Genes commonly deleted in childhood B-cell precursor acute lymphoblastic leukemia: association with cytogenetics and clinical features

Abstract

In childhood B-cell precursor acute lymphoblastic leukemia, cytogenetics is important in diagnosis and as an indicator of response to therapy, thus playing a key role in risk stratification of patients for treatment. Little is known of the relationship between different cytogenetic subtypes in B-cell precursor acute lymphoblastic leukemia and the recently reported copy number abnormalities affecting significant leukemia associated genes. In a consecutive series of 1427 childhood B-cell precursor acute lymphoblastic leukemia patients, we have determined the incidence and type of copy number abnormalities using multiplex ligation-dependent probe amplification. We have shown strong links between certain deletions and cytogenetic subtypes, including the novel association between RB1 deletions and intrachromosomal amplification of chromosome 21. In this study, we characterized the different copy number abnormalities and show heterogeneity of PAX5 and IKZF1 deletions and the recurrent nature of RB1 deletions. Whole gene losses are often indicative of larger deletions, visible by conventional cytogenetics. An increased number of copy number abnormalities is associated with NCI high risk, specifically deletions of IKZF1 and CDKN2A/B, which occur more frequently among these patients. IKZF1 deletions and rearrangements of CRLF2 among patients with undefined karyotypes may point to the poor risk BCR-ABL1-like group. In conclusion, this study has demonstrated in a large representative cohort of children with B-cell precursor acute lymphoblastic leukemia that the pattern of copy number abnormalities is highly variable according to the primary genetic abnormality.

Introduction

The cytogenetics of B-cell precursor acute lymphoblastic leukemia (BCP-ALL) is well documented, with specific chromosomal abnormalities used in risk stratification of patients for treatment.1,2 Genomic studies have shown that copy number abnormalities (CNA) of genes involved in B-lymphocyte development and differentiation, cell cycle control and those of significance in hematopoiesis are common in BCP-ALL.3–5 Notable deletions include PAX5, IKZF1 (Ikaros),6–11 and genes within the pseudoautosomal region (PAR1) of the sex chromosomes, resulting in the P2RY8-CRLF2 gene fusion and overexpression of CRLF2.12 Here there is particular interest in IKZF1 and CRLF2 in relation to outcome and their role as molecular targets for therapy. IKZF1 deletions have been associated with a poor prognosis in BCP-ALL,6–10 while the risk relating to CRLF2 has been variable and dependent on other features.13–15 Nevertheless, thus far these diverse findings have not led to any treatment changes. Studies have focused on small or selected cohorts and analyses have often been carried out independently from other genetic changes. Thus their accurate incidence, relationship to each other, and the major cytogenetic subgroups still have to be determined in order to understand their true clinical relevance.

Recently, we demonstrated that multiplex ligation-dependent probe amplification (MLPA) provided an accurate and reliable high throughput method to screen for CNA of the significant genes in BCP-ALL.16 In this study, we screened a cohort of 1427 childhood BCP-ALL patients from two consecutive treatment trials using the same MLPA approach. We report the frequency and type of CNA involving these genes, their associations with established chromosomal abnormalities, and other clinical features.

Design and Methods

Patients in this study were diagnosed with BCP-ALL and registered on UK treatment trials UKALL97/99 (April 1997-June 2002) for children aged 1–18 years17 and UKALL2003 (October 2003-July 2011) for children aged 1–25 years.18 Clinical details were provided by the Clinical Trial Service Unit (CTSU), Oxford, UK. All participating centers obtained local ethical committee approval and written informed consent from patients, parents or guardians in accordance with the Declaration of Helsinki. Risk was assessed using National Cancer Institution (NCI) criteria.

