Late relapses evolve from slow-responding subclones in t(12;21)-positive acute lymphoblastic leukemia: evidence for the persistence of a preleukemic clone

Childhood B-cell precursor (BCP) acute lymphoblastic leukemia (ALL) is a heterogeneous disease with a variety of distinct genetic abnormalities. Approximately 25% of cases harbor a t(12;21) with the molecular equivalent of a TEL/AML1 gene fusion.1-3 Clinically, patients with this type of leukemia have low-risk features. Nevertheless, well-treatable relapses occur in about 20% of cases, typically after a long first remission period.4-10 However, the relapse incidence varies and depends on the respective treatment protocol.3-10 

The clone-specific TEL/AML1 gene fusion is readily detected in neonatal blood spots in most children with TEL/AML1-positive leukemia, proving that it occurred already in utero.11 The long latency between this first hit and the clinical manifestation suggests that further events are required to trigger the stepwise progression of the clone.12,13 One obvious candidate for such a contributing factor is the concurring deletion of the second TEL allele as seen in approximately 75% of TEL/AML1-positive cases.14,15 

Relapses are assumed to represent the re-emergence of the initial leukemic clone. This notion is derived from the finding that in the majority of cases the initial and the relapse clone share clonotypic antigen receptor gene rearrangements.16-19 Antigen receptor genes of normal lymphoid cells are rearranged during early development. Each cell and its progeny have their individual immunoglobulin/T-cell receptor (Ig/TCR) rearrangements, and therefore, they can be used for the characterization of a specific lymphoid cell population, particularly for clonality studies. Thereby, it was seen that oligoclonality is common in acute lymphoblastic leukemias.20-22 

We have recently shown in 2 patients with TEL/AML1-positive ALL that off-treatment relapses were not derived from the dominant clone at diagnosis but from a sibling clone as characterized by antigen receptor gene rearrangements and 12p status. Thus, it was speculated that both presentations of the leukemia were derived from a TEL/AML1-positive pre/leukemic clone.23 We now demonstrate that the relapse and the initial clone are derived from an identical TEL/AML1 preleukemic clone. Further, the pre/leukemic relapse clone is already present at initial diagnosis, but has a slower response to therapy compared with the dominant clone. In contrast, this clone responds rapidly to chemotherapy when it becomes the relapse leukemia.

We selected 2 children in whom the late systemic relapse clone from a TEL/AML1-positive leukemia differed with regard to the immune receptor rearrangements. Leukemic cells or DNA from both leukemic manifestations, as well as from follow-up bone marrow cells, were required for inclusion into this study. Written informed consent was obtained from the patients or their guardians. The study was approved by the ethical committee of the Children's Cancer Research Institute and the St Anna Kinderspital.

Dual-color fluorescence in situ hybridization (FISH) was performed according to standard methods on methanol/acetic acid-fixed samples of cells that were cultured for 24 hours. For the analysis of the TEL/AML1 gene fusion, 2 cosmids (50F4 and cos664) were used that are specific for the TEL and AML1 locuses, respectively. The TEL probe covers 35 kb of intron 1 and exon 2. We screened at least 100 nuclei per sample and analyzed DNA preparation from fresh cells. Mononuclear cells (MNCs) were isolated from bone marrow (BM) samples using Ficoll-Paque (1.077 g/mL; Pharmacia, Uppsala, Sweden) density gradient centrifugation. DNA was extracted by QIAamp Blood Kit (QIAGEN, Chatsworth, CA). Integrity of the DNA was confirmed by amplification of albumin as previously described.24 By amplification of this control gene the amount of DNA was measured in all samples, compared with a standard, and corrected accordingly.

Polymerase chain reaction (PCR) amplification of incomplete and complete IgH rearrangements was performed using family-specific D_H and V_H primers, respectively, and one J_H consensus primer, as previously described.25 For the detection of the IgK–kappa deleting element (KDE) rearrangements we used 4 Vk primers and an intron recombination signal sequence (RSS) primer in combination with a KDE primer. Sequence of the primers and PCR conditions were used as described previously.26 PCR for the detection of incomplete TCRD (Dδ2-Dδ3, Vδ2-Dδ3) rearrangements and of TCRG rearrangements (VγI-Jγ.1/2.1, VγII-Jγ1.1/2.1, VγI-Jγ1.3/2.3, VγII-Jγ1.3/2.3) was performed as previously described.26 All primers were purchased from a commercial supplier (Invitrogen, Carlsbad, CA).

