A study on 289 consecutive Korean patients with acute leukaemias revealed fluorescence in situ hybridization detects the MLL translocation without cytogenetic evidence both initially and during follow-up
Dong Soon Lee, MD, Department of Clinical Pathology, Seoul National University College of Medicine, 28 Youngon-Dong, Chongno-Gu, Seoul, 110–744, Korea. E-mail: email@example.com
Summary. Translocations involving the MLL gene on the chromosome 11 (11q23) are frequently observed in acute leukaemia. The detection of this genetic change has a unique significance as a result of its implication of poor prognosis. To reveal the utility of fluorescence in situ hybridization (FISH) in detecting the MLL translocation, we analysed 289 consecutive Korean patients (children and adults) with acute leukaemias using both conventional cytogenetic analysis (CC) and FISH, placing an emphasis on the result discrepancies. Twenty-two of 289 patients (7·6%) had the 11q23/MLL translocation. In nine of 22 patients (41%), only FISH detected the translocation. In eight of these 22 patients, a total of 19 follow-up examinations were performed, of which FISH detected a significant level of leukaemic cells harbouring the MLL translocation in five patients (26%) without cytogenetic evidence. In addition to the MLL translocation, FISH detected submicroscopic amplification, partial deletion of the MLL gene and trisomy 11 in 12 patients without cytogenetic evidence. In summary, up to 41% of the MLL translocations at initial work-up and 26% during follow-up were detected by FISH without cytogenetic evidence. Thus, we recommend that MLL FISH should be performed in the diagnosis and monitoring of acute leukaemias in combination with CC.
Human leukaemia is now recognized as an acquired genetic disease. A large number of consistent chromosomal changes have been identified and some of these have provided unique insights into the understanding of the pathogenesis of the disease (Heim & Mitelman, 1995; Look, 1997). Karyotypic analysis to identify chromosomal abnormalities is now part of the routine work-up for diagnostic and risk-stratification studies for determining the appropriate therapy in newly diagnosed and relapsed leukaemia patients. Structural abnormalities involving the q23 band of chromosome 11 (11q23) are probably the most common of the many identified genetic abnormalities in haematological malignancies, and the majority of these involve the MLL (myeloid/lymphoid leukaemia or mixed lineage leukaemia) gene, which is also known as ALL-1, HRX and HTRX1 (Thirman et al, 1993). The MLL gene is located on 11q23 and contains more than 30 exons. Translocation involving this gene usually occurs between exons 5 and 11, which is known as the breakpoint cluster region (Djabali et al, 1992; Gu et al, 1992). The translocation of this gene occurs in approximately 5–8% of acute myeloid leukaemias (AML), 7–10% of acute lymphoblastic leukaemias (ALL) and in 60–70% of infant leukaemias, irrespective of the phenotype (Thirman et al, 1993; Pui et al, 1995). The 11q23/MLL translocation is known to be associated with acute leukaemia in infancy and with therapy-related leukaemia, and this translocation in acute leukaemia implies a poor prognosis although there is some disagreement on this issue (Cimino et al, 1995; Rubnitz et al, 1997). The term 11q23/MLL‘rearrangement’ is used to refer to reciprocal translocation (86%) and also to addition, duplication, inversion and deletion (16%) (Secker-Walker, 1998). With regard to translocations, at least 38 different partner chromosomes are involved.
