IDA-FLAG (idarubicin, fludarabine, cytarabine, G-CSF), an effective remission-induction therapy for poor-prognosis AML of childhood prior to allogeneic or autologous bone marrow transplantation: experiences of a phase II trial

Authors


Dr Gudrun Fleischhack Department of Paediatric Haematology/Oncology, University Bonn, Adenauerallee 119, D-53113 Bonn, Germany.

Abstract

A phase II trial was designed to explore the potential feasibility and efficacy of a reinduction therapy consisting of fludarabine, cytarabine, idarubicin and granulocyte colony stimulating factor (G-CSF) for acute myelogenous leukaemia (AML) patients with poor prognosis.

Twenty-three patients aged 1.2–17.5 years with refractory (n = 3), relapsed (n = 19) or secondary (n = 1) AML were treated with the IDA-FLAG regimen, a combination therapy of idarubicin (days 2–4, 12 mg/m2/d), fludarabine (days 1–4, 30 mg/m2/d), cytarabine (days 1–4, 2000 mg/m2/d) and G-CSF (day 0 up to ANC > 1 × 109/l, 400 μg/m2/d). They received a total of 37 courses of IDA-FLAG and/or FLAG (IDA-FLAG without idarubicin). 17/23 patients achieved a complete remission (CR) with a median duration of 13.5 months (1–39 months), one patient showed a partial remission, and five were nonresponders while in CR, 11 patients underwent bone marrow or PBSC (peripheral blood stem cells) transplantation. Overall, nine patients remain in continuous complete remission with a median duration of 17.5 months (9.5–39 months). The toxicity of the IDA-FLAG courses was more severe than for the FLAG courses with marked neutropenia and thrombocytopenia (for IDA-FLAG: median 22.5 and 25 d respectively; for FLAG: median 10.5 and 14 d respectively). Pulmonary infections were the main nonhaematological toxicity. One patient died in CR from invasive aspergillosis.

The IDA-FLAG regimen produced a CR of >12 months in more than half of the patients and can be recommended as a therapeutic option prior to allogeneic or autologous bone marrow transplantation.

Intensive chemotherapy has significantly improved the prognosis of patients with newly diagnosed acute myelogenous leukaemia (AML) of adults and children. About 60–80% of patients with de novo AML achieved a complete remission (CR) with induction regimens containing cytarabine and anthracyclines ( Hiddemann, 1991; Hurwitz et al, 1995 ; Behar et al, 1996 ). In paediatric clinical studies in the late 1970s and the 1980s, complete remissions were reported in 70–85% of patients. An intensive postremission therapy with or without maintenance therapy led to a 35–55% long-term survival ( Weinstein et al, 1992 ; Ritter et al, 1992 ; Grier et al, 1992 ; Wells et al, 1994b ; Hurwitz et al, 1995 ; Behar et al, 1996 ; Hann et al, 1997). 15–20 % of patients are refractory to firstline induction therapy and one quarter to one third of patients will relapse. Most of the relapses occur within 12 months of achieving CR ( Ritter et al, 1992 ; Hurwitz et al, 1995 ; Behar et al, 1996 ; Hann et al, 1997 ). The prognosis of refractory, relapsed or secondary AML is poor and effective reinduction regimens are rare. Allogeneic or autologous bone marrow transplantation (BMT or ABMT) offers a chance for long-term survival in cases of poor-prognosis AML ( Dini et al, 1996 ; Gale et al, 1996 ). Therefore newer therapeutic strategies as reinduction therapy prior to BMT or ABMT are needed.

