Current Knowledge of Microarray Analysis for Gene Expression Profiling in Chronic Lymphocytic Leukemia

Chronic lymphocytic leukemia (CLL) is characterised by the accumulation of mature CD5/CD19 B-lymphocytes in the blood, bone marrow, lymph nodes and spleen (Caligaris-Cappio & Hamblin, 1999). Although the role of cellular proliferation disorders in CLL may originally have been underestimated, the typical characteristic of the disease is still regarded as a failure of malignant cells to undergo apoptosis (Munk Pedersen & Reed, 2004). CLL is a heterogeneous disease and although it is relatively stable in some patients, it progresses rapidly in others (Caligaris-Cappio & Hamblin, 1999). The mutational status of immunoglobulin heavy chain variable gene segment (IGHV) and the expression of CD38 and/or ZAP70 are important prognostic factors of disease so their detection is very useful for stratification of patients into indolent or aggressive subgroups (Hamblin et al., 1999; Krober et al., 2002; Orchard et al., 2004).


Introduction
Chronic lymphocytic leukemia (CLL) is characterised by the accumulation of mature CD5/CD19 B-lymphocytes in the blood, bone marrow, lymph nodes and spleen (Caligaris-Cappio & Hamblin, 1999).Although the role of cellular proliferation disorders in CLL may originally have been underestimated, the typical characteristic of the disease is still regarded as a failure of malignant cells to undergo apoptosis (Munk Pedersen & Reed, 2004).CLL is a heterogeneous disease and although it is relatively stable in some patients, it progresses rapidly in others (Caligaris-Cappio & Hamblin, 1999).The mutational status of immunoglobulin heavy chain variable gene segment (IGHV) and the expression of CD38 and/or ZAP70 are important prognostic factors of disease so their detection is very useful for stratification of patients into indolent or aggressive subgroups (Hamblin et al., 1999;Krober et al., 2002;Orchard et al., 2004).
However, a more robust approach to subclassifying CLL is to identify the genomic changes in the malignant clone.The heterogeneity of the disease may result from different genetic abnormalities in distinct subclasses of patients.Furthermore, there is a strong relationship between specific genetic aberrations and the clinical course of the disease.On the basis of the mutational status of the variable region of the IGH, CLL can be divided into two subtypes.Somatic hypermutation of IGHV occurs in more than half of the patients and is associated with a more indolent clinical course.Additionally, deletions of the long arm of chromosome 13 or 11 and the short arm of chromosome 17, as well as trisomy of chromosome 12, are prognostically most important for the CLL patients.The most common abnormality in CLL, observed in more than 50% patients, is del(13)(q14) and, along with hypermutation of IGHV, this is linked with a good prognosis (Damle et al., 1999;Schroeder & Dighiero, 1994).

Gene expression profiling
Recent advances in genomics have transformed research on hematologic malignancies by improving molecular approaches to gene networks.New technologies have been designed to meet the need for methods to address the functional significances of nucleotides sequences.Microarrays have emerged as powerful tools for increasing the potential of standard methods through genome-wide biological studies.They have been focused mainly on gene expression profiling (GEP), but also on mutational screening, genotyping of polymorphisms and copy number analyses.

Contribution of microarray study to comprehension of CLL pathophysiology
DNA microarrays can be used to detect either DNA, as in comparative genomic hybridization, or to detect RNA, usually as complementary DNA (cDNA) after reverse transcription.The process of measuring gene expression via cDNA is called expression analysis or expression profiling (Schena et al., 1995).Alizadeh et al., (2000) investigated the construction of a commercial cDNA microarray (Lymphochip) for studies of normal and malignant human cells.They examined each stage of lymphocyte differentiation that can be defined by a characteristic gene expression signature.Genes that are coregulated by over hundreds of experimental conditions often encode functionally related proteins.GEP also provide an unprecedented ability to define the molecular and functional relationships between normal and malignant lymphocyte cell populations (Alizadeh & Staudt, 2000).
Using different microarray platforms such as oligonucleotide arrays, cDNA arrays printed on glass slides and on nylon membranes, Wang et al., (2004) found that several genes were consistently differently expressed between CLL and normal B-cell samples.The following 10 genes were shown to be expressed differently in CLL compared with tonsillar Blymphocytes and plasma cells: FCER2 ( CD23), FGR, TNFRSF1B, CCR7, IL4R, PTPN12, FMOD, TMEM1, CHS1 and ZNF266 (Zent et al., 2003).
The results of GEP tests on CLL cells indicated that their profile was more closely related to non-proliferating B cells, or memory B cells, than to cells from a naïve germinal centre (GC), mitogenically activated blood cells or CD5+ B cells (Klein et al., 2001).Over the last few years, global GEP has been revised and defined CLL as a tumor of antigen-experienced B cells.These could be marginal zone or memory B cells.Now we know, that CLL results not only from an accumulation of transformed B cells, due to an imbalance between cell generation and cell death rate, but also from a proliferation of B cells in particular microenvironments in the lymphoid tissues and bone marrow (Klein & Dalla-Favera, 2010).Based on these findings, it can be suggest, that the leukemic B cells are more complex mixture than we have hitherto expected.Other genes have been dubbed CLL signature genes because they are selectively expressed in CLL and not in normal cells or other types of B-cell malignancy (Rosenwald et al., 2001).The CLL signature includes genes already known to be characteristic for CLL, such as CD5, IL2Rα (CD25) and BCL2, and genes not previously known to be expressed in CLL, such as WNT3, TITIN, ROR1 and MRC-OX2.ROR1 and MRC-OX2 encode membrane proteins, so they might be useful for decisions concerning treatment with humanised monoclonal antibodies.WNT3 probably regulates B lymphocyte proliferation (Zent et al., 2003;Reya et al., 2000).A study by Zent et al., (2003) showed that the GEP of CLL lymphocyte is different from multiple myeloma (MM) cells.CLL expressed higher levels of tumour necrosis factor (TNF) and TNF receptor pathway genes (LTB, TRAF5, TNFRSF9, TNFSF7 and LITAF).The IAP family gene (BIRC1) and the XIAP antagonist (HSXIAPAF1) were expressed at higher levels only in CLL to MM, similar to BCL-2 expression.

Contribution to the identification of new genes that might be considered as prognostic factors
In the last few years, the development of cytogenetics and molecular biology has led to the release of new genetic prognostic markers such as IGHV mutational status, genomic aberrations and individual gene mutations.
To determine possible genetic and molecular abnormalities related to early clinical progression in CLL, Fernandez et al., (2008) investigated alterations in genomic and gene expression profiles in a series of samples sequentially obtained at diagnosis in early stage of the disease and at the time of clinical progression before treatment.A group of 58 genes was identified by supervised analysis comparing the initial and progressed samples: 37 were over-expressed while 21 were down-regulated.No significant differences were observed in the expression of these genes in samples from the three CLL cases with stable clinical decease.Functional analysis of the over-expressed genes showed that they are involved in different pathways, including cell cycle and cell growth (MCM4, RAPGEF2, OGG1, ESCO1, ESR1, ACTL6A, CENPJ, ATG5) and ion regulation (MYLC2PL, ADRB1, TRPV5, TMCO3).Interestingly, 6 of the 21 down-regulated genes were considered negative regulators of integrin-mediated cell adhesion and motility (PRAM1, CDC42EP4, COL4A2, PLCB2, RAPGEF1, FLNA).These findings suggest that in early stage CLL, clinical progression is associated with inactivation of tumour suppressor genes and modulation of the expression of a small number of genes that are inhibitors of cell adhesion and motility.Ferrer et al., (2004) performed gene expression profiling on 31 CLL cases and investigated the HV gene mutation status by nucleotide sequencing.The array data showed that the greatest differences between the unmutated (20 cases) and the mutated (11 cases) groups were observed in the expression of such genes as: ZAP70, RAF1, PAX5, TCF1, CD44, SF1, S100A12, NUP214, DAF, GLVR1, MKK6, AF4, CX3CR1, NAFTC1 and HEX.ZAP70 was significantly more highly expressed in the IGHV-unmutated CLL group, whereas all the other genes were more highly expressed in the IGHV-mutated cases.This study confirmed that ZAP70 expression can predict the HV mutation status and suggested that RAF1, PAX5 and other differentially expressed genes may be good markers for differentiating between these two groups and can serve as prognostic markers.

Deregulated apoptosis in poor-prognosis CLL
CLL is a heterogeneous disease with marked variability in its clinical course.With the aim of identifying genes potentially related to disease progression, Fält et al., (2005) performed gene expression profiling on CLL patients with non-aggressive disease or with progressive disease requiring therapy.The Affymetrix GeneChip U95Av2 technique was used in 11 samples obtained from CLL patients with stable and 10 patients with clinically progressive disease.To discriminate samples from progressive and stable disease, a group of genes was chosen as markers; two genes in particular, PPP2R5C and RBL2, were included among the best discriminators as both were expressed at lower levels in progressive than in stable CLL.These genes are known to be key regulators of both the cell cycle and the mitochondria/cytochrome c apoptotic pathway.This procedure allowed samples with progressive and stable disease to be identified with 70-90% accuracy.Stratowa et al., (2001) studied 54 peripheral blood lymphocyte samples obtained from patients with CLL to determine the expression levels of 1024 genes on a cDNA microarray and to correlate them with patient survival.Overall survival (OS) of CLL patients displaying low expression of genes coding for IL-1, IL-8 and L-selectin was shorter than for patients with high expression of these genes.However, high expression of TCL1 was connected with decreased patient survival.These findings suggest that CLL prognosis may be connected with a defect in lymphocyte trafficking, causing accumulation of leukemic B cells in the blood.Edelmann et al., (2008) used a microarray-based GEP (Affymetrix U95A) to study how the stroma modulates the survival of CLL cells in in vitro co-culture model employing the murine fibroblast cell line M2-10B4.CLL cells cultured in direct contact with the stromal layer (STR) showed significantly better survival than cells cultured in transwell (TW) inserts above the M2-10B4 cells.STR induced a more marked up-regulation of the PI3K/NF-B/Akt signaling pathway genes (INPP4A, NFKB2, REL and MAPKAPK2) than TW conditions and mediated a pro-angiogenetic switch in the CLL cells by up-regulating VEGF and OPN and down-regulating the anti-angiogenetic molecule TSP-1.The findings also suggest that TSP-1 expression in CLL cells may be related to both disease stage and CLL subtype as defined by ZAP70 and CD38 expression.OPN protein secretion may be correlated to disease progression in CLL.
GEP used to predict the prognosis in CLL is presented in Table 1 and Table 2.

Contribution of GEP microarray study to pharmacogenomics
Drug resistance remains a major problem of CLL treatment.Owing to their high adaptability to therapeutic conditions, malignant tumour cells frequently develop escape mechanisms in response to cytostatic drugs.It is very difficult to predict a tumour's reaction to drugs because it can deploy multiple cellular mechanisms such as enhanced DNA repair, elevated levels of drug transporters, over-expression of detoxifying enzymes or apoptosis inhibition, which are often involved in the development of drug resistance.To monitor the multiple alterations by which CLL may become drug-insensitive, highly parallel analyses such as the DNA microarray technique are required.This technique affords new ways of predicting resistance and sensitivity to therapy (Dietel & Sers, 2006).

In vitro experiments
There is now well documented that some genes induce apoptosis, whereas the others can inhibit this phenomenon (Table 3).It is also known that drugs used for therapy regimens can change GEP and modify apoptosis.However, the knowledge concerning the drug influence on GEP is still insufficient and demands further studies.
The study by Vallat et al., (2003) combined two series of microarray analyses (Hu-FL GeneChips, Affymetrix, 7,070 genes) with four sensitive and three resistant CLL samples and compared their gene expression patterns before and after in vitro irradiation-induced apoptosis.Sixteen differentially expressed genes (2-fold, specifically in resistant cells) were disclosed by data analysis.After the validation of the selected genes by quantitative RT-PCR on seven microarray samples, their altered expression level was confirmed on a further 15 CLL samples not previously included in the microarray analysis.Eleven patients with  (Edelmann et al., 2008;Fält et al., 2005;Fernandez et al., 2008;Ferrer et al., 2004;Stratowa et al., 2001)

