Treatment strategy and long-term results in paediatric patients treated in consecutive UK AML trials

Transcription

1 (2005) 19, & 2005 Nature Publishing Group All rights reserved /05 $ treated in consecutive UK AML trials BES Gibson 1, K Wheatley 2, IM Hann 3, RF Stevens 4, D Webb 3, RK Hills 2, SSN De Graaf 5 and CJ Harrison 6, for the United Kingdom Childhood Leukaemia Working Party, and the Dutch Childhood Oncology Group 1 Royal Hospital for Sick Children, Glasgow, UK; 2 Birmingham Clinical Trials Unit, University of Birmingham, Birmingham, UK; 3 Great Ormond Street Hospital for Children, London, UK; 4 Royal Manchester Children s Hospital, Manchester, UK; 5 Dutch Childhood Oncology Group, The Hague, The Netherlands; and 6 LRF Cytogenetic Group, Cancer Sciences Division, University of Southampton, Southampton, UK Between 1988 and 2002, 758 children with acute myeloid leukaemia (AML) were treated on Medical Research Council (MRC) AML 10 and AML 12. MRC AML 10 tested the role of bone marrow transplantation following four blocks of intensive chemotherapy and found that while both allogeneic bone marrow transplant (allo-bmt) and autologous bone marrow transplant (A-BMT) significantly reduced the relapse risk (RR), this did not translate into a significant improvement in overall survival (OS). A risk group stratification based on cytogenetics and response to the first course of chemotherapy derived from MRC AML 10 was used to deliver risk-directed therapy in MRC AML 12. Allo-BMT was limited to standard and poor risk patients and A-BMT was not employed. Instead, the benefit of an additional block of treatment was tested by randomising children to receive either four or five blocks of treatment in total. While the results of MRC AML 12 remain immature, there appears to be no survival advantage for a fifth course of treatment. The 5 year OS, disease-free survival (DFS), eventfree survival (EFS) and RR in MRC AML 12 are 66, 61, 56 and 35%, respectively; at present superior to MRC AML 10, which had a 5-year OS, DFS, EFS and RR of 58, 53, 49 and 42%, respectively. MRC AML trials employ a short course of triple intrathecal chemotherapy alone for CNS-directed treatment and CNS relapse is uncommon. Improvements in supportive care have contributed to improved outcomes and the number of deaths in remission fell between trials. Anthracycline-related cardiotoxicity remains a concern and the current MRC AML 15 trial tests the feasibility of reducing anthracycline dosage without compromising outcome by comparing standard MRC anthracycline-based consolidation with high-dose ara-c. MRC studies suggest that the role of allo-bmt is limited in 1st CR and that there may be a ceiling of benefit from current or conventional chemotherapy. (2005) 19, doi: /sj.leu Keywords: acute myeloid leukaemia; children; bone marrow transplantation; cytogenetics; prognostic factors Introduction The survival of children with AML treated on United Kingdom (UK) MRC trials has improved dramatically over the past 30 years and continues to improve (Figure 1). This has been achieved by a combination of increasingly intensive anthracycline- and cytosine-based chemotherapy and advances in supportive care, whcih have allowed intensive chemotherapy to be delivered with less morbidity and mortality. The role of allogeneic bone marrow transplant (allo-bmt) and autologous bone marrow transplant (A-BMT) in 1st complete remission Correspondence: Dr BES Gibson, Department of Haematology, Royal Hospital for Sick Children, Glasgow G3 8SJ, UK; Fax: þ 44(0) ; brenda.gibson@yorkhill.scot.nhs.uk Received 14 March 2005; accepted 20 April 2005 (CR 1) have been tested and a risk group stratification identified based on cytogenetics and response to the first block of chemotherapy. In the UK, children and adults have traditionally been treated on the same, or very similar protocols, within a single trial. There are approximately 70 new cases per annum of AML in children less than 15 years of age in the UK. Table 1 presents details of the most recent of these trials, including the number of centres involved, their average patient numbers and the patient numbers per trial. The protocols are shown schematically in Figure 2a and b and given in detail in Table 2. Treatment strategy of MRC AML trials Three MRC AML trials were conducted between 1978 and 1990: MRC AML 8 1 ( ), Joint AML study 2 ( ) and MRC AML 9 3 ( ): The recruitment of children was limited, but these trials provided important information (mainly from adult results), based on a combination of randomised and historical comparisons, which were carried forward in the design of MRC AML 10. Firstly, MRC AML 9 found the DAT 3 þ 10 regimen (daunorubicin, ara-c, thioguanine) to be superior to the DAT 1 þ 5 regimen of AML 8. The high remission rate for DAT 3 þ 10 (91% for children) was confirmed in the Joint AML study and the second course modified to DAT 3 þ 8; two courses of DAT 3 þ 10 having been shown to be too toxic. Secondly, both MRC AML 9 and the Joint AML study tested the role of high-dose consolidation therapy and allo-bmt in maintaining remission. New regimens, previously used in resistant disease 4 10 were piloted: MAZE (amsacrine, 5-azacytidine, etoposide) and ara-c 1 g/m 2 combined with daunorubicin and thioguanine. The former became MACE when ara-c replaced 5-azacytidine and the latter was further modified to MidAC, when the same dose of ara-c was combined with mitoxantrone. Thirdly, MRC AML 9 tested the value of maintenance therapy followed by late intensification (COAP regimen) and found it to be limited. Maintenance therapy was not employed in subsequent MRC trials. Finally, intrathecal chemotherapy was the only form of CNS treatment used in these studies and the CNS relapse rate was low suggesting that intrathecal chemotherapy and/or intensive systemic therapy were effective in preventing CNS disease. MRC AML 10 ( ): With excellent CR rates being achieved, MRC AML 10 concentrated on the prevention of relapse. DAT 3 þ 10 was compared with ADE 3 þ 10 þ 5 (ara-c,

