Acute Lymphoblastic Leukemia in Children: Treatment Planning via Minimal Residual Disease Assessment

Acute lymphoblastic leukemia (ALL) is the commonest malignancy in children (aged 1–18 years), accounting for almost 30% of all cancers in this age group. In Germany, every year four in every 100 000 children aged up to 15 develop ALL (Box). Almost all patients in Germany are treated according to the valid treatment algorithms developed by the two collaborative study groups ALL-BFM (Berlin–Frankfurt–Münster) and COALL (cooperative study group for childhood acute lymphoblastic leukemia), which are subdivided into three treatment phases (induction and consolidation of remission, re-intensification, maintenance treatment) and cover two years. Notable improvements in the treatment of childhood ALL have been made in the past four decades (1–6); compared with the late 1960s, when 30% of cases were cured, nowadays some 80% of patients remain in their first remission even after 10 years. The remaining 20% of children have recurrences, and the cure rate falls to 25–40%. Conversely, it is also subject to discussion that a proportion of the cured patients are currently overtreated and would benefit from a reduction in treatment for the sake of lower toxicity and fewer long-term sequelae. In addition to determining patient-specific risk factors (such as sex, age) and ALL-specific risk factors (such as leukocyte count, immunophenotype, chromosomal aberrations) when making the diagnosis, in vivo confirmation of responsiveness to chemotherapy has been found to be of particular relevance in predicting the likelihood of survival and risk of recurrence (7–9).

Box

Acute lymphoblastic leukemia (ALL) in children (basic facts)

Enlarge All figures

The standard method to gauge the therapeutic response is cytomorphology by means of microscopy. Its disadvantage is its low sensitivity, which in the best-case scenario allows detection of one leukemia cell among 100 normal cells. According to this diagnostic criterion, 95–98% of patients achieve complete remission within four weeks of starting therapy. This means that in substantial proportions, intensive polychemotherapy was administered after the leukemia had become undetectable. Two decades ago, a diagnostic breakthrough was reached when polymerase chain reaction (PCR) was introduced to identify minute amounts of malignant cells (minimal residual disease, MRD). This review article will highlight the great importance of MRD analysis for tailoring individual therapeutic approaches in children with ALL. The authors' own results were supplemented with a selective Medline search using the search terms “PCR-MRD”, “childhood ALL”, and with the therapeutic studies BFM (Berlin– Frankfurt–Münster) und AIEOP (Assoziazone Italiana Ematologia Oncologia Pediatrica).

Methods for diagnosing MRD

Three methods for identifying minute amounts of malignant cells (minimal residual disease, MRD) allow the identification of one leukemia cell among 10 000 to 1 million normal cells (10^–4 to 10^–6) and are thus 100–10 000 times more sensitive than conventional cytomorphological techniques. These methods differ with regard to their applicability (frequency of relevant target structures) in terms of their sensitivity and their expense in terms of time and money (10):

• Immunophenotyping using flow cytometry and multiple markers to identify leukemia-specific immunophenotypes,
• PCR analysis of fusion products of leukemia-specific chromosomal translocations (for example, BCR-ABL recombination in the context of the Philadelphia translocation-t [9; 22]), and
• PCR analysis of clone-specific chain joining regions of immunoglobulin (Ig) and T-cell receptor (TCR) rearrangements.

In the meantime the third method in this list has become most widespread in Europe in the context of European ALL studies because of its excellent sensitivity. This method is based on the insight that ALL always starts from individual lymphocyte precursors, and all leukemias are therefore characterized by an individual, clone-specific pattern on Ig rearrangements or TCR rearrangements.

The molecular basis is a DNA rearrangement in which different segments of the Ig genes or TCR genes are joined individually (11). In the context of these recombination processes, fewer or more nucleotides from the original DNA sequence are lost at the joining regions, and free nucleotides can be inserted into this region. The result is a unique DNA sequence for a particular lymphocyte, which can be identified on the basis of PCR analysis by using allele-specific DNA probes—a method that the authors and other working groups have been developing since the late 1980s in order to measure MDR in patients with ALL (12, 13).