Improved reinduction rates have reported using high-dose cytarabine (HD-ARA-C) in combination with anthracyclines ( Stahnke et al, 1992 ; Hiddemann et al, 1993 ; Wells et al, 1994a ; Archimbaud et al, 1995 ). In different leukaemic models a synergistic activity between the new cytostatic drug fludarabine and cytarabine, anthracyclines and G-CSF could be documented regarding the inhibition of several cellular pathways in DNA synthesis, DNA repair and RNA synthesis and the induction of apoptosis ( Rayappa & McCulloch, 1993; Gandhi et al, 1993a ; Plunkett et al, 1993 ; Tosi et al, 1994 ; Loughlin et al, 1996 ). These were the rationales behind several trials of combination therapies of fludarabine, HD-ARA-C with or without G-CSF and with or without anthracyclines in the treatment of relapsed, refractory, secondary and, finally, de novo AML ( Estey et al, 1993 , 1994; Visani et al, 1994 ; Huhmann et al, 1996 ; Clavio et al, 1996 ; Leahey et al, 1997 ; Dinndorf et al, 1997 ).

On the basis of these pharmacokinetic and clinical studies and the results of our pilot study ( Fleischhack et al, 1996 ) a phase II trial of a combination therapy with fludarabine, HD-ARA-C, idarubicin and G-CSF similar to a schedule previously published by Estey et al (1994 ) from the M. D. Anderson Cancer Center was designed to test the feasibility and efficacy in the treatment of poor-prognosis AML in childhood as a possible option prior to BMT or ABMT.

PATIENTS AND METHODS

Patient eligibility

The study was designed as an open comparative multicentre investigation and was started in February 1994. The diagnosis of AML or of the leukaemic transformation in MDS was confirmed in all patients by bone marrow aspiration. Patients with refractory, relapsed or secondary AML were eligible. Excluded were patients showing pre-existing severe life-threatening organ insufficiencies (renal, cardiac, pulmonary or hepatic failure) or infections (septic shock or multiorgan failure, W.H.O. scale 4).

Study mode

The study was conducted in accordance with the updated Declaration of Helsinki and approved by the local ethics committee of the University of Bonn. Prior to enrolment in the study all patients' parents and/or the patients were informed about the investigational character of the study, the potential risks of this regimen as well as the poor prognosis of relapsed, refractory or secondary AML, and gave their consent to treatment. A patient insurance was in effect. Regulatory approval was given by the participating institutions of Germany and Austria. The data of the patients were collected and documented in a database at the University of Bonn.

Patients' characteristics

In the current study, between January 1994 and July 1997, 23 patients were entered from eight departments of paediatric haematology/oncology in Germany and Austria. Patients' baseline characteristics at study entry are listed in Table I. Of the 23 patients aged 1.2–17.5 years (median 7.5 years), 17 had a first relapse, two patients had a second relapse, one patient had a secondary AML, and three patients had a refractory AML. According to the FAB classification or morphologic and immunophenotypic criteria, patients were classified as having M1 (n = 4), M2 (n = 2), M4 (n = 5), M5 (n = 8), M7 (n = 2), biphenotypic leukaemia (n = 1) or a CMML (chronic myelomonocytic leukaemia) in leukaemia transformation (n = 1). Beside bone marrow infiltration, two patients also had CNS involvement. Cytogenetic results performed on pretreatment bone marrow samples could be obtained from 16 patients and showed cytogenic abnormalities in nine ( Table II). A prognostic favourable cytogenetic abnormality t(9,11) was identified in only two patients (nos. 8 and 23) ( Martinez-Climent et al, 1995 ).

Table 1. Table I. Patient characteristics.Thumbnail image of
  • a

    FAB, French–American–British classification; CNS, central nervous system; CMML, chronic myelomonocytic leukaemia.

  • Table 2. Table II. Cytogenetic abnormalities.Thumbnail image of
  • a

    FAB: French–American–British classification.