2 daunorubicin, etoposide) as induction therapy. These regimens delivered the same doses of daunorubicin and ara-c and tested the comparative benefits of thioguanine with etoposide. MRC AML 10 delivered a total of four courses of treatment (DAT 3 þ 10 or ADE 3 þ 10 þ 5, DAT 3 þ 8 or ADE 3 þ 8 þ 5, MACE, MidAC) and investigated the role of BMT following this intensive chemotherapy. Allo-BMT was recommended for all children with a matched sibling donor. Children without an HLA compatible sibling were randomised to receive an A-BMT in 1st CR following four blocks of intensive chemotherapy (marrow harvested after 3rd course of chemotherapy), or no further treatment. While there was a reduction in relapse rate for both forms of transplantation, this did not translate into an overall survival advantage. MRC AML 10 identified a risk group stratification based on karyotype and response to the 1st block of treatment. MRC AML 12 ( ): The Dutch Children s Oncology Group joined MRC AML 12. This trial took forward the marginally better induction regimen from MRC AML 10 (ADE) and the standard MRC template became ADE, ADE, MACE, MidAC. MRC AML 12, investigated whether mitoxantrone 11 might be superior to daunorubicin and less cardiotoxic, by comparing ADE (daunorubicin, ara-c, etoposide) with MAE (mitoxantrone, ara-c, etoposide). MRC AML 10 risk group stratification was adopted in MRC AML 12 to deliver riskdirected therapy. Allo-BMT was restricted to standard and poor risk patients in whom relapse remained the main cause of treatment failure. In the absence of an advantage in OS, A-BMT, with its substantial morbidity in children, was omitted from MRC AML 12. The reduction in relapse risk, similar for A-BMT and allo-bmt, in MRC AML 10, suggested that the benefit might be one of additional treatment rather than a graft-versusleukaemia effect. This was tested by the introduction of a 5th Figure 1 MRC AML trials: OS: Age 0 14 years: course of therapy by randomisation. HD ara-c with asparaginase was chosen to avoid further anthracycline exposure. CNS-directed therapy in MRC AML 10 was with four courses and in AML 12 three courses of triple intrathecal chemotherapy (ara-c, methotrexate, hydrocortisone) at age-adjusted doses. Patients and methods Eligibility The paediatric upper age limit for MRC AML 10 was 14 years and for MRC AML 12 was 16 years, although only 22 (4%) and two (o1%) of the children who were entered into MRC AML 12 were aged 15 and 16, respectively. De novo AML, Down syndrome (DS), secondary AML and aggressive MDS (RAEB, RAEB-t) patients, for whom intensive-type therapy was considered appropriate, were eligible. Written informed consent from the patient or parent was required. Diagnosis The diagnosis of AML and its subtypes was established according to the FAB classification in each local centre and reviewed by a central morphology panel. Confirmatory immunophenotyping and cytogenetics were performed locally. Cytogenetics were coordinated by the LRF Cytogenetics Group. Remission marrows were not centrally reviewed. CNS disease was defined by the presence of /l leukaemia blasts in a CSF cytospin and/or the presence of neurological symptoms. Definitions and statistics Complete remission (CR) was defined as a bone marrow with o 5% leukaemic cells and evidence of regeneration of normal haemopoietic cells. Neutrophil and platelet parameters were not included in the definition. Partial remission (PR) was defined as a bone marrow with between 5 and 15% leukaemic cells and evidence of regeneration of normal haemopoietic cells. Resistant disease (RD) was defined as a bone marrow containing 415% leukaemic cells. Remission failures were classified as either induction deaths, that is, related to treatment and/or hypoplasia, or as resistant disease, that is, related to the failure of therapy to eliminate the disease (including partial remissions). When the clinician s evaluation was not available, deaths within 30 days of entry were classified as induction deaths and deaths at more than 30 days as resistant disease. OS was calculated from the date of entry into the trial, or from the date of randomisation, to death from any cause. Disease-free survival (DFS) for patients achieving remission was calculated from the date of remission to the date of the 2131 Table 1 Patient accrual and follow-up Trial Accrual of patients: time period Total no. of patients (n) Centres (n) No. of patients/centre (median/range) No. of patients/year Follow-up (median, range, years) MRC AML 10 5/88 3/ (1 48) (5 15) MRC AML 12 4/95 5/ (1 63) 78 3 (1 7)

3 2132 a MRC AML 10 Allo BMT DONOR D A T D A T R1 M A C E M id AC A D E A D E NO DONOR A B M T R2 N F T Anthracycline (mg/m 2 ) Ara-C (g/m 2 ) Etoposide (mg/m 2 ) DAT DAT or ADE ADE MACE MidAC CUMULATIVE DOSES An anthracycline conversion factor of 1.5 (mitoxantrone 1mg/m is equivalent to daunorubicin 5mg/m 2 ) has been used to calculate cumulative anthracycline doses at the instruction of the Editor. b MRC AML 12 MidAC RISK GROUP ASSIGNED GOOD R2 CLASP MidAC ADE ADE R1 MAE MAE MACE STANDARD + POOR DONOR R2 AlloBMT CLASP AlloBMT MidAC NO DONOR R2 CLASP MidAC Figure 2 Anthracycline(mg/m 2 ) Ara-C(g/m 2 ) Etoposide (mg/m 2 ) ADE ADE or MAE MAE MACE MidAC CLASP 24 - CUMULATIVE DOSES Anthracycline doses calculated as for MRC AML 10 Treatment schemas of MRC AML 10 (a) and AML 12 (b).

