Minimal residual disease prior to stem cell transplant for childhood acute lymphoblastic leukaemia

British Journal of Haematology, 2003, 122, 24–29

Minimal residual disease prior to stem cell transplant for childhood acute lymphoblastic leukaemia Nick Goulden, 1 Peter Bader, 2 Vincent Van der Velden, 3 John Moppett, 1 Marco Schilham, 4 Hans O. Masden, 5 Ondrej Krejci, 6 Hermann Kreyenberg, 2 Arjan Lankester, 4 Tom Re´ ve´ sz, 7 Thomas Klingebiel 8 and Jacques Van Dongen 3 1Royal Bristol Hospital for Sick Children, Bristol, UK, 2 Universita¨tsKinderklinik, Tu¨bingen, Germany, 3Erasmus MC, University Medical Centre Rotterdam, 4Leiden University Medical Centre, Leiden, the Netherlands, 5Rigshospitalet, Copenhagen, Denmark, 6University Hospital Motol, Prague, Czech Republic, 7University Medical Centre, Utrecht, the Netherlands, and 8Wolfgang Goethe Universita¨t, Frankfurt, Germany Received 26 November 2002; accepted for publication 23 January 2003

Summary. Allogeneic stem cell transplantation (SCT) is a highly effective therapy for childhood acute lymphoblastic leukaemia (ALL). Concerns about unnecessary toxicity and expense mean that SCT is currently largely reserved for children who cannot be cured with chemotherapy. Not surprisingly, many such children also fail SCT. Retrospective studies have shown that a single analysis of minimal residual disease (MRD) pre-SCT identified those at highest

The toxicity and expense of allogeneic stem cell transplantation (SCT) is such that in childhood acute lymphoblastic leukaemia (ALL) it is reserved for those deemed to be at an unacceptably high risk of recurrence after chemotherapy alone. Nevertheless, relapse remains the commonest cause of death after transplantation (Chessells, 1998, 2000). Since 1998, reports from three European centres have shown that minimal residual disease (MRD) burden prior to conditioning is the strongest single predictive factor for relapse post SCT (Knechtli et al, 1998; van der Velden et al, 2001; Bader et al, 2002). Recognition of the potential clinical impact of this test has led to the formation of a European Study Group on MRD detection in SCT for ALL. This group has now standardized MRD methods and set up a quality assurance network as the technical basis for collaborative clinical protocols. Consequently it is now appropriate to call for the introduction of standardized MRD measurement immediately prior to conditioning to be included in all SCT protocols in childhood ALL. Results may be used to evaluate different transplant protocols as Correspondence: Dr Nick Goulden, Department of Paediatric Haematology and Oncology, Bristol Royal Hospital for Children, Upper Maudlin Street, Bristol, BS2 8BJ, UK. E-mail: nick.goulden@ubht. swest.nhs.uk

24

risk of relapse. It is now appropriate to call for the universal incorporation of standardized MRD testing into SCT protocols as the next step to maximize the clinical impact of this technology in ALL. Keywords: MRD, BMT, childhood ALL, relapse, Ig/TCR gene rearrangement PCR.

well as to develop novel regimes within groups of children receiving homogeneous therapy. CURRENT INDICATIONS AND RESULTS OF SCT FOR ALL Advances in graft processing, tissue typing and supportive care over the last decade are such that it is now possible to offer SCT to almost all children (Heslop, 1999; Handgretinger et al, 2001). However, in each case the decision to perform an allograft reflects a balance between the biology of the disease and the risk of transplant-related death and long-term toxicity. At present, information about the biology of the disease is based on simple readily available prognostic factors, including white cell count, cytogenetics and morphological assessment of response to therapy. The risk of transplantrelated death is a function of human leucocyte antigen (HLA) disparity between donor and recipient, performance and remission status. Although the precise indications for SCT in first remission vary among the major European treatment co-operatives, the procedure is normally reserved for those with very highrisk cytogenetics and a poor early response to therapy as exemplified by the current Berlin–Frankfurt–Munster (BFM) approach (see Table I).
2003 Blackwell Publishing Ltd

Minimal Residual Disease Prior to SCT for ALL

25

Table I. Current indications for SCT in CR1 of ALL in childhood, as determined by the BFM group.

