Nature Medicine4, Special Web Focus: Breast Cancer (2001)
Published online: 16 April 2001; | doi:10.1038/20000258
Translational breast cancer research in the post-genomic world
Joe W. Gray
Recent progress in human genome research has resulted in efficient technologies for the analysis of gene copy number1,2, gene expression3-5 and methylation status6,7; useful biological reagents such as cDNA and genomic clone libraries8,9; and a nearly complete genome sequence for humans10,11. The use of these technologies, reagents and resources promises to revolutionize our understanding of events that occur during breast cancer initiation and progression, and to lead directly to improved strategies for early detection, prognostication and prediction of response to therapy and to new targets for therapeutic intervention. The challenge is to realize these promises.
Genome profiling studies in the last decade, including analyses of loss of heterozygosity12, measurement of genome copy number change using comparative genomic hybridization2,13,14 and metaphase chromosome analysis15-17, have demonstrated a remarkable spectrum of genomic changes that develop as breast cancers progress. In general, genomic changes occur early-in situ cancers already demonstrate most of the genomic changes found in invasive and metastatic cancers-and differ substantially between clinically similar tumors12,18 (Fig. 1). Breast tumor genomes evolve relatively slowly once they are established. In most cases, recurrent tumors are similar genetically to the original lesions19, and metastatic tumors are similar to the primary tumors from which they arise20. Paradoxically, analyses of individual tumor cells using fluorescence in situ hybridization indicate that the level of chromosomal instability is low before the development of in situ cancer but high thereafter when genomic evolution is slow17,21. Finally, genome profiles show many regions of recurrent abnormality that are associated with clinical behavior or histological features. Of course, mutations, translocations and epigenetic changes such as methylation are also essential in breast cancer development. These events have not yet been comprehensively assessed in breast cancer. However, mutation and/or methylation contribute to inactivation of several essential tumor suppressor genes, including those for p53 (ref. 22), p16 (ref. 23), CDH1 (previously known as E-cadherin; ref. 24) 24, estrogen receptor and breast cancer 1, early onset (BRCA1; ref. 25), and several DNA repair genes26.
Genome copy number abnormalities in clinically similar breast cancers. Genome copy number abnormalities were measured using comparative genomic hybridization (CGH). Ln (CGH ratios) are plotted as a function of distance along the genome with chromosome 1 to the left and X to the right. Vertical lines indicate chromosome boundaries (data courtesy of Dr. K. Chen).
Substantial progress also has been made in the analysis of gene expression. This has been accomplished using differential display27, serial analysis of gene expression28 and microarray-based expression profiling3. These analyses are important because they show the consequences of both genetic and epigenetic changes. Expression analyses so far indicate the same general picture of breast cancer progression as that gained from genomic studies. That is, expression profiles differ substantially between clinically similar tumors but change slowly during tumor progression or therapy. In addition, gene expression profiles allow patients to be stratified into groups that differ substantially in biological characteristics or outcome5,29.
Genome-scale analyses of breast cancers have demonstrated both the promise and the power of the new technologies for exploring the complex molecular nature of breast cancer development. Information that results from these studies should lead to improvement in all aspects of the management of the disease (Fig. 2).
Genome-scale analysis of breast cancer should influence disease management. Schematic diagram suggesting opportunities for improved breast cancer management that may arise from large-scale molecular analyses of disease progression. Studies of early stages of progression may yield information that facilitates identification of genes that influence cancer susceptibility, suggest targets for chemoprevention and/or allow development of markers that enable earlier cancer detection. Analyses of later stages of progression may reveal new therapeutic targets, or allow development of markers that predict response to therapy and/or can be used for sensitive detection of metastatic cells.
Breast cancer progression Analyses of the increase in cancer incidence with age have indicated that only five or six fundamental events are necessary for cancer formation30,31. However, breast tumor genomes and 'transcriptomes' demonstrate many more abnormalities than that. Most are likely to be consequences of tumor progression, not causes, so the challenge is to identify the essential or rate-limiting events. This will require detailed analyses of the interacting molecular events associated with progression in both human tumors. In vitro and rodent models will be essential to substantiate hypotheses developed from studies of human breast cancers.