4 Table 2 Treatment elements in the four AML Studies MRC AML 10 Induction Consolidation Course 1: Randomisation: DAT vs ADE DAT 3+10: Daunorubicin 50 mg/m 2 i.v. d 1, 3, 5; Ara-C 100 mg/m 2 /12 h i.v. d 1 10; 6-TG 75 mg/m 2 12-h orally d 1 10 ADE : Ara-C 100 mg/m 2 /12-h i.v. d 1 10; Daunorubicin 50 mg/m 2 i.v. d 1, 3, 5; Etoposide 100 mg/m 2 i.v. d 1 5 Course 2: DAT 3+8 vs ADE 3+8+5: Ara-C reduced to 8 days in both regimens and 6-TG to 8 days in DAT Course 3: MACE: Amsacrine 100 mg/m 2 d 1 5; Ara-C 200 mg/m 2 continuous infusion d 1 5; Etoposide 100 mg/m 2 /day i.v. d 1 5 Course 4: MIDAC: Mitozantrone 10 mg/m 2 /day i.v. d 1 5; Ara-C 1 g/m 2 12-h i.v. d 1 3 Age-adjusted intrathecal chemotherapy with methotrexate, Ara-C and hydrocortisone ( 4) Patients with a sibling donor were eligible for allo-bmt Patients with no sibling donor were randomised between A-BMT and no further treatment MRC AML 12 Induction Course 1: randomisation: ADE vs MAE ADE : Ara-C 100 mg/m 2 12-h i.v. d 1 10; Daunorubicin 50 mg/m 2 i.v. d 1, 3, 5; Etoposide 100 mg/m 2 i.v. d 1 5 MAE : Mitozantrone 12 mg/m 2 i.v. d 1, 3, 5; Ara-C 100 mg/m 2 12-h i.v. d 1 10; Etoposide 100 mg/m 2 i.v. d 1 5 Course 2: ADE vs MAE 3+8+5: Ara-C reduced to 8 days in both regimens Consolidation Course 3: MACE: Amsacrine 100 mg/m 2 d 1 5; Ara-C 200 mg/m 2 continuous infusion d 1 5; Etoposide 100 mg/m 2 d 1 5 Good risk patients randomised: MidAC vs CLASP-MidAC Standard and poor risk patients with no sibling donor randomised for MidAC vs CLASP-MidAC Standard and poor risk patients with sibling donor were eligible for allo-bmt and randomised for allo-bmt vs CLASP-allo-BMT MidAC: Mitozantrone 10 mg/m 2 i.v. d 1 5; Ara-C 1 g/m 2 12 h i.v. d 1 3 CLASP: Ara-C 3 g/m 2 12-h i.v. d 1, 2, 8, 9; Asparaginase 6000 U/m 2 s.c. d 2, 9 d ¼ Days. Age-adjusted intrathecal therapy with methotrexate, Ara-C and hydrocortisone ( 3) first event (either relapse or death in CR), or from the date of randomisation for post CR questions. Event-free survival (EFS) was calculated from the date of entry into the trial to the date of first event (failure to achieve CR, relapse or death from any cause failure to achieve CR was considered an event on day 1). RR for patients achieving remission was the cumulative probability of relapse censored at death in CR. Toxicity was assessed according to the WHO toxicity criteria. Randomised comparison results, including the Mendelian randomisation to assess allo-bmt, were based on an intention to treat analysis and included all randomised patients according to the eligibility criteria for the trial. Randomisations were balanced by minimisation. Remission rates and reasons for failure to achieve CR were compared using w 2 tests. Kaplan Meier life-tables were constructed for time to event data and were compared by means of the log-rank test. For the main mature randomised comparisons in MRC AML 10, odds ratios (OR) or hazard ratios (HR), with 95% confidence intervals (CI), are given. All P-values are two-tailed. In both MRC AML 10 and AML 12, some children were not registered at diagnosis. Such patients cannot contribute to estimates of overall outcome, since because they had to survive long enough to be registered, their inclusion will inflate the estimates slightly. In MRC AML 10, 341 out of 364 children were registered at diagnosis; the corresponding figure for MRC AML 12 was 527 out of 564. Also excluded, at the request of the editor, from analyses of overall outcome were patients aged 15 or 16 years, patients with secondary AML, patients with MDS and DS children. Four children rediagnosed after entry as having ALL (n ¼ 3) or bilineage leukaemia (n ¼ 1) were also excluded. Overall outcome data will be confined to the 758 children (MRC AML 10: 303, MRC AML 12: 455) who met the study criteria for this publication. 12 Data were analysed using a follow-up date of 1 April Results of MRC AML 10 and 12 trials This review will concentrate on MRC AML 10 and the preliminary results of MRC AML 12, which are already in the public domain because these are the only trials, that recruited all, or most children in the UK. Patient characteristics are given in Table 3 and results in Table 4. Overall outcome results are restricted to patients 0 14 years of age with de novo AML while randomised comparisons include all trial entrants. Survival, DFS and EFS have consistently improved and, most importantly, the improvement has been maintained between MRC AML 10 and AML 12 (Figures 1, 3, 4 and Table 4) MRC AML 10 In all, 93% (282 of 303) of patients achieved remission. 