Type of SCT Indication

Criteria

MFD

MUD

MMFD

Remission failure

Remission reached at a later stage

+

+

+

Poor prednisone response

+ + + + + +

+ + + + + (+)

(+)* (+) (+) (+) + (+)

+ (+)

+ (+)

(+) (+)

Good prednisone response

T-ALL pro-B-ALL M3 marrow on d 15 WBC > 100 · 109/l t(9;22) t(4;11)

+ t(9;22) + t(4;11)

*T-ALL patients with a poor prednisone response, plus MRD positivity at time point 2 (irrespective of MRD level) are candidates for an SCT with MFD or MUD. (+)In these situations there is insufficient evidence to support the use of SCT. The decision whether to transplant remains a joint one between the treating physicians and the transplant centre. MFD, matched family donor; MUD, matched unrelated donor, MMFD, mismatched family donor.

children with risk S3/S4 treated with chemotherapy alone in the absence of an appropriate donor, whereas 6-year event-free survival (EFS) in 65 children undergoing SCT was 43% (G. Henze, personal communication). Thus, SCT from any available donor is recommended for these patients. The second conclusion to be drawn from the BFM and MRC data is that the outlook for children suffering an isolated extramedullary relapse more than 6 months from the end of therapy (BFM S1) and treated with local radiation and chemotherapy alone is so good (77% 5-year EFS in MRC UKALL R1) that SCT is not indicated (Wheeler et al, 1998). Finally, the role of SCT in children who suffer a relapse in the marrow more than 6 months from the end of therapy (i.e. most patients in BFM S2) remains unclear. In

There is consensus among the treatment co-operatives that prognosis after relapse is dependent on the duration of first remission and the site of relapse (reviewed in Chessells, 1998). These factors are incorporated into the current BFM definition of risk in relapsed ALL, which has recently been adopted by the Medical Research Council (MRC) in the UK (Table II). The comparison of outcome in children treated either with chemotherapy alone or SCT in the BFM or MRC studies leads to three main conclusions. First, the prognosis for children relapsing in the bone marrow within 6 months of the end of therapy (included in the trial for relapsed ALL patients with a risk stratification of S3/4, ALL-BFM REZ ‘S3/4’) is very poor unless they receive SCT. In the BFM REZ 95/96 trial, there were no survivors in a group of 31 Table II. BFM and MRC adapted risk groups in relapsed ALL.

Non–T

Pre-T

Extramedullary

Combined

Marrow

Extramedullary

Combined

Marrow

Very Early Diag < 18 m Treat < 6 m

S2 Int

S4 High

S4 High

S2 Int

S4 High

S4 High

Early Diag ‡ 18 m Treat < 6 m

S2 Int

S2 Int

S3 High

S2 Int

S4 High

S4 High

Late Treat ‡ 6 m

S1 Std

S2 Int

S2 Int

S1 Std

S4 High

S4 High

S1, S2, S3, S4 refer to BFM risk groups. Standard (Std), intermediate (Int) and high refer to the adapted version as used by the MRC. BFM definitions of relapse timing are as follows: very early relapse: < 18 months after diagnosis (Diag) and < 6 months after completing therapy (Treat). This definition refers to the rare case, where a patient has had a very short front-line therapy (due to interruption of therapy or treatment according to B-cell non-Hodgkin’s lymphoma protocols). Early relapse: > 18 months after diagnosis but < 6 months after stopping therapy. Late relapse: > 6 months after stopping therapy.
2003 Blackwell Publishing Ltd, British Journal of Haematology 122: 24–29

26

N. Goulden et al

the MRC UKALL R1 study, more than 40% of such children are in long-term remission without SCT. Moreover, others have shown that treatment with chemotherapy followed by the salvage of therapy failures with SCT in third complete remission (CR3) maybe appropriate (Borgmann et al, 1997). On the basis of the above, many clinicians believe that SCT is superior to chemotherapy for a minority of children with very high-risk disease. Moreover, the gradual fall in transplant-related mortality might reveal a greater advantage for SCT in some subgroups. However, it is obvious that SCT is not a panacea. For example in the MRC UKALL R1 study, more than 80% of children receiving an allograft, for ALL relapsing in the marrow on therapy, relapsed again after SCT, suggesting that these children need different therapy. Equal concern must be attached to the fact that even in groups of patients where the benefit of SCT is considered to be clearly established (BFM S3) the risk of relapse is 50% (Seeger et al, 1998). Finally, data from the BFM and the MRC confirm that, at least in B-cell precursor ALL, as many as 40% of children who relapse in the marrow more than 6 months from the end of treatment can be cured without SCT (Wheeler et al, 1998; Eckert et al, 2001). As in front-line chemotherapy, there is now good evidence that MRD provides independent prognostic information within clinically homogeneous groups of children undergoing SCT. IMPACT OF MRD BURDEN PRE-SCT ON RISK OF RELAPSE Three published studies have examined the correlation between MRD burden prior to conditioning and relapse after SCT in more than 100 children transplanted for ALL in remission (Knechtli et al, 1998; Van der Velden et al, 2001; Bader et al, 2002). Knechtli et al (1998) used polymerase chain reaction (PCR) analysis of antigen receptor genes and allele-specific probing (a technique capable of detection of

