Comparative analysis of Ig and TCR gene rearrangements at diagnosis and at relapse of childhood precursor-B–ALL provides improved strategies for selection of stable PCR targets for monitoring of minimal residual disease

Several large prospective studies have clearly demonstrated the high prognostic value of minimal residual disease (MRD) monitoring in childhood acute lymphoblastic leukemia (ALL).1-4 Based on the sensitive measurement of early response to cytotoxic treatment, it is currently possible to identify not only patients at high risk for relapse but also a group of low-risk patients with an excellent relapse-free survival of more than 95%.4 Hence, MRD information provides a new definition of remission in childhood ALL, which justifies incorporation of MRD data in current treatment protocols to refine risk assignment.5 

Most MRD studies in pediatric precursor-B–ALL applied immunoglobulin (Ig) and/or T-cell receptor (TCR) gene rearrangements as polymerase chain reaction (PCR) targets for MRD detection. They can easily be identified in most patients at diagnosis with limited sets of PCR primers.6,7 Moreover, using these molecular targets, sensitivities of 10⁻⁴ to 10⁻⁶ (1 malignant cell within 10⁴ to 10⁶ normal cells) can be obtained routinely.7 With the advent of novel real-time quantitative (RQ) PCR techniques, Ig/TCR-based MRD techniques are now also quantitative.8-11 However, it is also known that Ig and TCR gene rearrangements might change during the disease course, owing to secondary rearrangement processes mediated via ongoing activity of the V(D)J recombinase enzyme system (reviewed by Szczepański et al12). This might lead to loss of the PCR target and consequently to false-negative MRD results. Such changes were most frequently described for Ig heavy chain genes (IGH) and to lesser extent for TCR genes and were found particularly in cases of precursor-B–ALL that already contained subclones at diagnosis.13-18 

Although the presence of clonal evolution phenomena is widely acknowledged, its actual impact on the effectiveness of MRD monitoring has not been defined. So far, studies assessing the stability of Ig and TCR gene rearrangements at diagnosis and relapse of ALL either did not compare junctional region sequences or were limited to particular gene loci.13-18 Therefore, we studied the stability of the currently used Ig/TCR rearrangements (IGH, Igκ light chain[IGK], TCRγ [TCRG], and TCRδ[TCRD] gene rearrangements) in a large series of 96 childhood precursor-B–ALL patients. This information is essential for reliable selection of MRD-PCR targets with minimal chance of false-negative results.

Bone marrow or peripheral blood samples from 96 childhood precursor-B–ALL patients were obtained at initial diagnosis and at relapse (91 patients) or at presumably secondary acute myeloid leukemia (AML) (5 patients). The age distribution ranged from 1 month to 183 months. Eight children (8%) were infants (age < 1 year). The diagnosis of precursor-B–ALL was made according to French-American-British and standard immunophenotypic criteria.19-21 Immunologic marker analysis of the precursor-B–ALL revealed that 6 (6%) were pro-B–ALL, 59 (61%) were common ALL, and 31 (32%) were pre-B–ALL. Seven patients were studied at 2 subsequent leukemia relapses. Cell samples of 52 patients were obtained from the cell bank of the Dutch Childhood Leukemia Study Group.

Comparative immunophenotypic analysis revealed intralineage switches in 21% (18 of 86) of precursor-B–ALL patients with available detailed immunophenotypic data at relapse, which is slightly higher than reported previously.22 

The rationale, methodology, and pitfalls of the stepwise molecular comparison of the Ig/TCR gene rearrangements between diagnosis and relapse of precursor-B–ALL were previously exemplified in the case report of patient 5498, also included in these series.23 A small subgroup of patients (n = 21) was studied before by comparative Southern blotting and PCR analysis of κ-deleting element (Kde) rearrangements at diagnosis and relapse.24 

