Monitoring minimal residual disease by quantification of genomic

Leukemia (2006) 20, 451–457 & 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00 www.nature.com/leu

ORIGINAL ARTICLE Monitoring minimal residual disease by quantification of genomic chromosomal breakpoint sequences in acute leukemias with MLL aberrations T Burmeister1, R Marschalek2, B Schneider2, C Meyer2, N Go¨kbuget3, S Schwartz1, D Hoelzer3 and E Thiel1 1

Medizinische Klinik III, Campus Benjamin Franklin, Charite´ Universita¨tsmedizin Berlin, Hindenburgdamm 30, Berlin, Germany; Institut fu¨r Pharmazeutische Biologie/ZAFES/Diagnostikzentrum fu¨r Akute Leuka¨mie (DCAL), Johann Wolfgang GoetheUniversita¨t, Marie-Curie Str. 9, Frankfurt/Main, Germany and 3Medizinische Klinik III, Johann Wolfgang Goethe-Universita¨t, Theodor Stern-Kai 7, Frankfurt/Main, Germany 2

An estimated 10% of acute leukemias carry mixed-lineage leukemia (MLL) fusion genes. Approximately 50 different fusion partners of the MLL gene have already been molecularly identified. These leukemias are commonly regarded as highrisk cases and are treated accordingly with intensified therapy regimens, including hematopoietic stem cell transplantation. However, a subset of patients may achieve long-term remissions with conventional therapy. Monitoring minimal residual disease (MRD) is undoubtedly of great value in clinical decision making, also in the pre- and post-transplant setting. Here, we describe a novel method for detecting MRD in leukemias with MLL aberrations. The method is based on monitoring patientspecific chromosomal breakpoint DNA sequences. This has several advantages over other methods that are based either on detecting specific RNA molecules of MLL fusion genes or on surrogate markers. An accurate and absolute quantification of the MRD level is possible. No reference to housekeeping genes is necessary and the target structure is much more stable than any mRNA fusion transcript. Leukemia (2006) 20, 451–457. doi:10.1038/sj.leu.2404082; published online 19 January 2006 Keywords: real-time PCR; pro-B ALL; 11q23; GMALL; MLL-PTD

Introduction The mixed-lineage leukemia (MLL) gene on chromosome 11q23 is involved in around 10% of chromosomal translocations in acute leukemias.1,2 In infant (age o1 year) acute lymphoblastic leukemia (ALL), the percentage of cases carrying MLL translocations is more than 50%.3 Fourty-nine MLL fusion genes have thus far been identified on the molecular level but at least a further 36 translocations have been described cytogenetically and await molecular characterization.4 Most MLL fusion genes are associated with an adverse clinical prognosis.1,5–7 Therefore, patients with these molecular abnormalities are usually treated as high-risk cases and have to undergo intensified therapy regimens including hematopoietic stem cell transplantation. An improved outcome has been accomplished in these patients under newer therapy regimens.8 It is not yet clear if all patients with MLL fusion genes should undergo stem cell transplantation. A subset could be cured by conventional intensified therapy.5,8 Under several aspects it appears worthwhile to investigate the level and course of Correspondence: Dr T Burmeister, Medizinische Klinik III, Charite´ Universita¨tsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, 12200 Berlin, Germany. E-mail: [email protected] Received 1 June 2005; revised 22 October 2005; accepted 17 November 2005; published online 19 January 2006