Cell adhesion and motility
The result of this study was defining clinically relevant new molecular markers specific to resistant CLL subtypes.Morales et al., (2005) investigated the regulation of apoptosis in B-CLL cells using cDNA microarrays (Human Apoptosis GEArray Q Series, Superarray) with 96 known genes.Data were obtained from and compared between two groups of CLL patients with either nonprogressive, non-aggressive, previously untreated disease in which the leukemic cells were sensitive to in vitro FA-induced apoptosis, referred to as sensitive B-CLL (sB-CLL), or progressive, chemotherapy-refractory disease in which the leukemic cells were resistant to in vitro FA-induced apoptosis, referred to as resistant B-CLL (rB-CLL).By performing a supervised clustering of genes that most clearly discriminated rB-CLL from sB-CLL, a small group of genes was identified.BFL1 was the most strongly discriminating gene, with higher expression in rB-CLL.This finding suggests that BFL1 may be an important regulator of CLL apoptosis, which could contribute to disease progression and resistance to chemotherapy, and could be a potential future therapeutic target.
Direct physical interaction of stromal cells with CLL cells and overexpression of RAD51 and LIG4 (DNA ligase IV) in the leukemic cells have been found.These genes code for DNA repair enzymes in mammalian cells (Edelmann et al., 2008).Given that RAD51 expression in CLL was previously reported to correlate with resistance to CHB.These findings may provide a molecular-level explanation of the capacity of stromal cells to protect CLL cells from drug-induced apoptosis (Christodoulopoulos et al., 1999).Segel et al., (2003) have used a cDNA microarray containing approximately 40,000 human gene sequences to obtain GEP for untreated and tetradecanoyl phorbol acetate (TPA)-treated B-CLL cells.Three genes, EGR1, DUSP2 and CD69, showed a 2-fold or greater increase in mRNA transcription in two studies.Several genes (PKC, N-MYC, JUN D and BCL2), previously reported to be overexpressed in CLL lymphocytes, were also overexpressed in these studies but were not altered by TPA treatment.These findings suggest that the products of these three genes may be central to early steps in the TPA-induced evolution of B-CLL cells to a plasma-cell phenotype.A variety of stimulators such as TPA, bryostatin, IL-2 and others can induce CLL lymphocytes to mature in vitro to an immunoglobulin-producing andsecreting phenotype.Such treatment corrects some metabolic defects such as impairment of the L-system amino acid transport, but not others such as diminished membrane gammaglutamyl transpeptidase (GGTP) activity.
GEP allows the study of a large number of genes and analysis of global pathways rather than single targets.Stamatopoulos et al., (2009) revealed the influence of valproic acid (VPA) on molecular changes in two key pathways in cancer: apoptosis and proliferation.The study was conducted on purified B cells obtained from 14 CLL patients.Microarray analysis was performed with an Affymetrix GeneChip Human Genome U133 Plus 2.0 array.Several genes (i.e.CD5, BCL2, CD23, LCK, PIM1) described as overexpressed in CLL by Wang et al., (2004) were downregulated by VPA in this study, whereas genes described by Wang et al., (2004) as underexpressed in CLL (i.e.BCLA1, C-MYC, DUSP2 and PEA15) were upregulated by VPA.The authors suppose that these results indicate that VPA could restore a more 'normal' epigenetic code and, in this way, could allow normal cellular processes that were silenced after malignant transformation.No differences among the GEP of ZAP + and ZAP − patients (poor and good prognosis, respectively) were found, indicating that VPA was acting independently of disease aggressiveness.It had also been observed that VPA acted on an important number of genes involved in apoptosis: BCL2, XIAP, FLIP, BCL-xL, AVEN and cIAP, which as a result, were significantly downregulated, whereas CASP 2, 3, 6, 8, 9, and BAX, BAK, APAF1 and P53 were all significantly upregulated.The ratio of anti-and proapoptotic genes determines the tendency towards cell death or cell survival.Moreover, a large number of cell-cycle genes were upregulated, not only CDK1, 2, 4, and 6, cyclin B1, B2, D1, D2, E1 and E2, but also inhibitors of cell cycle, such as P15, 16, 18, 19 and 21.The deregulated and simultaneous expression of all these genes is probably one of the reasons for proliferation inhibition (Stamatopoulos et al., 2009).
In our department, we identified differentially expressed genes in lymphocytes obtained from CLL patients and incubated with FA or cladribine (2-chlorodeoxyadenosine; 2-CdA) (Table 4).Among 93 studied apoptotic genes by means of 384 TaqMan Low Density Array (Applied Biosystems) most of them were downregulated, whereas such a few of them were upregulated: BAD, TNFRSF21, DAPK1 -in 2-CdA cultured group and CARD6 and CARD9 in FA cultured group.We have also noticed 4 genes (BAK1, BAX, FAS and PUMA) with about a 20-or more -fold decrease in gene expression with respect to control samples.Interestingly, in the above-mentioned genes we have found great differences in fold change value between FA and 2-CdA.The expression of two of them, BAX and PUMA, were considerably decreased when lymphocytes were incubated with FA.It may be hypothesized that the high ratio between anti-and proapoptotic gene expression might account for the failure to achieve complete response after purine nucleoside analogues (PNAs) therapy.Additionally, 2-CdA has inhibited to a lower extent the expression of PUMA and BID as compared to FA (Franiak-Pietryga I, Korycka-Wolowiec A, unpublished data), which might confirm the results reported by Robak et al., (2009) that 2-CdA, but not FA, is the most effective drug against P53-defective cells.At this stage of our knowledge, probably it is too far-fetched to make a suggestion that FA mostly triggers apoptosis in intrinsic pathways to caspase activation, while 2-CdA induces apoptosis via death receptor activation (extrinsic pathway) and by stress-inducing stimuli (intrinsic pathway).To confirm this hypothesis, further experiments are to be conducted in our department.Besides the in vitro experiments also in vivo studies play an important role in the increase of our knowledge on gene expression profiling.

