to a more intensive chemotherapy program. Third, patients with B lineage HR-ALL treated by the NOPHO ALL-92 protocol have a significantly higher relapse rate than patients with IR-ALL and SR-ALL. 31 However, patients with B lineage SR-, IR-, HR-ALL did not differ significantly in their MRD-PI, and HR patients had slightly higher levels even though they had received one extra dose of doxorubicin on day 8. This indicates that the subset of patients with a MRD-PI 10 4 or even 10 3 includes both patients with resistant disease who are at high risk for a subsequent 1069

5 1070 relapse and patients with a disease that responds slowly to chemotherapy but who have a good long-term outcome. For this reason treatment stratification by PCR quantification of MRD should probably include more than one time point as has been indicated by the data published by the International BFM study group, 14 since an early time point would identify the rapid responders with a good outcome, and later time points would distinguish between the patients with a more slowly disappearing disease and an intermediate outcome (low MRD levels) and the patients with resistant disease still having measurable and significant MRD and a very poor outcome. Finally, this is the first study that correlates in vitro drug resistance with the early response to chemotherapy as determined by a sensitive PCR technique. The correlations between in vitro prednisolone resistance and both the MRD-D15 and the MRD-PI indicate that the long-term prognostic significance of in vitro resistance to the drugs applied during induction chemotherapy at least partly reflect the actual early reduction of the tumor burden. 8 The previously published correlation between in vitro prednisolone resistance and the clinical response (determined by the blast count in peripheral blood) to a prednisolone prephase supports this. 6 Studies on the correlation between in vitro resistance to prednisolone and longterm prognosis have demonstrated that the patients most sensitive to prednisolone (LC g/ml) have an eventfree survival rate of more than In the present study all these patients also had MRD-PI 0.01% and none of these patients have so far relapsed. All of the eight patients in the present study who relapsed had either an MRD-PI 0.5% or a prednisolone LC g/ml. Although the numbers are small (and thus also the power of the statistical analyses), this is the first study that indicates an advantage of combined analyses of in vitro drug resistance and MRD quantification by PCR. An MRD level of 0.5% is 50 times higher than the average levels in children with ALL at the end of induction following treatment according to the NOPHO ALL-92 protocol. If this critical level (when combined with in vitro resistance assays) is confirmed in larger studies, very sensitive assays for MRD quantification may be unnecessary since MRD levels of 0.5% or higher can be identified using consensus primers located in conserved domains on each side of the immune genes followed by direct sequencing. In two large ongoing studies by the Nordic countries and the Dutch Childhood Study Group the correlation between in vitro drug resistance, residual leukemia by morphology evaluation days 15 and 29, and MRD quantification by both flow cytometry and PCR methods will be explored. In conclusion, the present study indicates that the in vitro resistance to prednisolone does not merely reflect the longterm prognosis in childhood ALL but is also a predictor of the early response to therapy as determined by sensitive PCR techniques. Ongoing and future clinical trials will determine the role of in vitro resistance assays and MRD quantification in combination with other factors to predict outcome. Acknowledgements The study has received financial support from The Danish Cancer Society (grant No and ), The Emil and Inger Hertz Foundation, The Kornerup Foundation, The Medical Research Council, Denmark (grant No ), The Queen Louise s Children s Hospital Foundation. References 1 Reiter A, Schrappe M, Ludwig WD, Hiddemann W, Sauter S, Henze G, Zimmermann M, Lampert F, Havers W, Niethammer D. Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients. Results and conclusions of the multicenter trial ALL-BFM 86. Blood 1994; 84: Miller DR, Coccia PF, Bleyer WA, Lukens JN, Siegel SE, Sather HN, Hammond GD. Early response to induction therapy as a predictor of disease-free survival and late recurrence of childhood acute lymphoblastic leukemia: a report from the Childrens Cancer Study Group. J Clin Oncol 1989; 7: Gaynon PS, Bleyer WA, Steinherz PG, Finklestein JZ, Littman P, Miller DR, Reaman G, Sather H, Hammond GD. Day 7 marrow response and outcome for children with acute lymphoblastic leukemia and unfavorable presenting features. Med Pediatr Oncol 1990; 18: Lilleyman JS, Gibson BE, Stevens RF, Will AM, Hann IM, Richards SM, Hill FG. Clearance of marrow infiltration after 1 week of therapy for childhood lymphoblastic leukaemia: clinical importance and the effect of daunorubicin. The Medical Research Council s Working Party on Childhood Leukaemia. Br J Haematol 1997; 97: Gaynon PS, Desai AA, Bostrom BC, Hutchinson RJ, Lange BJ, Nachman JB, Reaman GH, Sather HN, Steinherz PG, Trigg ME, Tubergen DG, Uckun FM. Early response to therapy and outcome in childhood acute lymphoblastic leukemia: a review. Cancer 1997; 80: Kaspers GJ, Pieters R, Van Zantwijk CH, van Wering ER, Van Der Does-Van