An aliquot of 500 ng DNA was used for each amplification reaction in a 50 μL volume of 1× buffer (Applied Biosystems, Foster City, CA) containing 200 nM of each deoxynucleotide triphosphate (dNTP), 30 pmol of each primer, and 0.5 U AmpliTaq Gold (Applied Biosystems). Amplifications were performed on a Thermocycler (PTC-200; Techne, Cambridge, United Kingdom). PCR products were then size-fractionated on 4% to 12% nondenaturing polyacrylamide gels (Invitrogen) and analyzed for heteroduplex formation in order to see if more than one clonal product of identical size was present and to separate clonal bands from polyclonal background.26 After elution in water, PCR products were directly sequenced (VBC Genomics, Vienna, Austria). All sequences were confirmed in different independent PCR reactions.

The involved gene segments were identified by the Imunogenetics Database (IMGT; http://www.ebi.ac.uk/imgt/), and all Ig rearrangements by comparison with sequences of all known human Ig genes obtained from the IGBlast (http://www.ncbi.nlm.nih.goc/blast/) or VBASE directory (http://www.mrc-cpe.cam.ac.uk/imt-doc/).

Breakpoint-spanning DNA fragments were amplified by nested long-range PCR. Primer sequences were deduced from gene bank sequence data covering the breakpoint cluster regions of both genes involved, TEL (U61375) and AML1 (AP001721), respectively, and designed with the help of OMIGA software (Oxford Molecular, Oxford, England). PCR was performed on an Eppendorf Mastercycler Gradient (Eppendorf, Hamburg, Germany). We used 2 nested primer sets located within exon 5 and intron 5 of TEL, respectively, together with 16 primer sets located within intron 1 of AML1 in all combinations possible. PCR was performed with a conventional kit (Gene Amp XL, Perkin Elmer, Foster City, CA) following the manufacturer's recommendations using 50 ng template DNA. Cycling conditions consisted of an initial denaturation step at 92°C for 2 minutes, followed by 30 cycles (92°C for 10 seconds, 68°C for 10 minutes and 30 seconds). The first-round PCR product (1 μL) was used as template for the nested reaction. Amplified products were excised from agarose gels, eluted, and directly sequenced. By using this approach, a breakpoint-spanning DNA sequence was obtained from patient 2 only. In patient 1, we assumed that the negative result was due to the breakpoint located within intron 2 of AML1, as is the case in about 20% of t(12;21) translocations. Intron 2 of AML1, however, is very rich in G/C residues. For this reason, a different set of 8 nested PCR primers was used to keep the size of possible PCR products lower. We combined 4 nested PCR primers binding within intron 2 of AML1 with 4 TEL primers. The PCR reaction mix was also modified to destabilize G/C bonds. PCR was performed in a total volume of 50 μL containing 100 ng template DNA, 1× buffer (Invitrogen), 12.5 μL enhancer-solution, 50 mM magnesium chloride, 10 mM of each dNTP (Biozyme, Heidelberg, Germany), 20 μM of each primer, and 1 U Taq polymerase (Taq DNA Polymerase; Invitrogen). Initial denaturation for 2 minutes at 94°C was followed by 35 cycles (94°C for 10 seconds, 68°C for 3 minutes) and a final annealing/extension step for 7 minutes. PCR primers giving rise to a product in both approaches are listed in Table 1. Primers for amplification of the derivative 21 breakpoint in patient 1 were selected on basis of the derivative 12 sequence.

Clone-specific oligonucleotide primers were designed homologous to the junctional regions of the Ig/TCR rearrangements and used for quantitative PCR, as previously described.24 Optimal thermocycling conditions were established for each patient-specific primer pair to detect the clone-specific sequence. The specificity and sensitivity of the primer combinations was determined for each clonal marker on DNA from either initial or relapse leukemia according to their presence in the screening PCR. Leukemic DNA was diluted (10⁻¹ to 10⁻⁷) in DNA from peripheral blood MNCs from healthy donors.

