GERM-LINE MUTATIONS AND CANCER RISK

With the many recent technical advances that make it easier to locate and identify genes and mutations have come increased efforts in the search for disease-causing genes. The current initiatives to sequence the human genome will no doubt further accelerate the study of familial susceptibility to all diseases, including cancer. The ultimate goal of this research is to identify individuals with an increased risk for cancer, who can then take action before cancer develops. Currently, that action primarily entails increased screening and surveillance, but ideally, in the future it should also include preventive strategies.

Traditionally, genes that confer a predisposition for cancer have initially been identified through standard epidemiological studies designed to detect familial clustering of specific cancers. Once family clusters have been identified, a process called “linkage analysis” is used to pinpoint the locations of the mutant genes. Markers throughout the genome are examined for coinheritance with the mutant phenotype because segments of DNA that are located close to a marker on the same chromosome will be inherited with the marker.

BRCA1 and BRCA2

Two examples of breast cancer genes originally identified in this way are BRCA1 and BRCA2. The search for these breast cancer susceptibility genes began more than 20 years ago (King et al., 1980). By the early 1990s, powerful new tools in molecular biology had been developed, and in 1994, positional cloning was used to identify BRCA1 on chromosome 17 (Miki et al., 1994). Mutations in this gene are now believed to account for 30 to 45 percent of the familial breast cancer cases and nearly 90 percent of the cases in families with high incidences of both breast and ovarian cancer (Easton et al., 1993; Ford et al., 1994). Because breast cancer in a significant number of families with high incidences of breast cancer appeared to be linked to genes other than BRCA1, the search continued for an additional breast cancer gene. In 1995, BRCA2 was identified on chromosome 13 by focusing on families with a high incidence of breast cancer in both male and female members (Wooster et al., 1995). Mutations in BRCA2 are thought to account for breast cancer in about 35 percent of families with a high incidence of early-onset breast cancer (Tavtigian et al., 1996).

Although there is no significant homology between the two large genes, the proteins that they encode may have some related activities in cells. Both appear to be multifunctional proteins that have been hypothesized to play a role in DNA repair pathways, cell proliferation, and transcriptional regulation (reviewed by Cortez et al., 1999 and Welcsh et al., 2000). Why germ-line mutations in these genes lead to breast cancer (or ovarian cancer) more frequently than other tumor types remains largely a mystery. However, knowledge of the gene sequences has allowed the development of genetic tests that may aid in determining a woman's risk for breast cancer.

A variety of tests for BRCA1 and BRCA2 mutations are now commercially available. The genes can be examined for specific mutations or sequenced in their entirety. Most of the currently available tests are labor intensive, but a recent technological advance known as “DNA microarrays” could potentially allow faster high-throughput analysis of samples. Microarrays, which first emerged in the mid-1990s, consist of thousands of different oligonucleotides spotted onto specific locations on glass microscope slides or silicon chips, which are then hybridized with labeled sample DNA (Figure 3-1). High-density arrays with more than 95,000 oligonucleotides have been used experimentally to identify mutations in exon 11 of BRCA1 (Hacia et al., 1996). However, the sensitivities and specificities of the various tests have not been fully determined, and thus, their clinical utility remains uncertain.

FIGURE 3-1

Example of a DNA array.

The decision as to whether a women should be tested for BRCA mutations is often made on the basis of the calculated risk that a family may be carrying a mutation. That risk is approximated by assessing the family history of incidence and age of onset, as well as the family's ethnic derivation (Parmigiani et al., 1998). However, breast cancer is a relatively common disease, so clustering could occur by random chance alone. Furthermore, there is significant heterogeneity among the mutations in these large genes that can predispose individuals to breast cancer, and each of the available tests has limitations in the types of mutations that it can reliably detect. Thus, selection of a particular test and interpretation of the results of that test can be difficult unless an affected relative has already been shown to carry a specific mutation.

Interpretation of negative results is further complicated by the fact that germ-line mutations in other genes may also confer an increased risk for breast cancer (see below). Environmental exposures could also play a role in familial clusters of breast cancer. Thus, a negative result may not be very meaningful to a woman with a strong family history of breast cancer, and it does not necessarily change her individual risk calculated before the test.

A positive test result carries many ramifications for the woman and her family. The ethical, legal, and psychosocial issues surrounding tests for genetic susceptibility are great. Women who carry BRCA mutations must deal with the psychological stress of knowing that they are more likely than women in the general population to develop breast cancer and that they could pass on this susceptibility to their children. Family and other personal relationships can be disrupted, and a woman could potentially face insurance or employment discrimination. A number of state and federal laws restrict some uses of genetic information,¹ but more could be done to ensure the privacy of the information and protection from discrimination. The National Action Plan on Breast Cancer, the National Human Genome Research Institute (HGRI), and others have developed recommendations for further restrictions on the use of genetic information (Department of Labor et al., 1998; Koenig et al., 1998; Rothenberg et al., 1997; Shalala, 1997). These issues are also of great importance to the continuing research on BRCA and other mutations (Fuller et al., 1999).

Genetic testing has not yet been shown to have an impact on breast cancer incidence or mortality, and unfortunately, there are relatively few data to guide a woman's plan of action once she has been identified as carrying a breast cancer susceptibility gene. The possibilities range from participating in standard screening programs, to enrolling in clinical trials for chemoprevention or alternate screening technologies, to bilateral prophylactic mastectomy (Hartman et al., 2000), but none of these interventions has yet been definitively proven to be of benefit among BRCA mutation carriers.