one leukaemic cell in 10 000 normal cells) to estimate MRD a median of 23 d pre-BMT in 64 children (19 CR1, 39 CR2, two CR3 or beyond) transplanted in a single centre predominantly from unrelated donors. The level of MRD was defined as either high positive [presence of a band on polyacrylamide gel electrophoresis (PAGE)], low (negative by PAGE but positive on radio-labelled probing) or negative. Five children died of transplant-related complications and were excluded from the analysis of correlation between preconditioning MRD and relapse. Thus, the 5-year EFS was 0%, 36% and 73% for high-level MRD positive (12 relapses, 12 patients), low-level MRD positive (five relapses, nine patients) and negative (eight relapses, 38 patients) respectively (see Fig 1A). Bader et al (2002) used the same MRD method to examine MRD pre-BMT in 41 children (seven CR1, 29 CR2, five CR3 or beyond) in Tuebingen. The majority of these patients (28 of 41) received unmanipulated grafts, although recipients of unrelated transplants did receive serotherapy. Three children died from transplant-related complications (one MRD negative, two high-level positive). Thus the correlation between MRD prior to conditioning and relapse could be discerned in 38 children. Five-year EFS was 23%, 48% and 78% for high-level MRD positive (11 relapses, 15 patients), low-level MRD positive (five relapses, 10 patients) and negative (two relapses, 13 patients) groups respectively (see Fig 1B). Finally a Dutch study (van der Velden et al, 2001) reported the results of real-time quantitative PCR (RQ-PCR) of antigen receptor gene rearrangements for pre-BMT MRD analysis for 17 patients (nine CR1, eight CR2) a median of 20 d (range 13–98 d) prior to conditioning. A mixture of T-replete and T cell-depleted transplants was performed. Eleven patients were found to be MRD negative (< 10)4) and six were MRD positive. EFS for the MRD-negative group was 80%, while that for the MRD-positive group was 33%. The two survivors in the MRD-positive group had disease below a level of 5 · 10)4.

Fig 1. Kaplan–Meier plots comparing event-free survival of patients with high-level positive, low-level positive and MRD-negative results preBMT, using data from (A) Knechtli et al (1998) and (B) Bader et al (2002).
2003 Blackwell Publishing Ltd, British Journal of Haematology 122: 24–29