  • The relapses had occurred 1–31 months (median 10 months) after achieving a first complete remission. In two patients with a second relapse the relapses occurred 2 months after achieving second complete remission. Most of the patients had a very early relapse (duration of first or second remission was leqslant R: less-than-or-eq, slant6 months; n = 7) or an early relapse (duration of first remission was 6–18 months; n = 9). The duration between the initial diagnosis and the diagnosis of first relapse was <12 months in 11/23 patients and <18 months in 18/23 patients. 22/23 patients had received first-line treatment polychemotherapy regimens according to the AML-BFM studies (AML-BFM-87 study: n = 4; AML-BFM-93 study: n = 18) ( Ritter et al, 1992 , 1993). One patient had a secondary AML 17 months after completion of treatment for a germ cell tumour (first-line therapy according to the MAKEI-89 study with a cumulative dose of 1200 mg/m2 VP16) ( Göbel et al, 1993 ). 7/23 patients received other relapse chemotherapies prior to IDA-FLAG or FLAG regimen. These were mostly combination therapies of HD-ARA-C with VP16 or mitoxantrone (n = 11) ( Stahnke et al, 1992 ). The cumulative equivalent doses of anthracyclines prior to the IDA-FLAG regimen were 100–680 mg/m2 (median 400 mg/m2) with no clinical and echocardiographic evidence of cardiomyopathy.

    The complete blood count (CBC) before therapy showed a median haemoglobin of 10 g/dl (range 7.3–16.9 g/dl), a median leucocyte count of 3.7 × 109/l (range 1.0–96.0 × 109/l) and a median platelet count of 58 × 109/l (range 7–232 × 109/l). No patient developed a tumour lysis syndrome before, during or after therapy.

    Treatment

    Treatment with IDA-FLAG involved the application of: (1) 30 mg/m2/d of fludarabine (Fludara®, Medac GmbH, Hamburg, Germany) by a 30 min intravenous infusion daily for 4 consecutive days, days 1–4; (2) 2000 mg/m2/d of cytarabine (Udicil®, Upjohn GmbH, Heppenheim, Germany) by 3 h intravenous infusion daily for 4 consecutive days, days 1–4, starting 4 h after the beginning of fludarabine; (3) 12 mg/m2 of idarubicin (Zavedos®, Pharmacia GmbH, Freiburg) by a 1 h intravenous infusion daily for 3 consecutive days, days 2–4, starting 1 h prior to cytarabine; and (4) 400 μg/m2 of G-CSF (Neupogen®, Amgen GmbH, Munich, Germany) subcutaneously daily from day 0 to the day of absolute neutrophil count (ANC) of >1.0 × 109/l. The FLAG course was a combination chemotherapy of fludarabine, cytarabine and G-CSF as in the IDA-FLAG regimen but without idarubicin.

    Eight patients received only one IDA-FLAG course and two patients only one FLAG course. 13 patients were treated with a second or a third course as [IDA-FLAG]–[IDA-FLAG] in three patients, as [IDA-FLAG]–[FLAG] in nine patients and as [IDA-FLAG]–[FLAG]-[FLAG] in one patient. The second or third course was started only if patients had achieved a second complete remission (CR) and the ANC was >1.0 ×109/l. The initial recommendation of a second IDA-FLAG course was changed for toxicity reasons. Thereafter idarubicin was omitted and only the FLAG course was allowed as second or third course. For nonresponders and patients in partial remission following the first reinduction course a second course was recommended, but no patient elected for this option.

    Prophylactic or therapeutic CNS therapy was administered in 12/23 patients in an age-dependent dose. Four patients received an intrathecal monotherapy with cytarabine and seven patients were treated with an intrathecal triple therapy consisting of cytarabine, methotrexate and prednisolone on day 0 or day 1 of each course. One patient with leukaemic CNS involvement was initially treated with four doses of an intrathecal monotherapy with cytarabine every 4 d and after persistence of blasts in the CSF he was treated six times with a weekly intrathecal triple therapy, after which the meningeosis subsided.

    Routine procedures and prophylactic measures used in all patients included (1) the implantation of an indwelling central venous catheter, (2) hydration, alkalinization and allopurinol advised for patients at risk for tumour lysis syndrome, (3) corticosteroid eye drops (dexamethasone, starting directly prior to the first cytarabine infusion, given every 6 h and discontinued 12 h after the last cytarabine infusion), (4) oral antimicrobial prophylaxis with cotrimoxazole (5 mg/kg body weight, three times a week), antifungal (nystatin 1 × 105 units 4 times daily or fluconazole 3 mg/kg body weight daily) and non-absorbable antibacterial antibiotics (paromomycin 4 × 10–20 mg/kg body weight daily) as well as (5) antiemetic prophylaxis.