Because of the uncertainties in interpreting negative results, as well as the implications of positive results, genetic testing for breast cancer susceptibility should be accompanied by genetic counseling before and after the test. However, no standard of care for such counseling exists. HGRI includes a branch known as ELSI (ethical, legal, and social issues) that is conducting clinical studies to determine the best approach for the counseling and education of women undergoing genetic testing. Professional societies, such as the American Society for Clinical Oncology, as well as other organizations like the Breast Cancer Working Group of the Stanford Program in Genomics, Ethics, and Society, have also produced guidelines and recommendations on counseling and genetic testing (American Society of Clinical Oncology, 1996; Koenig et al., 1998).

Such counseling and education are essential for informed consent before the test. Women need to understand the potential risks and benefits of undergoing testing, as well as the accuracy, efficacy, and limitations of the test. Different mutations could show various degrees of penetrance (the likelihood that affected individuals will develop cancer), which could potentially be further modified by other factors in the genetic background or the environment (E.L. Harris, 1999). The estimated cumulative risk of breast cancer by age 70 in the very high risk families originally studied during the search for BRCA1 was 80 percent (Easton et al., 1993), but some subsequent population-based studies have shown a lower cumulative risk. The risk for the Ashkenazi Jewish population, for example, is estimated to be 56 percent (Struewing et al., 1997).

The tests are offered primarily as a clinical laboratory service by Myriad Genetics Inc., which holds U.S. patents on both BRCA1 and BRCA2. The tests are not subject to Food and Drug Administration (FDA) regulation, and thus, the clinical validity and utility did not have to be documented before entry into the market. Rather, the quality of laboratories that provide genetic testing as a service is regulated under the Clinical Laboratory Improvement Amendments (CLIA) of 1988. CLIA requires the laboratories to demonstrate only that the tests can accurately and reliably measure the analytes that they are designed to assay (Holtzman, 1999). As a result, some advocacy groups such as the National Breast Cancer Coalition and the Alliance of Genetic Support Groups have proposed that the test should be made available only in a research setting. In response to such concerns, the Secretary's Advisory Committee on Genetic Testing (SACGT) was chartered in 1998 to advise the U.S. Department of Health and Human Services (DHHS) on the medical, scientific, ethical, legal, and social issues raised by the development and use of genetic tests. SACGT has recommended FDA review of all genetic tests, with particular attention to tests used for predictive purposes for diseases without an effective intervention. DHHS action on these recommendations is pending.

Myriad Genetics Inc. has recently launched testing services in Canada, Japan, Ireland, and the United Kingdom through exclusive licenses with laboratory service companies in those countries (Cancer Letter, March 2000). In the United States, Myriad recently signed a multiyear agreement with several large medical insurers to include the tests for BRCA in its list of covered services for its members (Cancer Letter, February 2000). A number of other major insurers and health care management organizations have taken similar actions. These national and international developments will make the test accessible to many more women, but coverage for counseling and follow-up care does not always accompany coverage for the tests.

Because of the exclusivity of Myriad's licensing agreements and the high cost of testing for BRCA, there are concerns that enforcement of Myriad's exclusive licensing agreements will limit access of consumers to the clinical testing provided by other laboratories (Reynolds, 2000). Separate concerns were also raised about the limitations of exclusive licensing with regard to publicly funded research. In response to this concern, some universities have obtained licenses from Myriad to conduct limited testing for research purposes. The company recently agreed to provide testing at a reduced fee to scientists at the National Institutes of Health, as long as the test is performed only for research purposes (Cancer Letter, February 2000). Proponents of the agreement are optimistic that this agreement will lead to increased research activity on the BRCA genes that will generate clinically useful information.

POLYMORPHISMS AND CANCER RISK

Much of the genetic variation among human populations is due to subtle DNA alterations that are shared by many people. Known as polymorphisms, these subtle differences can result in altered protein expression or changes in protein activity that may affect susceptibility to carcinogens and cancer promoters in the environment and that may contribute to the variability in individual responses to treatment. A major goal in studying polymorphisms in women with breast cancer is to more accurately predict which individuals are likely to develop breast cancer or to die from the disease.

Polymorphic sites in many genes have been studied to determine whether they are associated with an increased risk for breast cancer, and a number have been reported to confer elevated risk. A comprehensive analysis of all published studies found four polymorphic sites (in the CYP19, GSTP1, TP53, and GSTM1 genes) associated with a higher risk for breast cancer (Dunning et al., 1999). However, the investigators noted that there was insufficient statistical power to accurately determine the risk for some sites examined, and there are many more genes and polymorphisms that have yet to be studied. One recent study examined 10 polymorphic sites in the estrogen receptor gene but did not find any association between the polymorphisms and breast cancer (Schubert et al., 1999). Studies are in progress to identify and characterize additional susceptibility alleles, but precise estimation of the risks associated with genetic polymorphisms, as well as investigation of more complex risks arising from gene-gene and gene-environment interactions, will require studies much larger than those undertaken to date. New DNA microarraybased methods may speed the search for relevant polymorphisms in complex diseases like breast cancer by facilitating high-throughput analysis of many genes simultaneously (Hacia et al., 1999).