minimal residual disease (MRD) in patients with MLL fusion genes. Transplantation candidates could possibly be identified by an adverse MRD profile. On the other hand, the remission status in the post-transplant setting could be assessed. The currently used MRD detection methods are based on detecting the chimeric mRNA of the MLL fusion gene and thus have some disadvantages. First, no absolute quantification is possible and quantification can only be made with reference to a ‘housekeeping gene’. The concept of a housekeeping gene, however, is somewhat problematic and the assumption of a universally stably expressed gene can always be at best an approximation. Therefore, MRD data based on mRNA quantification are only to a limited extent comparable between different patients. Second, not only the housekeeping gene but also the MLL fusion gene itself may display a variable expression over time and its expression may be influenced by therapeutic procedures and agents, for example, certain cytostatics. Third, since MLL is found to be fused to a large set of different genes, a separate quantitative PCR has to be constructed for each of these genes. Even in one single entity, several different fusion transcripts may be possible. For example, in the case of t(4;11)(q21;q23), up to 10 different MLL-AF4 fusion mRNAs have been observed, depending on the precise location of the chromosomal breakpoint, although the detection has been somewhat simplified by the standardized BIOMED-1 and EAC protocols.9,10 In contrast, DNA-based MRD diagnostics allows a precise absolute quantification of the MRD level. In addition, DNA is much more stable over time compared to RNA. We report here a novel method for detecting MRD in acute leukemias based on the quantitative detection of MLL chromosomal breakpoint sequences. Using this approach we have measured prospectively the MRD profiles of high-risk pro-B ALL patients with MLL fusion genes since 2003 within the German Multicenter Acute Lymphoblastic Leukemia (GMALL) therapy study group.

Materials and methods

Patients Bone marrow and peripheral blood samples were obtained for diagnostic purposes within the framework of the GMALL therapy studies after written informed consent. Samples that showed a pro-B-ALL immunophenotype or expressed the NG2 antigen were further analyzed by RT-RCR.

Immunophenotyping FACS analysis and subclassification was carried out using standard procedures as described previously.11 Immunopheno-

MRD monitoring in leukemias with MLL fusion genes T Burmeister et al

452

typic classification was done according to the EGIL criteria.12 In particular, a pro-B immunophenotype was diagnosed in case of TdT þ , HLA-DR þ , CD10, CD19 þ , cyCD22 þ , CD79a þ , cyIg, and sIg. Divergent from the EGIL classification, we considered a pro-B subtype in patients with or without expression of CD24.

PCR primers and TaqMan probes Oligonucleotides and TaqMan probes were obtained from metabion Inc. (Martinsried/Germany) and HPLC purified. Primers and probes were preferably designed in a way that they did not anneal in Alu repeat regions. PCR product sizes ranged from 83 to 175 bp. Detailed sequence data, primers, and probes are given in the online supplement to this article.

RT-PCR RT-PCR for detecting the MLL-AF4 fusion gene mRNA was performed as described previously.11 In addition, samples that were MLL-AF4-negative were further analyzed by multiplex RTPCR for five other MLL fusion genes (MLL-AF6, MLL-AF9, MLLAF10, MLL-ENL, MLL-ELL) essentially as described previously.13 Samples revealing an MLL fusion gene by RT-PCR were further subjected to analysis on the DNA level to characterize the chromosomal breakpoint.

DNA isolation and preparation of a dilution series Great care was taken in the preparation of the DNA dilution series since this is a critical point in obtaining a valid standard dilution row. Pooled leukocytes from buffy coats of healthy blood donors were isolated by Ficoll gradient centrifugation. DNA was prepared from the buffy coat leukocytes using the NucleoSpin Blood XL kit (Macherey-Nagel, Du¨ren/Germany) which is suited for the isolation of DNA from larger quantities of blood and carefully adjusted by subsequent dilutions to a photometrically determined concentration of 60 ng/ml in TE buffer. The dilution was carried out over some days to allow a complete and homogeneous hydration of the high-molecular genomic DNA in the buffer. This buffy coat DNA was aliquoted in several portions of 225 ml and stored at 41C. DNA from patient’s samples was prepared using an alkaline-lysis-based method with subsequent isopropanol precipitation (Puregene, Biozym Diagnostik, Hessisch Oldendorf/Germany) and carefully and slowly adjusted to a final concentration of 60 ng/ml in TE buffer. To prepare a dilution row, 25 ml of patient DNA from the time point of primary diagnosis were added to 225 ml of buffy coat DNA and carefully mixed by pipetting up and down several times. After some hours of hydration time, 25 ml of this 101 dilution were added to a second 225 ml buffy coat DNA tube and so on up to a dilution of 105. This process of preparing the dilution row was carried out over some days to allow sufficient hydration of the genomic DNA.