DNA for Ig/TCR and TEL/AML gene fusions (500 ng and 100 ng, respectively) was used as a template in a 50-μL reaction volume. Negative controls were included in each experiment: one sample containing the reaction mix without DNA, and another sample containing DNA from peripheral blood MNCs from healthy donors as a polyclonal control. DNA obtained at diagnosis from each patient was used as a positive control. Samples and controls were always analyzed in triplicate. Precautions to avoid contamination were followed as previously described.27 

The 2 children had a late systemic relapse of their TEL/AML1-positive ALL. The clinical data as well as the duration of first remission are listed in Table 2. The children are still in second complete continuous remission (CCR). The children received chemotherapy for non–high-risk leukemias of the Berlin-Frankfurt-Münster (BFM) ALL 90 protocol in Austria.28 Remission induction chemotherapy consisted of one week prednisone (60 mg/m² per day) plus one intrathecal methotrexate application on day 1, followed by weekly administrations of intravenous vincristine (1.5 mg/m²) and daunorubicine per infusion (30 mg/m²). Asparaginase infusions (10.000 IE/m²) were given twice a week for 4 weeks. On day 33 the morphologic remission status was evaluated in the bone marrow. Relapses were treated according to the BFM Rez ALL 95 study.29 The first treatment block consisted of dexamethasone 20 mg/m² on days 1 to 5, triple intrathecal administration on day 1, vincristine 1.5 mg/m² on days 1 and 6, methotrexate 1000 mg/m²per infusion (36 hours) on day 1 and E coli–asparaginase 10.000 IE/m² on day 4 (infused over 24 hours). Both patients achieved a morphologic remission after the initial and relapse induction chemotherapy, respectively.

In the first patient, we found a total of 6 rearrangements, 4 at diagnosis and 2 different, completely unrelated ones at relapse. In the second patient, we identified 5 distinct rearrangements at diagnosis and 6 at relapse, one of which was found in both instances. Based on our results that in both leukemias a maximum of 2 different rearrangements per antigen receptor gene were detected, we concluded that the initial diseases and relapses were monoclonal. Thus, with regard to the antigen gene receptor gene rearrangements, the diagnostic and relapse samples of the first patient appeared as having arisen from 2 completely unrelated clones. This is in contrast to the second patient, in whom a clonal relationship between diagnostic and relapse leukemia was evident.

The first patient had a TEL/AML1-positive as well as TEL/AML1-negative cell population, but both with very similar secondary changes. At diagnosis, a deletion of one TEL allele was seen in 9 of 13 TEL/AML1-negative metaphases. A TEL/AML1 fusion was present in 81% of the 80 analyzable interphase cells. The following percentages always refer to the whole number of analyzed interphase cells. The 19% TEL/AML1-negative cells comprised some with a TEL deletion together with an additional AML1 signal (3.8%, indicative of the presence of an additional chromosome 21), whereas 7.6% had none of these changes. Another 7.6% had 2 additional AML1 signals, but no TEL deletion. The TEL/AML1-positive cell population consisted of 31% with and 50% without a TEL deletion. The former comprised 23% with 2 and 7.6% with 3 and the latter 3.8% with one, 31% with 2, and 15.3% with 3 extra AML1 signals. In the relapse sample we were able to screen 130 cell nuclei, of which 48% were TEL/AML1-negative and 15% TEL/AML1-negative, but with a TEL deletion. The 37% TEL/AML1-positive cells included 7% with only a TEL deletion, 10% with 2 extra AML1 signals, and 20% without a TEL deletion, but with one extra AML1 signal.

In summary, these FISH data suggest that in the first patient all TEL/AML1-negative and -positive clones were also present in the relapse sample, albeit with a different distribution. This clonal heterogeneity most likely indicates that the TEL/AML1 gene fusion and the subsequent clonal evolution have occurred in a transformed clone before the immune receptor rearrangements.

In the second patient, a colocalization of the TEL and AML1 signals was found in 62% of interphase cells from the diagnostic sample. Moreover, in line with the presence of an additional derivative chromosome 21, der(21), 7% of nuclei showed a second colocalization, whereas the absence of a noncolocalized TEL signal indicated that the residual TEL allele was deleted in this cell population. In the relapse sample, on the other hand, 74% of nuclei contained a TEL/AML1 gene fusion and 48% of them had 2 colocalizations, whereas the signal for the second TEL allele was always present. In this patient the relapse clone without TEL deletion could not have evolved from the diagnostic clone with a deleted TEL.