Since acute lymphoblastic leukemia can display subpopulations of an initial leukemia cell clone (oligoclonality) at the time of diagnosis and may undergo further Ig- and TCR-rearrangements during the disease course, which may then result in the loss of the DNA sequences used as probes, it is recommended for the PCR analysis to use two or more target sequences in order to avoid false-negative results. In the context of the BFM-ALL study, 51 PCR primer combinations are currently used for every patient, so as to be able to identify the IgH-, IgK-, TCRβ-, TCRγ- und TCRδ-recombinations that are characteristic of the respective leukemia. These joining regions are subsequently sequenced, and allele-specific oligonucleotide primers are generated on this basis, which are then examined for their sensitivity and specificity by using real-time quantitative PCR (RQ-PCR). This is mostly done by using serial dilutions, in which DNA amplificates of the leukemia cells are placed in control DNA of normal cells in dilutions of 10^–1 to 10^–5 and the limits are determined at which leukemia cells can be detected. Finally, RQ-PCR is used to determine the remission status at patients’ follow-ups. Figure 1 shows the time sequence from initial diagnosis to MRD-based risk stratification three months after the start of treatment.

Figure 1

Individual steps of MRD analysis in sequence (in weeks), from diagnosis to stratification into MRD-based risk groups

Enlarge All figures

The long road towards clinical application

The application of MRD analysis in the treatment of childhood ALL that is under discussion in this article provides an example of the careful, evidence-based introduction of new methods and marker systems of molecular medicine into clinical routine. The BFM Study Group recognized the clinical potential of this technique early on, which was developed further over the years, and in the early 1990s included a relevant clinical research project in the ALL-BFM 90 study protocol. From the very beginning, the emphasis was put on international agreement in the setting of the International BFM Study Group. Since the kinetics of residual leukemia cells under MRD criteria was not known, and neither was their clinical relevance, researchers agreed after extensive discussions (of the ethical implications in particular) that, after providing adequate information to and obtaining written consent from the participants, bone marrow specimens would be harvested (under general anesthesia) at nine time points during and after the end of the treatment, so as to determine the MRD status and correlate this with the clinical data. The results were evaluated retrospectively and during the project phase made available neither to the treating physicians nor to the patients, in order to prevent decisions that were not based on evidence.

The findings obtained from 240 patients were available in 1998 (14) and were confirmed simultaneously by other groups (15). MRD analysis was found to be superior to all conventional prognostic parameters. Puncture timings on days 33 and 78 after the start of treatment were of particular prognostic relevance; they provided essential indications of a positive clinical course or high risk of recurrence and made it feasible to restrict follow-up studies to these two control points. According to MRD criteria, three risk groups became obvious: patients at very low risk of recurrence of only 2% (43% of children included in the study), those with a very high risk of recurrence of 75% (15% of cases), and an intermediate group with a recurrence rate of 23% (43% of patients). These findings were still valid after 10 years of follow-up (16). Such a clear distinction of the different risk profiles had previously not been available. Also of note: Compared with conventional prognostic criteria, relevant shifts were observed in the individual risk profiles: Patients who had previously been allocated to the standard risk group were at intermediate or high risk of recurrence when MRD criteria were used, and vice versa.

From pilot project to general MRD-based therapeutic stratification

On the basis of these findings, a prospective therapeutic study of the Associazione Italiana di Ematologia Oncologia Pediatrica and the BFM-ALL Group (AIEOP-BFM ALL 2000) was set up that introduced for the first time the MRD-based risk stratification with only two remission samples, and which included some 4700 patients between 2000 and 2006. From the start, the internationally accepted standardization of this analytic technique was driven forward in this context (10, 17, 18), which was quality assured by means of regular round-robin tests. The evaluation of this globally largest study of its kind proved the logistical feasibility of this complex analytic technique in an international multicenter therapeutic study (16). MRD stratification using two or more sensitive markers that enabled detection of a leukemia cell among at least 10 000 normal cells (≤10^–4) was successful in 80% of patients. At least one sensitive marker was identified in 94% of B-cell leukemias and 82% of patients with T-ALL. Patients without an adequate MRD constellation were stratified using conventional parameters. Check points were Days 33 and 78 after the treatment was started; the MRD findings were available from the laboratory 7 to 14 days later, so that stratification into the three risk groups was complete 13 to 14 weeks after starting treatment.

Patients were stratified into the three MRD-based risk groups (SR, IR, HR—standard risk, intermediate risk, high risk) according to the following criteria:

• Patients whose results were MRD-negative at both control points (Day 33 and Day 78) for two independent MRD markers with a sensitivity of ≤10^–4 were classified as patients at standard risk (MRD-SR).
• Patients at intermediate risk (MRD-IR) had positive results on one or both control points, with the second MRD assessment value at <10^–3; Figure 2 shows an example of this.
• Patients with an MRD measurement of ≥10^–3 on Day 78 were classified as at high risk (MRD-HR).