3% of children had resistant disease and 4% an induction death (Table 4). The 5-year and 10-year OS, EFS and DFS are 58, 49, 53 and 56, 48, 51%, respectively. A total of 286 children (143 in each arm) were randomised to DAT vs ADE; a comparison of etoposide with thioguanine. There was no significant difference in CR rate (DAT 90% vs ADE 93%, OR ¼ 1.58, CI ¼ , P ¼ 0.3), induction deaths, the number of courses required to achieve CR, or resistant disease (DAT 4% vs ADE 3%, P ¼ 0.7). At 10 years, the OS, DFS and EFS were similar (DAT 57% vs ADE 51%, HR ¼ 0.83, CI ¼ , P ¼ 0.3; DAT 53% vs ADE 48%, HR ¼ 0.83, CI ¼ , P ¼ 0.3; DAT 48% vs ADE 45%, HR ¼ 0.89, CI ¼ , P ¼ 0.5, respectively). There was a nonsignificant excess of deaths in 1st CR during consolidation in the ADE

5 2134 Table 3 Initial patient data on studies MRC AML 10 and AML 12 (0 14 years with AML) MRC AML 10 MRC AML 12 n % n % Total number of patients entered Number of patients included in overall outcome analysis Gender (male) Age (years) WBC 10 9 /l o o X CNS leukaemia (yes) FAB types M M M M M M M M Cytogenetics Favourable t(8;21) t(15;17) inv(16) Intermediate Table 4 Results in studies MRC AML 10 and 12 (0 14 years with de novo AML) MRC AML 10 MRC AML 12 n % n % Number of patients Early deaths 4 4 Nonresponse 3 4 CR achieved Death in 1st CR (5 years) 10 6 Relapse (5 years) OS At 5 years At 10 years 56 F EFS At 5 years At 10 years 48 F DFS At 5 years At 10 years 51 F Adverse , 5, del(5q), abn(3q), complex karyotype Crude overall CNS relapse rate for both trials of 1.6% not expected to differ significantly from the cumulative CNS relapse rate. Unknown values are not included. Figure 3 MRC AML Trials: Survival from entry. arm (8 vs 15%, P ¼ 0.09). There was no evidence that children with monocytic involvement (FAB type M4 or M5) did better with ADE than with DAT. In this subgroup the CR rate with DAT was 83% (35/42) and with ADE 94% (32/34) (P ¼ 0.2), whereas 7-year survival was 52 and 55%, respectively (P ¼ 0.9). There was no significant OS advantage for A-BMT (A-BMT 70 vs Stop 58% at 10 years, HR ¼ 0.67, CI ¼ , P ¼ 0.2), but there was a decrease in RR (A-BMT 31% vs Stop 52%, HR ¼ 0.50, CI ¼ , P ¼ 0.03), with a significant improvement in DFS in the A-BMT arm (A-BMT 68% vs Stop 44% at 10 years, HR ¼ 0.50, CI ¼ , P ¼ 0.02). The benefit of A-BMT emerged at 2 years from diagnosis. The reduction in the RR did not translate into a significant improvement in OS because children who relapsed and who had not had an A-BMT were more likely to be salvaged with second-line treatment than those who had had an A-BMT (survival at 5 years from relapse A-BMT 7% vs Stop 27%, P ¼ 0.05). There were no deaths in CR after ABMT. The results of allo-bmt were analysed on whether or not a donor was available that is, intention to treat. Of the 85 children Figure 4 MRC AML Trials: EFS. with a donor, 61 received an allo-bmt and this included 12 of 21 good risk patients with a donor. Allo-BMT was associated with a significant reduction in RR from 45 to 30% at 10 years (HR ¼ 0.64, CI ¼ , P ¼ 0.02). The transplant-related

6 mortality for children with a donor was 11%, but 15% in those children who actually received an allo-bmt. There was no significant difference in survival between those children with or without a donor at 10 years (donor 68% vs no donor 59% HR ¼ 0.79, CI ¼ , P ¼ 0.3). The fewer relapses in the donor group were counterbalanced by procedure-related deaths (P ¼ 0.001). AML 10 allowed stratification of patients into a good (32%), standard (49%) or poor (19%) risk group based on karyotype and response to the first course of treatment. OS from CR at 10 years for good, standard and poor risk patients was 77, 58 and 30% (Po0.0001), respectively, while DFS at 10 years was 60, 52 and 25%, respectively, (Po0.0001). The RR for good, standard and poor was 35, 43 and 72%, respectively, (Po0.0001). Survival from relapse at 5 years was significantly different: survival for good, standard and poor risk was 57, 14 and 8% (P ¼ ) respectively, suggesting that bone marrow transplantation might be reserved for second CR in good risk patients at least. Results of MRC AML 12 MRC AML 12 was closed to recruitment in May 2002 and the results remain immature. In all, 92% (420 of 455) of patients who entered into MRC AML 12 achieved remission. Failure to achieve remission was equally due to early deaths (4%) and resistant disease (4%). The estimated probabilities of 5-year OS, EFS and DFS are 66, 56 and 61%, respectively. MRC