    Response evaluation and statistical methods

    Evaluation of antileukaemic response, especially the definition of complete remission (CR) and partial remission (PR), was based on CALGB criteria ( Yates et al, 1982 ). Nonresponders included all patients with resistant disease and persistence of blast cells of >25% in the bone marrow or evidence of persistent blasts at an extramedullary site. Treatment failure was classified as (1) early death, within 7 d after the completion of the first course of remission induction therapy, or as (2) therapy-related death due to complication of bone marrow aplasia or hypoplasia and no evidence of leukaemia.

    CR duration was calculated from the initial date of documented response to relapse, to death by any cause or to the last day of follow-up (31 December 1997) and overall survival duration from the first day of treatment with IDA-FLAG/FLAG to death or to the last day of follow-up.

    The haematological toxicity was evaluated for all patients who achieved CR or PR and was documented as duration of neutropenia <0.5 × 109/l, duration of leucopenia <1.0 × 109/l and as duration of thrombocytopenia <30 × 109/l. Nonhaematological toxicities were stated according to the World Health Organization (W.H.O.) grading system and were tabulated for all patients ( World Health Organisation, 1979).

    The data for this study were evaluated using descriptive statistical methods. Pretreatment and treatment patient characteristics were tabulated. Frequencies, median values, ranges or percentages were documented. With a total of only 23 cases the use of further statistical methods was limited.

    RESULTS

    Twenty-two of 23 patients entering the study received the planned dose of IDA-FLAG or FLAG courses without dose reduction. One course of IDA-FLAG in the patient suffering from a CMML in leukaemic transformation had to be interrupted on day 3 due to progressive respiratory failure. Two patients considered at risk for cardiomyopathy received only one FLAG course without further anthracyclines as induction therapy.

    Treatment response and survival

    All 23 patients were evaluable for response ( Table III. 17/23 patients (74%) achieved a CR, one patient showed a PR, and five patients were nonresponders. Using IDA-FLAG as induction therapy, 17/21 patients (81%) achieved a CR. Both patients who were treated with only one FLAG course were nonresponders. In regard to morphological classification CRs were observed in 4/6 patients (67%) with FAB-type M1/M2, in 10/13 patients (77%) with FAB-type M4/M5, and in one of two patients with the FAB-type M7. The complete remission rate showed a correlation with the duration of first remission. 4/7 patients (57%) with a very early relapse (duration of first CR < 6 months), 7/9 patients (78%) with an early relapse (duration of first CR ≥ 6 months and <18 months) and 4/4 patients (100%) with a late relapse (duration of first CR ≥18 months) achieved a CR. 7/11 patients (64%) with resistant disease, defined as induction failure of first-line treatment or relapse within 12 months after diagnosis of AML, achieved a CR. Of the two patients with a second relapse prior to IDA-FLAG regimen, one was a nonresponder and another showed a short third CR for 1 month. Most patients who achieved a second CR with an IDA-FLAG course as first reinduction therapy were treated with a second (n = 12) or third (n = 1) course of an IDA-FLAG or FLAG course or were rapidly transplanted with bone marrow (BMT) or peripheral blood stem cells (PBSCT) (n = 3). Overall 11/23 patients (48%) were transplanted in CR following the IDA-FLAG/FLAG regimen. Three patients received an autologous (ABMT: n = 2; APBSCT: n = 1) and eight patients an allogeneic transplantation (matched unrelated donor BMT/PBSCT: n = 4, matched sibling donor BMT: n = 4). All nine patients with detectable cytogenetic abnormalities achieved a second CR and five of them are in continuous CR after BMT/PBSCT (patients 8, 9, 21, 22 and 23). A correlation between results of cytogenetic analysis and treatment response could not be documented because of limited numbers.