Mononuclear cells were separated from bone marrow or peripheral blood samples by Ficoll-Paque centrifugation (density 1.077 g/cm³; Pharmacia, Uppsala, Sweden). DNA was isolated from frozen mononuclear cells, digested with the BglII enzyme, and blotted to nylon membranes as described previously.25,IGH and IGK gene configurations were analyzed with the IGHJ6, IGKJ5, IGKC, and IGKDE probes (DAKO, Carpinteria, CA).26,27 The configuration of the TCRD genes was analyzed with the TCRDJ1 probe (DAKO).28 The diagnosis and relapse samples of the 96 patients were always run in parallel lanes (Figure 1), except for 6 patients who were exclusively analyzed at diagnosis because of insufficient amounts of DNA from relapse samples. The Southern blot configuration of the Ig and/or TCR genes in 73 patients at diagnosis and in 30 patients at relapse has been reported previously.6,15 

Independent of Southern blotting, PCR analysis could be performed on both diagnosis and relapse samples in 89 patients, essentially as described previously.7,29 Four patients with a secondary AML and 3 precursor-B–ALL patients with insufficient remaining DNA at diagnosis were not analyzed. In each 50 μL PCR reaction, 50 ng DNA sample, 6.3 pM of the 5′ and 3′ oligonucleotide primers, and 0.5 U AmpliTaq Gold polymerase (PE Biosystems, Foster City, CA) were used. The sequences of the 25 oligonucleotide primer pairs used for amplification of IGH (5VH family–specific framework 1 primers and 7 DH family–specific primers with consensus JH primer),IGK-Kde (4 Vκ family–specific primers and intron recombination signal sequence [RSS] primer with Kde primer), TCRG (6 Vγ-Jγ primer combinations most frequently used in precursor-B–ALL), and TCRD gene rearrangements (Vδ2-Dδ3 and Dδ2-Dδ3 primer pairs) were previously published.7,30,31 PCR conditions were as follows: initial denaturation for 10 minutes at 94°C, followed by 40 cycles of 45 seconds at 92°C, 90 seconds at 60°C, and 2 minutes at 72°C using a Perkin-Elmer 480 thermal cycler (PE Biosystems). After the last cycle an additional extension step of 10 minutes at 72°C was performed. Appropriate positive and negative controls were included in all experiments.7 

For heteroduplex analysis, the PCR products were denatured at 94°C for 5 minutes after the final cycle of amplification and subsequently cooled to 4°C for 60 minutes to induce duplex formation.32 Afterward the duplexes were immediately loaded on 6% nondenaturing polyacrylamide gels in 0.5 × Tris-borate-EDTA buffer, run at room temperature, and visualized by ethidium bromide staining.32 

Relapse samples were at first analyzed with those primer combinations, which showed clonal PCR products at diagnosis. When the clonal PCR product was also found at relapse, its identity was subsequently compared with the PCR product found at diagnosis by means of mixed heteroduplex analysis, ie mixing of the diagnosis and relapse PCR products followed by heteroduplex analysis (Figure2).23 When clonal PCR products found at diagnosis were undetectable at relapse, the relapse sample was then analyzed with additional primer combinations for the involved gene loci.

Clonal PCR products discordant between diagnosis and relapse of precursor-B–ALL as found by mixed heteroduplex analysis were directly sequenced. Sequencing was performed using the dye-terminator cycle sequencing kit with AmpliTaq DNA polymerase FS on an ABI 377 sequencer (PE Biosystems) as previously described.31 VH, DH, and JH segments were identified using DNAPLOT software (W. Müller, H.-H. Althaus, University of Cologne, Germany) by searching for homology with all known human germline VH, DH, and JH sequences obtained from the VBASE directory of human Ig genes (http://www.mrc-cpe.cam.ac.uk/imt-doc/).33Vγ and Jγ gene segments were identified by comparison to germlineTCRG sequences as previously described.34 

Statistical analysis using the χ² test on a 2 × 2 table was performed to compare the frequencies of particular Ig/TCR gene rearrangements between different precursor-B–ALL patient subgroups. Pearson correlation coefficient was calculated to test an association between variables. P values less than or equal to .05 were considered statistically significant.

