Superior outcome of infant acute myeloid leukemia with intensive chemotherapy: results of the Japan Infant Leukemia Study Group

Infants with acute myeloid leukemia (AML) generally show monoblastic or myelomonoblastic features with hyperleukocytosis, extramedullary involvement, and MLL gene rearrangements by cytogenetic analysis.1,2 The outcome and prognostic factors in infants with AML remain generally obscure. Several studies have reported that the presence of MLL gene rearrangements was of prognostic significance, in addition to the high white blood cell (WBC) count at diagnosis in both adults and infants with AML.3,4 Pui and coworkers5 reported that male gender and a WBC count of more than 50 000/μL were significantly associated with an inferior outcome in infant AML. Molecular abnormalities of 11q23 translocations or MLL gene rearrangements in infant AML have been correlated with an M4 or M5 subtype in the French-American-British (FAB) classification and hyperleukocytosis.1,2,4-6 

The event-free survival (EFS) of infants with AML treated with chemotherapy has been reported to be 32% to 34%.7,8However, the presence of MLL gene rearrangements failed to correlate with treatment response,5,6 in contrast to the extremely poor outcome of infants with acute lymphoblastic leukemia (ALL) with the rearrangements.9-11 The clinical outcomes of infant AML with MLL gene rearrangements were also similar to those of childhood AML.6 The reason for this apparent discrepancy in the treatment outcome between patients with ALL and AML is not clear. Moreover, indications of stem cell transplantation (SCT) for infant AML are still controversial.

In the current study, we analyzed the outcome and prognostic factors of infants with AML who were treated with the same intensive chemotherapy. Because some of them received SCT after chemotherapy, the effect of SCT for infant AML was also assessed.

All infants with acute leukemia under 12 months old were registered in the Japan Infant Leukemia Study Group between December 1995 and December 1998. This study group has covered approximately 80% of infant leukemias in the whole country. Among them, those with ALL, unclassified or mixed leukemia, and natural killer (NK) cell leukemia were excluded from the study because the patients with these subtypes were treated with different strategies. For the infants with AML, in addition to routine clinical and laboratory examinations, characterization of leukemic cells including morphology, cytochemical stainings, immunophenotypes, cytogenetics, and molecular analyses were performed. AML was diagnosed when more than 25% myeloblasts were present in the bone marrow (BM). Leukemic cells in BM, peripheral blood, and cerebrospinal fluid (CSF) were stained using standard Wright-Giemsa and cytochemical stainings and were reviewed in the study center using the FAB classification. Central nervous system (CNS) involvement at diagnosis was defined as more than 5/μL mononuclear cells in CSF with obvious myeloblastoid morphology. The ANLL91 protocol, which was designed for newly diagnosed childhood AML in Japan, was also used in this study, because no appropriate strategy has been established for infant AML in Japan (Figure1).12 The institutional review board (Department of Pediatrics, Toho University School of Medicine) approved the protocols and informed consent was obtained from parents or guardians as appropriate.

Immunophenotypes were analyzed by an EPICS-PROFILE flow cytometer (Coulter Electronics, Hialeah, FL) at 2 study centers (Department of Pediatrics, Tokyo Medical and Dental University and Kyoto Prefectural University of Medicine) and reviewed by Drs Hiroji Okawa and Shigeyoshi Hibi. The monoclonal antibodies used were: anti-CD2 (pan-T), anti–cytoplasmic CD3 (Leu4), anti-CD13 (LeuM7), anti-CD15 (LeuM1), anti-CD33 (LeuM9), anti-CD34 (HPCA1), anti-CD56 (leu19), and anti-HLA-DR (HLA-DR), purchased from Becton Dickinson Immunochemistry System, Mountain View, CA; anti-CD7 (3A1), anti-CD14 (My4), anti-CD19 (B4), and anti-CD10 (J5) from Coulter Immunology (Hialeah, FL); anti-CD41 (TP80) from Nichirei (Tokyo, Japan). To detect cytoplasmic antigens, cells were pretreated with Dako Intra Stain (Dako, Glostrup, Denmark). Each antigen was considered to be positive when more than 30% of mononuclear cells showed a positive reaction.