In vivo studies
The study of CLL by Plate et al., (2000) was directed at understanding the signals that maintain viability in vivo and are lost when the leukemic cells are removed from the body, such that they immediately begin to undergo apoptosis ex vivo.Differences in gene expression between freshly isolated B-CLL cells and those maintained in vitro with and without FA were measured using the ATLAS apoptosis cDNA microarray (Clontech, Palo Alto, CA).Many genes, especially cyclin D1, were under-expressed after culturing.The anti-apoptotic genes BAG1 and AKT2 were over-expressed.The greatest positive effect of FA was the up-regulation of JNK1.Rosenwald et al., (2004) profiled gene expression in CLL leukemic samples obtained before and during FA administration using Lymphochip DNA arrays prepared from 17,856 cDNA clones.The procedure selected 27 microarray elements, 18 of which represented named genes while the other 9 represented novel genes of unknown function.In seven CLL samples, a consistent gene expression (GE) signature of in vivo FA exposure was identified.Many of the FA signature genes were known P53 target genes and genes involved in DNA repair (P21, MDM2, DDB2, TNFRSF10B, PCNA and PPMID).Because in vivo treatment with FA induces a P53-dependent GE response, it has the potential to select P53 mutant CLL cells, which are more drug-resistant and are associated with an aggressive clinical course.Therefore, treatment of CLL patients with FA has the potential to select for outgrowth of P53 mutant subclones that would be cross-resistant to several other chemotherapeutic agents.Moreover, the gene expression response to  radiation was highly similar to the response to FA.
The purine metabolism of B-CLL lymphocytes was studied by Marinello et al., (2006).Gene expression analysis was performed on samples obtained from 2 B-CLL patients.Data analysis revealed 17 genes whose expression varied at least 2-fold.Some purine metabolism genes expressed differently from controls were identified.Among the de novo enzymes, the Gars-Airs-Gart complex was over-expressed and IMPDH1 and APRT seemed under-expressed.An imbalance in the expression of the adenosine-related protein gene was also observed, with over-expression of CD26, CD38 and mtAK3, while ADORA 1 and cAK1 were under-expressed (Table 5).Simultaneous gene profiling of apoptosis-related factors and purine metabolism enzymes is of particular interest for drugs such as FA and 2-CdA, which are commonly used in CLL treatment.Three years later the above-mentioned data was confirmed on samples obtained from 5 B-CLL patients on a chip prepared with 57 genes.To the group of genes described previously some of new ones were added, including apoptosis-related proteins.CASP6, CASP8 and BCL2L1 (BCL-xL) were under-expressed, whereas IL-4, IL-18 were overexpressed.In contrast, less significant changes were observed in the expression of some other anti-or proapoptotic factors like BAX and BCL10, respectively.
To identify novel genes involved in the molecular pathogenesis of CLL, Proto-Siqueira et al., (2008) performed a serial analysis of gene expression (SAGE) in CLL cells and compared it with healthy B cells (nCD19).A gene ontology analysis revealed that TOSO, which plays a functional role upstream of the FAS extrinsic apoptosis pathway, was over-expressed in CLL cells.A positive correlation was observed between TOSO and BCL2, but not between TOSO and FLIP.The over-expression of TOSO and BCL2 might be responsible for BAX inhibition, which leads to the suppression of apoptosis and might be associated with poor prognosis in CLL.It is also known that bortezomib blocks BAX degradation in malignant B cells.TOSO might therefore be considered a possible target for small molecule therapy in combination with newer pro-apoptotic drugs such as bortezomib and lumiliximab.Giannopoulos et al., (2009) provided novel biological insights into the molecular effects of thalidomide and suggested the existence of a signature predictive of thalidomide response in CLL.GEP data on day 0 and 7, based on a paired supervised analysis, revealed a thalidomide-induced signature comprising 123 differentially expressed genes.Upon thalidomide monotherapy, an upregulation of genes, known to be involved in mediating thalidomide response, was observed.Such genes as FAS and CDKN1A, as well as novel candidate genes, such as STAT1 and IKZF1 were reported.Gene expression differences in responders as compared to nonresponders after thalidomide monotherapy on day 7 were determined.Responders showed lower expression of gene coding pro-survival cytokine such as IL-8 and lower level of TGFB1, whereas genes involved in apoptosis, i.e.CASP1, were more highly expressed than in nonresponders.Higher expression of ZAP70, as well as anti-apoptotic genes such as TRAF1, and genes involved in angiogenesis, (eg.ECGF1) was observed in nonresponders group.Thalidomide responders showed also lower JUN and CASP9 expression levels associated with deregulated insulin and RAS signalling pathways.
In CLL being induced by NFKB activation, IL-8 may function as an autocrine growth and apoptosis resistance factor promoting cell survival.
Our data depicts changes in apoptotic GEP in CLL patients treated with cladribine, cyclophosphamide and rituximab (CCR).The measurements were conducted by means of 384 TaqMan Low Density Arrays (Applied Biosystems).Data analysis pointed 20 out of 93 examined apoptotic genes, whose expression has significantly changed.Changes in GEP are mostly related to the intrinsic apoptotic pathway.The most significant differences in gene expression before, as opposed to after, treatment are demonstrated by antiapoptotic genes such as BCL2, BCL2L1, BIRC1, BIRC5 and BIRC8, whose expression is considerably decreased.Of the proapoptotic genes, NOXA, CASP10, ESRRBL1 and NFKBIZ are particularly distinguished, because they are significantly overexpressed (Table 4).Additionally, genes specifically clustered in terms of GEP, which was different in particular genes depending IGHV mutational status (Franiak-Pietryga et al., 2010).

Gen expression
Gen GEP may have a predictive value for the effectiveness of anti-cancer therapy.Although numerous experiments remain to be performed, it might become possible to predict chemoresistance and to avoid ineffective drugs.The possibility of pretherapeutic discrimination between responders and non-responders will further stimulate the development of an individualised therapeutic strategy using a personalised combination of drugs (Dietel & Sers, 2006).A list of the genes and their response to therapy and drug resistance is presented in Table 5.