The configuration of IGH, IGK, andTCRD genes was established with multiple Southern blot probes. This concerned all 96 patients at diagnosis of precursor-B–ALL and 91 patients at subsequent relapse or secondary AML (Figure 1). This gave us the unique opportunity to address the question of whether there are any differences in Ig/TCR gene configuration between the patients who relapsed compared with the total childhood precursor-B–ALL group. This comparison is summarized in Table 1, which shows that the Ig/TCR gene rearrangement patterns at diagnosis and at relapse in patients included in this study are largely comparable to each other and to previously published data derived from large series of random childhood precursor-B–ALL cases at initial diagnosis.6,27 Only 2 immunogenotypic features were more prevalent in ALL at relapse. The TCRD gene configuration at relapse was characterized by significantly less frequent Vδ2-Dδ3 and Dδ2-Dδ3 rearrangements and more frequent TCRDdeletions (P < .05), which reflects ongoing deletional rearrangements. Secondly, the frequency of IGH andTCRD oligoclonality at relapse was slightly less frequent, but this difference was only significant for TCRD(P < .05).

The configuration of Ig/TCR genes compared between 2 subsequent relapses in the 7 patients analyzed showed evidence for clonal evolution in only 2 cases (concerning 1 or 2 gene rearrangements), while in 5 of the 7 cases we observed some changes in gene rearrangement patterns between diagnosis and first relapse.

Southern blot analysis in 4 of 5 patients with a presumed secondary AML demonstrated the complete absence of clonal Ig/TCR gene rearrangements (Figure 3), which is in line with the AML diagnosis. However, in one patient (3991) 3 clonal Southern blot bands were preserved in the AML clone, and the sequence identity was confirmed by comparative PCR analysis for 2VH-JH gene rearrangements. Apparently the original ALL clone underwent a phenotypic switch to AML. This patient was included in our further comparative diagnosis-relapse analyses despite the phenotypic shift, while the other 4 patients were excluded.

A total of 362 clonal PCR products of different Ig/TCR gene rearrangements were identified at diagnosis in 87 (98%) of 89 patients, with an average of 4 targets per patient. In one patient no clonal gene rearrangements were detected by PCR, while Southern blotting showed a single weak rearranged IGH band, identical between the diagnosis and relapse sample. The second patient had an infant ALL and was fully oligoclonal at diagnosis, which precluded identification of clonal Ig/TCR markers for PCR-based MRD monitoring. Generally, IGH oligoclonality at diagnosis was more prevalent in infant ALL patients (6 of 8 cases), compared with the noninfant precursor-B–ALL cases (32 of 84 cases),P < .05.

A total of 256 (71%) of 362 clonal Ig/TCR gene rearrangements identified with heteroduplex PCR analysis at diagnosis in 87 patients were preserved at relapse. This concerned 99 (64%) of 155IGH, 54 (90%) of 60 IGK-Kde, 65 (75%) of 87TCRG, and 38 (63%) of 60 TCRD gene rearrangements (Table 2). In 3 additional patients, Southern blot analysis provided conclusive information about stability of gene rearrangement patterns.6,15 

In 36 patients (including patient 4616 studied exclusively by Southern blotting) all MRD-PCR targets identified at diagnosis were preserved at relapse (Figure 4A). In 38 cases (including patients 2665 and 4501 studied exclusively by Southern blotting) at least half of the targets remained stable during the disease course (Figure 4B). In another 10 patients (including patient 3991 with AML at relapse; Figure 3) most MRD-PCR targets were absent at relapse but at least one rearrangement was common for both diagnosis and relapse samples (Figure 4C). Consequently, at least one MRD-PCR target was preserved at relapse in 84 (93%) of 90 patients with available clonal MRD-PCR targets at diagnosis. In the remaining 6 patients all clonal markers found at diagnosis seemed to be lost. However, in 3 of these 6 cases the clonal relationship between diagnosis and relapse was confirmed by the identification of a common DH-JH stem shared by respective VH-JH gene rearrangements. In another 2 cases additional analyses35,36 showed at both disease stages identical DNA sequences of Vκ-Jκ (5771) and chimericMLL-AF4 fusion genes (5978), respectively (data not shown). Finally, in one patient (5282) there was no evidence for a common origin of diagnosis and relapse clones (Figures 1 and 4C). Because all original monoclonal rearrangements were lost in this case at relapse, we believe that this is in fact a secondary ALL.

Stability of MRD-PCR targets did not significantly correlate with age or white blood cell count at diagnosis or with remission duration, ie, time span between diagnosis and relapse.