To test our hypothesis that the clones from initial disease and relapse are derived from a common TEL/AML1-positive progenitor cell, we cloned the patient-specific TEL/AML1 and the reciprocal AML1/TEL fusion sites on the DNA level. In the first patient, the primer combinations P1/P3 and P2/P4 revealed a breakpoint-spanning PCR product of the AML1/TEL fusion on the derivative chromosome 12, der(12), in intron 5 of the TEL and in intron 2 of the AML1 gene. Sequencing was performed with primer P13. Sequence data were deposited in public databases (accession number AJ494735). TaqMan real-time PCR was performed with primers P16/P17 and the probe P18. In this patient, also a TEL/AML1 breakpoint-spanning DNA fragment on the der(21) was amplified with primers P5/P7 and P6/P8 and sequenced using primer P14. In the second patient, the der(12) genomic breakpoint was detected in the intron 5 of TEL and in intron 1 of AML1 with the primer combinations P9/P11 and P10/P12. Sequencing was performed with primer P15 (database accession number AJ494736). Real-time PCR was performed with P19/P20 and the probe P21. Primers P22/P23 and probe P24 were used to quantifiy the single copy gene APM-1 as a basis for calculating the number of TEL/AML1-positive cells in each sample. The breakpoint sequences, identified in the diagnostic and relapse samples, were clonotypic for each patient demonstrating unambiguously the common clonal origin of the leukemia.

The individual configurations of the immune receptor gene rearrangements enabled us to retrospectively confirm that the relapse clones were actually already present at diagnosis in both patients (Table 3). In the first patient, the relapse clone with its specific TCRGc and TCRGd gene rearrangements constituted only approximately 1 in 10⁴cells at the time of diagnosis. The relapse clone of the second patient, which was defined by unique IgHd, TCRDb, and TCRGb gene rearrangements, comprised 0.5% of the initial leukemic cell load. However, despite a high sensitivity (10⁻⁴) of the allele-specific PCR, we were unable to detect 2 of the relapse-specific rearrangements (IgHc and TCRGc) in this patient (Table 3). These rearrangements were also not found in samples that were obtained during the complete treatment phase (data not shown). We therefore presume that the relapse clone had evolved from a cell with ongoing recombinatorial potential that was related to the smaller subclone at diagnosis. We also used clone-specific PCR to find out whether the initial clone might have persisted (Table 3). However, in neither case were the respective clonotypic rearrangements detectable.

In both children the response of the different subclones to treatment was evaluated by quantitative PCR of the unique rearrangements from the initial and relapse leukemias, respectively (Figure 1). In patient 1, both leukemic subclones had an initial slow response after 2 weeks of treatment, although it was even slower in the minor subclone. One cell of the dominant clone was still detectable among 100 normal cells at day 15 as analyzed by the IgH and TCRD rearrangements, whereas the relapse clone, detected through its 2 TCRG rearrangements, persisted at similar low levels (10⁻⁴) as before treatment. These clones remained undetectable at all time points including 2 months after the end of chemotherapy. No further samples were obtained until relapse. The leukemia at relapse responded very promptly to chemotherapy as shown in Figure 1.

In patient 2, the dominant clone at initial diagnosis was not detected (< 10⁻⁴) anymore at day 15 through its clonotypic rearrangements IgHb, TCRDa, and TCRGa, whereas the relapse clone characterized by the IgHd, TCRDb, and TCRGb rearrangements had a slower response (10⁻³ at day 15, 10⁻⁴ at week 5) and persisted at low levels for 9 months. This slow response of the relapse clone to initial therapy contrasts with its rapid reduction by the relapse therapy, which resembles the fast response of the dominant clone to initial therapy. The common IgK-KDE rearrangement was traceable during follow-up at the same levels as the relapse rearrangements. None of the rearrangements was detectable in bone marrow samples obtained at 12 months and 18 months after initial diagnosis. At 2 months after the end of treatment (26 months after initial diagnosis) only one of the relapse-specific IgH rearrangements (IgHd) was detectable at a low level, whereas the clinical manifestation of the relapse occurred 15 months later.

To test whether quantification of the TEL/AML1 breakpoints would confirm the results obtained by immune receptor gene analysis, or detect a higher copy number of the TEL/AML1 fusion gene, we applied patient-specific TEL/AML1 real-time TaqMan PCRs using primers 16/17 and 19/20 and probes 18 and 21, for patient 1 and 2, respectively (Table1). We detected similar amounts of PCR products in all diagnostic and follow-up samples (indicated by an asterisk in Figure 1), thereby confirming the slower response and the persistence of the relapse clone, as well as the lack of amplification during later phases of the chemotherapy.

In 2 children with late systemic relapses of their TEL/AML1-positive leukemias we show that the initial and the relapse leukemias arise independently from a common TEL/AML1 precursor cell. At initial diagnosis, the dominant clone was sensitive to therapy, whereas the subclone, which later developed into the relapse leukemia, responded poorly. At the time of relapse, however, this clone was sensitive to treatment.