Minimal Residual Disease Prior to SCT for ALL Each of these studies has shown MRD burden pre-SCT to be an independent prognostic factor in prediction of relapse. Analysis identified MRD to be an independent prognostic factor in multivariate analysis in Bader et al (2002) and Knechtli et al (1998) (P < 0Æ01). Although formal metaanalysis is invalidated by differences in transplant protocol and MRD method, some clinically useful conclusions can be inferred from pooling the results of these studies. First, it is clear that EFS for the MRD-negative groups is uniformly good. Second, in the MRC and Dutch studies, SCT in children with high-level disease appears futile, whereas the BFM results suggest a minority of such children can be cured. It is intriguing to speculate whether differences in transplant protocol may explain this difference in outcome. A third observation common to these studies is the trend towards a high MRD level in children who relapse in the marrow after a short first remission, indicating that relative insensitivity to therapy underlies the high risk of relapse after SCT in such patients (Knechtli et al, 1998). DEVELOPMENT OF A STANDARDIZED METHOD FOR THE DETECTION OF MRD Incorporation of risk group assignment according to MRD levels, based on the preliminary findings discussed above, into large-scale clinical studies requires the development of uniform methods for the measurement of MRD. The method should be widely applicable, have a minimal risk of falsenegative results and be uniformly quantitative. Currently, three main methods are available for the measurement of MRD in childhood ALL: flow cytometric immunophenotyping using tumour-associated aberrant immunophenotypes, PCR of antigen receptor gene rearrangements and PCR of tumour-specific translocations. PCR analysis of antigen receptor gene rearrangements is applicable in virtually all childhood ALL patients and can reach sensitivities of at least 10)4 in the vast majority of patients. Consequently, the European Study Group (ESG) for MRD analysis in SCT for ALL has adopted this approach. It should be noted that the timing and sensitivity of MRD assays is critical to their use and interpretation. The prognostic value of any MRD result is dependant upon the treatment received both prior to and following the assay. Variations in the timing and threshold of detection can lead to significant changes in risk groupings (van Dongen et al, 1 1998; zur Stadt et al, 2001; Nyvold et al, 2002). The children with the best prognosis can only be identified preSCT if they are MRD negative using assays sensitive to at least 10)4. Those with the very worst prognosis can be identified using less sensitive assays (and simpler techniques) (Cave´ et al, 1998; Knechtli et al, 1998). The ESG MRD laboratories now use standardized primer combinations for screening of Ig/T-cell receptor (TCR) gene rearrangements (IGH, IGK-Kde, TCRD and TCRG) and standardized primer/probe sets for RQ-PCR-based detection of MRD (Verhagen et al, 2000; Szczepanski et al, 2002; van der Velden et al, 2002a, b). Residual disease is measured no more than 15 d prior to conditioning in all patients. Allelespecific oligonucleotide-primer-based RQ-PCR is applicable

27

to more than 95% of children with ALL, and two primers sensitive to 10)4 can be generated in 80% of children. RQ-PCR also has several advantages over the traditional dot-blot radio-labelled oligonucleotide probing method. The lack of post-PCR processing enables the results to be available much more rapidly, and the use of a standardized assay allows easy comparison of laboratory MRD results across international borders. A quality assurance programme has been initiated as part of a larger ESG on MRD detection in ALL (ESG-MRD-ALL; co-ordinators: J.J.M. van Dongen and V.H.J. van der Velden). This is backed up by combined meetings to discuss the quality assurance results, new technical developments and clinical results. TOWARD MRD-BASED PROTOCOLS FOR SCT IN ALL It is important to acknowledge that current evidence regarding the clinical significance of MRD is based on retrospective studies of selected groups of patients. However, the enormity of SCT in terms of toxicity, emotional trauma for patient and family, and cost is such that it is appropriate to rapidly assess the value of any technology that can improve outcome. The small numbers of children treated in any one centre argue strongly for both national and international co-operation. A major drawback of such studies in the past has been failure to agree common transplant protocols. This concern may be largely obviated by the adoption of a common protocol (methodology and timing of analysis) for the measurement of MRD pre-SCT. It will now be possible to compare the efficacy of different transplant protocols by contrasting the impact on MRD burden prior to conditioning. Future protocols will initially address two issues: novel therapies for those with high-level MRD pre-SCT and definition of a low-risk group who can be cured without SCT. Current Dutch, BFM and MRC proposals for such protocols are now presented. NOVEL STRATEGIES FOR THOSE WITH HIGH LEVEL MRD PRE-BMT The current Dutch protocol aims to overcome high-level MRD with graft-versus-leukaemia (GVL) effect. Thus, children with MRD pre-SCT of > 10)4 are subjected to withdrawal of cyclosporine 4–5 weeks after transplant and consecutive increasing doses of donor lymphocyte infusions. When any of these immunomodulating interventions leads to graft-versus-host disease (GVHD) grade II, the GVHD will be treated and future interventions will be cancelled. The BFM in Germany are also relying on GVL to overcome high-level MRD pre-SCT. Preliminary data from Bader et al (2002) suggested that the immune effect of a non-manipulated graft may be beneficial. In addition, rapid immune modulation in the context of increasing mixed chimaerism will also be employed. In the UK, pilot studies have investigated the value of additional pre-SCT therapy to minimize the MRD load at the time of transplant (Moppett et al, 2001). The next UK MRC