Cytogenetic analysis was performed in all patients at presentation. Mononuclear cells were separated from BM or peripheral blood; after 24 hours of unstimulated culture, samples were fixed in Carnoy fixative solution (3:1 methanol and acetic acid). Chromosomes were described according to conventions of the International System for Human Cytogenetic Nomenclature.13 

Both the fluorescence in situ hybridization (FISH) and Southern blot analysis were performed to detect the MLL gene rearrangements.14 Metaphase samples used for cytogenetic studies were used for the FlSH analysis. The probes used (S1363 and LB140 cosmid probes) covering the 5′ and 3′ end of the MLLgene were provided by Dr Misao Ohki (National Cancer Institute, Tokyo, Japan). The probes were labeled with biotin-11-dUTP or digoxigenin-11-dUTP (Boehringer Mannheim, Mannheim, Germany) using polymerase chain reaction (PCR) labeling after sequence-independent amplification, and hybridized to metaphase as previously described.15 The probes were detected by avidin fluorescein (Vector Laboratories, Burlingame, CA) or antidigoxigenin rhodamine (Boehringer Mannheim) and then counterstained with 4′6-diamidino-2-phenylindole dihydrochloride (DAPI). Images of the hybridized signals were captured using fluorescence microscopy (Olympus Optical, Tokyo, Japan). The results of FISH analysis were confirmed by the Southern blot analysis. High-molecular-weight DNA was extracted from the BM or peripheral blood mononuclear cells. Extracted DNA was digested with BamHI or HindIII, then electrophoresed and transferred to a nylon membrane (Hybond N+, Amersham, Buckinghamshire, United Kingdom). The transferred membrane was hybridized with the 880–base pair (bp) MLLcomplementary DNA (cDNA) probe covering the breakpoint cluster region of the MLL gene, which was provided by Prof M. Greaves (Leukemia Research Fund Centre, London, United Kingdom).

All infants with AML were treated with the protocol ANLL91 (Figure 1).12 Briefly, the combination of etoposide, cytarabine (AraC), and mitoxantrone was used as induction therapy. After achieving remission, 4 different courses of intensification therapy were prescribed for the patient: high-dose Ara C (HD-AraC), etoposide, and mitoxantrone for the first and second courses; AraC, etoposide, and tetrahydropyranyl-adriamycin (THP-ADR) for the third and fourth courses; HD-AraC, etoposide, and aclarubicin as the fifth and sixth courses; and HD-AraC, etoposide, and vincristine (VCR) as the seventh and eighth courses, respectively. As the prophylaxis for CNS leukemia, triple intrathecal therapy with methotrexate, AraC, and hydrocortisone was also used on the first day of each course of the induction or intensification therapy. The patients who achieved complete remission (CR; < 5% of leukemic cells in BM and no evidence of extramedullary disease) then received the subsequent intensification therapy. Each course of intensification therapy was prescribed when the WBC count reached more than 2500/μL. The duration of chemotherapy was 9 to 10 months, but varied among the patients. Some of the AML infants also received allogeneic SCT during the first remission, when an HLA-matched donor was available. Although the present study was designed to treat all infants with AML with chemotherapy, the application of SCT for each patient was decided in each institution.

Clinical, cytogenetic, and molecular findings were compared between patients with the M4/M5 and non-M4/M5 subtypes by the χ² test with the Yates correction for 2 × 2 tables and the Fisher exact test. Life-table estimates of survival were derived by the method of Kaplan and Meier and compared with the log-rank test. Overall survival (OS) was defined as the duration from the start of treatment to the end of the follow-up. EFS was defined as the time from achievement of CR to relapse, death, or last follow-up. Failure to enter remission was considered as an adverse event at time zero. To clarify the effect of chemotherapy for infant AML and compare the treatment outcome in this study with that in other studies, the patients who underwent allogeneic SCT were censored at the time of transplantation. The findings were updated on April 15, 2001.