Den Berg A, Veerman AJ. Prednisolone resistance in childhood acute lymphoblastic leukemia: vitro vivo correlations and cross-resistance to other drugs. Blood 1998; 92: Pieters R, Klumper E, Kaspers GJ, Veerman AJ. Everything you always wanted to know about cellular drug resistance in childhood acute lymphoblastic leukemia. Crit Rev Oncol Hematol 1997; 25: Kaspers GJ, Veerman AJ, Pieters R, Van Zantwijk CH, Smets LA, van Wering ER, Van Der Does-Van Den Berg A. In vitro cellular drug resistance and prognosis in newly diagnosed childhood acute lymphoblastic leukemia. Blood 1997; 90: Hongo T, Yajima S, Sakurai M, Horikoshi Y, Hanada R. In vitro drug sensitivity testing can predict induction failure and early relapse of childhood acute lymphoblastic leukemia. Blood 1997; 89: Wasserman R, Galili N, Ito Y, Silber JH, Reichard BA, Shane S, Womer RB, Lange B, Rovera G. Residual disease at the end of induction therapy as a predictor of relapse during therapy in childhood B-lineage acute lymphoblastic leukemia. J Clin Oncol 1992; 10: Brisco MJ, Condon J, Hughes E, Neoh SH, Sykes PJ, Seshadri R, Toogood I, Waters K, Tauro G, Ekert H. Outcome prediction in childhood acute lymphoblastic leukaemia by molecular quantification of residual disease at the end of induction. Lancet 1994; 343: Cave H, Guidal C, Rohrlich P, Delfau MH, Broyart A, Lescoeur B, Rahimy C, Fenneteau O, Monplaisir N, d Auriol L. Prospective monitoring and quantitation of residual blasts in childhood acute lymphoblastic leukemia by polymerase chain reaction study of delta and gamma T cell receptor genes. Blood 1994; 83: Cave H, van der Werff ten Bosch, Suciu S, Guidal C, Waterkeyn C, Otten J, Bakkus M, Thielemans K, Grandchamp B, Vilmer E. Clinical significance of minimal residual disease in childhood acute lymphoblastic leukemia. European Organization for Research and Treatment of Cancer Childhood Cooperative Group. N Engl J Med 1998; 339: van Dongen JJ, Seriu T, Panzer-Grumayer ER, Biondi A, Pongers- Willemse MJ, Corral L, Stolz F, Schrappe M, Masera G, Kamps WA, Gadner H, van Wering ER, Ludwig WD, Basso G, De Bruijn MA, Cazzaniga G, Hettinger K, Van Der Does-Van Den Berg A, Hop WC, Riehm H, Bartram CR. Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. Lancet 1998; 352: Coustan-Smith E, Behm FG, Sanchez J, Boyett JM, Hancock ML, Raimondi SC, Rubnitz JE, Rivera GK, Sandlund JT, Pui CH, Campana D. Immunological detection of minimal residual disease in

6 children with acute lymphoblastic leukaemia. Lancet 1998; 351: Jonsson OG, Kitchens RL, Scott FC, Smith RG. Detection of minimal residual disease in acute lymphoblastic leukemia using immunoglobulin hypervariable region specific oligonucleotide probes. Blood 1990; 76: Gustafsson G, Schmiegelow K, Forestier E, Clausen N, Glomstein A, Jonmundsson G, Mellander L, Mäkkiperna A, Nygard R, Saarinen-Pihkala UM. Improving outcome through two decades in childhood ALL in the Nordic countries: the impact of high-dose methotrexate in the reduction of CNS irradiation. 2000; 14: Nyvold C, Madsen HO, Ryder LP, Seyfarth J, Engel CA, Svejgaard A, Wesenberg F, Schmiegelow K. Competitive PCR for quantification of minimal residual disease in acute lymphoblastic leukaemia. J Immunol Meth 2000; 233: Kaspers GJ, Pieters R, Van Zantwijk CH, De Laat PA, de Waal FC, van Wering ER, Veerman AJ. In vitro drug sensitivity of normal peripheral blood lymphocytes and childhood leukaemic cells from bone marrow and peripheral blood. Br J Cancer 1991; 64: Pieters R, Huismans DR, Leyva A, Veerman AJ. Adaptation of the rapid automated tetrazolium dye based (MTT) assay for chemosensitivity testing in childhood leukemia. Cancer Lett 1988; 41: Pieters R, Loonen AH, Huismans DR, Broekema GJ, Dirven MW, Heyenbrok MW, Hahlen K, Veerman AJ. In vitro drug sensitivity of cells from children with leukemia using the MTT assay with improved culture conditions. Blood 1990; 76: Kaspers GJ, Veerman AJ, Pieters R, Broekema GJ, Huismans DR, Kazemier KM, Loonen AH, Rottier MA, Van Zantwijk CH, Hahlen K. Mononuclear cells contaminating acute lymphoblastic leu- kaemic samples tested for cellular drug resistance using the methyl-thiazol-tetrazolium assay. Br J Cancer 1994; 70: Pieters R, Huismans DR, Leyva A, Veerman AJ. Comparison of the rapid automated MTT-assay with a dye exclusion assay for chemosensitivity testing in childhood leukaemia. Br J Cancer 1989; 59: Siegel S, Castellan NJ. Nonparametric Statistics for the Behavioral Sciences. McGraw Hill: Singapore, Cox DR. Regression models and life-tables (with discussion). JR Stat Soc (B) 1972; 34: Kaplan EJ, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother 1966; 50: SPSS statistical software for Windows release SPSS Inc Lonsdale D, Gehan EA, Fernbach DJ, Sullivan MP, Lane DM, Ragab AH. Interrupted vs. continued maintenance therapy in childhood acute leukemia. Cancer 1975; 36: Rivera GK, Pinkel D, Simone JV, Hancock ML, Crist WM. Treatment of acute lymphoblastic leukemia. 30 years experience at St Jude Children s Research Hospital. N Engl J Med 1993; 329: Gustafsson G, Kreuger A, Clausen N, Garwicz S, Kristinsson J, Lie SO, Moe PJ, Perkkio M, Yssing M, Saarinen-Pihkala UM. Intensified treatment of acute childhood lymphoblastic leukaemia has improved prognosis, especially in non-high-risk patients: the Nordic experience of 2648 patients diagnosed between 1981 and Nordic Society of Paediatric Haematology and Oncology (NOPHO). Acta Paediatr 1998; 87: