Results of the first randomized multicentre trial on childhood acute ...

Leukemia (2008) 22, 1144–1153 & 2008 Nature Publishing Group All rights reserved 0887-6924/08 $30.00 www.nature.com/leu

ORIGINAL ARTICLE Results of the first randomized multicentre trial on childhood acute lymphoblastic leukaemia in Russia A Karachunskiy1,11, R Herold2,11, A von Stackelberg2, N Miakova1, A Timakow3, T Mahortih3, L Bajdun1, A Maschan1, L Fechina4, A Shamardina5, S Dudkin6, V Lebedev7, S Varfolomeeva8, V Timofeeva9, J Roumiantseva1, N Chipsanova10, A Rumjanzew1 and G Henze2 1

Russian Federal Clinical Research Centre for Paediatric Haematology, Oncology and Immunology, Moscow, Russia; Department of Paediatric Oncology/Haematology, Charite´FUniversita¨tsmedizin Berlin, Berlin, Germany; 3Department of Paediatric Haematology, Morozov Children’s Hospital, Moscow, Russia; 4Paediatric Oncohaematological Centre, Regional Children’s Hospital, Ekaterinburg, Russia; 5Department of Paediatric Haematology, Regional Children’s Hospital, Nizhni Novgorod, Russia; 6Department of Paediatric Oncology/Haematology, Regional Children’s Hospital, Novokuznetsk, Russia; 7 Department of Paediatric Oncology/Haematology, Regional Children’s Hospital, Sochi, Russia; 8Department of Paediatric Oncology/Haematology, Regional Oncological Hospital, Balashiha, Russia; 9Department of Paediatric Oncology/Haematology, Regional Children’s Hospital, Ulyanovsk, Russia and 10Department of Paediatric Oncology/Haematology, Regional Children’s Hospital, Syktyvkar, Russia 2

Until 1990, the survival of children with acute lymphoblastic leukaemia (ALL) in Russia was below 10%. To establish a protocol feasible under conditions there, ALL-MB 91 was designed to avoid prolonged bone marrow aplasia, thereby reducing needs for extensive supportive care, blood transfusions, long-lasting hospitalization and costs. High-dose therapies were avoided, anthracycline use was limited and CNS radiation therapy only foreseen in high-risk patients (about 30%). This was randomized against a modified BFM protocol. From 1995 to 2002, 834 patients of age up to 18 years were registered in 10 centres and 713 received after central randomization the allocated risk-stratified treatment. After a median follow-up of 7 years, the event-free survival (EFS) was 67±3% on ALL-MB 91 (N ¼ 358) vs 68±3% on ALL-BFM 90m (N ¼ 355). The overall survival (OS) was 71±3% vs 74±2%, respectively. Anaemia, thrombocytopenia, agranulocytosis 410 days and hospitalization (median 35 vs 68 days) were lower on ALL-MB 91 (Po0.01, N ¼ 197). While EFS and OS were similar with both protocols, ALL-MB 91 significantly incurred fewer toxicity and resource requirements and, therefore, has been increasingly used across Russia. Leukemia (2008) 22, 1144–1153; doi:10.1038/leu.2008.63; published online 27 March 2008 Keywords: acute lymphoblastic leukaemia; multiagent chemotherapy; toxicity analysis; children and adolescents; clinical trial; protocol comparison

Introduction Until the 1980s, the prognosis of children with acute lymphoblastic leukaemia (ALL) was extremely poor in Russia.1,2 There was no common standardized treatment protocol. First attempts to use modern intensive therapies gave poor results because of severe, mostly infectious complications and insufficient Correspondence: Professor Dr G Henze, Charite´FUniversita¨tsmedizin Berlin, Augustenburger Platz 1, 13353 Berlin, Germany. E-mail: [email protected] This study has been presented in part at the annual meeting of the International Society of Paediatric Oncology (SIOP) 2005 in Vancouver. 11 These authors contributed equally to this work. Received 19 October 2007; revised 22 January 2008; accepted 19 February 2008; published online 27 March 2008

experience to manage them adequately due to lack of supportive care, mostly due to financial constraints. Patients had to be hospitalized for long periods of time, and because of lacking accommodation facilities and long distances between their homes and the treatment institutions, the patients and their families were living in the hospital wards for months. Infections, in particular hepatitis B, transmitted by contaminated blood products, led to poor tolerance to treatment and thereby considerable delays. The protocol adherence varied largely, resulting in disparate survival rates.3,4 Hence, to offer these children a realistic chance for cure, there was an urgent need for a structured cooperative treatment approach, using a feasible protocol adapted to the specific requirements of the Russian conditions with consequent supportive care, accompanied by the advice and supervision of experienced paediatric haemato-oncologists, as well as offers to train Russian physicians and nurses abroad. After careful considerations, protocol ALL-MB 91 was designed based on data from international trials, including the German ALL-BFM 76/79, 81/83 and 86 trials,5–8 the Dutch DCLSG ALL-5 and ALL-6 trials9,10 and the Dana–Farber consortium trials DFCI 81-01 and 85-01.11–13 General aims were to use combination chemotherapy aiming at avoiding deep and long-lasting periods of bone marrow aplasia, thereby reducing the needs for extensive supportive care, the frequency of blood transfusions, long-lasting hospitalization and also costs. The cumulative doses of anthracyclines were limited, high- and intermediate-dose therapies avoided and preventive radiation therapy of the central nervous system (CNS) was only foreseen in high-risk patients (HRG, about 30% of the total population). During 1991 and 1995, protocol ALL-MB 91 was introduced in a single centre (Republican Children’s Hospital, Moscow, Russia) and found to be at least as feasible and similarly effective but accompanied with less side-effects than the ‘standard’ ALL-BFM 90m protocol.14,15 Since then, a network of 10 paediatric oncology and haematology centres in Russia was established and trained in the treatment of ALL in children within the context of these protocols. From 1995 until 2002, children with newly diagnosed ALL in these centres were randomized to be treated according to either protocol ALL-MB 91 or protocol ALL-BFM 90m in a multicentric comparison. The results of this study are described below.

Protocol ALL-MB 91 for treatment of childhood ALL in Russia A Karachunskiy et al

1145 Patients and methods

Patients From 1 August 1995 to 30 April 2002, 848 unselected consecutive patients up to 18 years of age with newly diagnosed precursor B- or T-cell ALL were assessed for eligibility in the 10 participating centres in Russia. Fourteen patients were ineligible because they died before accrual procedures (N ¼ 5), their family declined any treatment (N ¼ 6), leukaemia was a secondary malignancy (N ¼ 2) or a storage disease existed (N ¼ 1). Written informed consent for enrolment to this study was obtained from the patient’s guardians and the patient, if applicable, but was refused in 73 patients. After randomization of 761 enrolled patients, 713 patients (94%) were treated according to the allocated treatment arm (Table 1A) and are reported on in the following study. The study was carried out in accordance with the World Medical Association’s Declaration of Helsinki regarding the ethical principles for medical research involving human subjects. It had been approved before its start by the review board of the competent Russian Federal Clinical Research Centre for Paediatric Haematology, Oncology and Immunology.

Diagnosis The diagnosis of precursor B- or T-cell ALL required the presence of 425% lymphoblasts (L1 or L2 morphology) in the Table 1A

bone marrow, and it was made by cytomorphological and cytochemical examination of blood and bone marrow smears at the local centre. CNS involvement was defined as 45 nucleated cells per microlitre cerebrospinal fluid, with clearly detectable leukaemic blasts in the absence of major blood contamination (o17 erythrocytes per microlitre) or as leukaemic CNS infiltration. Immunophaenotyping of the leukaemic cells was performed by four reference laboratories according to standardized procedures if bone marrow blast portion exceeded 60%.16 Due to unavailability under the prevailing health care system, immunophaenotyping could only be performed in 566 (79%) and cytogenetic or molecular genetic tests in 180 (25%) out of the 713 patients. Posteroanterior and lateral chest radiographs were obtained in all patients before treatment initiation, and patients with a mediastinal mass or thymic involvement were regarded as having precursor T-cell leukaemia.17

Treatment After randomization, patients were stratified into the risk groups as defined in the protocols (Table 1B) and treated accordingly. It is of note that by definition, the standard risk group comprised about 70% of patients in ALL-MB 91 and only 32% in ALL-BFM 90m. Protocol ALL-BFM 90m was modified from the original German protocol ALL-BFM 9020 by reducing high dose of methotrexate (5 g m2) to intermediate dose (1 g m2), given as a

Patient randomization

Patients

Total

Eligible for enrolment Enrolment refused ALL diagnosis revised (misinclusion; for example, non-lymphoblastic leukaemia) Randomized to protocol Randomized protocol and any other treatment with curative intention refused Further treatment in centre not participating in ALL-MB 91 or ALL-BFM 90m protocols Received protocol as allocated by randomization

Protocol

834 73 6 755 33 9 713

ALL-MB 91

ALL-BFM 90m

5 379 18 3 358

1 376 15 6 355

Abbreviations: ALL, acute lymphoblastic leukaemia; BFM, Berlin-Frankfurt-Mu¨nster; MB, Moscow-Berlin. Table 1B

Protocol-specific stratification into risk groups ALL-MB 91 Risk group

Criteria18

Initial WBC o50 000 ml1 Age 41 year No CNS involvement No T-cell ALL No mediastinal mass

Standard risk group

PGR BFM-RF o0.8 No CNS disease No T-cell ALL No mediastinal mass

At least one of: Initial WBC X50 000 ml1 Age p1 year CNS involvement T-cell ALL Mediastinal mass

Intermediate-risk group

High-risk group

PGR BFM-RF X0.8 BFM-RF o0.8 and CNS disease T-cell ALL Mediastinal mass No PGR Ph+ ALL

Risk group

Criteria

Standard risk group

High-risk group

ALL-BFM 90m

Abbreviations: BFM-RF, BFM risk factor; BM, bone marrow; n.a, not applicable; PGR, prednisone good response; WBC, white blood cell count. PGR is defined as o1000 ml1 blasts in peripheral blood on day 8. BFM-RF was calculated at diagnosis as follows: BFM-RF ¼ 0.2 log10 (blast cell count (per ml)+1)+0.06 palpable liver size (cm)+0.04 palpable spleen size (cm).19 For definition and analysis of non-response, see ‘Statistical analysis and definition’. Leukemia


1146

36 h infusion followed by three doses of leucovorin (15 mg m2 i.v.); to compensate for the potentially less-effective CNS protection, cranial radiation therapy was performed in all intermediate- (ImRG) and high-risk group (HRG) patients over 1 year of age. Block-wise and cumulative dosages as well as timing schedules are presented in Table 2 for patients on ALL-MB 91 and non-HRG patients on ALL-BFM 90m. HRG patients on ALLBFM 90m were treated with 5-day alternating high-risk blocks

Table 2

Treatment protocols ALL-MB 91 (SRG and HRG) and ALL-BFM 90m (SRG and ImRG)

Drug

ALL-MB 91 Dose and route

Prednisone Dexamethasone Vincristine Daunorubicin L-asparaginase

Induction F 6 mg m2 PO 1.5 mg m2 i.v. 45 mg m2 i.v. 10 000 U m2 i.m.

Mercaptopurine Dexamethasone Vincristine Methotrexate L-asparaginase Daunorubicin Cytarabine

Consolidation I 50 mg m2p.o. 6 mg m2 p.o. 1.5 mg m2 i.v. 30 mg m2 i.m. 10 000 U m2 i.m. 30 mg m2 i.v. F

Cyclophosphamide

F

Mercaptopurine Dexamethasone Vincristine Methotrexate L-asparaginase Daunorubicin

Consolidation II 50 mg m2 p.o. 6 mg m2 p.o 1.5 mg m2 i.v. 30 mg m2 i.m. 10 000 U m2 i.m. 30 mg m2 i.v.

Dexamethasone Vincristine Mercaptopurine Thioguanine Methotrexate L-asparaginase Doxorubicin Cytarabine Cyclophosphamide

Consolidation III 6 mg m2 p.o. 1.5 mg m2 i.v. 50 mg m2 p.o. F 30 mg m2 i.m. 10 000 U m2 i.m. F F F

Methotrexatec

Intrathecal therapyb 6/8/10/12 mg

Cytarabine Prednisone

Leukemia

(initially as used in protocol ALL-BFM 9020 and from 1995 on as in protocol ALL-BFM 95, not tabulated). Supportive care measures had been standardized15 and were specified in the protocol, such as for prophylaxis and treatment of infections, gastritis and peptic ulcers, for management of adverse reactions to L-asparaginase, for blood component substitution and leukapheresis as well as for initial emergency situations, including tumour lysis syndrome, mediastinal tumour with trachea compression, superior vena cava syndrome,

Day(s)

1–28 8, 15, 22, 29, 36, 43 8 (22a) 36, 43

50–91 92–102 92, 99 50, 57, 64, 71, 78, 85 51, 58, 65, 72, 79, 86 (50, 71, 92)a

ALL-BFM 90m Dose and route Procotol Ia 60 mg m2 p.o. F 1.5 mg m2 i.v. 30 mg m2 i.v. 10 000 U m2 i.v. Protocol Ib 50 mg m2 p.o. F F F F F 75 mg m2 i.v. 2

1000 mg m

i.v.

Protocol M 106–147 25 mg m2 p.o. 148–158 F 148, 155 F 106, 113, 120, 127, 134, 141 1000 mg m2 107, 114, 121, 128, 135, 142 F (106, 127)a F Procotol II 10 mg m2 PO 1.5 mg m2 IV F 60 mg m2 PO 162, 169, 176, 183, 190, 197 F 163, 170, 177, 184, 191, 198 10 000 U m2 IV 30 mg m2 IV 75 mg m2 IV 1000 mg m2 IV

204–214 204, 211 162–203

Day(s)

1–28 8, 15, 22, 29 8, 15, 22, 29 12, 15, 18, 21, 24, 27, 30, 33 36–63

38–41, 45–48, 52–55, 59–62 36, 64 78–133 85, 99, 113, 127

148–169 155, 162, 169, 176 183–196 155, 158, 162, 165 155, 162, 169, 176 185–188, 192–195 183

Intrathecal therapyb 1, 8, 15, 22, 29, 43, 92, 148, 6/8/10/12 mg 1, 8d, 15, 22d, 29, 204, 260, 315, 372, 428 45, 59, 85, 99, 113, 127, 185, 192 SRG: 16/20/26/30 mg 1, 8, 15, 22, 29, 43, 92, 148, F HRG: 20/30/40/50 mg 204, 260, 315, 372, 428 4/6/8/10 mg 1, 8, 15, 22, 29, 43, 92, 148, F 204, 260, 315, 372, 428

CNS radiation therapy ALL-MB 91 Risk group

Age range

Weeks 32–33 Dosagee

ALL-BFM 90m Risk group

Age range

Weeks 30–32 Dosagee

SRG HRG HRG HRG

Any age o1 year 1–3 years 43 years

F F 12/12 Gy 18/18 Gy

Any risk group ImRG ImRG HRG HRG

o1 year 1–2 years 42 years 1–2 years 42 years

F 12/18 Gy 12/24 Gy 18/18 Gy 18/24 Gy


1147 Table 2

(Continued)

Maintenance treatment Drug

ALL-MB 91 Dose and route

Mercaptopurine (once daily)

50 mg m2 p.o.

Methotrexat (once weekly)

30 mg m2 i.m.

Dexamethasone (once daily)

6 mg m2 p.o.

Vincristine (on days 1 and 8)

1.5 mg m2 i.v.

ALL-BFM 90m Dose and route

Weeks 32–37, 40–45, 64–65, 72–77, 96–105 32–37, 40–45, 64–65, 72–77, 96–105 38–39, 46–47, 70–77, 78–78, 38–39, 46–47, 70–77, 78–78,

Weeks

48–53, 56–61, 50 mg m2 p.o. 80–85, 88–93,

30–105

48–53, 56–61, 30 mg m2 p.o. 80–85, 88–93,

30–105

54–55, 86–87, 54–55, 86–87,

62–63, F 94–95 62–63, F 94–95

Cumulative drug dosages ALL-MB 91 Drug (unit)

ALL-BFM 90m

SRG

HRG

SRG

ImRG

Intensive therapy Dexamethasone (mg m2) Prednisone (mg m2) Vincristine (mg m2) Methotrexate (mg m2) Daunorubicine (mg m2) Doxorubicine (mg m2) Mercaptopurine (mg m2) Thioguanine (mg m2) L-asparaginase (U m2) Cytarabine (mg m2) Cyclophosphamide (mg m2) Intrathecal therapy (times)

366 F 18 540 45 F 6300 F 200 000 F F 9

366 F 18 540 240 F 6300 F 200 000 F F 9

220 560 12 4000 120 120 2800 840 120 000 1800 3000 11 (13d)

220 560 12 4000 120 120 2800 840 120 000 1800 3000 11 (13d)

Maintenance therapy Mercaptopurine (mg m2) Methotrexate (mg m2) Dexamethasone (mg m2) Vincristine (mg m2) Intrathecal therapy (times)

18 900 1620 672 24 4

18 900 1620 672 24 4

26 600 2280 F F F

26 600 2280 F F F

Abbreviations: BFM, Berlin-Frankfurt-Mu¨nster; CNS, central nervous system; HRG, high-risk group; ImRG, intermediate-risk group; MB, MoscowBerlin; SRG, standard-risk group. ALL-BFM 90m 5-day alternating high-risk blocks (HRG) were given initially as in protocol ALL-BFM 9020 and from 1995 on as in protocol ALL-BFM 95, their blockwise and cumulative drug dosages are not shown due to small number of patients (Table 3) and variable number of blocks. ALL-MB 91: given on day 22 to all patients in HRG or to patients in SRG with 410% BM blasts on day 15. a ALL-MB 91: only patients in HRG are scheduled for daunorubicin (days 22, 50, 71, 92, 106 and 127) and for radiation therapy. b Intrathecal therapy: dosages given for age ranges o1 years/1–2 years/2–3 years/43 years. c ALL-MB 91: methotrexate given no more after CNS radiation therapy. d ALL-BFM 90m: given only in case of CNS involvement at diagnosis. e Dosages in patients without/with CNS involvement at diagnosis.

pericardial effusions, oliguria/anuria, extensive kidney infiltration, urate nephropathy and obstruction of urinary ways.

Study design and management The purpose of this study was to compare the new protocol ALL-MB 91 with respect to survival as well as toxicity and cost indicators against the control protocol ALL-BFM 90m. Event-free and overall survival probabilities were primary endpoints. Complete remission rate, frequencies and distributions of toxicity-related data were secondary end points. ALL-MB 91 was not expected to be superior to ALL-BFM 90m with regard to survival, but better manageable and less costly while maintaining acceptable survival rates. To support that the new protocol was not inferior to control, the confidence interval of the difference in their event-free survival (EFS) probabilities was considered according to Machin D and Gardner MJ21and Piaggio G et al.22 in advance of further analyses as specified in

the protocol. Risk-group-specific comparisons were not possible because definitions varied largely between the protocols. In a blinded fashion, enrolled patients were centrally randomized by the trial coordination centre (Research Institute for Paediatric Haematology and Oncology, Moscow) using a randomization list with equal probabilities for the two protocols. The coordination centre continuously monitored the trial, receiving reports on accrual, treatment course and events from participating centres, and requesting status updates and the resolution of pending queries at least every 6 months. Telephone consultation and email were used throughout the study to exchange information. For toxicity analysis, full on-site patient record analysis was performed for the first 197 of 713 (28%) patients receiving the allocated protocol. Patients in this analysis were included consecutively irrespective of site after completion of intensive treatment phases and representativity was assessed at study completion. Leukemia


1148

Statistical analysis and definitions Estimates of EFS, disease-free (DFS) and overall survival (OS) probability (±standard error) were calculated by Kaplan and Meier’s method23 and were compared using the log-rank test.24 Cumulative incidences were estimated according to Kalbfleisch and Prentice,25 and plots and statistical comparisons were produced using Gray’s methods.26,27 The accrual of 50 patients per protocol annually for 7 years, zero drop-in or dropout rates and an estimated EFS probability of 0.67 at 10 years would have provided for a power of 0.75 at a type I error of 0.05 to detect a clinically meaningful absolute survival difference of 10% or more.28 Results are reported for analyses in the per-protocol population (by protocol received as allocated by randomization, N ¼ 713; Table 1A and 1B), if not stated otherwise; for comparison, analyses in the intention-to-treat population (as randomized to protocol, N ¼ 755; Table 1A and 1B) were carried out. Events were non-response, relapse, second malignancy and death in case of EFS, relapse and death in case of DFS and death in case of OS. Survival times were calculated from attaining complete remission until event for patients who failed or until the date of last status report for patients who did not fail. Patients who did not attain complete remission, that is, those who died early and those with non-response were considered to have zero EFS time. Non-response was defined as more than 5% lymphoblasts in the bone marrow on day 36 (ALL-MB 91) or on day 33 (ALL-BFM 90m) of treatment, whereas the original definition in protocol ALL-BFM 90 would not be applicable until the completion of three further treatment blocks. Patients allocated to either protocol went off-study upon non-response, with an event, and are included in this study only in respect of OS, although some patients were further treated with BFM HRG blocks (see Treatment). Frequencies and distributions were compared with the w2 or Fisher’s exact test and with the U-test, respectively. For calculations, Graphpad Prism version 3.0 (GraphPad Software Inc., San Diego, CA, USA), Statistica version 6.0 (Statsoft Inc., Tulsa, OK, USA), and R version 2.4.0 were used on a database frozen on 1 March 2006.

Results

Patient characteristics Characteristics of the 713 patients receiving the protocol allocated by randomization and analysed in this study are detailed in Table 3. Time from diagnosis to randomization was median 0 days (maximum, 2 days). In patients without event, the median follow-up time was 6.6 (interquartile range (IQR), 3.9– 10.6) years on protocol ALL-MB 91 and 6.8 (IQR, 3.9–10.6) years on protocol ALL-BFM 90m. No protocol crossover occurred. Eight patients were lost to follow-up and censored in the analyses at 852, 1398, 1691, 2053, 2233, 2304, 2422 and 2462 days after diagnosis. The distribution of the baseline risk factors, such as sex, age, white blood cell count, CNS involvement and immunophaenotype (Table 3), were not different between patients on either protocol.

Outcome The estimates of the EFS probability projected at 10 years were 67±3% (N ¼ 358) on protocol ALL-MB 91 and 68±3% (N ¼ 355) on protocol ALL-BFM 90m (Plog-rank ¼ 0.66, Table 3, Figure 1a). The 95% two-sided confidence interval of this 1% difference in EFS probability estimates was 9 to þ 7%. At this Leukemia

level of confidence, an inferiority of ALL-MB 91 to ALL-BFM 90m by 10% or more of absolute EFS probability can be excluded. The estimates of OS probability projected at 10 years were 71±3% and 74±2% (Plog-rank ¼ 0.68, Table 3, Figure 1b) and the estimates of DFS probability were 75±2% and 75±3% (Plog-rank ¼ 0.94) in patients on protocol ALL-MB 91 and ALLBFM 90m, respectively. For comparison, in the intention-to-treat population, the estimates of DFS probability projected at 10 years were 74±2% (ALL-MB 91; N ¼ 339 patients, who had obtained remission) and 75±2% (ALL-BFM 90m; N ¼ 337; Plog-rank ¼ 0.63). The cumulative incidence curves of relapses under competing risks in patients who had obtained remission were not different by protocol (Figure 2, PGrayX0.3 for all relapse types). On protocol ALL-MB 91, isolated CNS relapses occurred in eight HRG patients (7.1% of HRG, 2.2% of protocol patients, Table 4A and 4B) and in only two standard risk group patients (0.8% of SRG and 0.6% of protocol patients). Among all 10 patients with isolated CNS relapses, 5 had an initial white blood cell count 4100 000 ml1.

Toxicity During the first days of treatment, two and five of the patients on protocols ALL-MB 91 and ALL-BFM 90m, respectively, died from hyperleukocytosis, encephalopathy or bleeding. During further induction treatment, 11 and 17 patients died from sepsis or pneumonia (9/15 patients on protocol ALL-MB 91/ALL-BFM 90m, respectively), necrotizing enterocolitis (1/1), chicken pox (0/1) and adrenal insufficiency (1/0). Causes of death in complete remission were infection (sepsis and pneumonia, 7/2 or necrotizing enterocolitis, 0/2), massive bleeding (2/2), necrotizing pancreatitis (1/1), catheter complications (1/0) or were unknown (1/0). There were no significant differences in the rates of early deaths, deaths in complete remission , relapses and secondary malignancies (one frontal lobe brain tumour) between the protocols (Table 4A). Toxicity as defined by grades of anaemia, thrombocytopenia and agranulocytosis of more than 10 days duration is summarized in Table 4B in relation with the intensive treatment phases. Notably, during induction therapy, no difference could be found for the above-mentioned parameters, whereas during Consolidation I/Protocol Ib and Consolidation II/Protocol M, the haematological toxicity was significantly lower in patients on ALL-MB 91. In Consolidation III/Protocol II of this protocol, toxicity was significantly lower in patients only with respect to platelet counts. The duration of hospitalization was markedly less for patients on protocol ALL-MB 91 (median 35 days; IQR, 19–61 days) compared to those on protocol ALL-BFM 90m (median 68 days; IQR, 40–99 days; PU testo0.001). Avascular bone necroses occurred in two patients on protocol ALL-BFM 90m and in no patient on ALL-MB 91.

Discussion The efficacy and toxicity were compared between a slightly modified version of the internationally established protocol ALL-BFM 90 and a new protocol ALL-MB 91 designed for the treatment of children with newly diagnosed precursor B- or T-cell lymphoblastic leukaemia in Russia in a randomized multicentre study. The study demonstrated that the EFS on protocol ALL-MB 91 was not inferior to that on protocol ALL-BFM 90m, at a 10%


1149 Table 3

Patient characteristics and associated outcome.

Protocol

N

EFS

OS

MB

BFM

PFischer

MB

BFM

Plog-rank

MB

BFM

Plog-rank

358

355

n.a.

0.67±0.03

0.68±0.03

0.81

0.71±0.03

0.74±0.02

0.69

SRG (MB), SRG (BFM)

246

115

n.a.

0.73±0.03

0.81±0.04

n.a.

0.77±0.03

0.89±0.03

n.a.

HRG (MB), ImRG (BFM) HRG (BFM)

112 n.a.

197 43

n.a. n.a.

0.54±0.05 n.a.

0.66±0.04 0.42±0.08

n.a. n.a.

0.58±0.05 n.a.

0.75±0.03 0.39±0.07

n.a. n.a.