    Table 3. Table III. Treatment response to IDA-FLAG/FLAG regimen.Thumbnail image of
  • a

    n, frequency; CR, complete remission; PR, partial remission; NR, nonresponse; DD, death from disease; TD, therapy-related death; CCR, continuous complete remission; OS, overall survival; BMT, bone marrow transplantation; PBSCT, transplantation of peripheral blood stem cells.

  • After achieving a second or third CR, 5/17 patients relapsed within 1, 3, 4, 7 and 34 months. Four of these patients previously had resistant disease. All patients who relapsed following the IDA-FLAG/FLAG therapy and those with PR and nonresponse died of progressive disease. After transplantation, two patients showed a relapse 2.5 months and 6 months following the autologous or allogeneic BMT. There was no treatment failure caused by early death. Three patient deaths were therapy-related, one patient died in CR caused by a fulminant Aspergillus sepsis during aplasia following a second IDA-FLAG course, and two patients died from transplantation-associated complications on day 66 and day 81 after BMT.

    Duration of remission and survival following reinduction therapy with the IDA-FLAG/FLAG regimen are shown in 4 Table IV. For 11/17 responders the duration of the second CR was longer than the duration of first CR with a median duration of 13.5 months and 10 months, respectively. Overall, nine patients remain in continuous complete remission (CCR) with a median duration of 17.5 months (range 9.5–39 months). Seven of them are in CCR following BMT or PBSCT with a median duration of 17.5 months (range 12.5–39 months). The median duration of overall survival for all 23 patients was calculated up to the last day of follow-up (31 December 1997) and was 10.5 months (range 1–46 months). The follow-up time for the nine survivors ranged from 10.5 to 40 months (median 13 months).

    Table 4. Table IV. Duration of remission and survival following reinduction therapy with IDA-FLAG/FLAG regimen.Thumbnail image of
  • a

    CR, complete remission; CCR, continuous complete remission; OS, overall survival.

  • Haematological toxicity

    After exclusion of nonresponders and of the patient who died due to therapy-related courses, 20/24 IDA-FLAG courses and 11/13 FLAG courses were evaluable for haematological toxicity. All patients experienced, as the main toxicity, bone marrow aplasia with a profound neutropenia and thrombocytopenia. For the FLAG courses the duration of myelosuppression was clearly shorter than for the IDA-FLAG courses. For IDA-FLAG courses the median duration of (1) ANC <0.5 × 109/l was 22.5 d (range 14–42 d), (2) leucocytes <1.0 × 109/l was 22 d (range 13–42 d), and (3) platelets <30 × 109/l was 26 d (range 5–90 d). The median duration of neutropenia, leucopenia and thrombocytopenia for the FLAG courses were 10.5 d (range 3–21 days), 8.5 d (range 0–20 d) and 14 d (range 2–28 d) in the IDA-FLAG courses and 19.5 d (range 2–28 d) in the FLAG courses. The time of haemopoietic recovery in responders was not influenced by the pretreatment degree of residual normal haemopoiesis when patients with <20% (11/17 responders) and with >50% haemopoiesis (6/17 responders) were compared.

    Nonhaematological toxicity

    Nonhaematological toxicities using the W.H.O. scale system are shown in 5 Table V. All patients and treatment courses (including patients with treatment failure and nonresponders) were evaluated for nonhaematological toxicities.

    Table 5. Table V. Nonhaematological toxicities of the IDA-FLAG/FLAG regimen.Thumbnail image of
  • a

    * W.H.O. grade.

  • Side-effects (nausea, haemorrhage, neurological symptoms, skin, cardiac, renal or hepatic disturbances) were rare and mild. Four patients showed a transient increase of hepatic transaminases (W.H.O. grade I) and one patient a reversible hyperbilirubinaemia with a peak serum level of 92 μmol/l. Mucositis or diarrhoea were observed following 11 IDA-FLAG courses and four FLAG courses and were more severe in patients who were treated with IDA-FLAG. The documented severe renal and cardiac complications were associated with septic or pulmonary infections. No persistent cardiomyopathy was seen at a median cumulative equivalent dose of anthracyclines of 580 mg/m2 (range 280–860 mg/m2) after the treatment with the IDA-FLAG regimen.

