Clinical features of childhood acute myeloid leukaemia with specific gene rearrangements

Article metrics


Specific gene rearrangements seem to distinguish distinct subsets of acute myeloid leukaemia (AML) with different features and prognosis, and some reports suggest that the epidemiological distribution of AML could vary among countries.1, 2 To date, cytogenetic examination has been used to study the frequency of these genetic alterations in a large series of children3, 4, 5 with AML; nevertheless, in a proportion of cases, these gene rearrangements may be cryptic and undetectable by conventional cytogenetic techniques.6, 7 To evaluate the frequency of specific gene rearrangements, the corresponding clinical morphological features at diagnosis and the potential prognostic impact on patients’ long-term survival, we screened by RT-PCR five different chimaeric transcripts (AML1-ETO, CBFβ-MYH11, PML-RARα, MLL-AF9 and BCR-ABL) in a series of 270 Italian children with AML, treated with AIEOP-LAM 87–92, BFM 83-93 and AIDA protocols between 1988 and 1998, and whose RNA were available and morphology had been centrally reviewed.

Of the cases, 45% were positive for one fusion gene. The frequency of AML1-ETO, CBFβ-MYH11, PML-RARα and MLL-AF9 were 14, 6.3, 20 and 4%, respectively, while BCR-ABL was observed in only two cases. The frequency of AML1-ETO, MLL-AF9 and CBFβ-MYH11 found was comparable with those of t(8;21), t(9;11) and inv(16) reported previously, suggesting that these gene rearrangements do not have geographic heterogeneity. On the contrary, PML-RARα-positive AML represented 20% of our cohort. This is a higher frequency than that reported in other studies of children from Northern Europe or the United States, but the increased frequency of promyelocytic leukaemia in Italian AML children has already been described.

Cytogenetic analysis was successfully carried out in 153 patients, while it failed or was not performed in 117 patients. Abnormalities were identified in 90/153 (59%), the frequency of t(8;21), inv(16), t(15;17) and t(9;11) was 9.8, 2, 17 and 2%, respectively; the 63 remaining cases (41%) had a normal karyotype. In the t(8;21)-positive AML cases, the most common associated alterations involved chromosome 9. Among patients with normal karyotype, RT-PCR identified 17/63 (27%) chimaeric transcripts, and six further cases with translocations undetected by cytogenetics were found among the 43 children reported to have an abnormal karyotype. Considering t(8;21), inv(16), t(15;17) and t(9;11) together, 130/153 (85%) patients, who had a successful analysis were correctly classified cytogenetically. Then, in our experience, cytogenetics showed comparable specificity (100%), but lower sensitivity (67%) than RT-PCR. A general agreement exists in considering RT-PCR more sensitive than conventional cytogenetics in identifying specific gene rearrangements and the fusion genes studied have been described in the absence of identifiable translocations. Nevertheless, the molecular technique is not capable of identifying associated abnormalities. Therefore, since the genetic studies in childhood AML allow for the identification of patient subgroups who probably can benefit from a more tailored therapy and the fusion gene identified can represent a useful target for the minimal residual disease study, ideally all cases should be studied by RT-PCR and cytogenetics: two complementary methods in the genetic characterisation of leukaemia.

Clinical data collected included FAB subtypes, gender, age, WBC count at diagnosis and extramedullary involvement, whenever available (Table 1). The low number of M7-AML was due to the scarce number of cells collected in megakaryoblastic leukaemia.

Table 1 Presenting clinical and biological features

The median age at diagnosis for all patients was 7.8 (range 0–19.5). Cases with AML1-ETO and PML-RARα fusion genes were older patients (median age 8.2 and 9.5 years, respectively), while the median age of patients with CBFβ-MYH11 and MLL-AF9 was 6.7 and 3.2 years, respectively.

A strong, although not exclusive, association between M3, M4 and M5 FAB subgroups and the presence of PML-RARα, CBFβ-MYH11 and MLL-AF9 gene rearrangements, respectively, was found. The AML1-ETO-positive cases represented 43% of M2 AML; however, this gene rearrangement was frequently found in M1 patients as well. As a whole, 92% (34/37) of AML1-ETO-positive AML were classified as FAB-M1 or FAB-M2.

