Chromosome copy number abnormalities (CNAs) are common in most cancers, with specific regions of amplification or deletion being associated with specific tumour types, stages or outcomes [16–18]. The introduction of metaphase CGH  to study these CNAs has eliminated the requirement to obtain metaphase spreads from the tumour samples, which was often challenging due to technical difficulties in culturing certain tissues. Though metaphase CGH allows for more samples to be examined, it provides a relatively limited resolution and still requires substantial cytogenetic experience to analyse the results. The recent development of array CGH has opened up the field of CGH research and permitted more laboratories to study an ever-increasing number of tumour types and stages. The application of BAC microarrays to analyse human samples is straightforward and their high-throughput nature makes them the method of choice for rapid detection of genetic alterations.
When analysing heterogeneous tumour samples, the gaCGH results obtained from "bulk extracted" DNA are likely to be inaccurate. However, with the use of LCM and MDA we have been to able obtain highly purified HPIN and CaP DNA and thus identify the particular chromosomal changes associated with the two disease stages.
In this study, we have used a 2,400-element BAC microarray with a resolution of ~1 Mb to study CNAs in a set of 15 patient samples comprised of 7 HPIN cases and 8 CaP cases. For the 7 HPIN cases, 41 genomic alterations (20 gains, 21 losses) were identified, in contrast to the 90 genomic alterations (38 gains, 52 losses) seen for the 8 CaP cases. As with other cancers, CaP development and progression is likely to be the outcome of a series of stepwise genetic changes. The accumulation of CNAs, which occurs during this process, although likely due to increased genetic instability, is non-random and may indicate chromosomal regions important in tumourigenesis. It is suggested that failure in the fidelity of homologous recombination within the repetitive sequences, that comprise the kinetochore complex, could lead to recurrent loss of 8p and gain of 8q by rearrangement of chromosome 8-specific alphoid centromeric sequences. Thus, the high fidelity process of homologous recombination can be the major DNA repair pathway, which is indispensable for the maintenance of genetic stability.
Examination of our array results indicates that aberrations involving parts or all of 1p, 6q, 7p, 7q, 8p, 8q, 10q, 13q 16p and 16q are most common, which is concordant with previous metaphase [19, 20] and array CGH [4, 21] results. It is therefore likely that the alterations in copy number are part of a programmed cycle of events that promote tumour development and progression as well having an impact on disease-specific survival.
A comparison of the CNAs present in the HPIN and CaP samples identified a significant increase in copy number for 7q (P<0.01, chi square test) and a significantly increased frequency of loss for 10q (P<0.01, chi square test) and 13q (P<0.0001, chi square test). In genotype/phenotype correlations, gain of chromosome 7q  and loss of 13q  have been associated with advancing tumour stage and aggressiveness, which is in agreement with the results presented here. However, gain of 8q  and loss of 16q  have also been linked to tumour progression, but our data do not show any significant difference for these CNAs in our HPIN and CaP samples. This would suggest that 8q and 16q CNA's are likely to be early events in tumourigenesis. In addition, they may also identify HPIN and CaP samples that are likely to progress.
Apart from the commonly reported CNAs, additional alterations that have been less frequently reported in earlier CGH studies have also been identified. These include gains on 12q (HPIN) and loss of 4q and 10q (HPIN and CaP). Whether the identification of 12q and 4q regions will provide additional insight into CaP progression is not yet clear. Further analysis using a platform such as tissue microarrays, which permits the screening of different disease stages from large patient cohorts, will help better identify their frequency and also the potential use of genomic imbalance in diagnosis of CaP. The value of genomic analysis in CaP was recently demonstrated by the discovery of a high frequency of chromosomal translocations leading to rearrangement of the fusion oncoproteins ERG or ETV1 with TMPRSS2 .
A common problem when analysing archival tissue is the availability of a suitable quality and quantity of RNA to study expression changes. Though RNA amplification techniques  can generate a sufficient quantity, obtaining RNA of the required quality is often challenging. Previous reports have demonstrated a relationship between alterations in chromosomal copy number and alterations in gene expression [28, 29]. The ability to distinguish these regions will point to genes, which may either directly (tumour suppressor gene loss or oncogene gain) or indirectly contribute to tumour development and progression. As a result, CGH can be used as a surrogate for gene identification. Candidate genes, which have previously been implicated in CaP have been highlighted in bold in Table 3. The roles of MYC [30–32], PSCA [33–35], and MDM2 [36–38] have all been well reported and alterations in gene dosage correlate well with their change in expression. However, other candidate genes have been less well studied. For example, EBAG9, whose increased expression in CaP is a negative prognostic indicator, has a potential role in progression by enabling cancer cells to evade the immune response . NBS1, which has been identified as a founder mutation causing an increased susceptibility to prostate cancer , is involved in processing/repair of DNA double strand breaks and in cell cycle checkpoints, thus its deregulation will likely contribute to chromosomal instability. FKHR, which is a member of the FOXO forkhead transcription factor family, is thought to play a regulatory role in several cellular functions including cell proliferation and survival . Loss of FKHR expression, as observed in CaP cell lines, is likely to abrogate this control leading to tumour cell growth. Though these genes have previously been implicated in CaP there are additional genes, including both oncogenes and tumour suppressor genes, which reside within all of the affected regions that may play an important role in the aetiology of the disease. Further analysis of these other candidates may identify their potential as molecular targets for diagnosis and treatment.
Several of the genes present within altered regions of the genome are associated with genetic pathways, indicating that these pathways are likely to be important in prostate tumourigenesis.