The mechanisms by which the genomic changes arise in breast cancers and the consequences thereof are fundamentally important to efforts to improve early cancer detection and/or prevention. Studies in vitro and rodent models have indicated that cell cycle checkpoint deregulation, leading to hyperproliferation, and eventual genome instability resulting from loss of telomere function are important early events in tumorigenesis32-34. However, the genes involved in deregulation are not well understood. Large-scale analyses of tissues from stages of progression from normal tissue to carcinoma in situ should demonstrate genes whose expression and/or translation are altered during tumorigenesis.
The mechanisms involved in cancer invasion and metastasis also remain unclear. Events that increase genome instability are one important aspect of this process. Defects that affect specific DNA repair genes, regulatory checkpoints and mitotic apparatus function are important. However, in most cases we do not understand how deregulation of these genes leads to instability. Large-scale profiling techniques should clarify this by providing a more complete assessment of the genes that are deregulated as cells become unstable.
The causes and consequences of genomic and epigenetic abnormalities also remain to be understood. Mapping recurrent abnormalities onto genomic DNA sequence using array-based methods should accelerate identification of the genes involved. Once discovered, candidate genes can be assessed functionally by comparing their DNA sequences with those for well-studied genes in model organisms such as yeast, Drosophila and mouse. Abnormality mapping also should facilitate identification of genome sequence features that influence aberration formation. These may include the location of fragile sites, repeat sequence density and chromatin structure. The genetic proximity of genes that influence tumor behavior may also be important. For example, amplification or deletion of a region may be enhanced if the region contains multiple genes that provide a positive or negative advantage when over- or under-expressed. In contrast, close proximity of genes that confer positive and negative advantages when differentially expressed may inhibit the formation of 'gene dosage' abnormalities in that region35.
Finally, the paradox that tumors with high genome instability evolve slowly in genotype also requires explanation. One possibility is that that most genomically unstable cells are not advantageous to tumor growth or metastasis. Coordinate analyses of single-cell genotype, proliferative rate and apoptotic rate in patient samples and in model systems should help resolve this.
Early detection Molecular differences between normal, premalignant and malignant discovered using genomic and 'proteomic' profiling can be developed into clinical assays for early and accurate breast cancer. There are several classes of molecular differences that prove useful in this, such as proteins that appear on the surface of early malignant cells but not on normal epithelial or stromal cells in the breast. Radiographically dense markers targeted to these proteins may increase the specificity and sensitivity of conventional mammography. Proteins that are secreted by early tumor cells but not present in normal serum are also useful. Mass spectrometry, enzyme-linked immunosorbent assay and antibody arrays seem well suited to detection of these proteins. Finally, protein or nucleic acid changes that occur early in cancer development and that are present on cells obtained by ductal lavage have utility36. Fluorescent in situ hybridization, methylation-specific polymerase chain reaction and immunohistochemistry may be useful for detection of these early changes in single cells.
Prognosis and prediction of response The stratification of patients into groups that differ in clinical outcome is already proving feasible using genome and expression profiling strategies. The weakness of correlative profiling studies, so far, is their limited statistical power. Typically, thousands of events per tumor are assessed in tens to hundreds of tumors. As a result, it is difficult to establish correlations or identify subpopulations with statistical confidence. Future work must focus on the confirmation of 'leads' that come from these large-scale profiling efforts. This may be accomplished by pooling data from many sources or by evaluating the most-informative markers in large populations. Tissue microarrays composed of collections of samples with diameters of less than a millimeter obtained from thousands of tumors will allow such markers to be tested efficiently37.
Cancer susceptibility Mouse model studies have demonstrated many genetic loci that are associated with susceptibility to cancer38. Most confer only a slight cancer risk when inherited singly, but the risk of developing cancer is often increased substantially when several are inherited together. It is likely that many such genes exist in humans, but finding them is challenging. In principle, this can be accomplished by establishing associations with the large set of single nucleotide polymorphisms that have now been discovered. In practice, such efforts are usually frustrated by the difficulty of assembling sufficiently large populations to make these studies statistically meaningful39. A possible way forward is to focus human association studies on regions of recurrent genomic abnormality that are syntenic with susceptibility loci already discovered in mice40.