Figure 2

RQ-PCR analysis of an ALL patient with an allele-specific IgH oligomere whose limits of detection were determined in a dilution series of 10^–1 to 10^–5 with one leukemia cell per 100 000 normal cells (10^–5)

Enlarge All figures

In total, 3184 patients with precursor-B-ALL were stratified by using MRD (19). Of these, 42% (1348) were MRD-SR patients with 7 years’ event-free survival (EFS, standard error) of 91.1% (1.2%), 52% (1647) were MRD-IR patients with an EFS of 76% (1.4%), and 6% (189) were MRD-HR patients with an EFS of 46.6% (5.1%) (p<0.001). The cumulative incidence rates of recurrences were similarly diverse (Figure 3).

Figure 3

Event-free survival (EFS) and cumulative incidences (CI) of recurrence after MRD-based stratification of 3184 patients with precursor-B-ALL

Enlarge All figures

Direct comparison with the conventional criteria (Table) shows the advantages of MRD-based risk stratification, such as was applied in the ALL-BFM-95 Study (20) (age, leukocyte count, cytogenetics, response to prednisone) (Figure 4). If these patients had been stratified according to conventional criteria, half of the patients at standard risk (SR 95) would have responded with a delay (medium and dark sections of the SR 95 bar) and thus been at higher risk of recurrence, whereas conversely, patients who had been stratified as at high risk according to conventional criteria (HR 95) responded better to treatment than had been assumed (dark and light sections on the HR 95 bar) and had a better prognosis. All three “classic” risk groups include patients at a particularly high risk of recurrence (MRD-HR). It is obvious that assessing MRD can describe the risk stratification in a more differentiated and precise way by using patient-specific and leukemia-specific risk factors. However, the clinical benefits will fully materialize only when more effective therapeutic elements are used for patients who are characterized by a slower or poor molecular therapeutic response, which are yet to be developed. Another aspect to this is the fact that the expectation that the excellent therapeutic result in the group at standard risk, which is characterized by a rapid response to treatment, can be realized by less aggressive treatment later on, has so far not been confirmed by preliminary results (M Schrappe, personal communication). In the AIEOP-BFM ALL 2000 Study, 464 T-ALL patients were stratified according to the MRD criteria explained above (21): 16% (75) of the patients were classified as MRD-SR, with a 7-year EFS of 91.1%; 63% (292) of patients were classified as MRD-IR, with a 7-year EFS of 80.6%; and 21% (97) of patients were classified as MRD-HR, with a 7-year EFS of 49.8% (p<0.001) (Figure 5).

Figure 4

Redistribution of all patients with precursor-B-ALL in the MRD-based study AIEOP-BFM-ALL 2000, after applying the “classic” criteria—for example in the ALL-BFM-95 Study

Enlarge All figures

Figure 5

Event-free survival (EFS) and cumulative incidence in recurrences after MRD-stratification of 464 patients with T-ALL (the results were originally published in Blood/American Society of Hematology)

Enlarge All figures

Table

Definition of risk groups

Enlarge All figures

It was found that both forms of leukemia had an equally good result (EFS) in the MRD-SR group, independently of the immunophenotype (B-cell or T-cell leukemia). This observation also applied to the other two risk groups. However, it should be noted that the MRD-SR group included a lower proportion of patients with T-ALL than precursor-B-ALL patients, and the distribution in the MRD-HR group is exactly the opposite. In patients with T-ALL the elimination of leukemia cells is therefore on average noticeably delayed compared with precursor-B-ALL. If patients with T-ALL present as MRD negative on Day 33, however, the recurrence rate of 8% makes for an excellent prognosis (Figure 5) (21). Furthermore, 32% of all T-ALL patients in the MRD-IR group who were MRD negative on Day 78 also had an excellent result; the seeming conclusion is that MRD positivity on Day 33 is of clinically lesser importance for T-ALL than for precursor-B-ALL. A large number of residual leukemia cells (≥10^–3) on Day 78 is the most important predictor for recurrence in T-ALL patients.