Introduction
Owing to a greater availability of the human genome sequence, the focus of research has now been shifted to identifying sequence polymorphisms.It is of utmost importance to understand how biological functions may be affected by these variations and be associated with heritable phenotypes.
A single nucleotide polymorphism (SNP) array is a type of DNA microarray that is used to detect polymorphisms within a population.SNPs are the most frequent type of variation in the genome.It is estimated that about 10 million SNPs have been identified in humans, an average of one SNP every 400-1000 base pairs (Botstein & Risch, 2003).Currently, about 5.6 million have been typed (dbSNP Build ID: 126), about half of which are estimated to have a minor allele frequency over 10% (Kruglyak & Nickerson, 2001).As SNPs are highly conserved throughout evolution and within a population, a map of SNPs serves as an excellent genotypic marker for research.SNPs from the whole genome form a genetic fingerprint.Although SNPs are spaced randomly throughout the genome and could therefore lie in coding sequences, only a small fraction has functional significance (i.e. are non-silent), such as those found in the transcribed or regulatory regions of genes (Mohr et al., 2002).SNPs on a small chromosomal segment tend to be transmitted as a block, forming a haplotype.This correlation between alleles at nearby sites is known as linkage disequilibrium (LD) and enables genotypes at a large number of SNP loci to be predicted from known genotypes at a smaller number of representative SNPs, called tag SNPs or haplotype tag SNPs (Gabriel et al., 2002;Dutt & Beroukhim, 2007).This reduction in the complexity of genetic variation among individuals enables an overall genotype to be determined much more efficiently and economically; roughly 500,000 tag SNPs are sufficient to genotype an individual with European ancestry (Dutt & Beroukhim, 2007;Nicolas et al., 2006).
The mechanisms of an SNP array and the DNA microarray are identical; the convergence of DNA hybridization, fluorescence microscopy and solid surface DNA capture.In order to study the genetic vulnerability of a germline to complex diseases, oligonucleotide arrays have been developed to interrogate such large numbers of SNP markers in multiple databases (Dutt & Beroukhim, 2007;Gunderson et al., 2005).

Genome-wide association studies
CLL and other B-cell lymphoproliferative disorders (LPDs) show clear evidence of familial aggregation, but the inherited basis is still largely unknown.To identify a susceptibility gene for CLL, Sellick et al., (2005) conducted a genome-wide linkage analysis of 115 families, using a high-density SNP array (GeneChip Mapping 10Kv1 Xba, Affymetrix) containing 11,560 markers.Multipoint linkage analyses were undertaken using both nonparametric (model-free) and parametric (model-based) methods.It confirmed that high LD between SNP markers could lead to inflated nonparametric linkage (NLP) and LOD scores (Dawn Tare & Barrett, 2005).After the high-LD SNPs were removed, a maximum NPL of 3.14 (p<0.0008) on chromosome (11)(p11) was obtained.The highest multipoint heterogeneity LOD (HLOD) score under both dominant (HLOD 1.95) and recessive (HLOD 2.78) models was yielded by the same genomic position.Moreover, four other chromosomal positions (5)(q22-23), ( 6)(p22), (10)(q25) and ( 14)(q32) displayed HLOD scores >1.15 (p<0.01).None of those regions coincided with areas of common chromosomal abnormalities frequently observed for CLL.These results support an inherited predisposition to CLL and related B-cell LPDs.Pfeifer at al., (2007) explored high-density 10k and 50k Affymetrix SNP arrays to assess genetic aberrations in the tumour B-cells of patients with CLL.Among the prognostically important aberrations, del(13)(q14) was present in 51%, trisomy 12 (+12) in 13%, del(11)(q22) in 13% and del(17)(p13) in about 6% of cases.A prominent clustering of breakpoints on both sides of the genes MIRN15A/MIRN16-1 indicated the presence of recombination hot spots in the 13q14 region.Patients with a mono-allelic del(13)(q14) had slower lymphocyte growth kinetics than patients with bi-allelic deletions.In four CLL cases with unmutated HV genes, a common minimal 3.5-Mb gain of 2p16 spanning the REL and BCL11A oncogenes was identified, implicating these genes in the pathogenesis of CLL.
New risk variants for CLL were identified by Crowther-Swanepoel et al., (2010).A genomewide association (GWA) study of 299,983 tagging SNPs (by means of HumanCNV370-Duo BeadChips, Illumina) was conducted with validation in four additional series totalling 2,503 cases and 5,789 controls.In 2008, the authors reported the results of a GWA study of CLL based on an analysis of 299,983 tagging SNPs in 505 cases and 1, 438 controls and through fast track analysis of SNPs, identified risk loci at 2q13, 2q37.1, 6p25.3, 11q24.1, 15q23 and 19q13.32 (Di Bernardo et al., 2008).The authors identified 4 new risk loci for CLL at 8q24.21 (rs2456449, TCF4), 2q37.3 (rs757978, FARP2), 15q21.3(rs7169431, NEDD4, RFX7) and 16q24.1 (rs305061, IRF8).The evidence for risk was found for two more loci: 15q25.2(rs783540, CPEB1) and 18q21.1 (rs1036935, CXXC1, MBD1).TCF4 binds to an enhancer for MYC, providing a mechanistic basis for this 8q24.21association.It had also been shown that variation in IRF4 influences CLL risk.There is a possibility that the effect of the other 8q24.21cancer risk loci is by MYC, which is a direct target of IRF4 in activated B-cells and this observation needs further study.
FARP2 is a gene connected with signalling downstream of G protein-coupled receptors.rs757978 is involved in the substitution of threonine for isoleucine at amino acid 260 (T260I), whereas rs305061 maps within a 30-kb region of LD at 16q24.1 locus and localises 19kb telomeric to IRF8, which regulates  and -interferon response.No connection between 17p deletion status and genotype was observed.Although there was evidence that the rs305061 risk genotype was associated with worse overall survival, IGHV-mutation status was highly correlated with rs305061, but risk genotype correlating with unmutated-CLL (Crowther-Swanepoel et al., 2010;Di Bernardo et al., 2008).
To identify genetic variants associated with outcome of CLL, Sellick et al., (2008) genotyped 977 non-synonymous SNPs (nsSNPs) in 755 genes relevant to cancer biology in 425 patients participating in a trial comparing the efficacies of FA and CHB ± C in first-line treatment.A total of 78 SNPs (51 dominantly acting and 27 recessively acting) were associated with progression-free survival (PFS), nine of them also affecting overall survival (OS) at the 5% level.These included SNPs mapping to the immunoregulatory genes IL16 P434S, IL19 S213F, LILRA4 P27L, KLRC4 S29I and CD5 V471A, as well as the DNA response genes POLB P242R and TOPBP1 S730L, which were all independently prognostic of IGHV mutational status.A total of five SNPs associated with PFS were common to patients treated with CHB or FA (DST L22S, LILRA4 P27L, SEC23B H489Q, XRCC2 R188H and ZAK S531L); three SNPs were common to patients treated with either CHB or FA with C (APBB3 C236R, ENPPS I171V, and C21orf57 S2L); and four were common to patients treated with either FA alone or FA with C (DDX27 G206V, DPYD S534N, WNT16 G72R and DHX16 D566G).The variants have proved to be invaluable prognostic markers of patient outcome (Table 6).6. Relationship between SNPs and drug response (Sellic et al., 2008) www.intechopen.comGunnarsson et al., (2008) compared platform dynamics, an in-depth analysis of copynumber alterations (CNAs) using four high-resolution microarray platforms: BAC arrays (32K), oligonucleotide arrays (185K, Agilent) and two SNP arrays (250K, Affymetrix and 317K, Illumina).Ten CLL samples were analysed.The evaluation of baseline variation and copy-number ratio response showed that the Agilent platform performed best and confirmed the robustness of BAC arrays.These platforms demonstrated more platformspecific CNAs.The SNP arrays showed more technical diversity, although the high density of elements compensated for this.Affymetrix detected more CNAs than Illumina, but the latter showed a lower noise level and a higher detection rate in the LOH analysis.Application of high-resolution microarrays will enhance the possibility of detecting new recurrent microevents in CLL leading to identification of new important subgroups, refining the prognostic hierarchy established by FISH.The whole-genome screening with SNP arrays (Affymetrix GeneChip Mapping 250K Nsp1) was conducted and a high frequency of known recurrent alterations in 203 newly diagnosed CLL patients was revealed (Gunnarsson et al., 2010).Moreover, the genome-wide analysis allowed detection of a novel combination of gain of 2p and del(11q), and additional large and small CNAs, which are important for the evaluation of overall complexity in CLL patients.The authors identified genomic complexity as a poor prognostic marker in the survival analysis.However, they noted that this characteristic was strongly linked to established poor-risk molecular markers.The small alterations were mostly non-overlapping.It seems unlikely that there are unknown recurrent CNAs > 200 kbp involved in the CLL pathophysiology detectable in this setting (Gunnarsson et al., 2010).Similar results have been presented by Kujawski et al., (2008), who reported a correlation between genomic complexity and a significantly shorter time to first and second treatment and presented the number of CNAs as an independent prognosis factor.