The 11q23/MLL translocations are identified by conventional cytogenetic analysis (CC) and by molecular studies, e.g. Southern blot analysis, reverse transcription-polymerase chain reaction (RT-PCR) and fluorescence in situ hybridization (FISH). Several investigators have performed comparative studies on the detection of the 11q23/MLL translocations using different methods (Caligiuri et al, 1994; Poirel et al, 1996; Uckun et al, 1998; Mathew et al, 1999; Andreasson et al, 2000; Ibrahim et al, 2000). For the 11q23/MLL translocation, CC is important because it can detect the partner chromosomes of 11q23 that are involved in the translocation. Despite its high sensitivity, RT-PCR has a limited role, because of not only a lack of standardization in quantitative analysis, but also because of the presence of many partner chromosomes. In addition, a recent report showed that the 11q23/MLL translocation can be encountered in normal subjects by RT-PCR, which further complicates the application of this method on a clinical basis (Uckun et al, 1998). Caligiuri et al (1994) and Ibrahim et al (2000) used the Southern blot method to analyse MLL status. Southern blot analysis is considered unsuitable for a routine use, and a recent report by Stanulla et al (1998) described ‘pseudo-MLL rearrangement’ in cells incubated ex vivo and detected by Southern blot analysis. This observation limits the interpretation of results and the use of Southern blot analysis in the evaluation of MLL status. On the other hand, FISH can yield quantitative data rapidly with a small sample, and the use of this method on interphase cells gives unbiased results compared with CC, which requires cell culture to obtain metaphase cells (Cremer et al, 1986). Mathew et al (1999) reported differences in the results obtained by CC and FISH in the detection of the 11q23/MLL translocation. However, they performed FISH retrospectively on the selected patients documented to have the 11q23 rearrangements by CC for the comparison of the results. Andreasson et al (2000) also employed FISH, but only on cases of childhood ALL. To determine the utility of MLL FISH as a routine test for acute leukaemia, we employed both CC and FISH on bone marrow cells from 289 consecutive Korean patients diagnosed with acute leukaemias, placing an emphasis on the discrepant results. Using this approach, we identified not only 11q23+/MLL– cases, but also a substantial number of cases with 11q23–/MLL+.
Materials and methods
Patient material. From January 1996 to December 2000, a total of 289 consecutive Korean patients (174 males and 115 females; median age, 19 years, ranging from 1 month to 81 years of age) diagnosed with acute leukaemia were analysed by both CC and FISH to detect the 11q23 and MLL translocations (Table I). There were 160 patients (55·4%) over 15 years of age (adults), and 129 patients were 15 years and younger (children), including 11 infants. For the years 1996 and 1997, only children were included. The diagnosis of acute leukaemia was based on the morphology, cytochemical staining profile, and immunophenotype of the leukaemic cells according to the French–American–British (FAB) and European Group for the Immunological Characterization of Acute Leukaemias (EGIL) classification (Bennett et al, 1976, 1985; Bene et al, 1995). One hundred and fifty-three patients were diagnosed with AML, 117 with ALL and 19 with acute biphenotypic leukaemia (ABL). Among the 117 ALL patients, 90 were B-lineage ALL.
Table I. The age and phenotype distributions of 289 patients with acute leukaemias.
Figures in parentheses are percentages of corresponding groups versus total number of patients (n = 289). Age, at initial diagnosis; AML, acute myeloid leukaemia; ALL, acute lymphoblastic leukaemia; ABL, acute biphenotypic leukaemia.
Number of patients (%)
Conventional cytogenetic analysis. Cytogenetic analysis was performed on heparinized whole bone marrow samples, according to standard protocols. Karyotypes were described according to the nomenclature system proposed by the International System for Human Cytogenetic Nomenclature (ISCN, 1995).
Fluorescence in situ hybridization. FISH analyses were performed with a commercially available MLL probe (LSI®MLL dual-colour break apart rearrangement probe; Vysis, Downers Grove, IL, USA) and the chromosome enumeration probe for chromosome 11 (CEP® 11 DNA FISH probe; Vysis), according to the manufacturer's instructions. Analysis was done with an Olympus fluorescence microscope (Olympus America, Melville, NY, USA) attached to a computer-based imaging system (Quips XL Genetics Workstation; Vysis) equipped with a triple-bandpass filter for 4,6-diamidino-2-phenyl indole (DAPI), fluorescein isothiocyanate (FITC) and Texas Red. The centromeric probe for MLL had an orange signal, the telomeric probe a green signal and the background chromatin showed as blue (Fig 1). In bone marrow smear slides, the erythrocytes were removed by sinking the slides into 70% acetic acid/methanol solution for 30 s at room temperature and washing twice in distilled water. The slide was then dehydrated in a 70%, 80% and 95% ethanol series and treated with cold acetone. Two hundred nuclei were counted in each sample. Interphase cell analysis was performed according to the manufacturer's instructions. Briefly, when the MLL probe hybridizes with DNA in normal interphase cells, the orange and green signals are juxtaposed to yield two fusion (partial or total) signals from the two chromosomes 11. In interphase cells with an MLL gene translocation, the normally adjacent orange and green signals become rearranged. In the case of translocation, one fusion signal splits to yield one orange and one green signal, thus producing the signal pattern of one fusion, one orange and one green. For the purpose of discussion, we refer to the centromeric MLL orange signal as O, the telomeric green signal as G and the O/G fusion signal as F. For scoring purposes, fusion signals were defined as merging or touching O and G signals.