Conclusions and outlook

The AIEOP-BFM ALL 2000 Study has set international standards for risk stratification in children with ALL. The MRD-based prognostic estimate has proved superior to all other parameters. This assessment has been confirmed in numerous studies, independently of which method was used. This holds for children (22, 23) as well as for adults (24, 25) which is notable since the biological basis of the pathogenesis of leukemia—including its molecular genetic pathogenesis—is very different in both age groups. However, it underlines the importance—independently of other factors—of this new therapeutic option.

MRD-supported therapy also has an important role for other hematopoietic neoplasias—for example, in chronic lymphocytic leukemia (26) and acute myeloid leukemia (27). Chronic myeloid leukemia provides a particularly instructive example. After the specific cytogenetic aberration, the Philadelphia chromosome, had been discovered in 1960 and the molecular genetic equivalent, BCR-ABL-recombination, had been figured out in the 1980s, this molecular target structure has been used to make the diagnosis and to guide MRD-based treatment. Furthermore, innovative therapeutic approaches that targeted this particular molecular defect in leukemia cells were developed in the shape of the tyrokinase inhibitors, and the treatment of CLL has been notably improved (28). MRD-status in the blood is measured for solid tumors too. But in this setting, the analytic approach is used for a different purpose: It is used to detect tiny amounts of disseminated tumor cells with the potential for metastasis (29).

MRD analysis in children with ALL was included in the diagnostic repertoire of the BFM-ALL Study that is covered by German statutory health insurers; previously this analytic technique was funded for 15 years by project grants from the German Cancer Aid (Deutsche Krebshilfe). Although it is hardly likely that significant improvement rates can be achieved for the large group of patients with an MRD-SR profile, whose cure rates are higher than 90%, the recurrence rate in the MRD-HR group remains unacceptably high—even though this group comprises only 6% of patients. The intermediate risk group is also in need of improved therapeutic options. Studies are needed to ascertain whether innovative therapeutic approaches that target the molecular defects in leukemia cells (“targeted therapy”) yield similar successes as have been reported for tyrokinase inhibitors in BCR-ABL-positive ALL. It should be borne in mind, however, that risk stratification of patients according to MRD criteria is inherently possible only several months after starting therapy. Whether current research projects of the genome-wide sequence analysis of children with ALL will identify new molecular parameters that would enable improved risk stratification at the time of diagnosis, remains to be seen. Any markers would have to be investigated for their validity in large prospective studies with adequate follow-up times. In the medium term, MRD-based risk stratification of patients will therefore remain the gold standard for all patients with ALL.

Acknowledgement

The initial development of the method for PCR-based Ig- and TCR-analysis was funded by the DFG, German Research Foundation (Bartram). Profound thanks are due to the German Cancer Aid for long years of comprehensive funding (Bartram) in introducing the MRD analysis. Furthermore, the German Cancer Aid has comprehensively funded the respective ALL-BFM Studies (Schrappe). The following scientists were crucially involved in the development and evaluation of the MRD diagnostic technique in Bartram’s working group (until 1995 division of molecular biology, department of pediatrics II, University Medical Center Ulm, subsequently institute of human genetics, University of Heidelberg): T E Hansen-Hagge, S Yokota, T Seriu, M Nakao, and T Flohr. We thank our patients and their families for their participation in the studies and all the doctors, nurses, and medical technicians who were involved. We also thank the study commission and members of our MRD laboratories for conceptual help and for conducting the study. Special thanks go to our international partners of more than two decades; names we wish to mention are J J M van Dongen (Netherlands), A Biondi and G Masera (Italy), R Panzer-Grümayer and H Gadner (Austria). We thank the initiator of the ALL-BFM studies, H Riehm, for his conceptual far-sightedness and emphatic support in introducing MRD analysis as a clinical diagnostic technique.

Conflict of interest statement
Professor Schrappe has received, and continues to receive, third-party funding from the German Cancer Aid. He has received conference delegate fees and fees from continuing medical educational events, as well as travel and hotel expenses from Medac and Eusapharma. He has received lecture honoraria and ringfenced funds into a third-part account from Medac. Professor Bartram has received third-party funding for projects from the German Research Foundation and the German Cancer Aid. Dr Schrauder and Dr Köhler declare that no conflict of interest exists.

Manuscript received on 12 January 2012, revised version accepted on
31 May 2012.

Translated from the original German by Dr Birte Twisselmann.

Corresponding author
Prof. Dr. med. Claus R. Bartram
Institut für Humangenetik, Universität Heidelberg
Im Neuenheimer Feld 366, 69120 Heidelberg
cr.bartram@med.uni-heidelberg.de