The median WBC count at diagnosis was significantly higher in CBFβ-MYH11-positive cases, while PML-RARα- and AML1-ETO-positive cases had the lowest values. Also, among M2 and M1 plus M2 FAB subgroups this value was significantly lower in AML1-ETO-positive AML than in negatives (16 000/μl vs 30 800/μl, P=0.027; 16 000/μl vs 34 200/μl, P=0.0005).

Extramedullary disease involved the liver (41%), spleen (37%), node (15%), CNS (8%) and other sites (9%). The FAB-M3 subtype and the presence of PML-RARα or AML1-ETO fusion genes were significantly associated with low frequency of extramedullary involvement (P=0.0002, P=0.002 and P=0.0047, respectively), while a correlation with the presence of extramedullary disease was found with the presence of MLL-AF9, M4 and M5 FAB subgroups and a WBC count at diagnosis higher than 20 000/μl (P=0.024, P=0.017, P=0.018 and P=0.046, respectively). CNS involvement was found in 19 patients and, in five of them, was the sole extramedullary site involved. A significant correlation was found only between CNS disease and a WBC count of 50 000/μl at diagnosis (P=0.002).

In our series, the genetic subgroups showed a characteristic clinical and morphological profile with regard to FAB subtype, age distribution, WBC count at diagnosis, extramedullary disease and CNS involvement. PML-RARα-positive cases showed a strong association with the FAB-M3 subtype, a progressive increase of frequency with age, a low WBC count and only occasional extramedullary and CNS involvement. AML1-ETO-positive leukaemia demonstrated a peculiar distribution with respect to specific age groups, with the highest incidence between 5 and 10 years, and a close, but not exclusive, correlation with the FAB-M2 subtype. Furthermore, a moderate increase of the WBC count at diagnosis seems to be a typical feature of t(8;21) AML; indeed, in our patients the median WBC count was 16 000/μl, similar to values reported by other authors, while, unlike the results of other studies, we did not observe an increased frequency of CNS involvement in this group. CBFβ-MYH11 AML were associated with the FAB-M4 subtype, even if occasionally found in M2 and M5 AML, and were characterized by a high WBC count at diagnosis with frequent CNS involvement. MLL-AF9-positive AML was the group with the highest percentage of infants and the lowest median age; these cases often showed extramedullary disease and CNS involvement even though the median WBC count at diagnosis was <20000/μl.

Complete remission was achieved in 84.5% of cases. In patients with AML positive for AML1-ETO, CBFβ-MYH11, MLL-AF9 and PML-RARα, the CR rate was 91, 87.5, 91, and 84.3%, respectively. The lowest CR rate was observed in patients less than 1 year (63%, P=0.007) and when the WBC count at diagnosis was higher than 50 000/μl (76.9%, P=0.029).

Univariate and multivariate survival analyses were carried out on a series of 200 patients, excluding secondary AML, Down syndrome and promyelocytic leukaemia since specific protocols, including ATRA, were implemented to treat APL over the years. Variables considered in univariate analysis were: gender, extramedullary disease, CNS involvement, presence of AML1-ETO/CBFβ-MYH11 gene rearrangements, FAB subtypes, median WBC count at diagnosis 20 000/μl or >20 000/μl and age 1 year or >1 year. The significant prognostic factors for both EFS and OS are listed in Table 2. Owing to the scarce number of cases and the presence of a case with Down syndrome, the MLL-AF9 gene rearrangement was not considered in the survival analysis; nevertheless, in our series 54.5% of the patients were still alive in remission. Multivariate analysis (Table 3) confirmed that the absence of the AML1-ETO and CBFβ-MYH11 gene rearrangements, a WBC count of >20 000/μl at diagnosis and an age 1 year are independent unfavourable prognostic factors. A general consensus exists in considering CBFβ-MYH11-positive AML as a group characterised by better CR and better survival rates than other AML, in spite of hyperleucocytosis at diagnosis and CNS involvement, while conflicting data exist with regard to AML1-ETO-positive AML. In these genetic subgroups, an improvement of the outcome in patients treated with high-dose cytarabine has been described.8 Therefore, the poor outcome occasionally reported might be related to the absence of high-dose cytarabine in the treatment regimens. In our series, many patients received high-dose cytarabine and, thus, the favourable outcome observed in patients with AML1-ETO- and CBFβ-MYH11-positive AML could depend on the treatment received.

