Diagnosis, classification and therapy development for human prostate cancer
Introduction
Prostate cancer has remained a mysterious disease in terms of its etiology, biological and genetic basis, natural history, clinical presentation and progression. Substantial controversies exist about issues such as screening, diagnosis and treatment. Research into the molecular basis of prostate cancer has revealed numerous clues to the disease process, but definitive solutions to clinical dilemmas are few in number, and no effective therapies exist for advanced prostate cancer.
We are now becoming more aware of the enormous complexity that underlies the molecular ‘wiring diagram’ of the cancer cells. In the light of this extreme complexity, it is hardly surprising that the few genes, such as the androgen receptor, that have so far been used as targets for therapy development have not yielded a cure. New tools, technologies and resources will be needed to unravel the mysteries of prostate cancer. Such technologies have now started to emerge with the launch of the post-genomics era and biochip technologies.
Now that the human genome sequence is published, we have at our disposal the building blocks of human life. Knowledge of our genes also forms the basis for a rational understanding of the cancer development process. Advances in translational genomics could profoundly impact upon prostate cancer research, early diagnosis, molecular classification and therapy development, and lead us towards personalized medicine.
Genome sequence, however, is only the first frontier, and much additional research will be needed to turn this basic research breakthrough into a deeper understanding of cancer development, and particularly into clinical benefits. The possibilities of basic science do not change medical practice overnight. More likely, this will take a decade for the progress to flow to the clinical arena, and will require active participation of both clinicians and basic scientists. Also, we will need to develop better tools to analyze not only the genome (DNA), but also the transcriptome (RNA) and proteome (proteins).
These data will have to be integrated to develop a comprehensive, global view of the cancer development process and to identify weak points as targets for drug development. In this chapter we review some of the possibilities and challenges that lie ahead as we move from genome sequence to research on functional genomics and proteomics and to molecular diagnostics and treatment tailored against the biological properties of the cancer cells. The focus of this review is describing technologies that may facilitate applied genomics research and clinical implementation.
Genomics: now we know it all, or do we?
A conservative estimate of the number of genes in the human genome is ~35 0001. This is only slightly more than the number of genes in lower organisms, such as fly (Drosophila) or worm (Caenorhabditis elegans). The relatively low number of human genes highlights a number of important features of human cell biology. First, non-coding regulatory sites in the genome are likely to be critical in directing the complex gene expression patterns of human cells.
Second, the process of RNA splicing determines how many proteins will be made from a given gene sequence. A single gene sequence may be transcribed and spliced into multiple different mRNA species, each of which encodes a different protein variant. Third, each protein may then undergo multiple, highly complex post-translational modifications, such as glycosylation, phosphorylation, etc. Therefore, a single gene may encode dozens, if not hundreds, of different proteins, often with differing or even opposing effects.
Fourth, several genes are not encoding proteins, but produce RNA molecules that have a regulatory role. Methods that would enable us comprehensively to survey all this complexity of the transcriptome and proteome do not even exist yet. However, substantial advances in exploring functional genomics of prostate cancer have taken place using cDNA microarray technologies.
Functional genomics: the ‘living genome’
Functional genomics refers to the analysis of gene expression changes throughout the genome using microarrays and other highthroughput tools. The expression levels of all human genes can be quantified simultaneously by hybridizing labelled cDNAs from a tumor sample against a microarray of oligos or cDNA clones representing all human genes. Such gene expression profiles are highly useful for diagnostic efforts. For example, two or more different tumor types can be objectively identified, and subclasses of tumors with a similar histological appearance, but different biological or clinical properties, can be distinguished.
Several applications of the technology in prostate cancer research have already been published. Luo and co-workers studied the differences in gene expression profiles between benign prostatic hyperplasia and prostate cancer. They reported that the patterns of gene expression in these samples were significantly different, facilitating diagnostic classification and understanding of the molecular basis of these diseases, as well as identifying markers for clinical diagnosis and therapy.