Patients were classified into eight cytogenetic subgroups according to the presence of the following chromosomal abnormalities: 1) t(12;21)(p13;q22)/ETV6-RUNX1 fusion; 2) high hyperdiploidy (51–65 chromosomes); 3) translocations involving 11q23/MLL rearrangements; 4) t(9;22)(q34;q11)/BCR-ABL1 fusion; 5) intrachromosomal amplification of chromosome 21 (iAMP21);19 6) t(1;19)(q23;p13)/TCF3-PBX1; 7) other abnormal (absence of abnormalities in subgroups 1–7 above); and 8) normal karyotype. Patients were classified into good, intermediate, and poor cytogenetic risk groups according to previously published data.20

DNA obtained from the presentation bone marrow sample was used to determine the copy number of IKZF1, CDKN2A/B, PAX5, EBF1, ETV6, BTG1, RB1, and genes within PAR1: CRLF2, CSF2RA, IL3RA, using the SALSA MLPA kit P335 IKZF1 (MRC Holland, The Netherlands), as previously described.16 In those patients entered on UKALL2003, deletions of IKZF1 and RB1 were confirmed and further characterized by the P202 IKZF1 and the P047 RB1 SALSA MLPA kits, respectively.

Deletions of genes within the PAR1 region identified by MLPA were confirmed as P2RY8-CRLF2 or unbalanced IGH@-CRLF2 translocations by interphase fluorescence in situ hybridization (FISH) as previously reported.12

Statistical analysis was carried out using Intercooled STATA v. 12.0 (StataCorp, USA), particularly Wilcoxon Rank Sum for non-parametric assays and χ for comparison of categorical variables.

Results

CNA among the entire cohort

In total, 1427 patients were included in this study. There was no difference between these patients and other trial participants with respect to sex, age, central nervous system (CNS) disease or NCI risk group. Tested patients were more likely to have a white blood cell (WBC) count over 10 × 10/L reflecting the increased possibility of surplus available material (Online Supplementary Table S1). Incidences of CNA for genes tested by MLPA are given in Table 1. Overall, 59% of patients showed an abnormality of at least one of these genes: 433 (30%) patients had one, 254 (18%) had two, 131 (9%) had three and 28 (2%) had four or more deletions. Overall, deletions of CDKN2A/B and ETV6 were the most frequent, while EBF1 deletions were rare. Patients classified as NCI high risk were significantly more likely to have a greater number of deletions compared to those classified as NCI standard risk (P<0.001) (Figure 1).

Figure 1.Bar chart showing the percentage of patients within NCI high and standard risk groups with increasing numbers of CNA.

Table 1.Association of copy number abnormalities with demographics and WCC.

Table 1 shows the distribution of abnormalities in relation to demographic and clinical features. The cohort comprised 665 (46%) females and 762 (54%) males. There was no shift in the gender balance within each subgroup according to the defined CNA. The median age of the cohort was five years (range 1–23) with 24% of patients being 10 years or older. Patients with IKZF1 and CDKN2A/B deletions were significantly older. The median age for IKZF1 and CDKN2A/B deletions was seven years (P <0.0001) and six years (P <0.0001), respectively, with 39% and 33%, respectively, of deletions occurring in patients aged 10 years old or older. The incidence of these deletions increased with age (Table 1), a trend that continued into adulthood as shown by the incorporation of MLPA data from the UK adult ALL treatment trial, UKALLXII21 (Figure 2). There was a peak in incidence of ETV6 deletions in children aged 2–4 years (data not shown), explained by their strong association with ETV6-RUNX1, which has a peak incidence in this age group.22 There was no significant change in incidence linked to age for deletions of the other genes tested, including ETV6 deletions in ETV6-RUNX1 negative patients.

The median WBC count of the cohort was 11.8 × 10/L with 44% of patients having a count of less than 10 × 10/L. Patients with IKZF1, PAX5 or CDKN2A/B deletions were more likely to have a WCC of over 50 × 10/L (each P<0.001) (Table 1). The association of these genes to age and WBC count meant that there was a significantly higher incidence of patients with IKZF1, PAX5 and/or CDKN2A/B deletions classified as NCI high risk compared to other patients (each P<0.001) (Table 2).

The frequency of each cytogenetic subgroup among 1351 patients with a successful cytogenetic result is shown in Table 3. Patients with ETV6-RUNX1, high hyperdiploidy and those classified as ‘other abnormal’ comprised the most common subgroups at incidences of 28%, 30% and 24%, respectively. Patients positive for ETV6-RUNX1 showed the highest number of CNA overall, followed by those in the other abnormal group. In contrast, CNA occurred at a lower than expected level in high hyperdiploid patients. The incidence of CNA was also low in the subgroup with MLL rearrangements and higher in the other poor-risk subgroups: BCR-ABL1 positive and iAMP21. The increasing numbers of CNA in each cytogenetic group are shown in Figure 3, while the incidences and distribution of the individual CNA within each cytogenetic group are shown in Figure 4.

Figure 2.Graphs showing the proportion of cases with IKZF1 and CDKN2A/B deletions according to age among BCR-ABL1 negative patients. As it is known that there is a strong association between IKZF1 deletions and BCR-ABL1 positive ALL7 and that the incidence of BCR-ABL1 positive ALL increases with age22, childhood BCR-ABL1 positive patients were excluded from this age association analysis of IKZF1 and CDKN2A/B for direct comparison with the MLPA tested adult BCR-ABL1 negative series21.

Table 2.Association of Copy Number Abnormalities with NCI and Cytogenetic Risk Group.

Table 3.Incidences of CNA according to cytogenetic subgroups.

Among ETV6-RUNX1 positive patients, in addition to a high incidence of wild-type ETV6 deletions, CDKN2A/B and PAX5 were each deleted in 22% of these patients, while the incidence of IKZF1 deletions was low. Although BTG1 deletions were rare throughout the cohort (n=87, 6%), they were frequently associated with ETV6-RUNX1 (15%) (P<0.0001). There was an association between BCR-ABL1 and IKZF1 with 64% having a deletion of IKZF1. Deletions of PAX5 and CDKN2A/B were also high in this subgroup at incidences of 45% and 48%, respectively.

Among iAMP21 patients, RB1 deletions and P2RY8-CRLF2 were observed at incidences (39% and 30%, respectively) significantly higher than expected (P<0.0001). They showed the second highest frequency of ETV6 deletions. In the TCF3-PBX1 subgroup, there was a high incidence of deletions of PAX5 and CDKN2A/B. In the group classified as ‘other abnormal’, deletions of CDKN2A/B, PAX5 and IKZF1, as well as P2RY8-CRLF2, were seen at higher than expected frequencies.

Figure 3.Bar chart showing the proportion of patients within each cytogenetic subgroup with increasing numbers of CNA.

Figure 4.Circos plots showing the relative incidence and distribution of CNA within the overall cohort and the individual cytogenetic groups. The proportions of cases with and without CNA are indicated in black and gray, respectively, around the perimeter and the ribbons representing the individual CNA are color coded. The width of the ribbons reflects the frequency of each CNA.

Individual abnormalities

A range of interesting observations were made in relation to the individual abnormalities. IKZF1: deletions of IKZF1 were present in 14% (n=196) of the cohort. The size of the deletion varied between patients (Online Supplementary Figure S1). The most frequent deletions involved either the whole gene (n=60) or were restricted to exons 4–7 (n=61). Other deletions occurred at lower frequencies: exons 2–3 (n=15), exons 2–7 (n=20), exons 4–8 (n=13), and miscellaneous deletions (n=27). Five patients had biallelic deletions; each showed a different pattern of loss, ranging from biallelic deletion of all exons to a subset of exons. No significant association was seen between the pattern of exon loss and cytogenetic subgroup, age or WBC count. Although unlikely to be significant, the biallelic deletions did not occur in association with any of the major cytogenetic groups. Whole gene deletions were associated with visible cytogenetic abnormalities of the short arm of chromosome 7 in 36 of 60 cases including: monosomy 7 (n=13), i(7)(q10) (n=7), dicentric chromosomes (dic) (n=6), balanced and unbalanced translocations involving chromosome 7 (n=10).