Copy number variation analyses
The discovery of microRNA and its biological functions is a significant step towards the understanding of the molecular bases of human physiology and pathology.MicroRNAs constitute a class of short, non-coding RNA molecules involved in the regulation of a number of important biological process including cell proliferation, differentiation and apoptosis by down-regulation of gene expression during the translation phase.On the basis of these findings, CLL is a genetic disease in which the main alterations occur in microRNAs (miRNAs).Down-regulation of MIR15A and MIR16 as a part of del(13)(q14) has been suggested as good prognostic factors.Both miRNAs negatively regulate BCL2 at a posttranscriptional level.In CLL cases with unmutated IGHV or high level of expression ZAP70 the overexpression of TCL1 was observed.This is due to low-level expression of MIR29 and MIR181, which directly targets this oncogene.The overexpression of TCL1 is correlated with del(11)(q22) and with the aggressive CLL.These miRNAs might be used to target BCL2 or TCL1 for therapy of the disease (Calin et al., 2007;Cimmino et al., 2005).et al., 2008).Further subdivision of del(13)(q14) type I cases into type Ia and type Ib is suggested by the occurrence of deletions that appear of relatively uniform length [del(13)(q14) type Ia] and that displays centromeric breaks within the vicinity of the MIR15A/MIR16 cluster.Bi-allelic del(13)(q14) type Ia lesions were associated with significant reductions in MIR15A/MIR16 expression levels.As opposed to Calin et al. (2007), this observation reveals that BCL2 levels were not correlated with MIR15A/MIR16 levels.In important recent discovery is that about 50% of all CLL cases with del(13)(q14) do not express the PHLPP gene.PHLPP dephosphorylates activated AKT and low or absent PHLPP expression may allow for sustained AKT signalling after proper cell surface stimuli (Ouilette et al., 2008).
Multiple, discrete, genomic alterations in the 13q region, including MIR15A/MIR16, Rb and others were also observed by Grubor et al., (2009).It might suggest greater complexity of lesions in the 13q region than already known.Moreover, they focused on intraclonal heterogeneity within CLL patients and they searched for genomic differences between CD38 + and CD38 -populations in the same patient.The study was conducted by means of a high-resolution CGH technique called representational oligonucleotide microarray analysis (ROMA).This method is very sensitive to examining the clonal heterogeneity of CLL within the same patient from mixed subpopulations.Copy number differences, in separated CD38 + and CD38 -fractions, were detected in 3 of 4 samples at various loci throughout the genome, some of clinical relevance (ie.ATM and TP53).With the exception of the del(6)(q21), reported major cytogenetic imbalances have been observed previously.The majority of lesions (315/419) were deletions and not amplifications, which is typical of CLL.Two novel regions were observed: del(8)(p21.2-p12) and del(2)(q37.1),including genes TRIM35 and SP100/110/140, respectively.The apparent on-going evolution of CLL clones in a patient may improve the understanding of the disease and the ability to identify patients at risk.The above-demonstrated capabilities offer opportunities for patient treatment individualisation and the identification of new therapeutic agents.Lehmann et al., (2008) performed molecular allelokaryotyping on 56 samples of early stage CLL using the 50k XbaI GeneChip from Affymetrix (50,000 SNP probes).Excluding the four common abnormalities [+12, del(17)(p13), del(11)(q22) and del(13)(q14)], SNP-chip analysis identified a total of 45 copy number changes in 25 CLL samples (45%).Four samples had del(6)(q21) that involved AIM1.UPD was detected in four samples, two of them involved the whole of chromosome 13, resulting in homozygous deletion of MIR15A/MIR16-1.The data suggests that genetic abnormalities including gain, loss and UPD of genetic materials frequently occur at an early stage of CLL.In addition to well-documented common genetic abnormalities, deletions of 5q, 6q and Xp were observed to be frequent in early-stage CLL.AIM1 was examined as a target of this deletion.In the study, expression levels of ZAP70 and the mutational status of IGHV were analysed.It was demonstrated for the first time that ZAP70 expression was correlated with del(11)(q22) in early-stage CLL.It was also observed that non-hypermutation of IGHV was correlated with +12, del(11)(q22) and del(13)(q14) in early-stage CLL.

Conclusion
Microarray technology provides comprehensive data on the expression patterns of thousands of genes in parallel, which positions this method in the centre of optimisation of diagnosis and the classification of leukemias.GEP may lead to the detection of new biologically defined and clinically relevant subtypes of chronic lymphocytic leukemia as a basis for specific therapeutic decision.If such testing is to be used as a routine method for diagnostic purposes in parallel with current standard methods, it is crucial to include GEP in future routine diagnostic applications and in clinical trials.With promising initial results, genome-wide association studies using SNPs are becoming increasingly well established as tools for discovering disease genes.SNP array is an important application in determining disease susceptibility, and consequently in pharmacogenomics, by measuring the specific effectiveness of a form of drug therapy for the patient.As each individual has many SNPs that together create a unique DNA sequence.SNPs may be performed to map disease loci, and hence determine individual-specific disease susceptibility genes.As a result, drugs can be personally designed to act efficiently on a group of individuals who share a common allele, or even a single individual.
up  -and downregulation 

Good-prognosis CLL Alterations in gene expression Gene description Cell cycle and transcription genes
Gene expression: upregulation Table 1.GEP in CLL, which may predict a good prognosis.
There is still no evidence for a direct role of NEDD4 in CLL, but it is a credible candidate gene because it has a role in regulating viral latency and pathogenesis of EBV.Particularly, NEDD4 regulates EBV-LMP2A, which mimics signalling induced by the B-cell receptor, altering B-cell development.CPEB1 plays a role in regulating cyclin B1 during embryonic cell division and differentiation.CXXC1 and MBD1 are involved in gene regulation.MBD1 expression in EBV-transformed lymphocytes correlated with risk genotype.Although MBD1 has no documented role in CLL, it can affect CLL development through translational control of MYC.

Table
Hernando et al., 2004.alleles could differentially affect the biology of CLL casesHernando et al., 2004.In the type I 198 genes were analysed.In this group reduced expression of FLJ11712, KCNRG, RFP2, RFP2OS and DLEU1 was identified.Many other genes have emerged as candidate differentially expressed genes by means of qPCR: LATS2, DFNA5, PHLPP, LPIN1, SERPINE2, ARHGAP20, CYTB5, SLA2, and AQP3.LATS2 RNA levels were lower in CLL cases with del(13)(q14) type I as opposed to type II cases or all other CLL cases without del(13)(q14).LATS2 is involved in cell cycle progression control.It is possible that Rb and LATS2 may be regulators [in nondel(13)(q14) cases] in different processes of CLL subsets (Ouillette cases] and consequently without such a loss as type I [60% of del(13)(q14) cases].Rb is a decisive regulator of cell cycle progression and genomic stability.