AML 12 compared mitoxantrone to daunorubicin (ADE vs MAE). In all, 251 children were randomised into each arm of the study and compliance was 99%. The CR (ADE 92% vs MAE 90%, P ¼ 0.3) and resistant disease (4 vs 4%) rates were similar for both regimens. Induction deaths were slightly, but not significantly higher for MAE than for ADE (MAE 6% vs ADE 3%); however, the DFS and RR were both superior for MAE compared to ADE (ADE 59%, MAE 68%, P ¼ 0.04; ADE 37%, MAE 29%, P ¼ 0.06, respectively). The estimated probability for 5-year OS is not significantly different (ADE 64%, MAE 70%, P ¼ 0.1). The second randomisation compared five courses of treatment with four; testing whether an additional course of treatment would reduce the RR. Compliance with randomisation to four courses was 93% and to five courses was 75%. There appears to be no benefit from an extra course of treatment at present, but analysis is ongoing; the estimated probability for 5-year OS for four courses is 81%, five courses 78%, P ¼ 0.5, DFS for four courses 65%, five courses 66%, P ¼ 0.8, and RR for four courses 33%, and five courses 32%, P ¼ 0.9. There was no difference in the number of deaths in CR: four courses 2% vs five courses 1%, P ¼ 0.3. The risk group stratification derived from MRC AML 10 remains prognostically significant in MRC AML 12. For good, standard and poor risk patients the estimated probability for 5-year survival from CR is 84, 76 and 47%; DFS 75, 62 and 41% and RR 19, 37 and 54%, respectively (all Po0.0001). Only 35 children underwent a sibling transplant in 1st CR. However, results from AML 10 and AML 12 can be combined to address the question of the benefit of allo-bmt in the 1st CR in AML in children. There was no heterogeneity for RR between the trials (P ¼ 0.3) and combined, they showed a significant reduction in RR (2P ¼ 0.02) which did not translate into a significant reduction in DFS (2P ¼ 0.06) or OS (2P ¼ 0.1). The numbers of children are too small for reliable interpretation, which will in due course be based on the entire population aged up to 44 years in AML 10 and AML 12. Table 5 AML 10 Presenting features Risk groups Results according to different risk parameters in MRC Total number of patients OS % at 5 years Cytogenetics Favourable t(8;21) t(15;17) inv(16) Cytogenetics Intermediate Cytogenetics Adverse FAB classification M M M M M M M M Status after course 1 Complete remission Partial remission Resistant disease Age (years) Risk group Good Standard Poor NB: data on status after course 1 and risk group is survival from start of course 2. The response after course 1 of treatment and diagnostic cytogenetics were strongly predictive of outcome in MRC AML For children with CR, PR (5 15% blasts) and RD (415% blasts), the 5-year survival from the start of course 2 was 67, 65 and 23% and the relapse rates were 34, 38 and 87%, respectively (both Po0.0001), and for favourable, intermediate and adverse karyotypic abnormalities, the survival was 76, 52 and 40% and the relapse rates were 34, 44 and 61%, respectively (P ¼ and 0.006, respectively) (Table 5). These factors combined to give three risk groups. Good risk patients are defined as those with t(8,21),inv(16),t(15,17) or FAB M3 morphology, irrespective of bone marrow status after course 1 or the presence of other genetic abnormalities; standard risk patients are those with neither favourable nor adverse genetic abnormalities or FAB M3, and not more than 15% blasts in the bone marrow after course 1; poor risk patients are those with more than 15% blasts in the bone marrow after course 1 or with adverse abnormalities of 5, 7, del(5q), abn(3q), complex karyotype and without favourable genetic abnormalities. The OS was intermediate for patients with 11q23 abnormalities and similar for those with t(9;11) compared to the other 11q23 translocations. (Harrison CJ et al. Blood 2003; 102: 98a) 2135