2003 Blackwell Publishing Ltd, British Journal of Haematology 122: 24–29

28

N. Goulden et al

ALL relapse protocol contains a window therapy arm for those found to be high-level MRD positive prior to conditioning. Currently, a combination of fludarabine, high-dose cytosine and liposomal daunorubicin is being used in this window. The effect of this therapy on MRD level and clinical outcome will be assessed. If this approach fails to produce improvements in survival, alternative therapies in the form of new chemotherapeutic agents or cytotoxic monoclonal antibodies will be examined. DEFINITION OF A LOW-RISK GROUP The prognosis for children with no evidence of MRD preSCT, using a technique sensitive to at least 10)4, was uniformly good in the three studies reviewed above. In the Bristol study (Moppett et al, 2001), many of these MRDnegative patients had suffered a marrow relapse more than 6 months after therapy (i.e. they were in the BFM S2 risk group). It is interesting to speculate whether all of these children really require transplantation. Such speculation is fuelled by data produced by the Berlin group, who showed that children in S2 treated according to REZ BFM 95/96 and who had MRD below 10)3 at 5 weeks of re-induction had a very good survival without recourse to SCT (Eckert et al, 2001). TOWARDS A COMMON MRD-BASED RELAPSE PROTOCOL Theoretically, the prognostic value of a given MRD result at a certain time point is determined by the treatment received up to that point and any subsequent treatment to be given, i.e. it is protocol dependent. However, the degree of concordance of pre-BMT MRD results for patients in CR2 between European centres argues that the effect of treatment differences prior to transplant for relapsed disease is not very large. Still, the development of a uniform approach to treatment will minimize confounding variables. To this end, the MRC and the International Berlin–Frankfurt– Mu¨nster Study Group (I-BFM-SG) have designed a protocol with a uniform definition of risk and allocation of therapy, a single method of MRD analysis, and identical time points of MRD measurement. REFERENCES Bader, P., Hancock, J., Kreyenberg, H., Goulden, N.J., Niethammer, D., Oakhill, A., Steward, C.G., Handgretinger, R., Beck, J.F. & Klingebiel, T. (2002) Minimal residual disease (MRD) status prior to allogeneic stem cell transplantation is a powerful predictor for post-transplant outcome in children with ALL. Leukemia, 16, 1668–1672. Borgmann, A., Baumgarten, E., Schmid, H., Dopfer, R., Ebell, W., Go¨bel, U., Niethammer, D., Gadner, H. & Henze, G. (1997) Allogeneic bone marrow transplantation for a subset of children with acute lymphoblastic leukemia in third remission: a conceivable alternative? Bone Marrow Transplantation, 20, 939–944. Cave´, H., van der Werff ten Bosch, J., Suciu, S., Guidal, C., Waterkeyn, C., Otten, J., Bakkus, M., Thielemans, K., Grandchamp, B. & Vilmer, E. (1998) Clinical significance of minimal residual disease