Polymorphisms that involve single-base-pair differences are called “single nucleotide polymorphisms” (SNPs). Many SNPs, perhaps the majority, do not themselves change protein expression or cause disease, but they may be closely linked on the chromosome to deleterious mutations. Because of this proximity, SNPs may be shared among groups of people with unknown disease-associated mutations and serve as markers for such mutations. Such markers may aid in the identification of the mutations and thus could contribute to the understanding of the molecular changes in diseases such as cancer.

The SNP² Consortium Ltd., a nonprofit entity consisting of several major pharmaceutical and technology companies and one large scientific trust, has taken on the challenge of identifying 300,000 SNPs and mapping at least 150,000 SNPs evenly distributed throughout the genome. The project started in the spring of 1999 and is anticipated to continue until the end of 2001. The data generated by this effort are being collected in a database that is freely available to scientists, with liberal licensing provisions for investigators. SNPs will also be deposited in a public database, dbSNP. This database, designed to serve as a central repository for both single-base nucleotide substitutions and short deletion and insertion polymorphisms, was established by the National Center for Biotechnology Information in collaboration with the National Human Genome Research Institute. The SNP Consortium is funding studies to determine the frequencies of at least 60,000 SNP alleles identified through the consortium's research. SNPs that occur in at least 20 percent of a major population (e.g., Caucasians, Asians, or African-Americans) are considered sufficiently common to be useful as genetic markers in the genome.

Recently, another collaboration between Celera Genomics and City of Hope Cancer Center was announced. The two organizations plan to specifically investigate associations between genetic polymorphisms and breast cancer (Cancer Letter, March 2000). In this case, all intellectual property developed through the collaboration will be jointly owned by the two organizations.

SOMATIC CHANGES IN BREAST CANCER

Initiation of sporadic (nonhereditary) cancers and the progression of all cancers occur via accumulation of changes in individual cells within the body. These changes ultimately lead to altered gene expression in those cells and may take many forms, including a variety of genetic alterations in the cell's DNA sequences (i.e., mutations) or epigenetic alterations that leave the DNA sequence intact but that nonetheless modify gene expression. Although many such changes have been observed in breast tumors, the functional relationship between the affected genes and cancer growth is still largely unknown. Indeed, the majority of breast cancers contain so many molecular changes that it is difficult to distinguish between those that are critical for tumor initiation and progression and those that are simply a product of cancer-associated genomic instability. Identification of the critical common events in breast carcinogenesis will therefore be essential for advancing the understanding of breast cancer biology and its etiology and for attaining the ultimate goals of improving means of detection and the ability to establish a prognosis. Most of the techniques described in this section are being used as research tools in studies with biopsy tissues, with the hope that the knowledge gained from this research will eventually be used to more accurately diagnose breast cancer and predict outcomes.

SOMATIC GENETIC ALTERATIONS

Somatic alterations in cancer cells include genetic changes such as amplification and deletion of DNA sequences, chromosomal rearrangements, and base change mutations. DNA amplification can affect any stretch of DNA, from a single gene (microduplications) to an entire chromosome (aneuploidy). Amplification can result in an increased level of expression of the affected gene(s) and thus is one of the major molecular mechanisms through which the oncogenic potential of proto-oncogenes is activated during tumorigenesis. DNA deletions, on the other hand, result in the loss of genes and the associated gene products. Tumor suppressors are often inactivated in cancer through deletions or insertions. In many cases, base change mutations that alter or inactivate protein function are found on one allele, and the second allele is lost via deletion, a mechanism known as “loss of heterozygosity” (LOH).

A technique known as “fluorescent in situ hybridization” (FISH) can detect common aneuploidies and chromosomal loss or rearrangements as well as microduplications and deletions (Mark et al., 1997). This technology relies on hybridization of chromosomes with labeled DNA probes that are specific for genes or chromosomes. Another related technology is comparative genomic hybridization. In this case, DNA from normal and cancerous cells is labeled with differently colored fluorescent tags that are then simultaneously hybridized to metaphase spreads of normal chromosomes. A gain or loss of chromosomal regions can then be identified by the color of the chromosomes. Such current applications could be adapted for use in the clinical cytogenetic laboratory if they prove to be useful as a means of identifying markers for diagnosis or prognosis. New sequence information derived from efforts to sequence the entire human genome, combined with new high-throughput technologies such as DNA microarrays similar to those designed to detect germ-line mutations and polymorphisms, may also make it easier to detect small somatic mutations in tumors in the future (Pollack et al., 1999; H. Yan et al., 2000).

LOH studies generally depend on polymerase chain reaction (PCR)-based methods that analyze polymorphic microsatellite loci³ as markers for DNA loss. Many studies have found significant rates of loss of heterozygosity at dozens of genetic loci in individuals with premalignant disease and early breast cancer (Table 3-2), as well as later-stage cancers, but to date it is not yet clear whether specific LOH events are associated with progression to invasive or metastatic cancer (Allred and Moshin, 2000).

TABLE 3-2

General Chromosomal Locations of Allelic Imbalances (Gains and Losses) in Premalignant Breast Lesions from Studies Assessing Loss of Heterozygosity and Comparative Genomic Hybridization Illustrating Their Tremendous Biological Complexity.