We identified characteristic gene expression patterns in human prostate cancer before, during and after androgendeprivation therapy. Characteristic gene expression patterns could be defined for all these states. Two distinct gene expression programs contributed to hormone therapy failure: first, reactivation of androge nresponsive genes, and second, activation of other, progression-associated genes. Finally, gene expression profiling can be used to identify critical target genes for therapy. For example, activation of the EZH gene, a polycomb group protein involved in transcriptional repression, was shown to be causally involved in prostate cancer growth and metastatic progression.
Comparative genomic hybridization microarrays: identifying primary gene targets that may drive cancer progression
The vast amount of data collected by cDNA microarrays raises the question whether these gene expression changes reflect primary genetic alterations driving tumor progression or secondary downstream changes. In leukemias and lymphomas, cytogenetic clues have played a major role in pinpointing critical primary cancer genes and therapeutic targets, whereas only limited progress in similar studies of solid tumors has so far been achieved.
Integration of the ‘functional genomic view’ of the cancer genome with the ‘genetic or chromosomal view’ could lead to the identification of primary genes playing a critical role in the development and progression of solid tumors.
New array-based comparative genomic hybridization (CGH) technologies enable the study of gene copy number changes (amplifications and deletions) throughout the genome virtually at a single gene resolution.
Furthermore, consequences that gene copy number changes have on gene expression patterns can be readily studied. This makes it possible directly to identify genes that are activated by DNA amplifications. As demonstrated for the HER-2 oncogene, which is amplified and overexpressed in breast cancer, such amplification target genes may provide invaluable starting points for anticancer drug development.
We hybridized differentially labelled genomic DNA from prostate cancer cell lines and normal reference to a microarray of cDNA clones (Wolf and colleagues, unpublished data). This made it possible to measure copy number changes of 15 000 genes throughout the genome at a time. Prostate cancer cells often have amplifications of the 10q chromosomal region, indicating that this region may harbor genes that play a role in prostate cancer progression.
High-resolution CGH microarray analysis indicated that there are several regions of amplification at 10q, and pinpointed the critical regions of involvement with virtually single gene precision. Furthermore, analysis of gene
expression changes of all genes along chromosome 10 indicated that the amplifications lead to the activation of numerous genes, some of which could possibly serve as candidate targets for therapy development. CGH microarray analysis may significantly help to narrow down the focus in the search of therapeutically important genes in the
genome.
Cell-based microarrays: functional studies of cancer cells
Cell microarrays or cell chips enable the analysis of consequences of gene up-regulation or down-regulation on cell functions, phenotypes or signalling pathways. Compared with the descriptive molecular profiling methods of genomics and proteomics, cell-based micro-arrays provide information on cause and effect relationships, and are therefore ideally applio able to the validation of targets for novel anticancer therapies. Cell-based arrays can be constructed using a number of methods. An ingenious method for this purpose was described by the Sabatini group at the Whitehead Institute.
Genes in expression vectors are printed as a microarray on a glass slide. Instead of using the array in a hybridization experiment, the microarray slide is placed in a cell culture flask, and living cells are plated to grow on top of the microarray. In a process called reverse transfection, the cDNA clones enter cells and are being translated to a protein, which may in turn have a measurable phenotypic or molecular effect on the cells. The process of reverse transfection can be substantially scaled up to analyze hundreds, if not thousands, of genes.
Similarly, using antisense oligonucleotides or small interfering RNAs, one could specifically turn down the expression of specific genes in the cells and explore the consequences on cell phenotypes. Modulating the expression levels of genes in living cells in a microarray format has powerful implications for exploring the cell signalling
pathways, identifying drug targets and facilitating a ‘systems biology’ approach to the studies of the ‘wiring diagrams’ of cancer cells.
Cell microarrays and other functional assay systems are fundamentally different from regular genomics and proteomics technologies that only enable analyses of gene and protein expression changes as well as correlations and patterns among these molecules. In contrast, cell-based microarrays make it possible to infer cause and effect associations and consequences of gene expression alterations.