The differentiation stage of a transformed lymphoid cell can be characterized by its immune receptor rearrangements. These rearrangements occur independently from the oncogenic process. Thus, they can be used as markers for the timing of molecular events during leukemogenesis as well as for determining the time point of an early event during ontogeny (fetal versus postnatal development).30-34 We propose an order of molecular events during leukemogenesis, which is based on a current model by M. Greaves.35,36 Accordingly, at least 2 “hits” are required for the manifestation of leukemia. The TEL/AML1 translocation might be such a “first hit” in a precursor cell, either before the onset of somatic recombination, or, at a later stage, when immune receptor genes rearrange. Thus, continuing rearrangement processes lead to a variety of immune receptor rearrangements in individual cells of the TEL/AML1-positive preleukemic clone, resulting in a heterogeneous population. A second crucial molecular hit, probably a deletion of the nonrearranged TEL allele, drives one of these cells to overt leukemia. The following chemotherapy eradicates that leukemia, while the preleukemic clone may survive. If one of these preleukemic cells acquires a second mutation another leukemia develops, which harbors an identical TEL/AML1 gene fusion, but may carry a different TEL deletion and also different Ig/TCR gene rearrangements. In both patients the identical TEL/AML1 gene fusion was found at initial manifestation of the leukemia and at relapse. This gene fusion presumably transformed a cell with its immune receptor genes in germ line configuration or with an IgK rearrangement in patients 1 and 2, respectively, which may, on some occasions, precede IgH rearrangements.37 The difference in the TEL deletion status between initial presentation and relapse in patient 2 concords with the proposed model, whereas the clonal heterogeneity in the first patient makes an interpretation and valid conclusion extremely difficult. At initial presentation of the leukemia, the relapse clones ranged from 10⁻⁴ to 5 × 10⁻³ in the bone marrow of the 2 patients. Although this seems comparatively large, it is nevertheless only in the same order of magnitude as that of the pre/leukemic clones, which are retrospectively detected in peripheral blood obtained after birth from children with ALL.11,38 These data further suggest that a preleukemic clone maintains its size during its entire life span until it either undergoes apoptosis or is eradicated by chemotherapy. This assumption is further supported by the finding that the size of a pre/leukemic or covert leukemic clone carrying the TEL/AML1 gene fusion in neonates who do not develop leukemia is similar to the preleukemic subclone at diagnosis from the 2 patients in this report.36 

Patient 2 from our study had a long persistence of the minor pre/leukemic subclone. Likewise, the persistence of a preleukemic clone carrying the leukemia-specific fusion gene during morphologic remission has been reported in patients with acute myeloid leukemia with the t(8;21) chromosomal translocation. Thereby, the AML/ETO fusion gene resides in a preleukemic multipotent stem cell, which is resistant to treatment and develops to overt leukemia due to secondary genetic events.39-42 

A rapid and sustained disappearance of the dominant leukemia clones and a slow response of the smaller clones to first chemotherapy were observed in both children. According to current concepts, the slow response of the minor clones results from a primary drug resistance. In-line, slow-responding subclones were described previously in children with TEL/AML1-negative ALL.43 These patients experienced an early disease recurrence, whereas the 2 patients from this report had long first remission durations of 48 months and 40 months, respectively. Typically, TEL/AML1-positive leukemias have late relapses as exemplified by the cases from this report. These relapses respond very well to relapse chemotherapy with persistent or long-lasting second remissions.8 Such biologic differences between TEL/AML1-positive and -negative leukemias allow an alternative interpretation, that is, that these minor—presumably preleukemic—subclones have a poor response to therapy because they have a slower cell cycle than the fully leukemic clones.23-44 These initially slow-proliferating subclones may change their proliferation kinetics due to the acquisition of molecular events that lead to the manifestation of the relapse, and thus become well treatable. An intrinsic effect of the relapse protocol, however, cannot be excluded.

The observation from the 2 cases reported here should, however, not suggest the exclusive use of breakpoint-specific sequences for MRD analysis—even though Ig/TCR rearrangements may not be conserved at relapse—because the presence of a TEL/AML1 fusion gene does not always indicate a malignant disease.36-45 Currently, we cannot yet distinguish preleukemic clones from leukemias.

In conclusion, we have shown that a minor subclone of TEL/AML1-positive ALL responds slowly to chemotherapy, whereas the major clone has a rapid response. This biologic behavior of the smaller subclone during first chemotherapy can be interpreted as a feature of the preleukemic stage, which contrasts its very rapid response when it has evolved to the relapse leukemia.

This study was performed using the network of the Biology Group within the I-BFM Group. We would like to thank Uli Monschein and Marion Zavadil for help in preparing the manuscript.