A total of 112 infants with acute leukemia were registered in the Japan Infant Leukemia Study Group between 1995 and 1998. Sixty-six were classified as having ALL, 5 as unclassified or mixed leukemia, and 2 as NK cell leukemia. A total of 39 patients were, therefore, registered as infant AML in this study. Among them, 4 patients were excluded from the study; 3 with Down syndrome having AML M7 subtype were treated with a different protocol, and 1 was classified as having chronic myelomonocytic leukemia. The remaining 35 patients were treated with the ANLL91 protocol. Age at diagnosis ranged from 0 days to 12 months, with a median age at diagnosis of 7.9 months. There was a boy-to-girl predominance of about 1.5 (21 versus 14). Ten patients had CNS involvement at onset. In the FAB classification, one was classified in the M1 subtype, 1 in M2, 11 in M4, 12 in M5, 2 in M6, and 8 in M7. A total of 23 (66%) patients were, thus, classified as monocytic or myelomonocytic leukemia. Immunophenotyping was performed in most of the patients. None or only a few patients expressed CD19 (0 of 33 patients, 0%), CD2 (2 of 29 patients, 7%), or CD10 (1 of 31 patients, 3%). In 1 of the 16 patients examined for cytoplasmic CD3, 35% of leukemic cells showed a positive reaction for this antigen. CD7, one of the T-lineage antigens, was positive in 8 of 33 (24%) patients, whereas the expression of myeloid antigens was high in these patients; 17 (50%) of 34 patients for CD13, 8 (24%) of 33 for CD14, 27 (79%) of 34 for CD33, and 7 (47%) of 15 for CD15. HLA-DR, CD34, and CD56 were also expressed in 23 (72%) of 32, 6 (18%) of 33, and 7 (32%) of 23 patients, respectively. CD41, a megakaryocyte marker, was positive in all 8 patients with the M7 subtype. When the incidence of positive antigens was compared between patients with the M4/M5 and non-M4/M5 subtypes, the expression of myeloid antigens was high in those with the M4/M5 compared with the non-M4/M5 subtype; 21 (95%) of 22 versus 6 (50%) of 12 for CD33, and 7 (70%) of 10 versus 0 (0%) of 5 for CD15, respectively.

The clinical and cytogenetic findings were also compared between patients with the M4/M5 and non-M4/M5 subtypes (Table1). The incidence of boys, younger age (< 6 months old), and hyperleukocytosis at onset tended to be higher in patients with the M4/M5 subtype, compared with those with the non-M4/M5 subtype, although a significant difference was not observed. In patients with the M4/M5 subtype, 13 (57%) of 23 patients showed 11q23 translocations: t(9;11) in 2, t(6;11) in 2, t(11;19) in 3, t(4;11) in 1, t(1;11) in 2, t(10;11) in 2, and t(11;12) in 1. In patients with the non-M4/M5 subtype, however, none showed 11q23 translocations in leukemic cells. From the molecular analysis, most patients (16 of 23, 70%) with the M4/M5 subtype showed the MLLgene rearrangements. In addition to all patients with 11q23 translocations, 2 with the normal karyotype and one with del(11)(q23) also showed the MLL gene rearrangements. In contrast, only one patient with an abnormal karyotype including chromosome 11 showed the MLL gene rearrangements in the non-M4/M5 subtype.

Thirty-two patients (91.4%) achieved CR after the induction therapy. Among them, 6 patients relapsed during or after intensification therapy and 1 died of fungal infection, but the remaining 25 have remained in first CR. Six patients underwent SCT during first CR; allogeneic or unrelated BM transplantation (BMT) in 2, allogeneic peripheral blood stem cell transplantation (PBSCT) in 2, and unrelated cord blood transplantation (CBT) in 2. The conditioning regimen was busulfan/melphalan in 5 and busulfan/cyclophosphamide/total body irradiation in 1. All 6 patients who received SCT at first CR have survived without disease, whereas 6 of the remaining 26 patients without SCT relapsed during or after chemotherapy. The OS and EFS rates at 3 years in all 35 patients were 76.0% (95% confidence interval [CI], 61.3%-90.7%) and 72.1% (95% CI, 56.4%-87.9%), respectively, with a median follow-up time of 1038 days (Figure2).

Prognostic factors contributing to the treatment outcome of the patients were analyzed (Table 2). The EFS rate of younger infants (< 6 months old) was similar to that of older infants (≥ 6 months old). The EFS was not different between boys and girls. There was also no difference in the EFS rate between patients with a high WBC count (≥ 100 000/μL or ≥ 200 000/μL) and those with a low WBC count (< 100 000/μL or < 200 000/μL). The treatment outcome of patients with the M4/M5 subtype tended to be better than that of the non-M4/M5 subtype (P = .105). The presence of MLL gene rearrangements was not correlated with outcome of infant AML in this study; when the patients with the M4/M5 subtype were analyzed, there was no significant difference in the EFS rate between patients with and without the rearrangements. The effect of SCT for infant AML was also analyzed. When all 6 patients who underwent SCT were not censored at the time of transplantation, the estimated EFS rate at 3 years was almost unchanged (74.1% versus 72.1%). In addition, the relapse occurred within 10 months and also before SCT in most patients. No benefit of SCT was, therefore, confirmed in this study.