Chromosomal breakpoint characterization in the MLL gene The method for the isolation of the chromosomal breakpoint has been described in detail recently.14 Briefly, genomic patient DNA from the time point of primary diagnosis was cut with BamHI, self-ligated, and subjected to an inverse long PCR method. The resulting PCR product sizes allowed a rapid identification of the chromosomal breakpoint by DNA sequencing. Usually less than 2 mg genomic DNA were sufficient for the Leukemia

whole analysis, and the identification of the breakpoint was usually complete within a few days.

Real-time quantitative PCR Real-time quantitative PCR was performed on a RotorGene RG3000 cycler (Corbett Research, Wasserburg am Inn/Germany) using basically the TaqMan chemistry. The method should therefore be applicable to other real-time PCR cycler systems without major changes. The Absolutet QPCR Mix (ABGene, Hamburg/Germany) with hot start enzyme Thermo-Starts DNA Polymerase and the following cycler program was used: 15 min 951C, 55 cycles (951C 15 s, 601C 60 s), 10 min 251C. PCR was performed in a duplex format. The reaction mix included four primers and two TaqMan probes: the two patient-specific MLL breakpoint primers with the 50 -FAM/30 -BHQ1-labelled patientspecific TaqMan probe and the primers (50 -30 ) hck-f TATTAG CACCATCCATAGGAGGCTT, hck-r GTTAGGGAAAGTGGAG CGGAAG, and probe hck-p HEX-TAACGCGTCCACCAAGGA TGCGAA-BHQ1. The latter primers amplified a 80 bp segment from the human hematopoietic stem cell kinase (hck) gene on chromosome 20q11–q12 as described previously.15 Primer and probe concentrations were the following: hck-f 50 nM, hck-r 30 nM, hck-p 200 nM, each MLL breakpoint primer 200 nM, the patient-specific TaqMan probe 400 nM. In total, 540 ng (9 ml) of samples DNA were added to each tube. For future experiments, we, however, recommend the use of slightly larger DNA quantities (e.g. 600–660 ng ¼ 10–11 ml) per experiment to further increase the reproducible sensitivity. The dilution row samples were investigated threefold and each follow-up sample was investigated fourfold per experiment. The detailed data of each PCR are shown in Table 1.

Principles of primer and probe design As illustrated in Table 1, there was a considerable variability in primer and probe melting temperatures and PCR product sizes. The Online Supplement to this article furthermore illustrates the variable location of the primers and probes with respect to the chromosomal breakpoints. For primer/probe design, we recommend the following: primer melting temperatures should be around 581C, probe melting temperatures at least 651C (both calculated as outlined in the legend of Table 1). The usual guidelines for primer/probe design apply (no hairpins, no dimers, etc.). The PCR product size should be as short as possible, but PCR products up to 175 bp (or perhaps even longer) may yield good results. The primers should anneal on separate chromosomes (although a significant overlap may be tolerable) and the probe should be located 3–8 bp 30 downstream of one primer on the same strand and should not contain a guanine at the 30 or 50 end.

Data interpretation Data interpretation basically followed the suggestions outlined by van der Velden et al.18 The designed primers and probe were tested with the standard dilution series and with two negative controls (buffy coat DNA as used for the preparation of the dilution series and water). If this yielded a satisfying result, follow-up samples were investigated together with the standard samples as described above. The MRD level in a follow-up sample was calculated from the averaged quadruplicate Ct value by substituting this value into the equation of the standard curve. The gene hck was always amplified in parallel in each PCR and detected in the HEX channel. It served as a control for DNA


453 Table 1 Sample no.

Real-time quantitative PCR details for all patients Slope of the curve

Correlation Reproducibility of Lowest DCt12 the 105 dilution coefficient R2 of amplified standard curve dilution (log10)

DCt23

DCt34

DCt45

Tm forw (1C)

4.07 3.76 3.67 4.38 3.06 4.34 3.37 3.52 4.08 4.28 3.37 4.08 3.98 3.91 3.06 3.15 4.70 3.25 4.28 3.57 3.64 3.29 3.58 3.08 3.33 3.56 3.54 3.98 3.43 3.48 3.92 3.94 3.68 3.45

3.19 3.55 3.86 4.40 3.77 3.75 4.30 5.56 3.01 3.69 3.50 3.43 3.16 6.02 3.30 3.89 4.95 3.34 3.18 2.79 3.83 3.47 3.71 3.27 2.82 4.21 3.73 3.77 3.68 2.61 3.99 3.62 3.49 3.35

3.76 3.60 2.32 2.48 2.69 2.37 3.71 4.09 3.11 2.90 3.10 2.42 2.52 5.20 4.99 2.62 2.51 3.36 5.08 2.73 3.36 1.77 2.96 3.61 2.25 2.61 2.38 3.39 1.41 3.46 1.67 1.97

58 57 58 58 56 59 58 56 58 57 58 57 57 57 54 58 58 56 56 60 60 58 58 59 56 58 59 58 58 59 57 58 58 57

57 57 57 59 56 59 59 61 59 56 55 59 60 58 57 57 58 56 58 56 55 55 56 56 63 58 57 57 57 60 60 56 58 57

64 63 65 66 64 67 62 64 62 66 66 65 67 64 63 66 66 64 64 64 64 62 65 66 66 68 67 67 65 67 68 66 69 66

99 120 128 92 103 105 175 93 134 117 96 86 107 84 110 113 86 110 100 107 86 92 100 103 110 110 101 153 114 83 94 170 94 112

57.6

57.6

65.2

108

3169 3402 3479 3483 3524 3602 3669 3729 3787 3812 3846 4006 4093 4159 4215 4249 4278 4279 4287 4415 4562 4606 4654 4697 4705 4777 4787 4793 4796 4841 4850 4867 4868 4898

3.59 3.46 3.53 3.58 3.37 3.90 3.66 4.30 3.53 3.89 3.65 3.45 3.38 4.61 3.69 3.37 4.11 3.35 3.78 3.49 3.52 3.35 3.20 3.28 3.24 3.48 3.66 3.82 3.53 3.26 3.56 3.94 3.19 3.14

5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 4 5 5 5 5 5 5 5 4 5 5 5 5 5 5 5

mean

3.58

4.94

1/3 1/3 1/3 2/3 1/3 1/3 1/3 1/3 1/3 1/3 1/3 1/3 2/3 2/3 2/3 2/3 1/3 2/3 0/3 1/3 1/3 3/3 2/3 1/3 1/3 1/3 0/3 2/3 1/3 1/3 1/3 1/3 3/3 2/3

0.99 0.98 0.99 0.99 0.99 0.98 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.98 0.99 0.99 1.00 0.99 0.99 0.99 0.99 0.99 0.99 1.00 0.99 0.99 0.99 0.99 0.99 0.98 0.99 0.98 0.99

3.34 3.96 3.55 3.31 3.61 3.47 3.86 3.89 3.69 4.14 4.19 3.23 3.58 2.93 4.10 3.37 3.38 3.54 3.73 3.51 3.42 3.21 3.22 3.79 3.48 3.01 3.76 4.34 4.11 3.79 3.86 4.66 3.67 3.54

1.32/3

0.989

3.65 3.70 3.71 3.01

Tm rev (1C)

Tm probe (1C)

PCR product (bp)

The table shows the slopes of the standard curves, the reproducibility of the 105 dilutions, the correlation coefficients of the standard curves, the difference in Ct values between successive dilutions in the standard curve, the melting temperatures (Tm) of the forward and reverse primers and the TaqMan probes and the PCR product sizes. Tm values were calculated using the oligonucleotide properties calculator. (http:// www.basic.northwestern.edu/biotools/oligocalc.html) based on the nearest neighborhood method as described by Breslauer et al.16 but with the values published by Sugimoto et al.,17 assuming [salt] ¼ 50 mM, [primer] ¼ 200 nM and [probe] ¼ 400 nM. In 31 cases (91%), at least 1 of the threefold 105 standard samples yielded a signal and the correlation coefficient of the standard curve was 40.975 in all PCRs. The theoretical difference between Ct values of 10-fold diluted samples is log2(10)E3.32. The slope of a standard curve using a decadic logarithmic scale should ideally approach 3.32 (if the dilutions are plotted in ascending order). The data in the Table show that this value is sufficiently approximated in most investigated samples. Samples 3729 and 4159 deviate most from this ideal value in retrospect. Recommendations regarding PCR efficiency are given under ‘Data interpretation’.