in childhood acute lymphoblastic leukemia. New England Journal of Medicine, 339, 591–598. Chessells, J.M. (1998) Relapsed lymphoblastic leukaemia in children: a continuing challenge. British Journal of Haematology, 102, 423–438. Chessells, J.M. (2000) The management of high-risk lymphoblastic leukaemia in children. British Journal of Haematology, 108, 204– 216. van Dongen, J.J.M., Seriu, T., Panzer-Gru¨mayer, E.R., Biondi, A., Pongers-Willemse, M.J., Corral, L., Stolz, F., Schrappe, M., Masera, G., Kamps, W.A., Gadner, H., van Wering, E.R., Ludwig, W.-D., Basso, G., de Bruijn, M.A.C., Cazzaniga, G., Hettinger, E., van der Does-van den Berg, A., Hop, W.C.J., Riehm, H. & Bartram, C.R. (1998) Prognostic value of minimal residual disease in acute lymphoblastic leukaemia. Lancet, 352, 1731–1738. Eckert, C., Biondi, A., Seeger, K., Cazzaniga, G., Hartmann, R., Beyermann, B., Pogodda, M., Proba, J. & Henze, G. (2001) Prognostic value of minimal residual disease in relapsed childhood acute lymphoblastic leukaemia. Lancet, 358, 1239–1241. Handgretinger, R., Klingebiel, T., Lang, P., Schumm, M., Neu, S., Geiselhart, A., Bader, P., Schlegel, P.G., Greil, J., Stachel, D., Herzog, R.J. & Niethammer, D. (2001) Megadose transplantation of purified peripheral blood CD34(+) progenitor cells from HLAmismatched parental donors in children. Bone Marrow Transplantation, 27, 777–783. Heslop, H.E. (1999) Haemopoietic stem cell transplantation from unrelated donors. British Journal of Haematology, 105, 2–6. Knechtli, C.J., Goulden, N.J., Hancock, J.P., Grandage, V.L., Harris, E.L., Garland, R.J., Jones, C.G., Rowbottom, A.W., Hunt, L.P., Green, A.F., Clarke, E., Lankester, A.W., Cornish, J.M., Pamphilon, D.H., Steward, C.G. & Oakhill, A. (1998) Minimal residual disease status before allogeneic bone marrow transplantation is an important determinant of successful outcome for children and adolescents with acute lymphoblastic leukemia. Blood, 92, 4072– 4079. Moppett, J., Goulden, N.J., Darbyshire, P.J., Steward, C.G., Hancock, J., Garland, R.J., Burke, G.A.A., Groves, S., Pamphilon, D., Cornish, J. & Oakhill, A. (2001) Additional cytoreduction pretransplant for MRD positive children with ALL in CR2: feasibility and outcome. Blood, 98, 386b. Nyvold, C., Madsen, H.O., Ryder, L.P., Seyfarth, J., Svejgaard, A., Clausen, N., Wesenberg, F., Jonsson, O.G., Forestier, E. & Schmiegelow, K. (2002) Precise quantification of minimal residual disease at day 29 allows identification of children with acute lymphoblastic leukemia and an excellent outcome. Blood, 99, 1253–1258. Seeger, K., Adams, H.P., Buchwald, D., Beyermann, B., Kremens, B., Niemeyer, C., Ritter, J., Schwabe, D., Harms, D., Schrappe, M. & Henze, G. (1998) Tel-AML 1 fusion transcript in relapsed childhood acute lymphoblastic leukemia. The Berlin–Frankfurt– Mu¨nster Study Group. Blood, 91, 1716–1722. zur Stadt, U., Harms, D.O., Schluter, S., Schrappe, M., Goebel, U., Spaar, H., Janka, G. & Kabisch, H. (2001) MRD at the end of induction therapy in childhood acute lymphoblastic leukemia: outcome prediction strongly depends on the therapeutic regimen. Leukemia, 15, 283–285. Szczepanski, T., van der Velden, V.H.J. & van Dongen, J.J.M. (2002) Real-time quantitative (RQ)-PCR for the detection of minimal residual disease in childhood acute lymphoblastic leukemia. Haematologica, 87, 183–191. van der Velden, V.H.J., Joosten, S.A., Willemse, M.J., van Wering, E.R., Lankester, A.W., van Dongen, J.J.M. & Hoogerbrugge, P.M. (2001) Real-time quantitative PCR for detection of minimal residual disease before allogeneic stem cell transplantation


2003 Blackwell Publishing Ltd, British Journal of Haematology 122: 24–29

Minimal Residual Disease Prior to SCT for ALL predicts outcome in children with acute lymphoblastic leukemia. Leukemia, 15, 1485–1487. van der Velden, V.H.J., Willemse, M.J., van der Schoot, C.E., van Wering, E.R. & van Dongen, J.J.M. (2002a) Immunoglobulin kappa deleting element rearrangements in precursor-B-acute lymphoblastic leukemia are stable targets for detection of minimal residual disease by real-time quantitative PCR. Leukemia, 16, 928–936. van der Velden, V.H.J., Wijkhuijs, J.M., Jacobs, D.C.H., van Wering, E.R. & van Dongen, J.J.M. (2002b) T cell receptor gamma gene rearrangements as targets for detection of minimal residual

29

disease in acute lymphoblastic leukemia by real-time quantitative PCR analysis. Leukemia, 16, 1372–1380. Verhagen, O., Willemse, M., Breunis, W., Wijkhuis, A., Jacobs, D., Joosten, S., van Wering, E., van Dongen, J.J.M. & van der Schoot, C. (2000) Application of germline IGH probes in real-time quantitative PCR for the detection of minimal residual disease in acute lymphoblastic leukemia. Leukemia, 14, 1426–1435. Wheeler, K., Richards, S., Bailey, C. & Chessells, J. (1998) Comparison of bone marrow transplant and chemotherapy for relapsed childhood acute lymphoblastic leukaemia: the MRC UKALL X experience. British Journal of Haematology, 101, 93–102.


2003 Blackwell Publishing Ltd, British Journal of Haematology 122: 24–29