PAX5 and CDKN2A/B: a total of 272 (19%) patients showed heterogeneous CNA of PAX5. The majority of patients had deletions including exon 1 involving the entire or part of the gene (n=141). The remaining patients had partial deletions excluding exon 1 (n=121) or intragenic amplifications of exons 2 or 5 (n=10), as previously illustrated.16

Deletions of CDKN2A/B represented the most frequent abnormality in the cohort (n=395), of which 143 showed visible cytogenetic abnormalities of the short arm of chromosome 9 (9p), 212 showed no 9p abnormality, while 40 failed cytogenetic analysis. Among those cases with a visible 9p abnormality, 103 showed concurrent CNA of PAX5. A further 59 patients with CNA of both genes had no visible 9p abnormality (n=41) or failed cytogenetic results (n=18). The types of cytogenetic abnormalities involving chromosome 9 associated with these CNA are shown in Online Supplementary Table S2. Dicentric chromosomes involving chromosome 9 were shown in 43 patients: dic(7;9) (n=7), dic(9;12) (n=16) and dic(9;20) (n=20). Interestingly, they showed heterogeneous exon loss from PAX5, ranging from the entire gene to loss of the telomeric exons (not including exon 1) and they variably included deletions of CDKN2A/B.

ETV6: ETV6 was frequently deleted throughout the entire cohort (n=312). As expected, ETV6 deletions were frequent among ETV6-RUNX1 positive patients (n=203). FISH results were available on 186 of these ETV6-RUNX1 positive cases. In 154 (83%), the results by FISH and MLPA were concordant. Among the remaining 32 cases, deletions were found by MLPA but not FISH, indicating the presence of focal deletions below the resolution of FISH. Conversely, 40 cases with ETV6-RUNX1 showed an ETV6 deletion by FISH only. These either represented small populations of cells containing the deletions (<25% of nuclei) below the level at which MLPA would be expected to detect loss, or deletion of the wild-type ETV6 allele with an associated gain of the derivative chromosome 21, thus producing a normal copy number for the ETV6 exons covered by the MLPA probes. These observations highlight the previously described advantages and disadvantages of these two techniques.16

In total, 109 ETV6-RUNX1 negative cases showed loss of ETV6; among 86 of these cases with FISH results available, 52 showed loss of the entire ETV6 by both FISH and MLPA, while 34 cases showed small intragenic deletions by MLPA which were below the resolution of FISH.

RB1: deletions of RB1 were present in 92 cases. These were of two types: 1) loss of the entire gene (n=60); 2) focal deletions, including exons 19–26 (n=28) (probes for these exons are included in the MLPA kit). An additional 4 cases had biallelic deletions of exons 19–26 as well as monoallelic loss of the remainder of the gene. Further studies using the P047-RB1 MLPA kit on 65 cases confirmed whole gene loss in 40 cases tested. These studies indicated that the deletions extended into the adjacent genes: ITM2B and RCBTB2 in all 40 cases, including DLEU in 38 of them. The relative location of these genes is shown in Online Supplementary Figure S2A. These results showed that RB1 loss in these patients is part of a larger deletion targeting several genes.

Further characterization of those cases with focal deletions, using the P047-RB1 MLPA kit with a higher probe density, showed the precise location of the deletion breakpoint to be between exons 17 and 18 in 19 of 21 cases tested, while in the 2 remaining cases the breakpoints were located between exons 16 and 17 and exons 18 and 19. In 8 of 21 cases with this focal deletion, the 5′ breakpoint included deletion of the RCBTB2 gene. The 4 patients with both deletions showed different patterns of exon loss: 2 showed monoallelic loss of ITM2B, RCBTB2 and DLEU. The other 2 showed normal copy number for ITM2B and DLEU with the biallelic loss extending to RCBTB2 in one case and monoallelic loss of this gene in the other. These results are illustrated in Online Supplementary Figure S2B.

From cytogenetic analysis, 52 cases in the entire cohort showed 13q abnormalities although only 52% (n=27) of these were associated with an RB1 deletion.