RNA Expression

Scientists have long sought to directly characterize gene expression in tumors compared with that in normal tissues and to correlate those differences with disease outcome or treatment response. However, that effort has been limited by a number of technical factors, including lack of gene sequence data and high-throughput technologies, inadequate access to specimen banks with appropriate patient information, and interpretive difficulties due to tumor heterogeneity. New genomic tools, such as complementary DNA (cDNA) microarrays, may offer the opportunity to make new advances.

Gene expression at the messenger RNA (mRNA) level has traditionally been examined by laborious methods such as Northern analysis (RNA-DNA hybridization) or, more recently, by PCR-based methods such as reverse transcription (RT)-PCR, in which only a small number of genes can be examined at one time. DNA microarray technology, in contrast, enables researchers to look at the expression of thousands of genes at once and obtain a tumor “signature.” In this case, the microarrays consist of cDNA clones corresponding to different genes. The microarrays are hybridized with differentially labeled cDNA populations made from the mRNAs of the samples to be compared (Figure 3-2). The primary data collected are ratios of label intensity, which are representative of the concentrations of mRNA molecules in each sample. Computer algorithms must then be used to identify differences in gene expression between the samples, as well as “clusters” of gene expression (Eisen et al., 1998). Designing the appropriate algorithms to make sense of all the data generated may in fact be the biggest challenge for this technology.

FIGURE 3-2

Example of a cDNA expression array. SOURCE: Adapted from CLONTECH Laboratories, Inc., 1999.

Breast tumor samples were recently separated into at least two categories on the basis of gene expression clusters (Perou et al., 1999). The investigators also identified expression clusters associated with some of the normal cell types that infiltrate tumors, such as lymphocytes and stromal cells, suggesting that one component of tumor cell heterogeneity could potentially be accounted for by using this technology. However, newer methods of isolating small populations of cells from a tumor sample, such as laser capture microdissection (Emmert-Buck et al., 1996), may improve the accuracy of the technology even more. Such techniques may also facilitate examination of normal breast epithelial tissue and earlier-stage cancers, including ductal carcinoma in situ (DCIS). Thus far, most studies have been done with relatively large tumors (generally greater than 2 centimeters) by using tissue taken from excisional biopsy specimens.

One goal of microarray analysis is to classify tumors on the basis of their complete gene expression patterns. For example, if shared gene expression patterns in breast tumors can be used to establish a prognosis or predict the treatment response with greater accuracy, they will yield classifications directly coupled to treatment and outcome. Much more research is needed before those goals can be attained, but a recent study that used cDNA microarrays did identify two molecularly distinct forms of lymphoma that could not be distinguished by traditional classification techniques. Remarkably, the patient groups with the two subtypes of cancer had significantly different survival times, suggesting that molecular classification could potentially be useful in the future for determining prognosis and the appropriate treatment regimen for this type of cancer, as well as others (Alizadeh et al., 2000). A similar attempt has been made to classify breast tumors at the molecular level. Using microarrays to examine differences in mRNA expression patterns, breast tumors could be classified into subtypes that related to physiological variation, but it is not yet known whether different subtypes are associated with different clinical outcomes or response to therapy (Perou et al., 2000).

The relative levels of mRNA species can be regulated at the stage of gene transcription or RNA degradation. A third stage of regulation, that of pre-mRNA splicing, can also produce variations in the resultant protein sequence and function. This is the stage at which RNA is processed after being produced on the DNA template but before it is exported from the nucleus to be translated into protein. Modulation of the cellular machinery responsible for removing intronic sequences and joining the exons into a readable mRNA transcript can lead to alternative splicing events that produce mRNA “splice variants.” Changes in splicing efficiency have been associated with malignant transformation and metastasis, suggesting that splice variants may be useful as tumor markers (reviewed by Cooper and Mattox, 1997). Perhaps the best-studied example of this phenomenon is the cell-surface-adhesion molecule CD44. Abnormal splice variants of this gene have been found in a variety of cancers, including breast cancer, and their presence has been correlated with metastatic potential (Cooper and Mattox, 1997; Kinoshita et al., 1999; Martin et al., 1997; Matsumura and Tarin, 1992). Many splice variants of the estrogen receptor have also been identified in breast cancer and have been hypothesized to play a role in resistance to anti-estrogen therapy (Tonetti and Jordan, 1997). Aberrant splicing can also result from mutations in the sequences at intron/exon junctions. For example, many of the mutations identified to date in the gene encoding the cell adhesion molecule E-cadherin are splice site mutations (Berx et al., 1998).

The variant mRNA species are currently detected primarily by RTPCR (Matsumura and Tarin, 1992), but if cDNA arrays were designed to include sequences specific for different splice variants, this component of variation in tumor gene expression could also be assessed using the high-throughput technology. In many cases, the resultant protein variants can also be detected in tumor tissue or serum (Kinoshita et al., 1999; Martin et al., 1997), as described below.