Febrile episodes associated with myelosuppression were observed in the majority of the patients after the each course of chemotherapy. Septic episodes were also documented in 4 and 5 patients during the induction and intensification therapies, respectively:Staphylococcus epidermidis in 2, α-Streptococcus in 1, Enterococcus in 2,Enterobacter in 1, Klebsiella in 1,Clostridium in 1, and unknown fungal infection in 1. Cutaneous abscess developed in 2 patients. Two patients suffered from viral pneumonia during the induction therapy, and 1 of them diagnosed as having respiratory syncytial virus pneumonia died before the achievement of CR. One patient suffered from neurotoxicity and died of disseminated fungal infection (aspergillosis) during the first CR. In other patients, septic or febrile episodes with myelosuppression were curable by supportive therapy with a combination of antibiotics or antifungal drug and the granulocyte colony-stimulating factor. One patient had hemorrhagic cystitis due to aclarubicin and one had hypocalcemia during the induction therapy, which were controlled by the supportive therapy. Although a few patients suffered from diarrhea, neither cardiac nor hepatic toxicities were observed in any patient.

The present study shows that the EFS of infant AML was improved by intensive chemotherapy. Table 3summarizes the results of treatment outcome of infant AML in recent studies. Our findings cannot be simply compared with those of other studies, because some of them include children aged 1 to 2 years8,16,17 or those treated with SCT.4,6,18The induction rate of AML infants was 91.4% in the present study. In the Medical Research Council AML10 trial, the induction rate of infant AML was also 92%.19 Although the induction rate of infant AML was not described in other studies, these 2 studies reported better results of the remission induction in infant AML by intensive chemotherapy. The EFS rates of AML infants treated with chemotherapy alone in 2 large study groups were 32% and 34%, respectively (Table3).7,8 Recently, another study showed a better outcome of infant AML; 58% of AML infants have been in first CR (Table3).20 The EFS of infant AML in our study was, therefore, superior, compared with these previous findings. Because the number of patients studied is relatively small and the follow-up duration is short, an increased number of patients and longer observation period are needed to confirm the superior outcome of AML infants.

Several factors contributing to the good prognosis in the present study should be considered. The incidence of the M4/M5 subtype and the 11q23/MLL gene rearrangements were similar between these studies,4-7 whereas the only difference was the male predominance in our study. Another factor is the different chemotherapeutic regimens in each study. In most clinical trials, the induction therapy consisted of daunorubicin, AraC, 6-thioguanine, and etoposide.8,16,17,19 In the present study, however, 2 different induction regimens were used: sequential use of etoposide and AraC, and introduction of mitoxantrone instead of daunorubicin. Another study that showed 58% of first CR in infant AML also included AraC, etoposide, and mitoxantrone in induction therapy.20 The combination of these drugs might induce a high induction and EFS rates in infant AML. In addition, HD-AraC and etoposide were mainly used in each course of intensification therapy in the ANLL91 protocol. Repetitive cycles or postinduction courses of HD-AraC have been effective for AML.21,22 Regimens including etoposide are also highly effective against acute monoblastic leukemia,23,24 which is frequently seen in infant AML. In children with AML who were treated with the same protocol, the disease- free survival rate was 62.4%,25 which was also superior compared with the findings of other studies of childhood AML (EFS rate was 34%-43%).8,16,17,19,20 Because the EFS rate is similar for infants and older children in most AML trials,16,19 the chemotherapeutic regimen is a more important factor contributing to the prognosis than other factors including age and clinical features of patients with AML. Intensified induction therapy can especially influence prognosis, regardless of postremission therapy.26 