Sex Boys Girls

186 172

201 154

0.21

0.70±0.03 0.64±0.04

0.64±0.03 0.74±0.04

0.198 0.037

0.72±0.04 0.70±0.04

0.70±0.03 0.81±0.03

0.35 0.085

Age o1 years X1 years o10 X10 years

10 261 87

9 255 91

0.87

0.40±0.15 0.70±0.03 0.61±0.05

0.44±0.17 0.72±0.03 0.59±0.05

0.71 0.64 0.78

0.50±0.16 0.75±0.03 0.61±0.06

0.44±0.17 0.80±0.03 0.62±0.05

0.86 0.36 0.68

Initial WBC (per nl) o10 X10 and o50 X50 and o100 X100

173 118 28 39

172 101 36 46

0.41

0.76±0.03 0.65±0.05 0.73±0.09 0.28±0.07

0.81±0.03 0.66±0.05 0.47±0.08 0.43±0.07

0.20 0.99 0.034 0.081

0.81±0.03 0.68±0.05 0.74±0.10 0.33±0.08

0.88±0.02 0.73±0.05 0.51±0.10 0.43±0.08

0.25 0.79 0.074 0.20

CNS status at diagnosis No involvement Involvement Unknown

316 32 10

322 21 15

0.13

0.69±0.03 0.53±0.09

0.70±0.03 0.57±0.11

0.72 0.75

0.73±0.03 0.54±0.09

0.76±0.03 0.60±0.11

0.78 0.64

Immuno-phaenotype Non T-cell ALL T-cell ALL Unknown

251 34 73

242 39 74

0.78

0.70±0.03 0.41±0.08

0.71±0.03 0.56±0.08

0.90 0.17

0.74±0.03 0.40±0.08

0.78±0.03 0.56±0.08

0.52 0.33

NCI criteriaa Standard risk High risk

158 87

151 84

1

0.75±0.04 0.64±0.05

0.81±0.03 0.53±0.06

0.19 0.17

0.81±0.03 0.64±0.06

0.92±0.02 0.57±0.06

0.026 0.32

311 34

0.17

0.71±0.03 0.41±0.07

0.73±0.03 0.41±0.08

0.63 0.88

0.75±0.03 0.48±0.08

0.80±0.02 0.39±0.09

0.45 0.53

227 22

0.65

0.74±0.03 0.38±0.10

0.75±0.03 0.50±0.11

0.64 0.47

0.78±0.03 0.52±0.10

0.84±0.03 0.50±0.11

0.19 0.83

79 9 10

0.044

0.66±0.06 0.50±0.12

0.68±0.05 0.11±0.10

0.83 0.14

0.67±0.07 0.49±0.12

0.70±0.05 0.11±0.10

0.81 0.09

296 44

0.0003

0.74±0.03 0.49±0.06

0.76±0.03 0.31±0.08

0.65 0.13

0.77±0.03 0.56±0.07

0.81±0.02 0.37±0.08

0.29 0.013

214 30

0.003

0.76±0.03 0.51±0.07

0.79±0.03 0.33±0.10

0.52 0.19

0.79±0.03 0.63±0.08

0.87±0.02 0.42±0.10

0.051 0.016

74 14 15

0.044

0.69±0.06 0.46±0.10

0.67±0.06 0.29±0.12

0.77 0.48

0.70±0.07 0.44±0.11

0.69±0.06 0.29±0.12

0.43 0.32

Total

Response day 8 (Blasts per ml) All patients 302 o1000 ml1 46 X1000 ml1 Age 1–9 years 1 229 o1000 ml 26 X1000 ml1 Age X10 years 1 66 o1000 ml 18 X1000 ml1 Unknown 10 Response day 15 (BM blast %) All patients o25% 265 X25% 83 Age 1–9 years o25% 198 X25% 57 Age X10 years o25% 59 X25% 25 Unknown 10

Abbreviations: ALL, acute lymphoblastic leukaemia; BFM, Berlin-Frankfurt-Mu¨nster; BM, bone marrow; CNS, central nervous system; EFS, eventfree survival at 10 years; HRG, high-risk group, ImRG, intermediate-risk group, MB, Moscow-Berlin; n.a., not applicable; OS, overall survival at 10 years; WBC, white blood cell count. Significant P-values (without correction for multiplicity) are underlined. a NCI criteria for non-T cell leukaemia.29

Leukemia


1150

Figure 1 Outcome of children with acute lymphoblastic leukaemia (ALL) treated with protocol ALL-MB 91 or ALL-BFM 90m in Russia. The (a) event-free survival and (b) overall survival are shown for comparing protocol ALL-MB 91 with ALL-BFM 90m. The event-free survival by protocolspecific stratification criteria (Table 1B) is shown for protocols (c) ALL-MB 91 and (d) ALL-BFM 90m. P refers to the log-rank test.

Figure 2 Cumulative incidences of different types of relapses in children with acute lymphoblastic leukaemia (ALL) by protocol. The incidence estimates of relapses over time are shown, comparing protocol ALL-MB 91 with ALL-BFM 90m: isolated bone marrow (BM) relapse (PGray ¼ 0.3), isolated central nervous system relapse (PGray ¼ 0.3) and other relapse (PGray ¼ 0.9). In brackets, estimates at 10 years. Risks competing with relapse: death in remission, second malignoma; censored: no event, lost to follow-up, non-response, death during induction. Leukemia

margin in EFS at 10 years. Moreover, the Kaplan–Meier curves for EFS and OS (Figure 1a and b) as well as DFS (not shown) are almost superimposed. Response rates at day 8 (peripheral blast count o1000 ml1) and day 15 (blast portion o25% in normocellular bone marrow) were slightly lower to significantly lower, in patients on protocol ALL-MB 91 compared with ALL-BFM 90m. Response at day 8 was associated with favourable EFS and OS (about 72 and 77%), but there were no differences in EFS or OS between both protocols. Response at day 15 was associated with slightly favourable to significantly favourable EFS and OS for patients on protocol ALL-MB 91 compared with ALL-BFM 90m, partly compensating for the slightly higher rate of non-response on protocol ALL-MB 91. During the further course of patients in remission, the time points and types of relapses were not different between the two protocols, as shown in the cumulative incidence curves (Figure 2). Characteristic features of the newly designed protocol ALLMB 91 are: the extended use of asparaginase, of dexamethasone and of prolonged intrathecal therapy instead of cranial radiation therapy for the majority of patients; the omission of intermediate- and high-dose methotrexate and cyclophosphamide and very low or moderate cumulative doses of anthracyclines for standard or high-risk patients, respectively. The superior antileukaemic efficacy of dexamethasone compared to prednisone was demonstrated in several trials.9,10,12 It can be explained by a higher affinity of the steroid receptor, less protein binding, more readily crossing of blood brain barrier, longer half-lives in serum and cerebrospinal fluid of dexamethasone.30,31 Its use in protocol ALL-MB 91 was


scheduled in analogy to the Dutch ALL-69,10 and the Children’s Cancer Group (CCG) 199219 trials at a dose of 6 mg m-2 day1, which had not been associated with thromboembolic events or avascular bone necrosis. Nevertheless, dexamethasone was not more effective on day 8 as randomly compared to prednisone in this study. The antileukaemic efficacy of weekly administered L-asparaginase was demonstrated in trials by the Dana–Farber Cancer Institute,11–13,32 especially for consolidation. In addition, its intramuscular injection, as in protocol ALL-MB 91, has been shown in a Children’s Cancer Study Group (CCSG) trial to be equally effective, but less toxic, than intravenous administration.33 Central nervous system-directed antileukaemic treatment in protocol ALL-MB 91 relied on extended dexamethasone use, on intensified L-asparaginase to more continuously deplete cerebrospinal fluid of asparagine, and also on prolonged use of triple intrathecal therapy. Compared to cranial radiation therapy, several trials have shown sufficient protection against CNS relapses by prolonged intrathecal therapy,33,34 except in children with T-cell phenotype or initial CNS involvement.20

Table 4A protocol

Adverse events and outcome results according to

Protocol

ALL-MB 91

ALL-BFM 90m PFisher

N Percentage N Percentage Received allocated protocol 358 Early death 13 Non-response 26 CR 319 Death in CR 11 Relapses (total) 66 Relapse BM CNS Testis Other BM and CNS BM and testis BM and other site Secondary malignancy Lost to follow-up Continuous CR

100 3.6 7.3 89.1 3.1 18.4

355 22 14 319 6 69

100 6.2 3.9 89.9 1.7 19.4

0.12 0.07 0.81 0.33 0.78

11.5 2.8 0 0.3 3.1 0.8 0 0 0.6 67.0

49 6 1 0 6 5 2 1 6 237

13.8 1.7 0.3 0 1.7 1.4 0.6 0.3 1.7 66.8

0.37 0.45 0.5 1 0.33 0.5 0.25 0.5 0.18 1

41 10 0 1 11 3 0 0 2 240

Abbreviations: BM, bone marrow; CNS, central nervous system, CR, complete remission.

Table 4B

1151 Central nervous system-directed antileukaemic treatment in protocol ALL-BFM 90m included cranial radiation therapy for standard and intermediate-risk patients to compensate for the reduction from high, as in original ALL-BFM 90, to intermediate dose of methotrexate required, because reliable facilities for prompt methotrexate serum level monitoring were not available. However, the specific contribution of high-dose methotrexate to the CNS-directed antileukaemic effect is not known because it had been introduced at the same time as a different dosing regime for intrathecal methotrexate into, for example, trial ALLBFM 86.8 There is intriguing evidence that basing the dosing regime for intrathecal methotrexate on age, rather than on body surface area, readily reduces CNS relapse rates from 8–27% to 2–6%.35 As a consequence, CNS-directed therapy differed markedly between the protocols. Only about 30% of patients on ALL-MB 91 but more then 67% on ALL-BFM 90m (ImRG and HRG) received cranial radiation therapy. The incidence of isolated CNS relapses on protocol ALL-MB 91 (2.8%) was slightly higher than that on ALL-BFM 90m, but it was only 0.8% in standard risk group patients, comparable with international results.36 Remarkably, this favourable rate of isolated CNS relapses in SRG patients (about 70% of patients) on ALL-MB 91 was obtained using neither intermediate- or high-dose methotrexate nor cranial radiation therapy. Applying the National Cancer Institute (NCI)/Rome criteria in precursor B-cell ALL, no statistically significant difference in EFS probability between protocol ALL-MB 91 and ALL-BFM 90m could be detected (Table 3). Due to unavailability of facilities and trained personnel in Russia at that time, results from immunophaenotyping and cytoor molecular genetic characterizations were available only in 79 and 25% of the patients, respectively, in this study. However, considering the low Ph þ ALL incidence of less than 3.5% among Caucasian children and adolescents with ALL,37 and their increased, but not universal failure to achieve complete remission (about 15%), the interpretation of our results would remain valid even if such patients had been over-represented in ALL-BFM 90m. Although protocol ALL-MB 91 did not introduce new drugs for the treatment of ALL in children, the chemotherapy design was completely newly composed according to the available efficacy data. These new induction, consolidation and maintenance blocks apparently allowed for a different course of leukaemic cell clearance and, possibly even more important, for less treatment-related compromise of haematological and

Haematological toxicity according to protocol Hb o7 g per 100 ml

Phase MB

BFM

Induction Consolidation I Consolidation II Consolidation III

Ia Ib M II

Thr o20 per nl

Long agranulocytosis

MB

BFM

PFisher

MB

BFM

PFisher

MB

BFM

PFisher

35/98 10/93 7/87 7/84

36/99 44/98 22/99 12/96

1.0 o0.0001 0.0085 0.46

41/98 9/93 2/87 6/84

45/99 30/98 19/99 13/96

0.67 0.0003 0.0003 0.22

64/98 44/93 37/87 27/84

75/99 67/98 67/99 70/96

0.12 0.004 0.001 o0.0001

Abbreviations: BFM, Berlin-Frankfurt-Mu¨nster; CTCAE, common terminology criteria for adverse events; Hb, haemoglobin; MB, Mascow-Berlin; Thr, thrombocytes. By protocol, the numbers of patients experiencing toxicity/numbers of patients exposed are given. The table is based on the first 197 patients (cf. Patients and methods section). Haemoglobin o7 g per 100 ml corresponding to anaemia of CTCAE grade 3 (o8 g per 100 ml) or 4 (o6.5 g per 100 ml). Thrombocyte counts o20 per nl corresponding to thrombocytopenia of approximately CTCAE v3.0 grade 4 (o25 per nl). Long-lasting (more than 10 consecutive days) agranulocytosis (neutrophils o0.5 per nl), corresponding to agranulocytosis of CTCAE v3.0 grade 4. Leukemia


1152 physiological function, as reflected by the response and toxicity data. In this study, toxicity analysis focussed on the acute haematological toxicity, because it determines feasibility and costs of intensive and supportive treatment; late effects are still under registration. Protocol ALL-MB 91 resulted in less toxicity compared to protocol ALL-BFM 90m during intensive treatment after induction. During induction, there was a trend towards a lower portion of patients suffering an early death or experiencing long-lasting agranulocytosis on protocol ALL-MB 91. Since toxicity during induction, however, is significantly resulting from disease-related bone marrow insufficiency, protocol ALLMB 91 appears favourable with respect to both, hazards during induction and haematological toxicity incurred after induction. When ALL-BFM 90m was adopted in Russia as a first ‘standard guideline’, significant treatment delays occurred in one third of patients because of treatment-related toxicity.4,38 However, along with this adoption, physician and nursing practices were greatly advanced, and inter-centre cooperation was initiated. EFS of children with ALL increased from 5–10%39,40 to about 60%. In the 10 centres participating in this study protocol, ALL-BFM 90m was performed with results approaching internationally obtained results.40 The randomized multicentric comparison of ALL-MB 91 with ALL-BFM 90m in this study demonstrated that both protocols were feasible and equally successful under conditions of the Russian health-care system. There were virtually no differences concerning the EFS curves over the range of 10 years (Figure 1) and rates of remission, course of relapses (Figure 2), and fatal complications were also in harmony between both protocols. Nevertheless, the supposed advantages of the new protocol ALL-MB 91 were confirmed, such as a lower haematological toxicity with diminished morbidity and hospitalization, which have been reported to translate into cost reductions (relative estimated costs of intensive treatment for non-HRG patients of ALL-MB 91 vs ALL-BFM 90m, 1:1.35).15 It is also reasonable to expect that protocol ALL-MB 91 will ultimately result in lesser frequencies of adverse long-term sequelae due to the reduced exposure of the CNS to radiation therapy and the much lower exposure of the heart to anthracyclines. In the mean time, the positive experience with ALL-MB 91 has led to a widespread acceptance of the protocol and the formation of a Russian ALL study group. ALL-MB 91 has become the basis for the current trial ALL-MB 2002 with 37 participating institutions across Belarus and Russia, recruiting between 400 and 500 patients per year. In most highly developed western countries, the treatment of childhood ALL has become more and more sophisticated with stratification based on molecular genetic findings, monitoring of minimal residual disease and rather complex multidrug chemotherapy schedules. The introduction of these techniques has led to a moderate further increase of cure rates but also to constantly increasing costs. The ALL-MB group aims at developing an effective and feasible treatment while keeping the costs affordable, using simple clinical stratification criteria and limiting acute and long-term toxicity. As shown, with this concept, it has been possible to achieve results approaching even to international standard protocols at that time.36,40,41 In our opinion, the ALL-MB 91 concept might be an interesting and attractive option for countries with less developed health systems and limited resources, which still constitute the larger part of the world population. Furthermore, the results support the assumption that less intensive therapy may be sufficient for a population of children with ALL and standard risk features even in countries with well-developed health-care systems. Leukemia

Acknowledgements We thank all the Russian paediatric oncologists and laboratory co-workers from the participating centres involved, especially E Samochatova, S Plyasunova, M Dubrovina, T Nasedkina, K Kondratchik, G Novichkova, O Fuchs, E Rogacheva and D Lavruchin (all from Moscow); E Inushkina (Balashicha); O Chlebnikova, O Streneva (both from Ekaterinburg); O Plaksina (Nishni Novgorod); M Goroshkova, A Zemlyanskiy, E Shabashkewich (all from Nowokuznezk); L Chazaryan (Sochi) and G Pavlova (Cheboksari). We also thank all our German colleagues from the Charite´FUniversita¨tmedizin Berlin for their help in the education of Russian doctors and nurses, in particular, senior consultant Ru¨diger Fengler, senior nurse Ina Arlt, and for help in statistics, Reinhard Hartmann. We also thank to Dr Martin Friedrichs, German Federal Foreign Office, and to all the members of the charity organization Kontakty-Kontakte e.V., Berlin, in particular project manager Eberhard Radczuweit, for continuous comprehensive support of our efforts to conduct this first multicentre trial in Russia. Meetings and travel expenditures of physicians, coworkers in the trial coordination center, technical equipment and drugs of this study were in part financed by the charity organization ‘Kontakty-Kontakte e.V.’, Berlin, Germany (http:// www.kontakte-kontakty.de/).

References 1 Alekseev N, Vorontsov I. Leukaemias in children. Leningrad 1988. 2 Kisliak NS, Makhonova LA, Koshel’ IV, Kurmashov VI, Maiakova SA, Rumiantsev AG et al. Late (after 5-year remission) recurrences of acute leukemia in children (a cooperative study). Gematol Transfuziol 1989; 34: 8–12. 3 Rumiantsev AG, Samochatova EV, Khamdan T. Treatment of acute lymphoblastic leukemia in children according to the BerlinFrankfurt-Mu¨nster Program. Pediatriia 1991; 11: 58–63. 4 Rumiantsev AG, Samochatova EV, Maschan AA, Kryzhanovskii OI. Supportive care in the treatment of childhood acute lymphoblastic leukemia according to BFM programme. Pediatriia 1992; 2: 68–73. 5 Henze G, Langermann HJ, Bra¨mswig J, Breu H, Gadner H, Schellong G et al. Ergebnisse der Studie BFM 76/79 zur Behandlung der akuten lymphoblastischen Leuka¨mie bei Kindern und Jugendlichen. Klin Pa¨diatrie 1981; 193: 145–154. 6 Riehm H, Gadner H, Henze G, Kornhuber B, Lampert F, Niethammer D et al. Results and significance of six randomized trials in four consecutive ALL-BFM studies. Haematol Blood Transfus 1990; 33: 439–458. 7 Reiter A, Schrappe M, Ludwig WD, Lampert F, Harbott J, Henze G et al. Favorable outcome of B-cell acute lymphoblastic leukemia in childhood: a report of three consecutive studies of the BFM group. Blood 1992; 80: 2471–2478. 8 Reiter A, Schrappe M, Ludwig WD, Hiddemann W, Sauter S, Henze G et al. Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients. Results and conclusions of the multicenter trial ALL-BFM 86. Blood 1994; 84: 3122–3133. 9 Veerman AJ, Hahlen K, Kamps WA, van Leeuwen EF, de Vaan GA, Vanwering ER et al. Dutch childhood leukemia study group: early results of study ALL VI (1984–1988). Haematol Blood Transfus 1990; 33: 473–477. 10 Veermann A, Ha¨hlen K, Kamps W, van Leeuwen E, de Vaan G, Solbu G et al. High cure rate with a moderately intensive treatment regimen in non-high-risk childhood acute lymphoblastic leukemia: Results of protocol ALL-VI from the Dutch Leukemia Study Group. J Clin Oncol 1996; 14: 911–918. 11 Clavell LA, Gelber RD, Cohen HJ, Hitchcock-Bryan S, Cassady JR, Tarbell NJ et al. Four-agent induction and intensive asparaginase therapy for childhood acute lymphoblastic leukemia. N Engl J Med 1986; 315: 657–663. 12 Barr RD, DeVeber LL, Pai KM, Andrew M, Halton J, Cairney AE et al. Management of children with acute lymphoblastic leukemia by the Dana–Farber cancer institute protocols. Am J Pediatr Hem Onc 1992; 14: 136–139.


1153 13 Schorin MA, Blattner S, Gelber RD, Tarbell NJ, Donnelly M, Dalton V et al. Treatment of childhood acute lymphoblastic leukemia: results of Dana–Farber institute/children’s hospital acute lymphoblastic leukemia consortium protocol 85-01. J Clin Oncol 1994; 12: 740–747. 14 Karachunskiy A, Stackelberg A, Miakova N. Comparative treatment results of childhood acute lymphoblastic leukemia according to protocols ALL-BFM 90 and ALL-MB 91. Pediatriia 1995; 2: 10–16. 15 von Stackelberg A, Karatchunsky A, Kudrjashova J, Miakova N, Belikova L, Rumiantzev A et al. Toxicity, supportive care and costs of two chemotherapy protocols for treatment of childhood all in Russia: BFM 90m and MB 91. Eur J Cancer 1999; 35: 1349–1355. 16 Bene MC, Castoldi G, Knapp W, Ludwig WD, Matutes E, Orfao A et al. Proposals for the immunological classification of acute leukemias. European Group for the Immunological Characterization of Leukemias (EGIL). Leukemia 1995; 9: 1783–1786. 17 Uckun FM, Sensel MG, Sun L, Steinherz PG, Trigg ME, Heerema NA et al. Biology and treatment of childhood T-lineage acute lymphoblastic leukemia. Blood 1998; 91: 735–746. 18 Langermann HJ, Henze G, Wulf M, Riehm H. Estimation of tumor cell mass in childhood acute lymphoblastic leukemia: prognostic significance and practical application. Klin Pa¨diatr 1992; 194: 209–213. 19 Bostrom BC, Sensel MR, Sather HN, Gaynon PS, La MK, Johnston K, et al., Children’s Cancer Group. Dexamethasone versus prednisone and daily oral versus weekly intravenous mercaptopurine for patients with standard-risk acute lymphoblastic leukemia: a report from the Children’s Cancer Group. Blood 2003; 101: 3809–3817. 20 Schrappe M, Reiter A, Ludwig WD, Harbott J, Zimmermann M, Hiddemann W et al. Improved outcome in childhood acute lymphoblastic leukemia despite reduced use of anthracyclines and cranial radiotherapy. Blood 2000; 95: 3310–3322. 21 Machin D, Gardner MJ. Calculating confidence intervals for survival time analyses. Br Med J 1988; 296: 1369–1371. 22 Piaggio G, Elbourne DR, Altman DG, Pocock SJ, Evans SJW. Reporting of noninferiority and equivalence randomized trials: an extension of the CONSORT statement. JAMA 2006; 295: 1152–1160. 23 Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: 457–481. 24 Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966; 50: 163–170. 25 Kalbfleisch J, Prentice R. The Statistical Analysis of Failure Time Data. Wiley: New York (NY), 1980. 26 Gray B. cmprsk: subdistribution analysis of competing risks (resource on the Internet). R package version 2.1–7 [cited 2007 Oct 12]. Available from: http://www.r-project.org, http://biowww. dfci.harvard.edu/Bgray. 27 Gray RJ. A class of k-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988; 16: 1141–1154. 28 Harrell Jr FE and with contributions from many other users [resource on the Internet] Hmisc: Harrell miscellaneous (cited 2007 Oct 12). Available from: http://biostat.mc.vanderbilt.edu/s/ Hmisc.

29 Smith M, Arthur D, Camitta B, Carroll AJ, Crist W, Gaynon P et al. Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukaemia. J Clin Oncol 1996; 14: 18–24. 30 Jones B, Freeman AI, Shuster JJ, Jacquillat C, Weil M, Pochedly C et al. Lower incidence of meningeal leukemia when prednisolone is replaced by dexamethasone in the treatment of acute lymphoblastic leukemia. Med Pediatr Oncol 1991; 19: 269–275. 31 Balis FM, Lester CM, Chrousos GP, Heideman RL, Poplack DG. Differences in cerebrospinal fluid penetration of corticosteroids: possible relationship to the prevention of meningeal leukemia. J Clin Oncol 1987; 5: 202–207. 32 Balis FM, Poplack D. Central nervous system pharmacology of antileukemic drugs. Am J Pediatr Hemtol Oncol 1989; 11: 74–86. 33 Nesbit M, Chard R, Evans A, Karon M, Hammond GD. Evaluation of intramuscular versus intravenous administration of L-asparaginase in childhood leukemia. Am J Pediatr Haematol Oncol 1979; 1: 9–13. 34 Kamps WA, Bo¨kkerink JP, Hahlen K, Hermans J, Riehm H, Gadner H et al. Intensive treatment of children with acute lymphoblastic leukemia according to ALL-BFM 86 without cranial radiotherapy: results of DCLSG protocol ALL-7 (1988–1991). Blood 1999; 94: 1226–1236. 35 Bleyer WA, Coccia PF, Sather HN, Level C, Lukens J, Niebrugge DJ et al. Reduction of central nervous system leukemia with a pharmacokinetically derived intrathecal methotrexate dosage regimen. J Clin Oncol 1983; 1: 317–325. 36 Clarke M, Gaynon P, Hann I, Harrison G, Masera G, Peto R, et al., Childhood ALL Collaborative Group. CNS-directed therapy for childhood acute lymphoblastic leukemia: childhood ALL collaborative group overview of 43 randomized trials. J Clin Oncol 2003; 21: 1798–1809. 37 Chessells JM, Hall E, Prentice HG, Durrant J, Bailey CC, Richards SM. The impact of age on outcome in lymphoblastic leukaemia; MRC UKALL X and XA compared: a report from the MRC Paediatric and Adult Working Parties. Leukemia 1998; 12: 463–473. 38 Rumiantsev AG, Samachatova EV, Karachunsky AI. Intensive chemotherapy of acute leukaemias (AL) in childhood with BFM protocols and Moscow experiences 1990–1992. First successes and failures. In: Bu¨chner et al. (eds). Acute Leukemias IV. Prognostic Factors. Springer: Berlin, 1993, pp 344–346. 39 Samochatova EV, Aslanyan KS, Belikova L. The retrospective analysis of the treatment results of childhood acute lymphoblastic leukemia according to modified BFM protocol in the 9 Russian clinics (1991–1996). Pediatriia 1997; 4: 34–37. 40 Miakova NV, Aleinikova OV, Hartmann R, Rudneva AE, Trubina NM, Fechina LG et al. Treatment of acute lymphoblastic leukemia in children according to the ALL-BFM 90m protocol in the Russian Federation and the Republic of Belarus. Vopr Onkol 2002; 48: 347–351. 41 Schrappe M, Reiter A, Zimmermann M, Harbott J, Ludwig WD, Henze G et al. Long-term results of four consecutive trials in childhood ALL performed by the ALL-BFM study group from 1981 to 1995. Berlin-Frankfurt-Mu¨nster. Leukemia 2000; 14: 2205–2222.

Leukemia