Table 2 Univariate survival analysis: significant variables
Table 3 Multivariate analysis by Cox's model

Finally, we performed long-term survival analysis to evaluate the events time distribution considering that the main cause of failure in AML was relapse (Figure 1). A time interval from diagnosis of 18 months or longer was significantly more frequent among AML lacking AML1-ETO and CBFβ-MYH11 transcripts than in positive cases (12.3 vs 2%, respectively, P=0.02). As a consequence, the survival probability showed a progressive decrease in the first group of patients so that their corresponding 10-year survival probability was 31.7±4.8%. On the contrary, patients with AML1-ETO- and CBFβ-MYH11-positive AML achieved a plateau at 4 years for both EFS and OS. Among infants, all events occurred within 14 months from diagnosis and were mainly due to induction death. Thus, the presence of early relapses seems to distinguish the AML1-ETO/CBFβ-MYH11-positive cases from the negative ones. These data have been not previously reported and suggest that the molecular study of minimal residual disease, performed during the first 2 years from diagnosis, could allow for the timely identification of early events in AML with rearrangement of the core binding factor genes AML1 and CBFβ, since relapses are the main cause of failure in these patients. In addition our study suggests that a long-term follow-up could be necessary to evaluate the true probability of survival in patients at high risk of late relapse.

Figure 1

Event-free survival. Patients by molecular markers.


  1. 1

    Forestier E, Heim S, Blennow E, Borgstrom G, Holmgren G, Heinonen K et al. Cytogenetic abnormalities in childhood acute myeloid leukaemia: a Nordic series comprising all children enrolled in the NOPHO-93-AML trial between 1993 and 2001. Br J Haematol 2003; 121: 566–577.

  2. 2

    Biondi A, Rovelli A, Cantù-Rajnoldi A, Fenu S, Basso G, Luciano A et al. Acute promyelocytic leukaemia in children: experience of the Italian Pediatric Hematology and Oncology Group (AIEOP). Leukemia 1994; 8: 1264–1268.

  3. 3

    Martinez-Climent JA, Lane NJ, Rubin CM, Morgan E, Johnstone HS, Mick R et al. Clinical and prognostic significance of chromosomal abnormalities in childhood acute myeloid leukemia de novo. Leukemia 1995; 9: 95–101.

  4. 4

    Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, Harrison G et al. The importance of diagnostic cytogenetics on outcome in AML: analysis of 1612 patients entered into MRC 10 trial. Blood 1998; 92: 2322–2333.

  5. 5

    Raimondi SC, Chang MN, Ravindranath Y, Behm FG, Gresik MV, Steuber CP et al. Chromosomal abnormalities in 478 children with acute myeloid leukemia: clinical characteristics and treatment outcome in a cooperative Pediatric Oncology Group Study – POG8821. Blood 1999; 94: 3707–3716.

  6. 6

    Krauter J, Peter W, Pascherberg U, Heinze B, Bergmann L, Hoelzer D et al. Detection of karyotypic aberrations in acute myeloblastic leukaemia: a prospective comparison between PCR/FISH and standard cytogenetics in 140 patients with de novo AML. Br J Hematol 1998; 103: 72–78.

  7. 7

    Rowe D, Cotterill SJ, Ross FM, Bunyan DJ, Vickers SJ, Bryon J et al. Cytogenetically cryptic AML1-ETO and CBFβ-Myh11 gene rearrangements: incidence in 412 case of acute myeloid leukaemia. Br J Haematol 2000; 111: 1051–1056.

  8. 8

    Byrd JC, Dodge RK, Carrol A, Baer MR, Edward C, Stamberg J et al. Patients with t(8;21)(q22;q22) and acute myeloid leukaemia have superior failure-free and overall survival when repetitive cycles of high-dose cytarabine are administered. J Clin Oncol 1999; 17: 3767–3775.

Download references


We thank Professor G Valsecchi for her helpful suggestions, Ms S Disarò for excellent technical assistance, Dr A Leszl for cytogenetic data and Dr C Case for linguistic consultancy. We are grateful to all the clinicians of the participating AIEOP centres for clinical data. We also thank S Fenu, L Sainati, A Cantù Rajnoldi and G Basso for the morphological review. This research was supported by Fondazione Città della Speranza and and Prog.Fin. 2001/01/X/000177 Ministero Salute OBG, MIUR 60-40% and CNR Progetto Oncologia.

Author information

Correspondence to E Frascella.

Rights and permissions

Reprints and Permissions

About this article

Further reading