To determine the positive cut-off level of MLL FISH, control experiments were performed on the bone marrow samples from 20 patients free of haematological diseases. Mean + 3 SD (standard deviation) of 2·18% (0·6 + 0·53 = 3) was set as the normal range.
Among the 289 patients, karyotype analysis was not possible in 25 (8·7%) because of insufficient or absent mitotic cells. On the other hand, the FISH signal proved inadequate for analysis in only one patient. Twenty-two of 289 patients (7·6%) were found to have the 11q23 translocation by CC (designated as 11q23+) or the MLL translocation by FISH (designated as MLL+) (Tables II and III). Of these 22 patients, 20 (91%) were MLL+ and 12 (55%) were 11q23+: FISH detected the MLL translocation in nine patients (41%) in whom CC could not detect the 11q23 translocation (11q23–). CC identified the partner chromosomes in 12 patients with 11q23+(Tables II and IV). Twenty-two patients with 11q23+ and/or MLL+ consisted of 15 AML and seven ALL. Among the AML cases, 11 (73·3%) were AML M4 (three patients) or AML M5 (eight patients). Of seven ALL patients, four had ALL L1 and three had ALL L2. All ALL patients were B-lineage ALL except for one patient (patient 5), who was T-lineage ALL. Among a total of 19 patients with ABL, none had the 11q23/MLL translocation. Sixteen of 22 patients (72·7%) were children (15 years or less), including nine infants (41%). Four patients had a history of chemotherapy as a result of prior malignancies (Table V).
Table II. The summary of 22 patients with an 11q23/MLL translocation detected either by conventional cytogenetics or by fluorescence in situ hybridization.
BM diagnosis (Dx)
BM blast (%)
Karyotype (11q23 translocations shown in bold)
Patients with previous histories of chemotherapy.
The ages of patients 6 and 7 are at initial diagnosis while conventional cytogenetic analysis and fluorescence in situ hybridization studies were done with BM samples at relapsed state.
FISH results for the MLL translocation were considered negative (nuclei with split signals were less than 2·18% of cut-off level).
Patients with a lineage switch at relapse: in patient 3, AML M1 relapsed as ALL L1 and in patient 16, ALL L2 relapsed as AML M7 (second relapse).
All cases of ALL were of B-lineage except patient number 5. m, month(s); y, years; BM, bone marrow; CC, conventional cytogenetic analysis; FISH, fluorescence in situ hybridization.
Table IV. The partner chromosomes of the 11q23/MLL translocations in 22 patients with acute leukaemias detected by conventional cytogenetic analysis and/or FISH.
Number of patients
CC, conventional cytogenetic analysis; FISH, fluorescence in situ hybridization; AML, acute myeloid leukaemia; ALL, acute lymphoblastic leukaemia; CTx, chemotherapy; ?, unidentifiable because the 11q23 translocation was not detected by CC.
Including a case with negative FISH result
Cases with previous history of CTx
Negative MLL translocation by FISH
Revealed at follow-up analysis Patient with previous history of CTx
Negative 11q23 translocation by CC
Table V. The characteristics of the four patients with an 11q23/MLL translocation and with previous histories of chemotherapy.
m, months; y, years.