Protein Expression and Function

Knowledge of the RNA expression patterns of cells is not sufficient for determination of cellular behavior. The RNA expression level is not necessarily indicative of protein levels, because protein expression can be modulated at various stages, from the regulation of mRNA translation into protein to the targeting of the protein for degradation pathways. Traditionally, protein levels in tumors have been evaluated by methods such as immunohistochemistry with tissue sections (Figure 3-3). A plethora of proteins in breast tumors has been examined by this approach in the search for prognostic markers. The most commonly used protein markers to date include the estrogen and progesterone receptors and, more recently, the erbB2 receptor, all of which may be considered in the decision-making process for therapy. Although a number of proteins have shown some correlation with breast cancer progression, it has become clear that the identification of a single marker that can accurately predict disease progression is unlikely. In fact, very few tumor markers have been recommended as part of routine clinical care because it is quite difficult to determine the clinical utility of markers. For this reason, inves tigators developed a system for assessing the use of tumor markers (Hayes et al., 1996, 1998). The Tumor Marker Utility Grading System established an investigational agenda for the evaluation of tumor markers that is analogous to the system used to evaluate new therapeutic agents, which is quite standardized.

FIGURE 3-3

Representative examples of prognostic and predictive factors that are commonly assessed in breast cancers by immunohistochemistry, including estrogen receptor (ER), progesterone receptor (PgR), membrane overexpression of the erb B-2 oncoprotein, nuclear (more...)

Consideration of the functional state of the proteins could add another level of complexity to protein analysis. Protein function can be regulated on many different levels, including through biochemical modifications such as phosphorylation and glycosylation (addition of phosphate or sugars) and through associations with other proteins. In fact, protein-mediated signal transduction pathways that initiate such critical activities as cell division, cell death, or cell movement can in many cases be activated without the synthesis of new proteins. Thus, a complete understanding of the molecular changes in cancer may require a functional analysis of pathways and circuits in cells and tissues, known as proteomics.

The term “proteome” was first coined in 1994 to refer to all proteins expressed by a genome. Traditionally, such protein analysis has required the labor-intensive method of two-dimensional gel electrophoresis, in which proteins are separated by size in one direction and electrical charge in the second direction. Each protein species migrates to a reproducible spot on the gel, and the proteins at these spots can be isolated and sequenced for identification. However, this method requires large amounts of protein and thus has been limited to cultured cells or homogenized tissues that contain a variety of cell types. Recent technological advances, including laser capture microdissection, new methods for cell sorting and mass spectroscopy⁴ and improved bioinformatics may soon allow high-throughput analysis of the specific cell populations within tissues and tumors (Liotta and Petricoin, 2000). For example, one recent study identified a number of differences in the protein profiles of two different cell types in normal breast tissue (Page et al., 1999). Other recent technical advances suggest that the creation of protein arrays (the protein equivalent of DNA arrays) may also soon be feasible (Macbeath and Schreiber, 2000; Service, 2000).

GROWTH OF BREAST CELLS IN CULTURE

The biology of mammary gland development and tumorigenesis has historically been studied with rodent model systems (Amundadottir et al., 1996; Russo and Russo, 1996; Welsch, 1987). However, it has been shown that profound differences in the development and transformation of mammary tissue exist between rodent species and humans (Russo and Russo, 1987). It would therefore be advantageous to study normal breast development and function, as well as breast cancer etiology, progression, and treatment, with human mammary cells. A major goal in this area of research is the isolation of specific cell lines derived from a variety of relevant human tissue types, including normal breast epithelium, atypical hyperplasia DCIS, early and late stages of breast cancer, and histologically normal tissue adjacent to the tumor. Such cell lines would allow investigation into the physiological, morphological, and genetic changes that occur during the process of breast tumorigenesis, as well as during subsequent tumor progression. One example of a model human cell line that can be used to study the evolution of breast cancer from premalignant proliferative breast disease is MCF10AT (Dawson et al., 1996; Heppner et al., 1999). The cell line was derived from tissue taken from a patient with hyperplastic growth of the breast epithelium.

Unfortunately, though, primary cultures of human breast epithelial tissue have been notoriously difficult to grow (Bergstraesser and Weitzman, 1993; Smith et al., 1981). A review of the literature suggests that the success rate for the establishment of cell cultures from breast tumors is no more than 10 to 15 percent (Engel and Young, 1978). Thus, much research has been done with a small number of established tumor cell lines that have the limitations of being clonally evolved and generally available only at high passage numbers (Engel and Young, 1978). Furthermore, most of these cell lines were derived from pleural metastases of patients who had been heavily treated with chemotherapuetic drugs. As a result, the cell lines probably represent a very small subfraction of breast tumor cell types. Although valuable information has been gained from these cell lines, much effort has been made to develop more physiologically relevant models.

To date, the success of human mammary epithelial cell (HMEC) culture has been hindered by limited knowledge of the specific factors that are required for the maintenance of epithelial cell function, growth, and differentiation. Although defined media that contain many of the known factors such as steroid hormones and a variety of peptide growth factors have been developed for HMEC culture, none to date have supported the growth of HMECs in primary culture for more than 2 or 3 weeks (Taylor-Papadimitriou and Stampfer, 1992). Furthermore, breast tissue contains several different cell types (stromal fibroblasts, the luminal epithelium lining the ducts and lobules, myoepithelial cells, and adipocytes) whose interaction and communication may be vital for cell growth, survival, and function. New methods for separation of the various cell types from breast tissue have been developed, but thus far, the purified cell populations have been grown in culture only for very short periods of time (Clarke et al., 1994; Monaghan et al., 1995).