7 2136 Toxicity The treatment-related mortality for AML10 was 14% but decreased in the latter half of the trial from 18 to 10% (P ¼ 0.03) and fell to 10% in AML 12; 4% during induction in both trials. The main cause of death was infection, with fungal infection accounting for 23% of the infection-related deaths in AML Haemorrhage occurred most commonly early in children with an initial WCC of greater than /l (P ¼ 0.001), and M4 and M5 morphology. In AML 10, there were nine deaths due to cardiac failure in children still receiving treatment and in the majority (seven of nine) this occurred during an episode of sepsis. Cardiac monitoring was not part of the study and therefore subclinical reduction in cardiac function may have been overlooked, and when exacerbated by infection proved fatal. Cardiac deaths were only reported from course 3 onwards and after a cumulative anthracycline dose of 300 mg/m 2. The incidence of long-term cardiac toxicity for this study is unknown. Acute promyelocytic leukaemia FAB M3 APML represented 8% of children entered into MRC AML 10 and MRC AML 12. Extended-course ATRA (ATRA given during induction chemotherapy until remission was achieved, or for a maximum of 60 days) was found to be superior to shortcourse ATRA (ATRA given for 5 days only prior to starting induction chemotherapy) both at a dose of 45 mg/m 2 in MRC AML 10; CR rate (Po0.001), induction deaths (P ¼ 0.02), resistant disease (P ¼ 0.03), RR (P ¼ 0.04) and OS (P ¼ 0.005). Extended-course ATRA became standard treatment in MRC AML 12. The OS, DFS and RR for children with APML who received extended ATRA in AML 10 and AML 12 are 75, 64 and 31%, respectively; 87, 79 and 18%, respectively, for those with a WCC of o /l. Down syndrome Treatment for children with DS in MRC AML 10 was not modified but in MRC AML 12 they were allocated four courses of chemotherapy only and were not eligible for BMT. Approximately 6% of children entered into MRC AML 10 and AML 12 had DS. In all, 34% of DS children and 4% of non-ds children had FAB M7 morphology. There was a strikingly different age distribution between DS and non-ds AML children, 41 of 46 patients with DS were 1 3 years of age. In this age group, the OS, DFS and EFS at 5 years was somewhat better for DS compared to non-ds children for those treated on MRC AML 10 and AML 12; 76 vs 63% (P ¼ 0.4), DFS 84 vs 59% (P ¼ 0.02) and EFS 76 vs 56% (P ¼ 0.08), respectively. There were more induction and remission deaths in DS children; 10 vs 4% (P ¼ 0.2) and 16 vs 7% (P ¼ 0.05). The relapse rate was significantly lower for DS children in this age group 0 vs 36% (P ¼ ). The most common additional cytogenetic abnormality was trisomy 8 and favourable karyotypes were absent. Infants o1 year of age All infants treated in MRC AML 10 and AML 12 had chemotherapy arbitrarily reduced by 25% as a protocol recommendation. Younger children had a high incidence (35%) of 11q23 abnormalities. The favourable cytogenetic abnormalities of t(8,21) or t(15,17) were not seen, but there was one case of inv(16). The EFS at 5 years was 58% (56% for normal cytogenetics, 55% for 11q23 and 64% for other abnormalities), which was not significantly different from that of older children (P ¼ 0.6), nor were DFS (65% at 5 years, P ¼ 0.2) or RR (31%, P ¼ 0.3). There were more induction deaths in infants (12 vs 3%, P ¼ ). Despite the absence of favourable cytogenetics, the EFS is very encouraging for infants. Discussion While acknowledging the problems of nonrandomised comparisons, MRC AML 10 results are comparable with, or better than those of other trial groups conducted during the same period. The dosage, duration and total number of courses of treatment were more intensive in MRC AML 10 than that used by most other groups and this is the likely explanation for its superior outcome. The hypothesis that more treatment is better was further tested in MRC AML 12. The results of this trial are immature, but while they suggest continuing improvement, this cannot be explained by the benefit of additional therapy. The reason for the observed improvement continues to be investigated, but is likely to be related to improvements in supportive care and increased clinical experience. The intensity of both induction and post remission treatment, in terms of dosage and scheduling, is important in AML. The CCG 2861 study 15 compared standard timing during induction (days 0 3 and or later on count recovery) with intensive timing (days 0 3 and irrespective of count recovery). At 3 years there was no difference in CR rate (70 vs 75%), but a significant EFS advantage for intensive timing (42 vs 27%, Po0.0005), although at the expense of excess early deaths in the intensive timing arm (11 vs 4%). In MRC AML 10, A-BMT reduced relapse but did not significantly improve long-term survival. There are three other randomised paediatric trials that have shown no survival advantage for A-BMT, and a meta-analysis 19 concluded that there was insufficient data to determine whether A-BMT is superior to nonmyeloablative chemotherapy. Therefore, A-BMT does not appear to have a major role in paediatric AML in first remission, particularly when the acute and long-term morbidity is considered. The lack of benefit found in MRC AML 10 for allo- BMT mirrors the experience of the BFM, 20 but not that of CCG who reported a significant survival advantage for allo- BMT compared to A-BMT (P ¼ 0.002) and chemotherapy (P ¼ 0.05) as postremission treatment, but no advantage for A-BMT over intensive chemotherapy. However, the CCG analysis was not a true intention to treat comparison since, while all donor patients were included in the analysis whether or not they received allo-bmt, not all patients without a donor were included only those who were randomised between A-BMT and chemotherapy. 