    The most common regimen-related toxicity were infections during bone marrow aplasia, predominantly with pulmonary involvement. Infectious complications were more frequent and severe in IDA-FLAG than in FLAG courses. Fever of unknown origin occurred during myelosuppression following 13/24 IDA-FLAG courses and 5/13 FLAG courses. In nine IDA-FLAG courses and only one FLAG course clinically, radiologically or microbiologically documented severe infections were observed. Nine patients showed pulmonary involvement with life-threatening or fatal pneumonia in three patients, of whom one patient was a nonresponder and died from progressive disease and another patient developed a lethal Aspergillus sepsis following a second IDA-FLAG course. Of the other six patients with pneumonia, two were also suspected to have fungal infections. Microbiologically documented infections without pulmonary involvement were seen in three patients (enterococcus bacteraemia, candidaemia, parvovirus B19 infection); all of them were treated successfully. As a late complication one patient developed a pulmonary tuberculosis in CR 7 months after treatment with two IDA-FLAG courses. Persistent T-cell depletion with a marked decrease of T-helper cells was documented 5 and 11 months after completion of treatment. With normal numbers of total leucocytes and total lymphocytes in peripheral blood, the absolute T-helper-cell counts were 0.146 × 109/l and 0.333 × 109/l, respectively. The patient recovered from this complication with tuberculostatic treatment.

    DISCUSSION

    Despite significant progress in the treatment of newly diagnosed AML of childhood during the past 20 years, currently one half of patients will relapse after first-line therapy according to the BFM protocols AML-87 and AML-93 study ( Ritter et al, 1992 ; Creutzig et al, 1995 ). Patients refractory to first-line treatment, as well as early relapsed patients and patients with a secondary AML, are poor responders to classic salvage reinduction therapy ( Kantarjian et al, 1993 ; Stahnke et al, 1992 ).

    During the past decade new drugs have been investigated for their antileukaemic activity in the treatment of refractory and relapsed AML of childhood and have reached complete response rates of 34–54% ( Hakami et al, 1987 ; Miller et al, 1991 ; Santana et al, 1992 ; Steuber et al, 1996 ; Kantarjian et al, 1996 ). Higher second remission rates of 73% and of 60% in adults were reported using a combination therapy consisting of high-dose ARA-C/mitoxantrone and of intermediate ARA-C/etoposide/mitoxantrone, respectively ( Wells et al, 1994a ; Archimbaud et al, 1995 ), but in all studies most patients had a short second or third remission and relapsed early.

    At the beginning of the 1990s several trials were designed on the basis that HD-ARA-C was one of the most effective drugs in reinduction therapy of poor-prognosis AML and that fludarabine administered prior to ARA-C enhanced the ARA-CTP accumulation in the blast cells and acted synergistically with ARA-C, anthracyclines and G-CSF ( Avramis et al, 1990 ; Gandhi et al, 1993b ; Rayappa & McCulloch, 1993; Tosi et al, 1994 ; Loughlin et al, 1996 ). Combination therapies of fludarabine and HD-ARA-C with or without G-CSF tested in relapsed and refractory AML of adults produced CR rates of 50–60% ( Estey et al, 1994 ; Visani et al, 1994 ; Clavio et al, 1996 ). Using idarubicin in combination with continuous infusion of fludarabine and cytarabine CR rates of 67% (n = 15) and 80% (n = 10) were reported in paediatric patients with a refractory or relapsed AML ( Leahey et al, 1997 ; Dinndorf et al, 1997 ). In our study, with the IDA-FLAG regimen as single-course reinduction treatment, a CR rate of 74% was documented, although, for the first-line or previous reinduction therapy according to the BFM protocols, 35% of patients had already received idarubicin and 78% of patients had received HD-ARA-C/mitoxantrone.