Recent advances in tissue engineering, due largely to the commercial availability of extracellular matrix substrata (Hall et al., 1982; Petersen et al., 1992) and an awareness of concepts related to cell communication, have led to the development of a novel cell culture system with the ability to grow primary cell cultures from most breast tissues, both normal and neoplastic (U.S. patent no. 6,074,874).⁵ The cultures of normal breast tissue, which have mixed cell morphologies, are long lived as primary cultures and grow and differentiate into organotypic architectures that persist for at least 3 to 4 months. The cultures progress and differentiate from three-dimensional domes or “mammospheres” to de novo luminal branching ducts. Tumor cells under the same culture conditions do not form an epithelial architecture but show a more chaotic behavior.⁶ Often, tumor cells manifest autonomous, single-cell behavior and seem to avoid contact with one another. In addition, significant variability among similarly staged tumors has been documented by using time-lapse digital movies of the cell in culture. Studies are under way to determine whether more aggressive tumors demonstrate more aggressive behavior in culture. If such a correlation is found, the technology may perhaps be useful for prediction of metastatic potential, recurrence, and outcomes, but the clinical utility of this technology is currently unknown.

COLLECTION AND ANALYSIS OF BREAST FLUIDS

Adult breast tissues secrete fluid into the breast ductal system even in the absence of pregnancy and lactation, and this fluid can be aspirated using breast massage and a modified breast pump. A number of studies have been undertaken to determine whether such nipple aspiration fluid (NAF) specimens might be useful for breast cancer screening and diagnosis, but the method is still confined to experimental protocols. Nipple aspiration was first proposed as a potential breast cancer screening technique by Papanicolaou and colleagues in the 1950s when they reported on the diagnosis of a small number of unsuspected cancers as a result of studying NAF specimens from 2,000 women (1958). Cells exhibiting nuclear changes characteristic of hyperplasia, atypia, or malignancy can be observed in NAF, and a more recent prospective study has shown that the incidence and relative risk of breast cancer were positively correlated with increasing severity of the cytological changes (Wrensch et al., 1992). Furthermore, the relative risk of breast cancer associated with hyperplastic cells in NAF was similar to the relative risks calculated in other studies for women with hyperplastic changes identified by traditional biopsy methods. Another recent study, in which more than 95 percent of samples were sufficient for cytological evaluation regardless of menopausal status, also found a correlation between abnormal NAF cytology and increased risk of breast cancer (Sauter et al., 1997). However, the current sensitivity (less than 50 percent) of the test is not high enough to reliably determine whether a woman has breast cancer (Sauter et al., 1999). Thus, the proponents of the technology suggest that it could be a useful adjunct to mammography, especially for women for whom mammography has limitations, such as young women at high risk, those with dense breast tissue, or women whose breasts have been irradiated for prior cancers. As in the case of serum markers, this technology would most likely have to be used in conjunction with an imaging method to pinpoint the lesion responsible for the abnormal finding in NAF.

One technical limitation to the clinical use of NAF has been the difficulty in obtaining samples sufficient in volume for analysis from a significant number of women. Although the ability to obtain sufficient samples volumes may increase with practitioner experience, an alternative approach to NAF collection, known as ductal lavage, has also recently been developed (Love et al., 2000). A catheter is used to flush cells from the breast ducts with saline, and the morphologies of the cells are examined by cytology, just as in the NAF procedure. Pro•Duct Health, Inc., is conducting multicenter trials of this approach.

Molecular biology-based assessment of cells obtained by NAF or breast lavage could potentially identify genetic or epigenetic changes, in addition to cellular morphology, that could perhaps be predictive of breast cancer. Such assessment could include examination for chromosomal abnormalities or the use of DNA arrays to identify changes in the DNA of the cells. This approach has more commonly been investigated with other body fluids such as urine and saliva, with mixed results. One difficulty that must be overcome is the extreme sensitivity needed to identify genetic changes in the exceedingly small number of abnormal cells in such samples. One potential approach to overcoming this obstacle may be to focus on mutations in mitochondrial DNA rather than mutations in nuclear DNA (Fliss et al., 2000). Mitochondrial DNA mutations are common in many cancers, including breast cancer, and each cell contains 1,000 to 10,000 copies, making it easier to detect the mutations in small samples. However, this method is at a very early stage of development and has not been studied at all with NAF or breast lavage samples. Another possibility is to culture the collected cells to expand their number, but this can be quite difficult technically.

Collection of breast fluid samples can also facilitate measurement of protein markers such as growth factors and tumor-specific antigens, which are likely to be more concentrated in the breast fluid than in serum. To date, no markers have been demonstrated to reliably predict breast cancer, but several are under investigation.

SERUM MARKERS AND CELLS IN THE CIRCULATION

Many diseases can be detected and monitored by blood tests, and significant efforts have been made to develop similar tests for the detection and monitoring of cancer. In the case of breast cancer, these efforts have met with limited success. There are two basic approaches to the development of such tests. The first is to measure tumor-specific proteins or other biomolecules in the serum, and the second is to identify and analyze tumor cells themselves in the circulation.

Currently, a few serum markers for breast cancer are mainly used to monitor the course of disease after diagnosis and treatment, although their usefulness in that setting has also been questioned (American Society of Clinical Oncology, 1996; Fitzgibbons et al., 2000; Hayes et al., 1996, 1998). Thus far, the best-established markers are CA 15-3, a polymorphic epithelial mucin, and carcinoembryonic antigen. These markers are unlikely to be used for breast cancer screening or diagnosis because they are accurate only in situations in which the tumor burden is relatively high. Many more potential markers are under development, including growth factors like Her2, oncoproteins such as c-myc and mutant p53, cytokeratins, and markers of angiogenesis and bone metabolism (reviewed by Cheung et al. [2000]). This approach is generally dependent on immunological detection techniques such as enzyme-linked immunosorbent assay and radioimmunoassay.