Langerhans cell histiocytosis
Prednisolone, vinblastine, methotrexate and cyclophosphamide
Cyclophosphamide, doxorubicin, 5-fluorouracil and tamoxiphen
Vincristine, dactinomycin and cyclophosphamide, Cisplatin, doxorubicin, methotrexate, ifosfamide and etoposide
Bleomycin, cisplatin and etoposide
A total of 19 follow-up examinations by CC and FISH were performed in eight patients among the 22 patients with 11q23/MLL translocation (Table II). Four of them (21·1%) showed MLL+ in a significant proportion of bone marrow cells despite 11q23–. Patient 3, with 2·3% FISH positivity for the MLL translocation (slightly over the positive cut-off level of 2·18%), was morphologically considered to be in remission with 2·6% blasts in his bone marrow, but relapsed 7 months later. At the time of this relapse, CC still showed a normal karyotype.
Abnormal copy number of MLL
In 16 of 289 patients (5·5%), MLL FISH signals suggested an abnormal MLL copy number (Table VI). As previously described, MLL FISH yields two fusion signals in a cell with two normal chromosomes 11. One or more than two fusion signals without spilt signals suggests a change in the copy number of MLL with or without a change in the number of chromosome 11 rather than the MLL translocation. To differentiate between these two possibilities, FISH was applied with chromosome enumeration probe 11 (CEP 11). CEP 11 FISH produces the same number of signals as the total number of chromosome 11 in each cell. As a result, of nine patients with three MLL fusion signals, eight were confirmed to have trisomy 11 either by CC or by CEP 11. It was of interest that among these eight patients, CC detected trisomy 11 in only one (patient 24). In patients 27 and 28, multiple (three or more) MLL fusion signals suggested the presence of amplification of the MLL gene (Fig 2). In all three patients with a single MLL fusion signal, the lost signal was not caused by the loss of chromosome 11, suggesting submicroscopic deletion of MLL.
Table VI. Comparison of the copy number of the MLL gene determined by MLL FISH with the number of chromosome 11 determined by chromosome enumeration probe (CEP) 11 FISH.
MLL copy number revealed by MLL FISH
Number of chromosome 11 by CEP 11
rem, remission; NA, not available; BMIML, bone marrow involvement by malignant lymphoma.
4 in 85·0%
4 in 75·5%
M2 In relapse Reinduction 39 d, in rem Reinduction 44 d, in rem In persistence
The design scheme of the commercial FISH probe used in the present study was similar to the cosmid probes used by van der Burg et al (1999) and should detect all MLL translocations irrespective of the partner loci, which is important as there are at least 38 partner chromosomes. In addition, the probe is single locus-specific (targeted to MLL on 11q23), double-coloured and also acts upon interphase nuclei. This strategy has been shown to overcome the low specificity at detecting the translocation, that is, a frequent false-positive signal, which is known to be a major drawback of the FISH method (cut-off limit of mean + 3 SD, 2·18% in MLL FISH in the present study). Recently, Cuthbert et al (2000) evaluated the sensitivity of FISH with the same commercial probe as that used in this study, using material from 29 patients and four cell lines, which had all been confirmed to harbour the MLL translocation both by CC and Southern blot analysis. They found 100% sensitivity in detection of the translocation.
In the present study, the frequency of the 11q23/MLL translocation in 289 consecutive patients with acute leukaemias was 7·6% with a predilection in infancy (81·8%). In terms of phenotype, the translocation was most frequent in AML (9·8%), followed by B-lineage ALL (6·7%) and T-lineage ALL (3·7%). It was of interest that there was no patient with ABL harbouring the translocation. The partner chromosomes revealed by CC in 13 patients were largely similar to those reported by Secker-Walker (1998) in a study of 550 patients with acute leukaemias and myelodysplastic syndromes (Table IV). Among the 22 patients with the 11q23/MLL translocation, 12 (54·5%) patients were revealed to have other chromosomal abnormalities by CC (Table I). In these patients, the additional chromosomal abnormalities might further increase the poor prognostic implication of the 11q23/MLL translocation, considering that complex karyotypic abnormalities per se are recognized as an independent adverse prognostic factor.