    Previous studies documented a relationship between the duration of first remission and both the rate and the duration of second CR. First remission duration >18 months is related with a higher CR rate and a more favourable outcome of relapsed AML in adults and children. In contrast, both rate and duration of second CR is poor in patients with refractory or very early relapsed (within 6 months) leukaemia ( Kantarjian et al, 1988 ; Keating et al, 1989 ; Hiddemann et al, 1990 ; Achimbaud et al, 1995 ). However, the IDA-FLAG regimen achieved in 6/10 patients with refractory AML or with very early relapses a CR with a duration of 3–34 months (median 11.5 months); three of them are alive in continuous CR, two after BMT or PBSCT. Despite an apparently higher response rate in patients with late relapses in our series, no clear difference could be documented because of the limited numbers. Overall, the median duration of second CR was longer than of the first CR (13.5 v 10 months) for all responders.

    In the absence of post-remission therapy, most patients relapse again within 12 months of achieving a second CR. Recent alternatives of therapeutic strategies are allogeneic or autologous BMT or PBSCT ( Hurwitz et al, 1995 ; Gale et al, 1996 ; Dini et al, 1996 ). The precise comparison of the different postinduction approaches in our study was impossible because of the limited patient number and the lack of randomization. 11 patients were transplanted in second CR and seven of them are now in continuous CR (CCR). Only two of six patients with post-remission consolidation and maintenance therapy are in CCR.

    The main toxicity of the IDA-FLAG regimen was a long-term myelosuppression associated with a high incidence of infections (76% of courses). Severe infections (W.H.O. scale > II) were observed in 27% of therapy courses and involved predominantly the lung. The high rate of pulmonary infections could have been caused by a toxic injury of the lung epithelial cells by the cytostatic drugs and the long-term neutropenia involving a risk for fungal infection ( Anderlini et al, 1996 ). The neutropenia of the FLAG courses was clearly shorter than for the IDA-FLAG courses and was associated with a lower infection rate. A second IDA-FLAG course given in three patients seemed to be more toxic with a prolonged myelosuppression and led to one toxic death. A relationship between the pretreatment residual haemopoiesis and the time of haemopoietic recovery after the IDA-FLAG could not be detected and needs further evaluation in a larger number of patients. The reinduction/consolidation regimen [IDA-FLAG]–[FLAG] was tolerated well and the advised option in the continuation of the trial. The other side-effects were mild and reversible, but no severe neurotoxicity or haemolytic anaemia were observed as reported in previous studies ( Kornblau et al, 1993 ; Di Raimondo et al, 1993 ). Persistent T-cell depletion led to pulmonary tuberculosis, 7 months after IDA-FLAG therapy in one patient. Purine analogues such as fludarabine or cladribine produce long-term T-helper-cell depletion involving a high risk of opportunistic infections ( Goodman et al, 1996 ). Thus a long-term antimicrobial prophylaxis following the IDA-FLAG regimen is recommended.

    Based on our data, the IDA-FLAG regimen with IDA-FLAG as reinduction course and FLAG as consolidation therapy is an effective therapy prior to allogeneic and autologous BMT or PBSCT. With regard to the unfavourable prognosis of refractory or relapsed AML of childhood, the toxicity of the IDA-FLAG reinduction regimen is acceptable. Further evaluation in a phase III trial is needed to determine the efficacy and toxicity in comparison to other reinduction therapies, especially those which contain high-dose ARA-C.

    Acknowledgements

    We thank the following colleagues for entering patients in this study: Stefan Müller-Weihrich (München, Germany), Wolfram Scheurlen (Mannheim, Germany), Joachim Kühl (Würzburg, Germany), Herbert Breu (Dortmund, Germany), Wolfgang Dörffel (Berlin, Germany), Werner Havers (Essen, Germany). Appreciation is also expressed for support from the companies Medac Hamburg, Amgen/Roche München and Pharmacia Erlangen.

    Ancillary