Therapy The identification of therapeutic targets is of both academic and commercial interest. Genes that differ from normal genes in DNA sequence, genome copy number, methylation status, transcription and translation are attractive candidates. However, the number of potential targets is dauntingly large because of the many genetic and epigenetic changes that occur during breast cancer development. Thus, the challenge is to recognize targets that have a substantial inhibitory or toxic effect when therapeutically altered. Many of these will be identified through a more detailed understanding of the molecular events associated with tumor development. Agents that counter the molecular changes that enable hyperplasia or that reduce the probability of telomerase reactivation may be useful in chemoprevention. The molecular changes that are essential to progression in subsets of breast cancers are obvious targets for therapy. Not so obvious is how to discover the essential events. Some clues may come from the identification of recurrent changes in gene expression and association with adverse clinical behavior and function in cell processes fundamentally associated with cancer such as signaling, regulation of cell proliferation, apoptosis and angiogenesis. However, it is likely that 'rules' for the identification of subsets of tumors that will respond to specific inhibitors will only become apparent only after comprehensive analyses of the biological and molecular responses of genetically and genomically diverse tumors.
Conclusion Comprehensive analyses of genomes, expression patterns and protein profiles in breast tumors, normal epithelium and associated stroma are now possible as a result of the work of the human genome project. The application of these discovery tools will provide much more detailed information about the interacting events involved in cancer genesis and progression. However, making sense of the data that result is a daunting undertaking and will require new ways of visualizing and integrating biological and clinical data. The reward for success will be substantial improvements in cancer prevention, detection and treatment.
REFERENCES
Albertson, D.G. et al. Quantitative mapping of amplicon structure by array CGH identifies CYP24 as a candidate oncogene. Nat Genet25, 144-6. (2000). | PubMed |
Pinkel, D. et al. High resolution analysis of DNA copy number variation using comparative genomic hybridization to microarrays. Nat Genet20, 207-11. (1998). | PubMed |
Brown, P.O. & Botstein, D. Exploring the new world of the genome with DNA microarrays. Nat Genet21, 33-7. (1999). | PubMed |
Diehn, M., Alizadeh, A.A. & Brown, P.O. Examining the living genome in health and disease with DNA microarrays. Jama283, 2298-9. (2000). | PubMed |
Perou, C.M. et al. Molecular portraits of human breast tumours. Nature406, 747-52. (2000). PubMed
Smiraglia, D.J. et al. A new tool for the rapid cloning of amplified and hypermethylated human DNA sequences from restriction landmark genome scanning gels. Genomics58, 254-62. (1999). | PubMed |
Costello, J.F. et al. Aberrant CpG-island methylation has non-random and tumour-type-specific patterns. Nat Genet24, 132-8. (2000). | PubMed |
Osoegawa, K. et al. A Bacterial Artificial Chromosome Library for Sequencing the Complete Human Genome. Genome Res11, 483-496. (2001).
Osoegawa, K. et al. Bacterial artificial chromosome libraries for mouse sequencing and functional analysis. Genome Res10, 116-28. (2000).
Venter, J.C. et al. The Sequence of the Human Genome. Science291, 1304-1351. (2001). | PubMed
Lander, E.S. et al. Initial sequencing and analysis of the human genome. International Human Genome Sequencing Consortium. Nature409, 860-921. (2001). | PubMed |
O'Connell, P. et al. Analysis of loss of heterozygosity in 399 premalignant breast lesions at 15 genetic loci. J Natl Cancer Inst90, 697-703. (1998). | PubMed |
Kallioniemi, O.P. et al. Comparative genomic hybridization: a rapid new method for detecting and mapping DNA amplification in tumors. Semin Cancer Biol4, 41-6. (1993).
Gray, J.W. et al. Molecular cytogenetics of human breast cancer. Cold Spring Harb Symp Quant Biol59, 645-52 (1994).
Schrock, E. et al. Multicolor spectral karyotyping of human chromosomes. Science273, 494-7. (1996). | PubMed |
Macville, M. et al. Spectral karyotyping, a 24-colour FISH technique for the identification of chromosomal rearrangements. Histochem Cell Biol108, 299-305. (1997). | PubMed |
Pandis, N. et al. Chromosome analysis of 97 primary breast carcinomas: identification of eight karyotypic subgroups. Genes Chromosomes Cancer12, 173-85. (1995). | PubMed |
Gray, J.W. & Collins, C. Genome changes and gene expression in human solid tumors. Carcinogenesis21, 443-52. (2000). | PubMed |
Waldman, F.M. et al. Chromosomal alterations in ductal carcinomas in situ and their in situ recurrences. J Natl Cancer Inst92, 313-20. (2000). | PubMed |
Nishizaki, T. et al. Genetic alterations in primary breast cancers and their metastases: direct comparison using modified comparative genomic hybridization. Genes Chromosomes Cancer19, 267-72. (1997). | PubMed |
Burbano, R.R. et al. Cytogenetics of epithelial hyperplasias of the human breast. Cancer Genet Cytogenet119, 62-6. (2000). | PubMed |
Nayak, B.K. & Das, B.R. Mutation and methylation status of p53 gene promoter in human breast tumours. Tumour Biol20, 341-6. (1999). | PubMed |
Foster, S.A., Wong, D.J., Barrett, M.T. & Galloway, D.A. Inactivation of p16 in human mammary epithelial cells by CpG island methylation. Mol Cell Biol18, 1793-801. (1998). | PubMed |
Hiraguri, S. et al. Mechanisms of inactivation of E-cadherin in breast cancer cell lines. Cancer Res58, 1972-7. (1998). | PubMed |
Fearon, E.R. BRCA1 and E-cadherin promoter hypermethylation and gene inactivation in cancer-association or mechanism? J Natl Cancer Inst92, 515-7. (2000). | PubMed |
Esteller, M. Epigenetic lesions causing genetic lesions in human cancer. promoter hypermethylation of DNA repair genes. Eur J Cancer36, 2294-300. (2000). | PubMed |
Lisitsyn, N.A. et al. Comparative genomic analysis of tumors: detection of DNA losses and amplification. Proc Natl Acad Sci U S A92, 151-5. (1995).
Nacht, M. et al. Combining serial analysis of gene expression and array technologies to identify genes differentially expressed in breast cancer. Cancer Res59, 5464-70. (1999). | PubMed |
Perou, C.M. et al. Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A96, 9212-7. (1999). | PubMed |
Pierce, D.A. & Mendelsohn, M.L. A model for radiation-related cancer suggested by atomic bomb survivor data. Radiat Res152, 642-54. (1999). | PubMed |
Day, N.E. The Armitage-Doll multistage model of carcinogenesis. Stat Med9, 677-9. (1990). | PubMed |
Li, S. et al. Functional link of BRCA1 and ataxia telangiectasia gene product in DNA damage response. Nature406, 210-5. (2000). | PubMed |
Romanov, S. et al. Normal human mammary epithelial cells spontaneously escape senescence and acquire genomic changes. Nature409, 633-637 (2001). | PubMed |
DePinho, R.A. The age of cancer. Nature408, 248-54. (2000). | PubMed |
Collins, C. et al. Comprehensive genome sequence analysis of a breast cancer amplicon. Genome Research (In press)(2001).
Love, S.M. & Barsky, S.H. Breast-duct endoscopy to study stages of cancerous breast disease. Lancet348, 997-9. (1996). | PubMed |
Kononen, J. et al. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med4, 844-7. (1998). | PubMed |
Balmain, A. & Nagase, H. Cancer resistance genes in mice: models for the study of tumour modifiers. Trends Genet14, 139-44. (1998). | PubMed |
McCarthy, J.J. & Hilfiker, R. The use of single-nucleotide polymorphism maps in pharmacogenomics. Nat Biotechnol18, 505-8. (2000). | PubMed |
Balmain, A. & Harris, C.C. Carcinogenesis in mouse and human cells: parallels and paradoxes. Carcinogenesis21, 371-7. (2000). | PubMed |
Acknowledgment This work was supported by National Cancer Institute Grant CA58207.
Massachusetts General Hospital Cancer Center
and Harvard Medical School
Boston, Massachusetts 02114