Clonal IGH gene rearrangements were detected by PCR in 75 of the 90 studied childhood precursor-B–ALL patients. In 66 cases (88%) at least one IGH MRD-PCR target was preserved at relapse. In 12 additional patients PCR analysis did not result in identification of clonal IGH rearrangements, while Southern blot data suggested the preservation of at least one target in all 12 patients (fully identical IGH configuration in 9 cases). In 3 patients deletions of JH region were identified by Southern blotting both at diagnosis and at relapse of ALL. The Southern blot results of these 15 patients were not used for calculating the stability of theIGH PCR targets.

The stability of the IGH PCR targets was markedly different between oligoclonal and monoclonal patients, ie, at least one MRD-PCR target was preserved in 76% and 98% of patients, respectively (Table2). This significant difference was even more pronounced at the MRD-PCR target level, with 63 (85%) of 74 monoclonal IGH gene rearrangements being stable compared with only 36 (44%) of 85 oligoclonal rearrangements. Taking into account the type ofIGH gene rearrangement, complete VH-JHrecombinations were characterized by a higher stability compared with incomplete DH-JH rearrangements, with 69% versus 43% of targets preserved, respectively.

A total of 60 IGK deletional rearrangements were detected in 40 childhood precursor-B–ALL patients at diagnosis: 44 Vκ-Kde, 15 intron-Kde, and 1 rarely occurring Vκ-intron RSS recombination. At least one of the rearrangements was preserved in 37 cases (93%). In fact, IGK-Kde recombinations represented the most stable MRD-PCR targets, with 90% of all targets preserved. Most (55 of 60) Kde rearrangements were monoclonal and highly stable (52 targets preserved; 95%), while only 2 of 5 oligoclonalIGK-Kde targets were found at relapse. No significant difference in stability was found between Vκ-Kde and intron-Kde rearrangements.

A total of 87 TCRG gene rearrangements were detected in 53 precursor-B–ALL patients at diagnosis, and in 44 cases (83%) at least one MRD-PCR target was preserved at relapse. Because accurate oligoclonality detection in TCRG locus is rather complex, even by Southern blotting,37 we did not evaluate whether there were any differences in MRD-PCR target stability between monoclonal and oligoclonal patients.

A total of 60 clonal TCRD gene rearrangements (43 Vδ2-Dδ3, 16 Dδ2-Dδ3, and 1 Vδ3-Jδ1) were identified by PCR in 39 precursor-B–ALL patients. At least one of the clonal rearrangements was preserved in 27 (69%) of 39 cases. Once again we observed striking differences in stability between monoclonal and oligoclonal patients, ie, at least one target was preserved in 88% and 36% of patients, respectively (Table 2). Only 26% of oligoclonal targets were preserved compared with 86% of monoclonal targets. Moreover, in 10 patients Southern blot data suggested the presence of a clonally rearranged band corresponding to Vδ2-Dδ3 rearrangements, while heteroduplex PCR analysis of these rearrangements showed oligoclonality or even polyclonality.6 

Based on combined Southern blot, PCR, and sequence analysis, it was possible to follow the patterns of clonal evolution leading to disappearance of rearrangements that were originally present at diagnosis.

Owing to clonal evolution phenomena, 62 IGH targets in 36 patients were lost. We could determine the exact patterns of clonal evolution in 8 patients with monoclonal IGH gene rearrangements and in 26 patients with an oligoclonal rearrangement pattern.

In 7 patients (2 monoclonal and 5 oligoclonal) Southern blot rearrangement patterns between diagnosis and relapse were identical, while PCR analyses showed disappearance of a single rearrangement. This might be explained by the disappearance of minor subclones undetectable by Southern blotting. In 3 patients with monoclonal IGHconfiguration, one of the rearrangements was changed both in Southern blotting and PCR, while sequence comparison showed VHreplacement with a preserved VH-N-DH-N-JH junction. In another 2 monoclonal patients we observed clonal “regression” of one of the rearrangements to germline. Finally, in a single patient (5282) both monoclonal IGH rearrangements were replaced by 2 new unrelated rearrangements; this patient was suspected of having developed a secondary ALL (Figure 4C).