Four patients with the 11q23/MLL had a history of previous chemotherapy (Table V). They all received alkylating agents and in three patients, topoisomerase II inhibitors were co-administered (patients 18, 19, and 20). Patient 1 had been diagnosed with Langerhans cell histiocytosis and had received chemotherapy, including cyclophosphamide, at 4 months of age. Three months later, he was diagnosed with AML. Considering that alkylating-agent-related secondary acute leukaemia has an incubation period of about 5 years and that the 11q23/MLL translocations occurs in 60–70% of acute leukaemias in infancy, it would seem reasonable not to regard the AML in this patient as therapy related. As shown by the results of the initial work-ups, in as many as 41% of patients with the 11q23/MLL translocation, FISH detected the MLL translocation, whereas CC could not detect the 11q23 translocation (11q23–/MLL+). This apparently shows a higher sensitivity of FISH than CC in the detection of the MLL translocation. Of course, we cannot overlook the importance of CC in detecting the translocation partner chromosomes that may affect the prognostic implication (Martinez-Climent et al, 1995; Mrozek et al, 1997). Thus, CC and FISH should be performed in every patient with acute leukaemia in initial work-ups. There are many reports on the discrepancy between CC and MLL FISH (11q23–/MLL+) (Harrison et al, 1998; Lillington et al, 1998; Martineau et al, 1998; Moorman et al, 1998; Mathew et al, 1999). A possible explanation for this might be the cryptic translocation. In addition, a methodological bias could contribute because CC requires culturing to obtain metaphase cells, and malignant cells with the MLL translocation might not proliferate in in vitro culture. On the other hand, in two patients (patients 6 and 11), the situation was reversed, i.e. 11q23+/MLL–. As for this discrepancy, one can assume that the translocations occurred in 11q23 outside the known breakpoint cluster region of the MLL gene (Tanaka et al, 2001). We performed Southern blot analysis to verify this, but the results were inconclusive because of insufficient amounts of DNA. Our study also demonstrated that FISH was very sensitive at detecting the MLL translocation both at the initial work-ups and during follow-up. In addition, MLL FISH detected MLL rearrangements other than translocation and also trisomy 11, without cytogenetic evidence, showing that CC failed to detect trisomy 11 in a significant number of patients (seven of eight, 88%). CEP 11 FISH was useful for revealing the origin of the extra/lost MLL signal(s) in these patients.
In 25 out of 289 patients (8·7%), CC failed as a result of inadequate cells in mitosis. This failure rate further increased in B-lineage ALL patients up to 17·8%. By comparison, only one of 289 patients showed inadequate FISH signals. Thus, the low failure rate of FISH indicates another technical advantage over CC.
A major drawback of CC lies in the need for a culture procedure to obtain cells in the metaphase. In the bone marrow of patients with acute leukaemias, normal haemopoietic cells co-exist with leukaemic cells. These normal cells and leukaemic cells may differ in terms of their proliferability in a given condition. In other words, during culture, the selective proliferation of specific clones might occur, which can produce a result that does not represent the true karyotype of the malignant cells. In support of this, the percentages of MLL-positive cells by FISH in archival cultured bone marrow slides for CC obtained during the period from 1996 to 1999, tended to be lower than the respective percentages of bone marrow blasts (Table II). On the other hand, this tendency decreased in the results from the specimens obtained during the year 2000, which were direct preparations for FISH. We also found that the percentages of MLL+ cells by interphase FISH were more continuous and closer to those of bone marrow blasts than by CC, which were extremely deviant (Table II).
In conclusion, our results showed that MLL FISH detected up to 41% of the MLL translocations in acute leukaemia at initial work-ups and 26% during follow-up, without cytogenetic evidence. Thus, MLL FISH should be regarded as an indispensable tool in patients with acute leukaemias for their diagnostic and monitoring work-ups. We believe the application of MLL FISH is very useful, considering the current trend to include the MLL translocation in the classification of the acute leukaemias.
This study was supported by a grant from the 1999 Korean National Cancer Control Program, Ministry of Health & Welfare, R. O. K.