A previous study described that molecular alterations of MLLin infant AML may correlate with a poor prognosis.2However, in the present study the presence of the MLL gene rearrangements was not correlated with the treatment outcome of infant AML. Several other studies also described that there was no significant difference in the EFS between patients with and without 11q23/MLL gene rearrangements.4-7 Children with t(9;11) were reported to have a favorable prognosis.17,27,28 Our study also showed that the younger infants (< 6 months old), even though they had the 11q23/MLLgene rearrangements, did not show a poor outcome. The gender and WBC count at diagnosis also lacked a prognostic impact on the OS and EFS rates. Only the patients with the M4/M5 subtype tended to have a better prognosis than those with other subtypes. Leukemic cells with the 11q23/MLL gene rearrangements are usually seen in the M4 or M5 subtype.2,17,28,29 Pui and colleagues5initially reported that male gender and hyperleukocytosis were significantly associated with an inferior outcome in infant AML. In their additional analysis, however, only 2 factors predicted a favorable prognosis: M4 or M5 leukemia and the t(9;11).7These different findings may, in part, be explained by the inclusion of children 1 to 2 years old or the difference in the chemotherapeutic regimens. As shown in the present study, the effect of these prognostic factors can disappear by intensifying the chemotherapy.

In the ANLL91 protocol, the accumulation doses of mitoxantrone, THP-ADR, and aclarubicin were 45 mg/m², 90 mg/m², and 60 mg/m², respectively. Etoposide and AraC were used at a total of 5.75 g/m² and 95.4 g/m², respectively. Green and colleagues30reported that treatment with ADR was significantly associated with a shorter survival time due to cardiac mortality during childhood. The relationship between the cumulative dose of etoposide and the risk of secondary neoplasms has been reported.31-33 Therefore, further observations concerning long-term sequelae in these AML infants are required, and the treatment outcome of infant AML with reduced doses of each chemotherapeutic drug should be evaluated in further studies.

The effect of allogeneic SCT is still unclear in infant AML because of the lack of a randomized study. In an early study, infants with AML were more likely than older children to have an aggressive disease associated with an inferior outcome, which indicated the possibility of SCT.34 In fact, allogeneic BMT has yielded long-term survival in some infants with AML.35,36 In the present study, all AML infants who received SCT survived without disease. Woolfrey and coworkers37 reported that the OS and EFS rates of infants with AML who received BMT at first CR were 54% and 38%, respectively. A recent study showed that the AML infants who received SCT had a 5-year disease-free survival of 73%; a short interval between remission and SCT was only associated with a better prognosis in that study.38 However, allogeneic SCT was used in only 4 patients, whereas the remaining received autologous SCT. In the Pediatric Oncology Group study, treatment of childhood AML with either autologous SCT or intensive chemotherapy prolonged EFS equally.39 These findings suggest that the EFS can be improved by intensifying the chemotherapy for infant AML. Moreover, SCT has resulted in significant late neuropsychological and growth sequelae in infants with cancer, especially when total body irradiation was included as a conditioning regimen.40 Taken together, infants with AML can be cured or can survive with an appropriate intensive chemotherapy regardless of the presence of the MLLgene rearrangements and without rescuing SCT. It will be important to select subgroups that require more aggressive therapy including SCT in the future.

Institution department investigators who registered infants with AML in this study: S. Yoshida, Hokkaido University, Pediatrics; T. Oda and T. Kudo, Sapporo Medical University, Pediatrics; M. Tsuchida, Ibaragi Children's Hispital, Hematology; I. Komori, Matsudo City Hospital, Pediatrics; Y. Tsunematsu, National Children's Hospital, Hematology; H. Onishi and Y. Hayashi, University of Tokyo, Pediatrics; R. Hosoya, St Duke International Hospital, Pediatrics; A. Kinoshita, Keio University, Pediatrics; H. Mugishima, Nihon University, Pediatrics; K. Isoyama, Showa University Fujigaoka Hospital, Pediatrics; K. Koike, Shinshu University, Pediatrics; K. Horibe and S. Kojima, Nagoya University, Pediatrics; H. Kawasaki and Y. Komada, Mie University, Pediatrics; T. Takimoto, Otsu Red Cross Hospital, Pediatrics; J. Hara, Osaka University, Pediatrics; M. Yoshikawa, Osaka Prefectural Hospital, Pediatrics; Y. Kosaka, Kobe University, Pediatrics; T. Tsutsui, Kobe City General Hospital, Pediatrics; K. Hamamoto, Hiroshima Red Cross Hospital, Pediatrics; K. Ayukawa, Yamaguchi University, Pediatrics; Y. Nagatoshi and J. Okamura, National Kyushu Cancer Center, Pediatrics; H. Eguchi and H. Inada, Kurume Univertsity, Pediatrics; M. Migita, Kumamoto University, Pediatrics; H. Kuroda and T. Sugimoto, Miyazaki Medical College, Pediatrics.