quality and quantity and all investigated (follow-up and standard) samples in an experiment were required to yield similar (72) Ct values in the HEX channel. The absolute Ct value depends on the threshold setting and varied between different PCRs. An example is shown in Figure 1. Regarding PCR efficiency, van der Velden et al.18 suggested for the standard curve that ‘a slope beween 3.0 and 3.9 will probably be acceptable as long as the correlation coefficient is 40.95’. These recommendations were given for a monoplex real-time quantitative PCR. For our duplex PCR, we would recommend the following for future experiments: if the correlations coefficient is 40.95, the DCt value between subsequent dilutions should be beween 2.5 and 4.5, if all triplicates are amplified. This variation may naturally be larger if only one or two of the triplicates are amplified. The slope of the curve should be between 3.0 and 4.5. This lower stringency in PCR efficiency is, in our view, justified because a duplex PCR cannot

be generally expected to show the same PCR efficiency (i.e. slope of the standard curve) as a monoplex PCR, as two different PCR products are generated simultaneously. Table 1 shows that most of the samples met this standard. The two samples which deviate most (3729, 4159) showed a linear standard curve but very varying DCt values. These samples were among the first investigated in this study and were investigated under slightly different conditions (without a hot start technique) which may explain this deviation.

Results The chromosomal breakpoints within the MLL gene were successfully characterized in 71 adult patients with ALL (Figure 2). All nucleotide sequence data were submitted to the EMBL/Genbank/DDBJ database and are available under the Leukemia


454

Figure 1 Real-time duplex PCR for MRD detection. (a) The standard dilution row samples (threefold) and a follow-up sample (fourfold). (b) The standard row with location of the Ct values of the follow-up sample on the standard curve. The MRD level corresponds to a niveau of 5.4 105. (c) The detection of the hck gene in the HEX channel (standard row and follow-up samples).

Leukemia


455

Figure 2 Chromosomal breakpoints in the major breakpoint region of the MLL gene of 71 pro-B-ALL patients. Exon numbering according to Nilson et al.,19 base pairs are given in parentheses. Breakpoints of patients where follow-up material was available are indicated by open circles (J) with patient numbers, the remaining with closed circles (K). Detailed sequence data of the former are given in the Online Supplement. Sixtyfive patients had a MLL-AF4, 5 a MLL-ENL and one a MLL-AF9 fusion. The distribution of breakpoints was the following: intron 9: 34, exon 10: 3, intron 10: 16, exon 11: 3, intron 11: 14, intron 12: 1. Breakpoint sequences are accessible in the EMBL/Genbank/DDBJ database (Acc. nos. AM050737–AM050807).

accession numbers AM050737–AM050807. In 37 samples (35 MLL-AF4, one MLL-ENL, one MLL-AF9), the chromosomal breakpoint was determined retrospectively to test for the practicability and reliability of the breakpoint identification method but no follow-up investigations could be performed because of lacking follow-up samples. In the remaining 34 cases (30 MLL-AF4-, four MLL-ENL-positive), samples were prospectively investigated and a patient-specific quantitative PCR was established. The applicability of the constructed patient-specific real-time PCR was evaluated in each single case by testing it with the respective dilution series obtained from the time point of primary diagnosis. In all except four cases, the designed PCR primers and probe yielded excellent results with linear correlation coefficients above 0.975 and no further optimization was necessary. In one case, new primers and a new probe had to be designed because of unsatisfying results. In two cases, unspecific amplifications with negative controls at Ct values 440 were observed and new primers and probes were designed. These new primers and probes worked well afterwards. In a forth case, the result was improved by modifying concentrations of patient-specific forward and backward primer. The real-time quantitative PCR was always performed in a duplex format with parallel amplification of a control gene (hck on 20q11–q12). This guaranteed the integrity and appropriate quantity of the investigated sample DNA. It was technically possible to construct a highly sensitive quantitative patient-specific breakpoint PCR in each investigated case. This is not trivial as the MLL breakpoint region often involves Alu repeats and highly repetitive sequences (see the Online Supplement to this article). In all 34 cases, the reproducible sensitivity was at least 104 and in 32 cases (94%) at least one of the threefold samples of the 105 dilution in the standard dilution row also yielded a PCR signal. The overall maximal sensitivity (in the terms of van der Velden et al.18) was thus near the theoretical optimum. In 32 cases, follow-up samples were available and could be investigated. Altogether 103 evaluable follow-up samples were investigated for the MRD level in the above-described manner

(average: 3.2 follow-up samples per patient). All investigated patients were pro-B ALL patients enrolled in the GMALL 07/ 2003 therapy study. This therapy study includes MRD measurements at certain time points (days 11, 24, 44, 71, weeks 16, 22, 41, 52). Originally, MRD investigations within this therapy study were only required for standard-risk patients. High-risk pro-B ALL patients are recommended to undergo allogeneic stem cell transplantation in case of an available donor after completion of induction and consolidation cycles on day 71. Follow-up samples were available after day 71 in 16 patients. Examples of MRD profiles are given in Figure 3, and the MRD profiles of all thus far investigated patients are given in the online supplement to this article. Evaluation of clinical data was not carried out due to a short follow-up and the limited number of available samples which currently permit a meaningful analysis. The limited number of available follow-up samples in some cases was mostly due to logistical reasons. However, the astonishingly variable MRD profiles in different patients further underline the usefulness of MRD monitoring in these high-risk leukemia cases.

Discussion Assessing MRD is of great value for monitoring therapy response or long-term engraftment after allogeneic transplantation. In principle, MRD monitoring methods based on DNA targets are preferable over methods based on RNA targets since they allow a more accurate quantification of the MRD level. The simplest real-time quantitative PCR method for MRD detection is the use of the DNA-intercalating dye SYBR Green I.18 However, unspecific PCR products may also be detected with this approach. This unspecificity can be reduced by performing a melting curve analysis after amplification. When using individual patient-specific PCR primers, the PCR products may differ significantly in length and melting temperatures, thus making the discrimination between specific and unspecific amplificates Leukemia


456 Our method is in principle also applicable to MLL partial tandem duplications (PTDs) that can be found in 5–10% of AML cases.23,24 These MLL PTDs appear to be present in low copy number in a large percentage of healthy individuals which led some authors to the conclusion ‘that nested PCR for MLL duplication is not suitable for the detection of minimal residual disease’.23 We did not investigate cases with MLL PTDs because this target is currently not included in our routine diagnostics but a quantitative PCR based on genomic breakpoint sequences should be capable of distinguishing the malignant clone from a potential low copy MLL PTD ‘background’. MRD monitoring should thus be possible in these cases with our method. In summary, the method presented here allows a rapid, highly sensitive absolute quantification of the MRD level in patients with various MLL fusion genes. As it is based on DNA and not RNA it does not depend on housekeeping genes and appears much more robust than RNA-based methods. The obtained data show that MLL breakpoint junctions are suitable and reliable PCR targets for quantitative MRD analysis and demonstrate the practicability of the method in patients with various MLL aberrations. Given the advantages of a DNA-based method, we furthermore believe that the concept of MRD monitoring based on chromosomal breakpoint sequences will become more widely applied also in leukemia subtypes with other recurrent chromosomal translocations as soon as appropriate and standardized methods for the rapid identification of the respective chromosomal breakpoints have been developed, as is the case for the MLL gene.

Acknowledgements Figure 3 Examples of different MRD profiles in MLL-AF4-positive patients. (a) A patient with a continuous reduction in the MRD level who reaches MRD negativity after allogeneic transplantation after day þ 71, illustrating the value of MRD monitoring after and before transplantation. The patient in figure (b) shows an initial response to therapy but increasing MRD under continued therapy. This patient is an early non-responder to therapy and could have been identified as refractory to therapy at least several weeks before the relapse became clinically apparent.

