Although the frequency and size of the copy number aberrations varied greatly, samples from both accelerated and blast stage CML showed substantially larger numbers of genome imbalances than chronic phase. These aberration frequencies obtained at a resolution of 33 K by oligonucleotide aCGH are broadly in agreement with earlier investigations of CML samples using BAC [10, 11] and SNP array CGH .
Altogether 435 loci were identified as significantly different in lymphoid and myeloid BC samples. All loci were located on the short arms of chromosomes 7 & 9, with the exception of 4 from chromosomes 14, 2 from chromosome 2 and a single chromosome 22 probe. Most of the probes mapped to known genes and a number were in regions associated with immune response. We show that deletion of certain regions within TCR (alpha/delta and gamma) (11/12) and IGH (12/12), are almost universal while being accompanied by deletions within IKZF1 (10/12) and, to a lesser extent, CDKN2A and sequences within several other genes in the short arm regions of chromosomes 7 and 9 (7/12). These genome losses were not seen in any of 31 CP, 6 accelerated and 12 BCM samples or controls. Furthermore, they were present in a BCL sample (Case 199) when DNA from the patient's CP was used as reference for the aCGH analysis, thus proving their secondary origin.
A recent genome wide single nucleotide polymorphism (SNP) array analysis of Ph positive samples - 43 ALL (paediatrics and adults) and 23 CML (11 CP, 9 BCM and 3 BCL)  - revealed the presence of three common deletions affecting the IKZF1, PAX5 and CDKN2A loci. These deletions, confirmed by FISH and qPCR, also detected in 2 out of 3 CML BCL samples, were not always co-existent. How the reduced activity of IKZF1 and PAX5 genes, that play a key role in regulating B lineage commitment, collaborates with BCR/ABL1 to induce Ph(+) ALL remains unclear [14, 16]. The frequency of the genome loss described by Mullighan et al., in the Ph(+) ALL, is in overall agreement with our findings in CML lymphoid blast crisis. In particular, CDKN2A deletions are 53.5% in Ph(+)ALL. 58.3% in CML/BCL and PAX5 loss is seen in 51.% of Ph(+)ALL vs. 58.3% in CML/BCL, while deletions of IKZNF1 are seen in 20 of 22 adults with Ph(+)ALL (90.9%)  and our aCGH analysis at 1 K resolution detected loss in 10/12 CML/BCL samples (83.3%). In contrast with the Ph(+) ALL results, we found that the loss of IKZF1 always precedes the deletions of CDKN2A and that losses of RNF38 were always accompanied by loss of the 9p13-p23.1 region that houses, among others, CDKN2A, IFNA, RNF38, PAX5 and MLLT3. In addition, we demonstrated by FISH the presence of clonal evolution leading to the concomitant 7p/9p loss as seen in less than half of patients (5/12 samples) (see typical aCGH profiles in Additional file 10 : Figure S7). Importantly, the cell clone bearing the concomitant loss in the studied cases was dominant and hence, detectable by aCGH.
No imbalances of IGH and/or TCR loci were reported in Ph(+) ALL study by Mullighan et al  either alone or in combination with IKZF1, CDKN2A or PAX5. Substantial parts of the IGH and TCR regions are known to show copy number variations (CNVs) in disease free individuals http://projects.tcag.ca/variation, but whereas CNVs are characterised by amplifications and deletions of a specific locus within the sample population, we found a set of probes in the IGH and TCR regions to be universally deleted in exclusively lymphoid BC samples, thus showing features typical of secondary genomic aberrations. That they are not somatic aberrations is illustrated by Case 199, where identical features emerged when the BCL sample was hybridised against patient's DNA from the CP. These deletions characteristic of lymphoid BC were also identified in the 2 samples with mixed immunophenotype (Figure 1).
There are two possible explanations for the origin of these deletions. Either they represent genome losses associated with cross lineage rearrangements known to occur in both ALL and AML, or they represent a bona fide event with oncogenic potential. IGH/TCR rearrangement patterns are widely used to monitor minimal residual disease (MDR) in a range of haematological malignancies, the levels of which significantly correlate with clinical outcome . The immunophenotype data for five of our 12 CML/BC samples that carry concomitant 7p/9p loss (cases 184, 192, 193, 195 & 200, Additional file 12 : Table S4) are consistent with early B cell origin of the blast cells (TdT+, CD10+, CD19+, CD22+, CD7 neg, so it could be that the recurrent loss of the TCR G/D region in B cells with legitimate IGH rearrangements represents the lineage infidelity as reported previously in blast phase CML [18–21]. However, the similarity of the deletions within both immunoglobulin genes in BCL does not comply with the notion that IGH and TCR configurations are unique (clonal by origin and patient specific). The common deleted regions could indicate a preferential VDJ rearrangement in the clone that ultimately becomes dominant in the BCL patients. Such preferential VDJ usage has been reported before with a preference for the VH4 family segment in about 35% of children with B-ALL . In our case the deletions in IGH and TCR loci possibly reflect the clonal origin of the malignancy and the stage at which the additional chromosomal aberrations that we observed were initiated. Further work needs to be done to relate these possibilities to the waterfall of changes that characterise blast crisis.
So, how does the loss of a relatively small ~100 Kb region from IGH and TCR contribute to the development of the lymphoid blast transformation? RAG1 and RAG2 (RAG) are the key components of the V(D)J recombinase machinery that catalyses the somatic gene rearrangements of antigen receptor genes during lymphocyte development. B cell progenitors undergo a development program involving ordered IGH gene rearrangements conducted by the RAG enzyme system. Recent evidence suggests that BCR/ABL1 induces expression of activation-induced cytidine deaminase (AID), found exclusively in early B cells. It participates in the class switch recombination and has been shown to lead to single strand breaks in IGH regions in Ph(+)ALL . Therefore, it would appear that the deregulation of AID silencing by transcriptional or epigenetic factors is the key event in compromising the V(D)J recombination/RAG impairment. This could lead to the TCR rearrangements in B cells, which have already undergone IGH rearrangements and are expressing the CD10, CD19 and CD22 antigens. Thus genome losses identified in the BCL samples represent the consequence of cross lineage rearrangement i.e. DJ joining occurring in early B cells and aberrant incomplete recombination in the TCR complex.
This view is supported by Mullighan et al  who showed that the mechanism responsible for IKZF1 deletions in Ph(+) ALL appears to involve aberrant RAG mediated recombination as heptamer signal sequences were found internal to the deletion breakpoints. Similarly, CDKN2A loss was linked to RAG activity in lymphoid leukemia [24, 25]. Furthermore, the impairment of the RAG system in Ph positive cells could be a direct consequence of the genetic damage caused by BCR/ABL1 [5, 23] either through inflicting direct damage via reactive oxygen species or by compromised DNA repair [4, 7]. It is also possible that the reduced/abolished expression of c-ABL  could further contribute to the impaired V(D)J recombination . In summary, a chain of events initiated by BCR/ABL1 impairs the RAG enzyme system. The compromised V(D)J recombinase machinery is thus instrumental in creating, by means of cross lineage rearrangements, clonal populations of B-cell progenitors carrying IGH and TCR deletions.