21 It is possible that any benefit for allo-bmt may be protocol dependent and may only appear when less intensive treatment is delivered than that used by the MRC. Whatever the benefit of allo-bmt, good and standard risk children have a good outcome when treated with chemotherapy alone, suggesting that BMT in CR 1, particularly with alternative donor, with its associated morbidity and mortality, should probably be restricted to poor risk patients. When MRC AML 10 and AML 12 are combined to increase numbers, there is no statistically significant survival benefit for allo-bmt in 1st CR for any risk group in children.

8 The BFM report that cranial irradiation not only reduces CNS relapse, but also marrow seeding and relapse. MRC AML trials have never employed cranial irradiation, but delivered intrathecal chemotherapy alone and the low incidence of isolated and combined CNS relapses challenges the use of cranial irradiation with its long-term sequelae. The reported benefit of cranial irradiation in BFM-87 in reducing bone marrow relapse occurred in nonrandomised patients (randomised patients P ¼ 0.34, non randomised patients P ¼ 0.01). 22 The MRC strategy is to continue decreasing CNS therapy in a step-wise manner. MRC protocols include high dose ara-c, which may afford additional CNS protection. The MRC risk group stratification has been confirmed to be highly discriminatory in children treated in AML 10 and AML 12. In the UK, the coordination of cytogenetics facilitated a high success rate in achieving an overall cytogenetic result of 88% (93% of entrants with a specimen/information), with an abnormality detection rate of 78% among successful cases. In MRC AML 10 and AML 12, children received a relatively high cumulative dose of anthracycline, but acute cardiotoxicity was rare. 23 However, anthracyclines are important in AML and while it may be possible and desirable to reduce anthracycline exposure, this must not be done at the expense of disease control and without the safety of reduction being proven by clinical trial. Further improvement in outcome, particularly for poor risk patients, requires new approaches. Monoclonal antibodies, drugs targeted at specific fusion genes and multidrug-resistant modifiers are being tested. MRD monitoring, gene expression profiling and continued cytogenetic analysis may further refine risk groups and allow treatment to be more tailored to risk and biology of the disease. Toxicity will be limited by restricting transplantation and reducing anthracycline dosage if the latter is shown not to result in an increased relapse rate. Acknowledgements The MRC/UK Childhood Leukaemia Working Party acknowledge the very significant contribution to MRC AML 10 and subsequent trials of Dr Richard Stevens (Deceased). For Acknowledgements refer to Supplementary Information. Supplementary Information Supplementary Information accompanies the paper on the website ( References 1 Rees JKH, Gray RG, Swirsky D, Hayhoe FGJ, for the Medical Research Council s Working Parties for Therapeutic Trials in Adult and Childhood Leukaemia. Principal results of the Medical Research Council s 8th Acute Myeloid Leukaemia Trial. Lancet 1986; ii: Marcus RE, Catovsky D, Prentice HG, Newland AC, Chessells JM, Stevens RM et al. Intensive Induction and Consolidation Chemotherapy for Adults and Children with Acute Myeloid Leukaemia (AML) Joint AML Trial In: Büchner Th, Schellong G, Hiddemann W, Urbanitz D, Ritter J, (eds). Acute Leukaemias. Prognostic Factors and Treatment Strategies, Haematology and Blood Transfusion, Vol. 30. Berlin-Heidelberg-New York, Tokyo: Springer-Verlag, Rees JKH, Gray RG, Wheatley K, on behalf of the MRC Leukaemia in Adults Working Party. Dose intensification in acute myeloid leukaemia: greater effectiveness at lower cost. Principal report of the Medical Research Council s AML9 study. Br J Haematol 1996; 94: Worsley AM, Catovsky D, Goldman