CRLF2 gene rearrangements: deletions within PAR1 were detected by MLPA in 4% (n=63) of the cohort, of which 54 cases had fixed cells available for FISH investigations. The P2RY8-CRLF2 fusion was confirmed by FISH in 49 cases, while the remaining 5 cases were found to be IGH@-CRLF2 translocations with associated deletion within the PAR1 region. FISH testing for the presence of IGH@ rearrangements had been carried out on 57% (n=807) of the entire cohort. Twelve cases, with a normal result by MLPA, were shown to have balanced IGH@-CRLF2 translocations by FISH. Collectively, these patients were described as CRLF2 rearranged, accounting for 5.3% of the cohort.

P2RY8-CRLF2 was most common among the iAMP21 subgroup (30%) and patients classified as ‘other abnormal’ (10%). Although P2RY8-CRLF2 was rare in association with other cytogenetic subtypes, it was found in all subtypes in this series except the MLL rearranged group. The positive cases included 3 ETV6-RUNX1 positive patients and one each of TCF3-PBX1 and BCR-ABL1.

Among these patients with CRLF2 rearrangements, 48% also showed loss of IKZF1. However, IKZF1 deletions were more significantly associated with IGH@-CRLF2 than P2RY8-CRLF2, occurring in 82% and 37%, respectively (P=0.001). IGH@-CRLF2 patients were more likely to be classified as NCI high risk than P2RY8-CRLF2 (65% and 29%, respectively, P=0.008). However, this result did not translate into a correlation between IKZF1 deletion status and risk among CRLF2 rearranged patients, as 56% of high risk and 42% of standard risk CRLF2 rearranged patients also had IKZF1 deletions (P=0.22).

Discussion

In this study, we present the findings from a detailed retrospective analysis of CNA in significant genes involved in B-cell development, cell cycle control and hematopoiesis among a large consecutive series of pediatric BCP-ALL patients treated on UK ALL treatment trials. Although 59% of patients showed an abnormality of at least one of these genes, 41% showed none. The number of CNA occurring simultaneously in the same patient was low. Thus, these observations, in association with cytogenetic data, confirm that the genomic profiles of childhood BCP-ALL are not generally complex. Although the involvement of other genes not covered by the MLPA kit cannot be ruled out, data from SNP arrays have shown the incidence of other recurrent sub-microscopic abnormalities to be infrequent.3–5 From MLPA studies, it is not possible to gain information on the temporal order in which these events arose in terms of karyotypic evolution or identify which abnormalities were the potential ‘drivers’ of leukemogen-esis. However, it was possible to examine associations between these abnormalities and demographic and clinical features, as well as with cytogenetics.

In relation to the individual abnormalities, IKZF1 showed heterogeneity in the size of the deletion, as previously demonstrated.3 The majority of patients either showed deletion of the entire gene, often seen as a visible cytogenetic change involving 7p, or restricted to exons 4–7. A range of other deletion types were also observed of which a small number were biallelic. No significant association was found between the pattern of exon loss and cytogenetic subtype. IKZF1 is transcribed in several isoforms as a result of alternative splicing, essentially altering the expression of exons 3 to 5 that encode the N-terminal DNA-binding domain. Deletions of exons 4–7 result in expression of a dominant negative IKZF1 isoform, Ik6, that lacks the N-terminal DNA binding zinc finger and shows oncogenic activity.11,23 Deletion of exon 2, which harbors the translational start site, will inhibit protein translation. Loss of exon 8 will have an effect on dimerization of IKZF1. Thus, deletions involving these exons are likely to have the same impact as whole gene deletions. Loss of the non-coding exon 1 only is likely to be of no significance. Accurate characterization of these heterogeneous deletions of IKZF1 is important before we can begin to understand their prognostic relevance.

The extent of PAX5 deletions was variable, ranging from whole gene loss to loss of the telomeric exons, confirming previous SNP data3 and our earlier observations based on FISH.24 Those with deletions of the entire or part of the gene including exon 1 are predicted to result in reduced PAX5 expression. Those with partial deletions not involving exon 1 are predicted to express a mutant allele.25 Ten patients showed intragenic amplifications of exons 2 or 5. These amplifications have been previously reported as a rare occurrence and are similarly predicted to express mutant alleles.25 In a number of cases, PAX5 deletions occurred as the result of dicentric chromosomes involving chromosome 9. Interestingly, among those cases with dic(9;12), 4 were associated with ETV6-RUNX1 fusion, while the remaining 5 were classified as ‘other abnormal’. Although the number of cases was small, there was a distinct pattern of exon loss between the two groups of dic(9;12) cases. The other abnormal cases showed loss of ETV6 exons 1–2 and PAX5 exons 5–10. This finding was consistent with previously published data in which this abnormality was associated with expression of an ETV6-PAX5 fusion protein.26 The dicentric chromosomes associated with ETV6-RUNX1 showed larger deletions of ETV6 and loss of the entire PAX5, indicating that these dic(9;12) translocations do not result in ETV6-PAX5 fusion.

Deletions of CDKN2A/B represented the most frequent abnormality in the cohort, which were often associated with visible abnormalities of 9p and concurrent loss of PAX5. Interestingly, 8 of the 10 cases showing intragenic amplification of PAX5 also showed deletion of CDKN2A/B.

RB1 deletions were homogeneous compared to other deletions, being restricted to two types: 1) those including the entire gene, as well as the adjacent genes: ITM2B, RCBTB2 and DLEU; and 2) focal deletions including exons 18–26 in all but one case. They were occasionally biallelic with different sized deletions on the two alleles. Some of the larger deletions were visible at the cytogenetic level. Deletions of 13q have been associated with increased risk of relapse.20 However, this study showed that only approximately 50% of visible 13q abnormalities were associated with an RB1 deletion, indicating that in at least some of these cases RB1 is not the target of the deletion. RB1 deletions of exons 18–27 have been previously reported in lymphoma.27 The molecular consequence of this recurring deletion is still not well understood. Expression of a truncated protein with altered function or deletion of LPAR6/P2RY5, located within RB1, may be one of the consequences.

With the exception of patients with MLL rearrangements, the presence of CRLF2-P2RY8 has now been reported in association with all cytogenetic subtypes. In this study, we identified a small number of cases among patients with ETV6-RUNX1, TCF3-PBX1 and BCR-ABL1; the latter two have not been previously reported. We showed CRLF2 rearrangements to be present in 5.3% of the cohort. Several groups have reported a strong association between CRLF2 overexpression and IKZF1 alterations.13,14,28–30 Although this study is restricted to the detection of CRLF2 rearrangements, with the exception of rare mutations of the gene,31 without measure of expression, we confirmed this association. We also confirmed that IGH@-CRLF2 occurred at a higher incidence in NCI high-risk patients, while P2RY8-CRLF2 was seen at a higher frequency in NCI standard risk.

We have previously reported the high incidence of P2RY8-CRLF2 among iAMP21 patients.32 We confirmed an increased incidence of ETV6 deletions as previously reported;33 however, the frequent occurrence of RB1 deletions in these patients is shown here for the first time.

There was some correlation between the distribution of CNA and patient age. Patients with IKZF1 and CDKN2A/B deletions were older and their incidence increased with age. Patients with IKZF1, PAX5 and CDKN2A/B deletions had significantly higher WBC count than patients with the other CNA. These associations with older age and higher WBC count explain why these deletions occur at a higher frequency in the NCI high-risk group, that is defined by age and WBC count. Previous studies have been inconclusive as to the prognostic relevance of CDKN2A/B in both childhood and adult ALL.34 The association with NCI risk defined here, coupled with observations that patients with CDKN2A/B deletions have a shorter time to relapse than other relapsed patients,35,36 suggests that further studies are warranted in order to clarify the prognostic relevance of CDKN2A/B deletions.

The frequency of each cytogenetic subgroup among this patient cohort was the same as that previously reported by us for a single UK childhood ALL treatment trial, ALL97/99.15,20 We showed that the incidence of CNA varied according to cytogenetic subtype, although the number did not correlate with the cytogenetic risk. For example, the good-risk cytogenetic groups (ETV6-RUNX1 and high hyperdiploidy) as well as the poor risk (BCR-ABL1 and MLL rearranged) showed a high and low number of CNA, respectively.

The high number of CNA in the ETV6-RUNX1 positive group suggests that CNA rather than point mutations may be the drivers of leukemia in this subgroup, as other studies have shown ETV6-RUNX1 positive leukemia to harbor a modest number of point mutations.37 Interestingly, although BTG1 deletions were rare, they were most often associated with ETV6-RUNX1, as previously reported.38 As BTG1 has been reported to be associated with glucocortoid receptor autoinduction,39 these patients require follow up to determine whether these deletions affect their overall survival.

The lower than expected level of deletions in high hyperdiploid patients is unlikely to be an artefact of analyzing CNA in the context of ploidy change, as neither the genes tested nor the reference probes in the MLPA kit are located on chromosomes commonly gained in high hyperdiploidy. Point mutations have been found at an increased level in this subgroup, indicating that disease progression in high hyperdiploid patients may be driven by mutations rather than deletions, at least among the genes tested.40 Although the incidence of CNA was also low in patients in the poor-risk subgroup with MLL rearrangements, this finding might be expected from the known potency of this abnormality as a driver of leukemogenesis.41 From these observations, it is evident that future studies assessing the prognostic value of these CNA must include cytogenetic data in order to gain a clear picture of their association with outcome.

As previously shown, there was a strong association between BCR-ABL1 fusion and IKZF1 deletions.7 Deletions of PAX5 and CDKN2A/B were also high in the BCR-ABL1 positive group. In the group classified as ‘other abnormal’, deletions of CDKN2A/B, PAX5 and IKZF1 were also frequent. The striking similarity in CNA profiles between these two groups is clearly shown in Figure 4. The poor-risk group described as BCR-ABL1-like,6,41 defined as sharing the same gene expression profile as well as the same poor risk as BCR-ABL1 positive patients, is most likely to be found among those patients with ‘other abnormal’ karyotypes as no distinctive karyotypic features have yet been described to define them. Thus the occurrence of deletions of CDKN2A/B, PAX5 and IKZF1, as well as deregulated CRLF2, among patients in this ‘other abnormal’ group may provide a pointer to the BCR-ABL1-like subgroup, as indicated by others.42

This study represents the largest trial-based screen for abnormalities in selected genes of significance in the development of BCP-ALL. It has confirmed findings of previous studies of associations between copy number abnormalities and particular cytogenetic subgroups. It has shown the heterogeneous nature of deletions such as PAX5 and IKZF1 and the recurrent nature of RB1 deletions. The association of IKZF1 and CDKN2A/B deletions with NCI high risk is of interest, and within the ‘other abnormal’ cytogenetic group, the presence of these deletions and/or rearrangements of CRLF2 may point to the poor-risk BCR-ABL1 like group.

Acknowledgments

The authors are grateful to the members of the UK Cancer Cytogenetics Group for contribution of data and samples to this study, and to members of the Leukaemia Research Cytogenetics Group past and present for help in establishing this data set. They also wish to thank the Clinical Trial Service Unit, Oxford, for the contribution of clinical data and the trial co-ordinators for permission to publish these data. Primary childhood leukemia samples used in this study were provided by the Leukaemia and Lymphoma Research Childhood Leukaemia Cell Bank working with the laboratory teams in the following centers: Bristol Genetics Laboratory, Southmead Hospital, Bristol; Molecular Biology Laboratory, Royal Hospital for Sick Children, Glasgow; Molecular Haematology Laboratory, Royal London Hospital, London; Molecular Genetics Service and Sheffield Children’s Hospital, Sheffield. Thanks to Leukaemia Lymphoma Research for financial support and the Kay Kendall Leukaemia Fund for funding additional FISH screening of the IGH@ gene.

Footnotes

• The online version of this article has a Supplementary Appendix.
• Authorship and Disclosures Information on authorship, contributions, and financial & other disclosures was provided by the authors and is available with the online version of this article at www.haematologica.org.

• Received January 25, 2013.
• Accepted March 8, 2013.