In 11 oligoclonal patients, the IGH configuration at relapse was monoclonal, which is suggestive of clonal selection during the treatment. In another 10 oligoclonal patients, IGH was also oligoclonal at relapse, with several rearrangements lost and with emergence of new (sub)clones. Sequence comparison was fully completed in 12 oligoclonal patients with changes at relapse, indicating ongoingVH to DH-JH joinings with preservation of common DH-JH stems in 5 patients and possibly secondary rearrangements in 5 other cases. In the remaining 2 oligoclonal patients, the IGH sequences were unrelated and suggestive of the development of diagnosis and relapse clones from a common clonal progenitor via independent secondary rearrangements.

Clonal evolution of TCRG gene rearrangements was observed in 19 patients, leading to loss of 22 MRD-PCR targets. In 9 patients this concerned “regression” of clonal rearrangements most probably to germline configuration. In 5 patients the new rearrangements at relapse contained upstream Vγ and downstream Jγ segments as compared with the Vγ-Jγ rearrangements at diagnosis, which is suggestive of ongoing recombination with Vγ-Jγ replacement. In the remaining 5 patients the sequence comparison of Vγ-Jγ rearrangements at diagnosis and at relapse excluded secondary rearrangements and indicated the emergence of a clone related to the initial (pre)leukemic clone but different from the predominant clone at diagnosis.

Clonal evolution in the TCRD locus resulted in the loss of 22 MRD-PCR targets in 17 patients (5 monoclonal and 12 oligoclonal cases). In 6 patients (including 4 patients with monoclonal Vδ2-Dδ3 rearrangements) ongoing deletions were observed. In contrast, in 8 patients (including 1 patient with monoclonal Vδ2-Dδ3) the rearrangement pattern “regressed” to germline configuration. Finally, in the remaining 3 cases new rearrangements were found at relapse, including 1 Vδ2-Dδ3 joining and 2 unidentified rearrangements to the Dδ3-Jδ1 region. Interestingly, 5 of 10 cases with oligoclonal/polyclonal Vδ2-Dδ3 rearrangements at diagnosis had a monoclonal TCRD configuration at relapse with a single Vδ2-Dδ3 joining. In the remaining 5 patients, ongoingTCRD deletion was assumed in 2 cases, 2 patients demonstrated “regression” to germline configuration, and only a single patient preserved the oligo/polyclonal Vδ2-Dδ3 rearrangement pattern at relapse.

Our comparative Southern blot, PCR, and sequencing analyses of childhood precursor-B–ALL at diagnosis and relapse have provided detailed insight in the stability and changes of Ig and TCR gene rearrangements during the disease course. This information is essential for reliable application of Ig/TCR gene rearrangements as MRD-PCR targets in childhood ALL. However, one should be cautious with extrapolating these data to adolescent or adult precursor-B–ALL patients, because the immunogenotype of adult precursor-B–ALL has more immature features.29 

The Ig/TCR gene rearrangement patterns at diagnosis in relapsed patients appeared to be comparable to those in a random series of newly diagnosed pediatric precursor-B–ALL, implying that the various Ig/TCR gene characteristics at diagnosis have no prognostic value. This is in contrast to the previously reported strong predictive value of the Vδ2-Dδ3 gene rearrangement or the overall clonal diversity.38,39 The overall Ig/TCR gene configuration patterns at relapse were largely comparable to those at diagnosis but were characterized by less oligoclonality and more frequentTCRD gene deletions. These 2 differences fit with the concept of ongoing clonal selection and continuing rearrangements.

The detailed molecular analyses proved the clonal relationship between diagnosis and relapse in 88 of 89 childhood precursor-B–ALL with identified MRD-PCR markers at diagnosis. In only one patient (5282) were the Ig/TCR gene rearrangement patterns at diagnosis and relapse completely different with unrelated junctional region sequences (Figures 1 and 4C). Consequently, the presumed ALL relapse in this child (3.5 years after diagnosis) might in fact represent a secondary leukemia. This single patient confirms previous observations that ALL rarely occurs as second malignancy after previous cytotoxic treatment.40 This is in contrast to secondary AML, which affects approximately 4% of children treated for ALL with cytotoxic regimens containing topoisomerase II inhibitors.41 We indeed proved the presence of secondary AML with germline Ig/TCR genes in 4 (4%) of the 96 patients (Figure 3). However, in a fifth patient we demonstrated the clonal relationship between the precursor-B–ALL at diagnosis and the presumed secondary AML, which in fact represented a phenotypic switch.

The comparison of Ig/TCR gene rearrangement patterns between diagnosis and relapse showed marked heterogeneity in the occurrence of clonal evolution phenomena. In 40% of patients all PCR-identified clonal Ig/TCR rearrangements were present at relapse (Figure 4A), and in another 42% of cases at least half of the identified gene rearrangements remained stable at relapse (Figure 4B). Extreme clonal evolution with differential outgrowth of subclones characterized the remaining 18% of cases (Figure 4C), in whom most or even all clonal Ig/TCR gene rearrangements identified at diagnosis were lost during disease course. Interestingly, in contrast to the frequent occurrence of clonal evolution between diagnosis and relapse, we did not observe major clonal instability of Ig/TCR genes between 2 consecutive relapses (7 cases), which is in line with previous observations.14 

Previous studies suggested that the risk of changes in Ig/TCR rearrangement patterns increases with time.15,42 We did not find a significant correlation between remission duration and target stability in this extensive study. This is in line with the report that clonal selection processes can already occur in early treatment phases and the reports on clonal identity between diagnosis and very late relapse of precursor-B–ALL.43-45 

In this extensive molecular study, we wished to identify the factors associated with the occurrence of clonal evolution and therefore the increased risk of false-negative MRD-PCR results. It is entirely clear from our study that discrimination between monoclonality versus oligoclonality at diagnosis is the most powerful predictor of clonal evolution during the ALL disease course. All other variables, such as age, white blood cell count, and immunophenotype at diagnosis, failed to identify patients prone to clonal evolution of their Ig/TCR gene rearrangements. Monoclonal MRD-PCR targets were characterized by high stability, with 89% of all targets detectable at relapse. In contrast, only 40% of the oligoclonal MRD-PCR targets were preserved at relapse. Therefore, it is probably important to discriminate between monoclonal and oligoclonal Ig/TCR rearrangements, which requires a combined Southern blot and PCR approach. Southern blotting is particularly informative for detection of oligoclonality in IGH andIGK gene rearrangements, whereas heteroduplex PCR analysis in combination with Southern blotting is informative for detection of oligoclonal TCRD gene rearrangements. Southern blotting needs more DNA and is more labor intensive and time consuming than PCR techniques. However, with a single BglII restriction enzyme digestion it is possible to detect oligoclonality in IGH, IGK, and TCRD loci.26-28 Judging clonality solely from the number of PCR products per gene would result in marked underestimation of oligoclonality; eg, at least one third of the oligoclonal IGH targets (29 of 85, including 15 lost MRD-PCR markers) would have been classified as monoclonal.

The herein presented detailed comparison of Ig/TCR gene rearrangement patterns provides important information for appropriate selection of PCR targets for MRD monitoring. It is already accepted that preferably 2 MRD-PCR targets should be used per patient. Furthermore, our data show that monoclonal targets should be chosen as first option. As previously suggested, monoclonal Kde rearrangements were characterized by the best stability (95%), owing to their end-stage character.24,30 In addition, approximately 85% of monoclonal IGH and TCRD gene rearrangements remained stable at relapse (Table 2). In monoclonal VH-JHrearrangements, it is particularly attractive to position the patient-specific primers/probes at the VH-DHpart of the junctional region, which is a preferred strategy in current RQ-PCR based strategies.9-11 Identification of preferably 2 monoclonal MRD-PCR targets (IGH, IGK, and/orTCRD) was possible in 67 (77%) of 87 patients. When applying these monoclonal targets, the detection of relapse would have been possible in 65 patients, but false-negative results would have been obtained in 2 patients: one with presumably secondary ALL and an infant case characterized by extensive clonal Ig/TCR evolution. The second choice for target selection should concern oligoclonalIGH gene rearrangements. Although these rearrangements are particularly prone to ongoing and secondary recombination processes, they are the sole MRD-PCR targets in approximately 10% of childhood precursor-B–ALL patients.46 In the case of oligoclonalIGH targets, the patient-specific primers/probes should preferably be positioned at the D-N-J junctions. Moreover, all identified clonal DH-JH stems should be followed because restriction to 2 targets would increase the risk of false-negative MRD results. In our series this approach could have been used in an additional 17 (20%) of 87 patients with available MRD-PCR targets and should have resulted in detection of relapse in 15 cases (false negativity in patients 2308 and 5978 with secondary IGH gene rearrangements). Finally, successful MRD detection in the remaining 3 patients could have been accomplished by usage of TCRG gene rearrangements, which were sole MRD-PCR targets in these patients. One could argue for the preferred usage of TCRG gene rearrangements instead of oligoclonal IGH targets. However, in at least 2 of our patients (4910, 5661) usage of common DH-JHstems would have been superior to Vγ-Jγ targets (Figure 4C). More importantly, TCRG gene rearrangements are generally less sensitive markers in RQ-PCR analyses, owing to their limited combinatorial diversity and the abundance of polyclonal Vγ-Jγ joinings in normal T-cells in postinduction follow-up samples.47 Finally, our study clearly shows that oligoclonal Vδ2-Dδ3 and Dδ2-Dδ3 rearrangements should not be used as MRD-PCR targets, because most of them (about 75%) would be modified through clonal evolution processes. The Southern blot and PCR-based strategy for MRD-PCR target selection (priority order: 2 monoclonal IGH, IGK, and/or TCRDtargets, followed by DH-JH stems of oligoclonalIGH rearrangements, followed by TCRG targets with full exclusion of oligoclonal IGK and TCRDtargets) would enable successful detection of relapse in 83 (95%) of 87 patients with the currently available MRD-PCR targets (Figure5).

The above-presented strategy is based on combined Southern blot and PCR analyses for discrimination between monoclonal and oligoclonal Ig/TCR gene configuration. However, many MRD-PCR laboratories do not routinely perform Southern blotting, implying that they will underestimate the occurrence of oligoclonality in IGH (44% of patients in this series), TCRD (36%), and IGK (8%) genes. In an exclusively PCR-based strategy for MRD target selection, Kde rearrangements should be chosen as first option. When applying all available Kde targets, the detection of relapse would have been possible in 37 patients in our series (43%), but false-negative results would have been obtained in 2 patients. The second choice for target selection should concern IGH gene rearrangements. Using all identified IGH gene rearrangements with patient-specific oligonucleotides positioned at the DH-JHstems should have resulted in detection of relapse in 43 cases (49%), with false-negative results in 2 patients. If the design ofDH-JH oligonucleotides is not successful, one might decide to design VH-DH-JH oligonucleotides supplemented with the usage of PCR-based monoclonal TCRD targets. Finally, successful MRD detection in the remaining 3 patients could have been accomplished by usage of TCRG gene rearrangements (Figure5). Similarly to the combined Southern blot/PCR–based approach, the exclusively PCR-based approach would enable successful detection of relapse in 95% of patients. Thus, the lack of Southern blot information for discrimination between monoclonal and oligoclonal PCR targets might be compensated by monitoring of a higher number ofIGH and TCRD targets (Figure 5).

Both described strategies for selection of MRD-PCR targets have their advantages and limitations, which should be carefully weighed in the context of the facilities and experience of each MRD-PCR laboratory. Nevertheless, each strategy would enable successful detection of relapse in 95% of patients. If one assumes that the actual relapse rate in childhood precursor-B–ALL is 25% to 30%, the current Ig/TCR-based MRD-PCR methodology should be “effective” in 97% to 98% of cases with identifiable MRD-PCR targets at diagnosis.

We are grateful to Prof dr R. Benner and Prof dr D. Sońta-Jakimczyk for their continuous support, Mrs F. J. Weerden for assistance in statistical analyses, Mrs W. M. Comans-Bitter and Mr T. van Os for preparation of the figures, and Mrs J. A. Boon for her secretarial support. We thank the Dutch Childhood Leukemia Study Group for kindly providing precursor-B–ALL cell samples. Board members of the Dutch Childhood Leukemia Study Group are P. J. van Dijken, K. Hählen, W. A. Kamps, E. Th. Korthof, F. A. E. Nabben, A. Postma, J. A. Rammeloo, G. A. M. de Vaan, A. J. P. Veerman, and R. S. Weening.