One recent example of a potential new serum biomarker for breast cancer is the riboflavin carrier protein (RCP). Some vitamin carrier proteins are overexpressed in patients with cancer, and RCP was investigated as a serum marker for breast cancer because its expression is induced by estrogen (Rao et al., 1999). The small prospective study found that RCP levels were significantly elevated in women with breast cancer and that the RCP assay could predict the presence of breast cancer with a sensitivity of 92 percent, a specificity of 88 percent, a positive predictive value of 89 percent, and a negative predictive value of 92 percent. However, the results of this very preliminary study have not yet been validated, and the test has not been examined in the setting of breast cancer screening.

Measurement of serum marker levels may also be helpful in determining the risk for breast cancer. For example, two retrospective studies have found significantly higher serum insulin-like growth factor (IGF) type 1 (IGF1) levels in women with breast cancer than in controls, especially premenopausal women (Pollak, 1998). More recently, a study with prospectively acquired blood samples has provided more direct evidence that the serum IGF level is related to the risk of premenopausal breast cancer (Hankinson et al., 1998). However, it remains to be determined whether higher IGF1 levels during the premenopausal years may also influence the risk of breast cancer after menopause.

Methods that rely on the collection and characterization of tumor cells in circulation are still largely in the experimental stages. The cancer cells must first be separated from the normal blood cells in circulation, which outnumber the cancer cells by many orders of magnitude. This can be accomplished by techniques such as flow cytometry or magnetic separation after the cells have been immunologically labeled with molecules that bind to epithelium- or cancer-specific markers on the surfaces of the cells. Once the cells have been isolated, they can be further characterized by the tests described above to measure genetic changes or gene expression.

Because these methods are being developed primarily for use for the monitoring of treatment response and disease progression, it is unclear whether they will ever be sensitive enough for use for “early detection” or how they might be used for screening or early diagnosis and prognosis. In any case, serum markers are unlikely to replace imaging technologies for the diagnosis of breast cancer because current therapy for early disease requires localization of the primary lesion in the breast, which generally depends on breast images.

OBSTACLES TO BE OVERCOME IN DEVELOPMENT OF BIOLOGICAL DETECTION METHODS

In addition to technology needs, a number of infrastructure needs and other impediments to progress in the development of biomarkers for breast cancer detection have been identified. For example, samples from relatively large tumors obtained by biopsy have traditionally been the specimens most widely available to researchers, but this has limited the study of smaller, earlier lesions. Recently, more effort has been devoted to the development of specimen banks with samples from the entire continuum of malignant and premalignant lesions of the breast, but the small sizes of these lesions make it difficult to share samples for multiple investigations. Attempts have also been made to examine samples obtained by core-needle biopsy or even fine-needle aspirates, but in most cases the technologies are not yet sensitive enough for accurate assessment of these small samples. Because of their small size, materials obtained by needle biopsy are also less likely to be made available to researchers through established tissue banks.

Once the tissues have been collected, other impediments to research can arise. For example, concerns have been raised about informed consent and patient confidentiality (Anderson, 1994; National Bioethics Advisory Commission, 1999). A common approach to obtaining consent is to use a very general consent form that will allow future, unspecified research to be conducted without the need to reacquire consent for every subsequent study. As with all research with human subjects and materials, institutional review boards can then approve or reject specific study designs as they are proposed, but this may lead to variability in the types of studies approved at different institutions. Confidentiality, always a concern in health care and biomedical research with human subjects, could be especially problematic with regard to studies of hereditary genetics (Daly et al., 2000). As a result, NCI has proposed methods for protecting the identities of tissue donors while still maintaining links to data on clinical outcome.⁷

Other organizational challenges also must be overcome to establish and maintain useful specimen banks (Burke and Henson, 1998; Grizzle et al., 1998). Effective use of patient samples depends critically on the ability of the specimen bank to acquire, organize, and disseminate samples and associated information in a timely manner and to standardize sample collection and reporting across different institutions. These activities require substantial monetary support and staff resources, which are often not available at adequate levels . A new high-throughput method know as “tissue microarray” (Kononen et al., 1998)⁸ may offer one potential means by which the organization, dissemination, and analysis of tissue specimens could be streamlined and automated, but such an approach would certainly require a broad and general consent form for sample collection.

A recent report by NCI's Breast Cancer Progress Review Group (BC-PRG) concluded that NCI study sections have historically given tissue banking efforts and the associated correlative clinical studies such low priority that they have been unfundable (Breast Cancer Progress Review Group, 1998). The charge to the BC-PRG was to identify and prioritize scientific needs and opportunities critical for progress against the disease and to compare these priorities with the current portfolio of the NCI research program. The group recommended that NCI increase funding support for tissue banks through mechanisms separate from the traditional grants to principal investigators.

Recently, concerns over intellectual property issues associated with biomedical research tools and resources have also been raised. Although some specimen banks freely share samples with investigators, some institutions are beginning to demand a share of any profits derived from technologies that may be developed as a result of research conducted with samples derived from their banks. NCI could potentially alleviate such impediments by requiring specimen banks supported with NCI funding to forego intellectual property right claims to technologies derived from research with their archived samples. There is already precedent for such restrictions. For example, investigators funded through CGAP are required to rapidly add their data to publicly accessible databases and must therefore relinquish any patent rights to their discoveries.

Concerns about funding and intellectual property rights have also been raised in regard to high-throughput technologies. The BC-PRG concluded that the academic research community has inadequate funds to purchase and operate the new high-throughput technologies that are likely to advance the field and recommended increased funding for this purpose, perhaps through shared core facilities. However, patents on gene sequences may keep the price of DNA arrays and related technologies out of reach for many researchers. Some companies may also be limiting the research community's access to new high-throughput technologies through control of the intellectual property rights to future products based on discoveries made with the new technologies.

Another obstacle to be overcome stems from the need for new bioinformatics approaches to make sense of all the data that are being generated by these high-throughput technologies. The field of bioinformatics is relatively new, and thus, the recently developed training programs have not kept pace with the demand for individuals with the necessary experience to tackle these issues. Given the enormous number of genetic and epigenetic changes already identified and the vast heterogeneity within and among breast tumors, this may indeed be the greatest challenge of all.

SUMMARY

To substantially reduce morbidity and mortality from breast cancer, basic and clinical research must lead to improvements in the understanding of breast tumor biology and etiology as well as improvements in the ability to detect early lesions. To optimize advances made in detection and diagnostic technologies, knowledge about the biology of the lesions detected should ideally play a role in the decision-making process that occurs after detection. In particular, research on the biology of premalignant disease and early breast cancer is crucial for understanding and predicting the progression of breast lesions and for the development of more targeted and effective treatments for those lesions that are likely to become lethal. Such knowledge would also allow women with clinically insignificant lesions to forego unnecessary treatments and the associated side effects.

Identification of meaningful biomarkers of breast cancer could significantly advance the field of functional imaging (as discussed in Chapter 2) and thus lead to improved methods of screening and diagnosis. Better understanding of the genetics and biology of breast cancer can also be expected to open new avenues in the future for assessing the risk of developing cancer as well as aiding in treatment planning once a tumor has been detected. Some of the developing technologies described in this chapter, such as the identification and assessment of biomarkers in serum or breast fluids and tissues, offer promise as screening procedures but will require further study and development before their use as screening modalities can be evaluated. In many instances these technologies could potentially identify fundamental changes in the breast that appear before a lesion can be identified. Thus, they may identify women at high risk of developing breast cancer (or, more importantly, women at high risk of dying from breast cancer). Increased efforts in these areas should also be a priority.

Some of the obstacles to attaining these goals include lack of funding for and accessibility to the resources and tools necessary to move this field forward. For example, much research has focused on late-stage breast cancer due to the predominance of these tissues in the specimen banks and cell line repositories. Thus, the study of early breast cancer has lagged. Furthermore, funding for the establishment and maintenance of specimen banks has traditionally not been a high priority.

NCI has launched several new funding initiatives in the last year aimed at increasing the understanding of breast cancer initiation and progression, in part as a result of its external progress review process. Clearly, much work remains to be done in the field of breast cancer biology. Making sense of the many molecular changes in breast tumors will be extremely challenging, but the end result could potentially have an enormous impact on reducing the burden of breast cancer.

Footnotes

1

The Health Insurance Portability Accountability Act of 1996 (Public Law No. 104-191, 701, 110 Stat 1936) prevents group health plans from labeling genetic information as a preexisting condition.

2

The SNP Consortium's members include the Wellcome Trust; 10 pharmaceutical companies including AstraZeneca PLC, Aventis Pharma, Bayer AG, Bristol-Myers Squibb Company, F. Hoffmann-La Roche, Glaxo Wellcome PLC, Novartis Pharmaceuticals, Pfizer Inc, Searle, and SmithKline Beecham PLC; Motorola, Inc.; International Business Machines Corp; and Amersham Pharmacia Biotech. Academic centers including the Whitehead Institute for Biomedical Research, Washington University School of Medicine in St. Louis, the Wellcome Trust's Sanger Centre, Stanford Human Genome Center, and Cold Spring Harbor Laboratory are involved in SNP identification and analysis. Orchid BioSciences performs third-party validation and quality control testing on SNPs identified through the consortium's research.

3

Microsatellite loci are stretches of repetitive DNA that are located throughout the genome and thus serve as useful markers for genetic analysis.

4

Cyphergen Biostystems, Inc. (Palo Alto, California) has developed a mass spectroscopy technology known as “surface-enhanced laser desorption/ionization” that rapidly analyzes native proteins at the femtomole level without the use of labels.

5

Latimer, J.J. Epithelial cell cultures useful for in vitro testing, 1998.

6

See www​.pitt.edu/∼rsup/mgb/latimer​.html.

7

http://www​.nci.nih.gov/confidentiality​.html.

8

By this method, small cylinders of tissue are punched from 1,000 individual tumor biopsy specimens embedded in paraffin. These cylinders are then arrayed in a large paraffin block, from which 200 consecutive tissue sections can be cut, allowing multiple, rapid analysis of the arrayed samples by immunohistochemistry or in situ hybridization.