Tissue microarrays: clinical translation of genomics and proteomics
Often the molecular targets of interest need to be studied in the context of cancer or normal tissues, integrating morphological information with the molecular level data and expanding the studies to large cohorts of patients, perhaps hundreds or thousands of cases. For such analyses, assay costs may become prohibitively expensive when dozens of targets need to be explored. Furthermore, sufficient quantities of fresh-frozen tissues are often not available for molecular analyses.
Tissue microarray (TMA) technology is based on the arraying of hundreds of tissue specimens at high density on microscope slides7. TMA technology enables rapid visualization of molecular targets in thousands of arrayed tissue specimens at a time, at the DNA, RNA or protein level.
This facilitates rapid translation of molecular discoveries to clinical applications, particularly when large cohorts of patient samples from retrospective tissue archives, tissue banks or clinical trial materials need to be investigated. By revealing the cellular localization, prevalence and clinical significance of candidate target molecules in tissues, TMAs are ideally suitable for genomics-based diagnostic and drug target discovery, validation and prioritization.
TMAs have a number of advantages compared with conventional techniques of analyzing tissue samples. The speed of molecular analysis is increased by up to 100-fold, precious tissues are not destroyed and a large number of different molecular targets can be analyzed from consecutive TMA sections.
The ability to study archival tissue specimens is an important advantage, as such specimens are usually more readily available in existing tumor banks in large quantities, often with associated demographic, pathological, clinical, treatment and follow-up data. Furthermore, such tissues are not at analysis of TMAs can be automated, increasing the throughput further.
There are many applications of TMA technology in cancer research, including:
(1) Identification of the inter- and intracellular molecular target distribution in tumor samples;
(2) Analysis of the frequency of molecular alterations in large tumor materials;
(3) Exploration of tumor progression by including samples from normal tissues, hyperplasia, invasion, metastasis and therapy failure;
(4) Identification of predictive or prognostic factors and validation of newly discovered genes as diagnostic and therapeutic targets.
TMAs provide a high-throughput methodology for microscopic examination of tissue specimens in the post-genome era. Using TMA technology, we showed how amplification of the androgen receptor (AR) gene and overexpression of the insulin-like growth factor-binding protein-2 (IGFBP-2) were common in hormone-refractory end-stage prostate cancers, but infrequent in untreated primary tumors.
These kinds of clinical associations could be rapidly ascertained by constructing a prostate cancer ‘progression TMA’ that included dozens of tissue samples from all stages of prostate cancer development, starting from normal prostate, and progressing through benign prostatic hyperplasia, prostatic intraepithelial neoplasia and localized clinical cancer to metastatic and hormone-refractory end-stage cancer. Perrone and coworkers10 reported differences in tumor proliferation rates between matched prostate cancer cases from Caucasians and African-Americans using TMA technology.
This indicates how demographic, genetic or other etiological risk factors can be rapidly associated with molecular or phenotypic characteristics of cancers. Dhanasekaran and co-workers identified and validated prognostic biomarkers in prostate cancer (hepsin transmembrane protease and a serine-threonine kinase PIM-111 as well as the EZH2 gene).
Conclusions
Progress in genome sequencing has initiated a new wave of research based on the comprehensive molecular profiling of prostate cancers. Genome sequence-based research has led to ‘functional genomic’ and, most recently, ‘proteomic’ research strategies that all produce an overview of the molecular fingerprint of the cancer genome. The challenge for the future is how to turn the enormous quantities of research data obtained using these tools into knowledge of the signalling cascades and molecular wiring diagrams.
This should be highly useful to identify ‘weak links’ in the cellular signal processing network that could be targeted therapeutically. The second enormous challenge is how to translate these data to the clinical setting. The tools, technologies and strategies described in this review of translational cancer genomics research may facilitate this goal. All the various technologies will need to be integrated with sample acquisition and pathological and clinical characterization of the tumors and patients, as well as with the process of clinical trials, to identify the role that this new post-genomic era ‘molecular medicine’ could have on the management of prostate cancer patients in the future.