JM, Galton DAG. New combination chemotherapy for relapsed acute myeloid leukaemia. Lancet 1984; I: Preisler H. High-dose Ara-C as the initial treatment of poor risk patients with acute non-lymphocytic leukaemia. A leukaemia intergroup study. J Am Soc Clin Oncol 1987; 5: Herzig RH, Lazarus HM, Wolff SN, Phillips GL, Herzig GP. Highdose Ara-C therapy with and without anthracycline antibiotics for remission re-induction of acute non-lymphoblastic leukaemia. J Clin Oncol 1985; 3: Brito-Babapulle F, Catovsky D, Slocombe G, Newland AC, Marcus RE, Goldman JM et al. Phase II study of mitoxantrone and cytarabine in acute myeloid leukaemia. Cancer Treat Rep 1987; 71: Hiddemann W, Kreutzmann H, Straif K, Ludwig WD, Mertelsmann R, Donhuijsen-Ant R et al. High dose cytosine arabinoside and mitoxantrone: a highly effective regimen in refractory acute myeloid leukaemia. Blood 1987; 69: Peters WG, Willemze R, Colly LP. Results of induction and consolidation treatment with intermediate and high-dose ara-c and m-amsa-containing regimens in patients with primarily failed or relapsed acute leukaemia and non-hodgkin s lymphoma. Scand J Haem 1986; 36: Willemze R, Zwaan FE, Colpri G, Kenning JJ. High-dose cytosine arabinoside in the management of refractory acute leukaemia. Scand J Haem 1982; 29: Arlin Z, Case DC, Moore J, Wiernik P, Feldman E, Saletan S et al. Randomized multi-centre trial of cytosine arabinoside with mitoxantrone or daunorubicin in previously untreated adult patients with acute non-lymphoblastic leukaemia (ANLL). Leukaemia 1990; 4: Kaspers GJL, Creutzig U. Pediatric acute myeloid leukaemia: International progress and future directions. Leukaemia 2005; 19: Wheatley K, Burnett AK, Goldstone AH, Gray RG, Hann IM, Harrison CJ, et al, on behalf of the United Kingdom Medical Research Council s Adult and Childhood Leukaemia Working Parties. A simple, robust, validated and highly predictive index for the determination of risk-directed therapy in acute myeloid leukaemia derived from the MRC AML 10 trial. Br J Haematol 1999; 107: Riley LC, Hann IM, Wheatley K, Stevens RF, on behalf of the MRC Childhood Leukaemia Working Party. Treatment-related deaths during induction and first remission of acute myeloid leukaemia in children treated on the Tenth Medical Research Council Acute Myeloid Leukaemia Trial (MRC AML10). Br J Haematol 1999; 106: Woods WG, Kobrinsky N, Buckley J, Neudorf S, Sanders J, Miller L et al. Intensively timed induction therapy followed by autologous or allogeneic bone marrow transplantation for children with acute myeloid leukaemia or myelodysplastic syndrome: a Children s Cancer Group pilot study. J Clin Oncol 1993; 8: Amadori S, Testi AM, Arico M, Comelli A, Giuliano M, Madon E et al. Prospective comparative study of bone marrow transplantation and post remission chemotherapy for childhood acute myelogenous leukaemia. The Associazione Italiana Ematologia ed Oncologia Pediatrica Cooperative Group. J Clin Oncol 1993; 6: Ravindranath Y, Yeager AM, Chang MN, Steuber CP, Krischer J, Graham Pole J et al. Autologous bone marrow transplantation versus intensive consolidation chemotherapy for acute myeloid leukaemia in childhood. N Engl J Med 1996; 334: Woods WG, Neudorf S, Gold S, Sanders J, Buckley JD, Barnard DR et al. A comparison of allogeneic bone marrow transplantation, autologous bone marrow transplantation, and aggressive chemotherapy in children with acute myeloid leukaemia in remission: a report from the children s cancer group. Blood 2001; 97: Bleakley M, Lau L, Shaw PJ, Kaufman A. Bone marrow transplantation for paediatric AML in first remission: a systematic review and meta-analysis. Bone Marrow Transplant 2002; 10:

9 Creutzig U, Reinhardt D, Zimmermann M, Klingebiel T, Gadner H. Intensive Chemotherapy vs BMT in pediatric AML a matter of controversies. Blood 2001; 97: Wheatley K. Current controversies: which patients with acute myeloid leukaemia should receive a bone marrow transplantation? A statistician s view. Br J Haematol 2002; 118: Creutzig U, Ritter J, Zimmermann M, Schellong G. Does cranial irradiation reduce the risk for bone marrow relapse in acute myelogenous leukaemia. Unexpected results of the Childhood Acute Myelogenous Leukaemia Study BFM-87. J Clin Oncol 1993; 11: Stevens RF, Hann IM, Wheatley K, Gray RG, on behalf of the MRC Childhood Leukaemia Working Party. Marked improvements in outcome with chemotherapy alone in paediatric acute myeloid leukaemia: results of the United Kingdom Medical Research Council s 10th AML trial. Br J Haematol 1998; 101: