Identification of critical gene regions in chronic lymphocytic leukemia using high-resolution array comparative genomic hybridization

With an objective to identify novel copy number alterations (CNAs) critical to CLL pathogenesis, we performed array comparative genomic hybridization (array CGH) on genomic DNA extracted from purified leukemic cells of 51 CLL patients. We used the 1.4 Mb HumArray platform, and analyzed data using a combination of three different statistical algorithms and in full awareness of the copy number variations (CNVs) found in normal populations. Use of reference DNA extracted from matched patient neutrophils in four cases also assisted a more comprehensive interpretation of constitutional versus CLL-specific copy number alterations. A total of 108 clones showed alterations in >10% of the cases. Of these, 25 clones corresponded to known recurrently acquired CNAs for CLL, representing +12 in 8 cases (16%) and del(13)(q14) in 26 cases (51%). For the remaining 83 clones, frequent deletions corresponded to loci within cytobands 17q12 (71% of CLL cases), 9q32 (55%), 8p23 (47%) and 2q21 (33%), and frequent gains corresponded to loci within cytobands 18q22.3 (51%), 20p12 (33%) and 15q13 (31%), most of which have been reported in normal population cohorts. We further analysed associations between CNV status across the 108 frequently affected clones and CD38 expression status. Consistent with previous findings, loss of 13q14 (represented in clone RPI-269F22) and gain of chromosome 12 (clones RP11-64J22, RMC12P001 and RP11-282G15) were associated with low and high CD38 expression, respectively. We also found that loss within the beta-defensin (DEFB) locus at 8p23, and gains across clones CTB-120N12 (17q25) and RP11-118K20 (13q31.1) were associated with the poor prognostic CD38+ subgroup (p<0.05). Further studies are needed to assess biological relevance of genes located within these regions and prognostic outcomes in CLL. At the time this study was initiated, array CGH was a relatively new technique that was still under development. In order to establish robust and reliable array CGH protocols, extensive optimization was carried out. Different array platforms (cDNA and BAC arrays) and hybridization protocols were compared, and diverse methods for statistical analysis were explored. After optimization, the University of California San Francisco (UCSF) 2.5K BAC arrays and BlueFuse software for Microarrays were used to analyse CNVs in 51 CLL DNA samples. Genomic DNA was extracted either directly from peripheral blood mononuclear cells (PBMCs), or after further purification of CD19+ B-lymphocytes using a magnetic bead system. Different data analysis approaches that provided varying degrees of sensitivity and specificity were applied for more effective detection of CNVs. Identification of CLL-specific CNVs was maximized by performing array CGH using a single female reference DNA across all experiments, and by using reference DNA extracted from matched patient neutrophils in four cases. Patients were recruited into the study from a cohort of consenting patients referred for routine assessment to the Clinical Haematology Department at Christchurch Hospital, New Zealand. Diagnosis of CLL follows clinical assessment for commonly associated symptoms including physical weakness, enlarged spleen or lymph nodes, loss of weight, low immunological tolerance and night sweats. Follow-up pathological assessment typically includes a repertoire of blood tests (white blood cell count, platelet count, erythrocyte sedimentation rate, blood differential, cytochemistry), a bone marrow aspirate and sometimes lymph node biopsy assessment. At the time of this study, flow cytometry analysis of whole blood collected in either ethylenediaminetetraacetic acid (EDTA) or heparin tubes typically included the following antigens: CD5, CD19, CD20, CD22, CD23, CD11a, CD38, and membrane surface immunoglobulin lambda and kappa light chains. For CLL cases where flow cytometry results were not available at the time of sample collection, flow cytometry was performed to determine leukaemic cell purity. Conventional G-banded karyotype information was not available for patients recruited to this study. This study was approved by the New Zealand Upper South Regional Ethics Committee.