not always easy. Hence, a sequence-specific detection with the use of double-labelled (TaqMan) probes appears more advantageous. A well-established method for detecting MRD in lymphoid neoplasms is based on the clonally rearranged immune receptor genes.20 Most ALL cases with MLL aberrations are of pro-B immunophenotype and pro-B-ALL often shows only an incomplete (i.e. DHJH) rearrangement of the immunoglobulin heavy chain locus and evidences oligoclonality.18,21 This oligoclonality may be difficult to detect if one of the clones is dominant. Even if the VDJ rearrangement has been completed, further rearrangements at least of the VHDH part may be possible by ‘V gene editing’ with replacement of the V region (reviewed in Darlow and Stott22). Thus it is generally recommended that MRD monitoring based on rearranged immune receptor genes should always be carried out using at least two different rearranged immune gene loci (Ig and TCR) to at least partially compensate for potential oligoclonality at diagnosis and the possible loss of the MRD target by secondary rearrangements.18,21 In contrast, MRD monitoring based on the genomic breakpoint sequence of the MLL fusion gene is not endangered by further rearrangements. The chromosomal break appears to be a unique molecular event and the resulting fusion gene furthermore plays a key role in the induction and/or maintenance of the malignant phenotype of the leukemic cell clone. Leukemia

We thank all participating clinics of the GMALL therapy study for their support. We are particularly indebted to Ms M Molkentin (Berlin) for skilful PCR analysis, and Ms B Komischke and R Lippoldt (both Berlin) for immunophenotyping. Grant numbers and sources of support: Grant 10-1988-Bu1 (Deutsche Krebshilfe) to TB and SS, grant 70-2657-Ho2 (Deutsche Krebshilfe) to DH and ET, grant 2002.032.1 (Wilhelm Sander Stiftung) to RM.

References 1 Pui CH, Relling MV, Downing JR. Acute lymphoblastic leukemia. N Engl J Med 2004; 350 (15): 1535–1548. 2 Schoch C, Schnittger S, Klaus M, Kern W, Hiddemann W, Haferlach T. AML with 11q23/MLL abnormalities as defined by the WHO classification: incidence, partner chromosomes, FAB subtype, age distribution, and prognostic impact in an unselected series of 1897 cytogenetically analyzed AML cases. Blood 2003; 102 (7): 2395–2402. 3 Gilliland DG, Jordan CT, Felix CA. The molecular basis of leukemia. Hematology (Am Soc Hematol Educ Program) 2004, 80–97. 4 Meyer C, Schneider B, Jakob S, Strehl S, Attarbaschi A, Schnittger S et al. The MLL recombinome of acute leukemias. 2005, submitted. 5 Pui CH, Chessells JM, Camitta B, Baruchel A, Biondi A, Boyett JM et al. Clinical heterogeneity in childhood acute lymphoblastic leukemia with 11q23 rearrangements. Leukemia 2003; 17 (4): 700–706. 6 Wetzler M, Dodge RK, Mrozek K, Carroll AJ, Tantravahi R, Block AW et al. Prospective karyotype analysis in adult acute lymphoblastic leukemia: the cancer and leukemia Group B experience. Blood 1999; 93 (11): 3983–3993. 7 Cimino G, Elia L, Mancini M, Annino L, Anaclerico B, Fazi P, et al., for the GIMEMA group. Clinico-biologic features and treatment outcome of adult pro-B-ALL patients enrolled in the GIMEMA 0496 study: absence of the ALL1/AF4 and of the BCR/ ABL fusion genes correlates with a significantly better clinical outcome. Blood 2003; 102 (6): 2014–2020.


457 8 Ludwig WD, Rieder H, Bartram CR, Heinze B, Schwartz S, Gassmann W et al. Immunophenotypic and genotypic features, clinical characteristics, and treatment outcome of adult pro-B acute lymphoblastic leukemia: results of the German multicenter trials GMALL 03/87 and 04/89. Blood 1998; 92 (6): 1898–1909. 9 van Dongen JJ, Macintyre EA, Gabert JA, Delabesse E, Rossi V, Saglio G et al. Standardized RT-PCR analysis of fusion gene transcripts from chromosome aberrations in acute leukemia for detection of minimal residual disease. Report of the BIOMED-1 Concerted Action: investigation of minimal residual disease in acute leukemia. Leukemia 1999; 13 (12): 1901–1928. 10 Gabert J, Beillard E, van der Velden VH, Bi W, Grimwade D, Pallisgaard N et al. Standardization and quality control studies of ‘real-time’ quantitative reverse transcriptase polymerase chain reaction of fusion gene transcripts for residual disease detection in leukemia – a Europe Against Cancer program. Leukemia 2003; 17 (12): 2318–2357. 11 Schwartz S, Rieder H, Schla¨ger B, Burmeister T, Fischer L, Thiel E. Expression of the human homologue of rat NG2 in adult acute lymphoblastic leukemia: close association with MLL rearrangement and a CD10()/CD24()/CD65s(+)/CD15(+) B-cell phenotype. Leukemia 2003; 17 (8): 1589–1595. 12 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. 13 Andersson A, Ho¨glund M, Johansson B, Lassen C, Billstro¨m R, Garwicz S et al. Paired multiplex reverse-transcriptase polymerase chain reaction (PMRT-PCR) analysis as a rapid and accurate diagnostic tool for the detection of MLL fusion genes in hematologic malignancies. Leukemia 2001; 15 (8): 1293–1300. 14 Meyer C, Schneider B, Reichel M, Angermueller S, Strehl S, Schnittger S et al. Diagnostic tool for the identification of MLL rearrangements including unknown partner genes. Proc Natl Acad Sci USA 2005; 102 (2): 449–454. 15 Fehse B, Chukhlovin A, Kuhlcke K, Marinetz O, Vorwig O, Renges H et al. Real-time quantitative Y chromosome-specific PCR (QYCS-PCR) for monitoring hematopoietic chimerism after sexmismatched allogeneic stem cell transplantation. J Hematother Stem Cell Res 2001; 10 (3): 419–425.

16 Breslauer KJ, Frank R, Blocker H, Marky LA. Predicting DNA duplex stability from the base sequence. Proc Natl Acad Sci USA 1986; 83 (11): 3746–3750. 17 Sugimoto N, Nakano S, Yoneyama M, Honda K. Improved thermodynamic parameters and helix initiation factor to predict stability of DNA duplexes. Nucleic Acids Res 1996; 24 (22): 4501–4505. 18 van der Velden VHJ, Hochhaus A, Cazzaniga G, Szczepan˜ski T, Gabert J, van Dongen JJM. Detection of minimal residual disease in hematologic malignancies by real-time. Quantitative PCR: principles, approaches, and laboratory aspects. Leukemia 2003; 17: 1013–1034. 19 Nilson I, Lochner K, Siegler G, Greil J, Beck JD, Fey GH et al. Exon/intron structure of the human ALL-1 (MLL) gene involved in translocations to chromosomal region 11q23 and acute leukaemias. Br J Haematol 1996; 93 (4): 966–972. 20 van Dongen JJ, Langerak AW, Bru¨ggemann M, Evans PA, Hummel M, Lavender FL et al. Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4-CT98-3936. Leukemia 2003; 17 (12): 2257–2317. 21 Szczepan˜ski T, Willemse MJ, Brinkhof B, van Wering ER, van der Burg M, van Dongen JJM. Comparative analysis of Ig and TCR gene rearrangements at diagnosis and at relapse of childhood precursorB-ALL provides improved strategies for selection of stable PCR targets for monitoring of minimal residual disease. Blood 2002; 99 (7): 2315–2323. 22 Darlow JM, Stott DI. V(H) replacement in rearranged immunoglobulin genes. Immunology 2005; 114 (2): 155–165. 23 Schnittger S, Wo¨rmann B, Hiddemann W, Griesinger F. Partial tandem duplications of the MLL gene are detectable in peripheral blood and bone marrow of nearly all healthy donors. Blood 1998; 92 (5): 1728–1734. 24 Do¨hner K, Tobis K, Ulrich R, Fro¨hling S, Benner A, Schlenk RF et al. Prognostic significance of partial tandem duplications of the MLL gene in adult patients 16 to 60 years old with acute myeloid leukemia and normal cytogenetics: a study of the Acute Myeloid Leukemia Study Group Ulm. J Clin Oncol 2002; 20 (15): 3254–3261.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia