Breast and Ovarian Cancer
Table of Contents+, http://www.100md.com
PDF of this article+, http://www.100md.com
PDA version of this article+, http://www.100md.com
Editor's Summary+, http://www.100md.com
CME Exam+, http://www.100md.com
Related Letters to the Editor+, http://www.100md.com
Find Similar Articles in the Journal+, http://www.100md.com
Notify a friend about this article+, http://www.100md.com
Articles citing this article+, http://www.100md.com
Ethical, Legal, and Social Implications of Genomic Medicine+, http://www.100md.com
Ellen Wright Clayton, M.D., J.D.+, http://www.100md.com
Table of Contents+, http://www.100md.com
PDF of this article+, http://www.100md.com
PDA version of this article+, http://www.100md.com
Editor's Summary+, http://www.100md.com
CME Exam+, http://www.100md.com
Related Letters to the Editor+, http://www.100md.com
Find Similar Articles in the Journal
Notify a friend about this article5h}{&q, 百拇医药
Articles citing this article5h}{&q, 百拇医药
Services 5h}{&q, 百拇医药
Add to Personal Archive5h}{&q, 百拇医药
Download to Citation Manager5h}{&q, 百拇医药
Alert me when letters appear5h}{&q, 百拇医药
Alert me when this article is cited5h}{&q, 百拇医药
ISI Web of Science5h}{&q, 百拇医药
Related Articles5h}{&q, 百拇医药
Citing Articles (8)5h}{&q, 百拇医药
Medline 5h}{&q, 百拇医药
Related Articles in Medline5h}{&q, 百拇医药
Articles in Medline by Author:5h}{&q, 百拇医药
Clayton, E. W.5h}{&q, 百拇医药
Medline Citation5h}{&q, 百拇医药
Genetics5h}{&q, 百拇医药
Legal Issues in Medicine5h}{&q, 百拇医药
Medical Ethics5h}{&q, 百拇医药
Related Chapters at Harrison's Online
As detailed in other articles inthe Genomic Medicine series, genomics has contributegreatly to our understanding of the molecular basis of disease and, to a lesser but growing extent, to the development of effective interventions. Clinicians and society at largewever, are concerned about the effect geneticknowle will have on the well-being of individual persons and groups. Much effort is being devoted to trying to anticipate, understand, and address the ethical, legal, social, and political implications of genetics and genomics..x?@^, http://www.100md.com
The inquiry is complex. Understanding the social effects of genomics requires an analysis of the ways in which genetic information and a genetic approach to disease affect people individually, within their families and communities, and in their social and working lives. Genomics presents particular challenges with respect to clinicians' ethical and professional responsibilities, including the appropriate use of genomic information in the health care setting. In this article, I examine public concerns about genetic information and discuss a few recent cases in some depth to highlight a few of the dilemmas presented by genomics and emerging solutions.
What Genetic Information Is and What People Are Worried About|85w}pw, 百拇医药
Genes affect virtually all human characteristics and diseases. These influences can be ascertained in individual patients through a review of the family history, physical examination, and the use of medical diagnostics. In some conditions, such as cystic fibrosis and sickle cell disease, the specific molecular mechanisms are largely understood, but in many, including such common chronic diseases as diabetes mellitus and hypertension, the relevant genes — and there are often many — are only beginning to be identified. Given the variety of these effects and the limits of our knowledge, it is not surprising that the term "genetic information" is used in different ways at different times. Sometimes it is used to mean the influence of the entire genome, but more often it is used to refer to recognized, single-gene disorders or, even more narrowly, the results of DNA-based tests. These various meanings may make sense in context, but confusion can occur unless the speaker and listener are defining the term in the same way.
The most commonly expressed fear is that genetic information will be used in ways that could harm people — for example, to deny them access to health insurance, employment, education, and even loans. Part of that concern is fueled by the growing recognition that health information is not entirely private, despite people's expectations and desires to the contrary. In fact, both federal and state governments have been actively engaged in discussions about who ought to have access to health information and under what conditions.1,2 This debate is informed appropriately by the recognition that limiting access to the medical record to the patient and the treating clinician is neither possible nor unequivocally desirable.31rs$\(, http://www.100md.com
People tend to see genetic information as more definitive and predictive than other types of data, in the sense that "you cannot change your genes" and that "genes tell all about your future." This notion of genetic determinism, however, includes an unwarranted sense of inevitability, because it reflects a fundamental failure to understand the nature of biologic systems. The DNA sequence is not the Book of Life. Human characteristics are the product of complex interactions over time between genes — both a person's own and those of other organisms — and the environment. Both germ-line and somatic cells undergo mutations, the latter being a primary way in which cancer develops. Moreover, a pathogenic mutation does not doom one to ill health; many diseases can be treated. As is true for so many conditions in medicine, clinicians have a variable but usually limited ability to predict when, how severely, and even whether a person with a genetic predisposition to a certain illness is going to become ill.
One might be tempted to conclude that the way to allay people's fears about genetics is simply to give them a more realistic understanding of the informative power of these tools. Given the optimistic predictions about genetics that pervade the media and public opinion today, that path is unlikely to succeed in the short term. A more promising approach to addressing the social implications of genetics requires us to consider both how genes are perceived in the real world and what is actually known about their function.m(@p1cd, http://www.100md.com
The Problem of Discriminationm(@p1cd, http://www.100md.com
The question of whether genetic information should ever be used to affect one's access to health and other forms of insurance has been a dominant issue of public concern in the past decade. People cite fear of losing insurance as a major reason to avoid genetic testing.4 Others argue that discrimination by insurance companies is not a problem, often pointing out that few of these cases, which are difficult for employees to win, have been filed.5 Insurers assert that they do not perform tests to obtain genetic information but argue that they should be free to use such information if it is available, citing the need to avoid "moral hazard" — the risk that people who know they will become ill or die soon will try to obtain insurance at regular rates.6 In response to consumer pressure, many states have passed laws in this area .7,8 In passing the Health Insurance Portability and Accountability Act (HIPAA),9 Congress specifically banned certain uses of genetic information in determining insurance eligibility, but it placed no limits on rate setting.10 Vigorous debate about optimal solutions is ongoing,11 and bills have been introduced in every recent session of Congress.12
fig.ommtted:]y/|6c, http://www.100md.com
Summary of Statutes Regarding Discrimination on the Basis of Genetic Information and the Privacy of Such Information.:]y/|6c, http://www.100md.com
The complexity of the issues surrounding discrimination can be illustrated more generally by examining a case involving Burlington Northern Santa Fe Railroad (BNSF). Allegedly relying on the advice of its company physician, who in turn had apparently relied on the representations of a diagnostic company, BNSF began obtaining blood for DNA testing from employees who were seeking disability compensation as a result of carpal tunnel syndrome that occurred on the job. The employees were reportedly not told the purpose of the tests, which was to detect a mutation associated with hereditary neuropathy with liability to pressure palsies.13 The company's motive for pursuing testing was never made clear, but it seems reasonable to suspect that BNSF would have tried to deny disability benefits to any employee who had such a mutation, arguing that the mutation, and not the job, caused the carpal tunnel syndrome. When the company's practice came to light, it was almost immediately stopped by the federal Equal Employment Opportunity Commission,14 and shortly thereafter, the company settled claims brought by its employees for an undisclosed amount of money.15
What lessons can be learned here? One is that the company's effort to find mutations for hereditary neuropathy with liability to pressure palsies made little sense. This disorder is very rare, affecting about 3 to 10 persons per 100,000, and more important, although carpal tunnel syndrome can be a part of hereditary neuropathy with liability to pressure palsies, it has not been reported as the sole symptom. The injuries these employees sustained were not the result of an epidemic of hereditary neuropathy with liability to pressure palsies. Getting the biologic process correct is a critical step in making decisions about genetic testing.:&o3, http://www.100md.com
Another important lesson is that identifying a genetic predisposition to carpal tunnel syndrome would not have been the end of the discussion in the eyes of the law. The company got in trouble because its practice violated numerous laws forbidding discrimination in the workplace. In particular, the Americans with Disabilities Act permits employers to require a medical evaluation only under clearly specified circumstances.16 Testing employees after they were disabled without their informed consent clearly fell outside the bounds of this and other antidiscrimination laws.
The actions of BNSF led to widespread criticism and, not surprisingly, to calls to ban genetic discrimination in the workplace.17 Although some states have enacted laws the need for federal action has grown as the Supreme Court has progressively narrowed the protection provided under the Americans with Disabilities Act.18,19 The answer, however, is not simply to forbid employers to use genetic information or to require genetic testing.7, 百拇医药
The first step in developing an appropriate response is to determine how the use of genetic information fits within the broader framework of antidiscrimination laws, which were passed to create a certain kind of society, one in which people must be included regardless of race, sex, or disability, even at some cost to employers. Biology alone does not determine the social outcome. To use an analogy, an employer cannot exclude women from the workplace, even if he or she believes, with some justification, that women are more likely than men to take time off to care for family members. At the same time, employers are not required to bear unlimited costs to promote these social goals — the employee, male or female, who misses months of work at a time to care for sick relatives can still be fired.
A similar debate about social goals and the limits of our pursuit of them must occur with regard to genetic discrimination. The Equal Employment Opportunity Commission recently awarded damages to Terri Sergeant, who was fired from her job as an office manager for an insurance broker because she required extremely expensive medication to treat her at-worst mildly symptomatic alpha1-antitrypsin deficiency.20 A person's need for expensive health care is not sufficient reason to fire that person or to refuse to hire him or her in the first place. The fact that the costs may cause the employer to go under or to decide not to provide health insurance simply underlines the inherent weakness of employment-based health insurance.g, http://www.100md.com
At the same time, one can imagine a genetic condition that might affect a person's ability to perform a job in ways that could not be accommodated with reasonable efforts. Suppose a person with a recurrent and untreatable cardiac arrhythmia that leads to loss of consciousness, owing to an inherited ion-channel defect, is seeking employment as a long-distance truck driver. Because of the risk to third parties, such a person would not even be able to get a driver's license in many jurisdictions. The more difficult question — and the one posed particularly with respect to genetics — would arise if an asymptomatic person had a predisposing, but incompletely penetrant, mutation for the same disorder. Deciding what to do about such predispositions will require close attention both to the true, as opposed to the feared, likelihood that symptoms will develop and to the complex weighing of the interests of the individual, the employer, and society.
A similar calculus must be applied to every question regarding who can obtain and use genetic information to distinguish, or discriminate, among people in ways that affect their ability to obtain social goods, such as health insurance and education .21 If, as is likely, some uses are deemed to be appropriate, the challenge for clinicians will be to discuss with their patients the potential adverse social consequences of testing so that the patients can make informed choices about whether or not to proceed with testing.&^, http://www.100md.com
fig.ommtted&^, http://www.100md.com
Elements to Be Considered in Decisions about the Use of Genetic Information.&^, http://www.100md.com
The Challenge of Genomic Medicine with Respect to the Physician–Patient Relationship&^, http://www.100md.com
Consider the case of a man who died of colon cancer in the 1960s. When the same disease developed in his daughter approximately 25 years later, she obtained her father's pathology slides, discovered that he had had diffuse adenomatous polyposis coli, and sued the estate of her father's surgeon, alleging that the physician should have warned her about her 50 percent risk of having the disorder. An intermediate appellate court in New Jersey ruled that the physician had a duty to warn the daughter directly (she would have been a child at the time of her father's death), perhaps even over her father's objections.22
This is only one court's view in one case, but given how much attention it received, it is important to ask whether this was a good result. Two central tenets of Western medicine are that physicians should focus on the interests of their patients and that they should protect the confidentiality of their patients' medical information. Yet the tools of genomic medicine often reveal information about health risks faced not only by patients but also by their relatives. What should clinicians do? It seems clear that they should tell their patients about the risks faced by family members. The harder questions are whether physicians are ethically permitted to contact the relatives themselves, in contravention of traditional patient-centered norms, and whether they should be legally required to do so.'0, 百拇医药
This issue must be viewed in the light of the fact that the duty to protect confidentiality is not absolute. Physicians are required to report numerous infectious diseases,23 and they have been held liable for failing to warn people whom their patients have specifically threatened with violence.24 The question then becomes more complex: are genetic risks sufficiently similar to these existing exceptions to the requirement of confidentiality that they warrant an exception as well? Over the years, numerous prominent advisory bodies have said no, opining that physicians should be permitted to breach confidentiality in order to warn third parties of genetic risks only as a last resort to avert serious harm.25,26,27
These learned opinions, however, are not the end of the matter, in part because they lack the force of law. In fact, as the case above illustrates, relatives have sued the primary patients' physicians for failing to warn them of their own genetic risks — and won limited victories, although none have been awarded monetary damages. The decisions in the colon-cancer case and a similar one in Florida28 have been criticized for both their legal reasoning and their deviation from ethical guidelines, but they have not been overturned and, in the tradition of the common law, may be persuasive to other courts. Physicians who breach their patients' confidentiality and warn family members are not likely to incur substantial liability, even under HIPAA.29 As a result, physicians might understandably conclude that warning relatives is the least risky option.6s, 百拇医药
The existing directives are thus in conflict: "expert consensus," ethical analysis, and the HIPAA regulations argue for honoring confidentiality, whereas at least one legal opinion holds that physicians fail to warn a patient's relatives at their peril. Given the press of other business, legislators are not likely to resolve this conflict soon. In this setting, clinicians should inform their patients about the risks their relatives face, discuss the appropriateness of sharing this information and offer assistance, trust — usually realistically — that patients will in turn tell their relatives who are at risk, and hope that the courts will get it right in the future.
Genomic Medicine and Public Health(z, 百拇医药
When Sierra Creason underwent state-mandated newborn screening, she had abnormally low levels of both thyroxine and thyrotropin, findings consistent with the presence of congenital hypothyroidism. Her physician was not notified of these results, however, because the state had chosen not to divulge the actual values and, instead, to report as abnormal only results in which thyroxine levels were low and thyrotropin levels were high.30 As a result, the diagnosis of congenital hypothyroidism and subsequent treatment were seriously delayed, resulting in permanent harm. When the child's family sued the state, however, the California Supreme Court ruled that the state program could not be held liable, in part to avoid diverting funds that would have been used for other state purposes. By contrast, had a private diagnostic laboratory given the same report, especially without providing the actual results, which would have enabled the child's physician to make an independent assessment, it almost certainly would have been held responsible.
Complex questions arise when the government requires testing and interventions. State-mandated screening of newborns for metabolic and genetic disorders was described by Khoury et al. in an earlier article in this series.31 Governments undertake many activities to promote health — universal screening of newborns for phenylketonuria, for example, is generally considered a resounding success — but it is worth asking in each case whether there is sufficient justification to pursue mandatory as opposed to voluntary action or to place such activities in the public rather than the private sector. Requiring public health agencies to assume such responsibilities has advantages, such as more transparent accountability to the public and greater uniformity in access and results. Relying on public health entities in matters that directly affect the health of individual persons, however, entails certain risks as well. Physicians and patients count on receiving accurate and informative results regardless of whether a private or a public entity is doing the testing. Permitting state agencies to avoid financial responsibility when their actions harm patients like Sierra Creason is unjust and should raise questions about the wisdom of proposals that would dramatically expand newborn screening.
A public health analysis of genomics, of course, involves more than state-run testing. The broadest question is whether the public's health is actually improved by the knowledge derived. A major determinant is access to testing and to the medical interventions that may be warranted as a result. In our current multipayer system of health care, people will have widely differing levels of access to these forms of technology. One cannot assume that everyone will reap the benefits of this knowledge.93@%i, 百拇医药
From a public health perspective, it might do to go one step further and ask whether people will actually use the test results to alter their behavior in ways that improve health. Some people whom testing identifies as predisposed to cancer subsequently decline to undergo surveillance or other interventions for psychological reasons or because of other demands on their time. Some preventive or therapeutic measures are more likely to be pursued than others; most people find it difficult to take medications for a lifetime or to maintain major lifestyle changes, no matter how important such approaches are for their health.
Public health agencies exist not only to identify barriers to health but also to improve health and health care. Efforts to determine when genetic tests are reliable enough for routine clinical use are quintessential public health activities.32,33 The Secretary's Advisory Committee on Genetic Testing and its successor, the Secretary's Advisory Committee on Genetics, Health, and Society, were formed to provide such guidance.34 The development of strategies to educate health care providers and patients about genomic medicine, a long-standing goal of the Human Genome Project, and to decrease obstacles to health-promoting behavior also falls comfortably within this rubric.(lvwuz2, http://www.100md.com
Conclusions(lvwuz2, http://www.100md.com
This brief discussion illustrates public expectations and fears about the effect of genomics, challenges to the goals of antidiscrimination laws and to the nature of the physician–patient relationship, and the contrasting perspectives and legal rules that apply to personal medical care and public health. Acknowledgment and examination of these complex issues are critical for identifying the appropriate ethical principles that should be applied and for creating the necessary legislative and regulatory responses.
Supported in part by a grant from the National Institutes of Health (5R01 HG 01974-02).87quxo;, http://www.100md.com
I am indebted to William O. Cooper and Jay Clayton for their helpful comments on earlier drafts and to Vanessa M. Spencer for her research assistance.87quxo;, http://www.100md.com
Source Information87quxo;, http://www.100md.com
From the Center for Genetics and Health Policy, Vanderbilt University, Nashville.87quxo;, http://www.100md.com
Address reprint requests to Dr. Clayton at Vanderbilt University, 507A Light Hall, Nashville, TN 37232-0165.87quxo;, http://www.100md.com
References87quxo;, http://www.100md.com
Services 87quxo;, http://www.100md.com
Add to Personal Archive87quxo;, http://www.100md.com
Download to Citation Manager87quxo;, http://www.100md.com
Alert me when letters appear87quxo;, http://www.100md.com
Alert me when this article is cited87quxo;, http://www.100md.com
ISI Web of Science87quxo;, http://www.100md.com
Related Articles87quxo;, http://www.100md.com
Citing Articles (9)
Medline np/mr*, 百拇医药
Related Articles in Medlinenp/mr*, 百拇医药
Articles in Medline by Author:np/mr*, 百拇医药
Wooster, R.np/mr*, 百拇医药
Weber, B. L.np/mr*, 百拇医药
Medline Citationnp/mr*, 百拇医药
Breast Cancernp/mr*, 百拇医药
Geneticsnp/mr*, 百拇医药
Oncologynp/mr*, 百拇医药
Women抯 Healthnp/mr*, 百拇医药
Related Chapters at Harrison's Onlinenp/mr*, 百拇医药
Despite years of intensive study and substantial progress in understanding susceptibility to breast and ovarian cancer, these diseases remain important causes of death in women. However, several recent critical advances — sequencing of the human genome and the development of high-throughput techniques for identifying DNA-sequence variants, changes in copy numbers, and global expression profiles — have dramatically accelerated the pace of research aimed at preventing and curing these diseases. We review some of the important discoveries in the genetics of breast and ovarian cancer, ongoing studies to isolate additional susceptibility genes, and early work on molecular profiling involving microarrays.
Susceptibility to Breast and Ovarian Cancer44!)0, 百拇医药
In the United States, 10 to 20 percent of patients with breast cancer and patients with ovarian cancer have a first- or second-degree relative with one of these diseases.1 Two major genes associated with susceptibility to breast and ovarian cancer — breast cancer susceptibility gene 1 (BRCA1) and breast cancer susceptibility gene 2 (BRCA2) — have been identified to date.2,3 Mutations in either of these genes confer a lifetime risk of breast cancer of between 60 and 85 percent and a lifetime risk of ovarian cancer of between 15 and 40 percent.4,5 However, mutations in these genes account for only 2 to 3 percent of all breast cancers,6,7 and susceptibility alleles in other genes, such as TP53, PTEN, and STK11/LKB1, are even less common causes of breast and ovarian cancer .44!)0, 百拇医药
fig.ommtted44!)0, 百拇医药
The Genetics of Breast Cancer.
BRCA1 and BRCA2 mutations occur in approximately 20 percent of families with evidence of inherited susceptibility to breast cancer. Germ-line mutations in TP53 cause the Li–Fraumeni syndrome and account for no more than 1 percent of cases of familial breast cancer, but women who survive the childhood cancers associated with the Li–Fraumeni syndrome have as much as a 90 percent risk of breast cancer.8 Mutations in the cell-cycle–checkpoint kinase gene (CHEK2) account for about 5 percent of all cases of familial breast cancer (defined by the diagnosis of breast cancer in two or more family members before the age of 60 years), but the risk for individual mutation carriers is probably less than 20 percent.9 All other cases of breast cancer are presumed to be due to an undefined number of additional susceptibility genes with various degrees of penetrance, exposure to hormonal and environmental factors, and stochastic genetic events.[k/k, 百拇医药
The prediction that there are common DNA-sequence variants that confer a small but appreciable enhanced risk of cancer has been validated with the recent discovery of the 1100delC mutation in the cell-cycle–checkpoint kinase gene (CHEK2).9 This mutation was found in 1.1 percent of women without breast cancer, 1.4 percent of women with a personal but no family history of breast cancer, and 4.2 percent of index patients from 718 families in which two or more members had been given a diagnosis of breast cancer before the age of 60 years but in which there was no detectable BRCA1 or BRCA2 mutation. This mutation doubles the risk of breast cancer among women and increases the risk among men by a factor of 10. CHEK2, an important component of the cellular machinery that recognizes and repairs damaged DNA, is activated after phosphorylation by the checkpoint gene ATM and in turn activates BRCA1. The role of ATM mutations in the predisposition to the early onset of breast cancer remains controversial, but some missense mutations do appear to increase susceptibility to breast cancer in humans10 and mice.11
There is convincing evidence that additional high-penetrance genes that increase susceptibility to breast cancer exist. In contrast, it has been suggested that, other than BRCA1 and BRCA2, high-penetrance genes that confer susceptibility to ovarian cancer do not exist.12 An ovarian-cancer–susceptibility locus on chromosome 3p22–25 has putatively been identified, but this finding has yet to be confirmed by an independent group.13s?z+./, http://www.100md.com
Many additional genetic variants in low-penetrance susceptibility alleles may moderately increase the risk of breast cancer, ovarian cancer, or both. These genetic variants are much more common in the population than are high-penetrance gene mutations and, thus, in aggregate may make a substantially greater contribution to breast and ovarian cancer in the population than mutations in high-risk genes.14 However, genetic heterogeneity and the rarity of high-penetrance genes make both high- and low-penetrance genes difficult to identify.
Identification of Genes That Increase Susceptibility to Breast and Ovarian Cancer|\1tn, 百拇医药
Identification of High-Penetrance Genes|\1tn, 百拇医药
Genetic linkage was used to identify the BRCA1 and BRCA2 loci on chromosomes 17q and 13q, respectively.15,16 In both cases, no information was initially available on the location, structure, or function of the genes, and they were identified through positional cloning. Loss-of-heterozygosity mapping was of little assistance in either search; however, a homozygous deletion on chromosome 13 in a pancreatic adenocarcinoma helped identify the location of BRCA2.17 Finally, critical data in the search for BRCA2 came from studies of breast cancer in Iceland, whose population derives from a small group of settlers from Norway and Ireland.18,19 Such populations share more genetic information than large, admixed populations and have been used successfully many times in gene mapping.20 After the BRCA1 locus was identified, it took almost four years to isolate the gene and involved several labor-intensive strategies.2 By contrast, the BRCA2 locus was one of the first genomic intervals to be systematically sequenced as part of the Human Genome Project. These data, together with other information about the genes in the region, reduced the time it took to isolate BRCA2 to two years.3 Thus, information from the human genome sequence greatly enhances the utility of linkage analysis for gene identification .
fig.ommtted*5f2$, 百拇医药
Effect of Sequencing the Human Genome on Gene-Discovery Strategies.*5f2$, 百拇医药
The annotated DNA sequence of the human genome can be used to locate genes, repeat sequences, and other features and has revolutionized the identification of cancer genes. A sequence without annotation is of limited utility (Panel A). As shown in Panel B, an annotated sequence shows genetic markers such as CA and GT repeats along with other data, such as CpG islands, known genes, genes predicted to exist on the basis of computational models, and Basic Local Alignment Search Tool (BLAST) matches. Using publicly available data (http://www.ensembl.org, http://www.ncbi.nlm.nih.gov, and http://www.genome.ucsc.edu), it is possible to jump from a genetic region of interest to the identification of candidate genes in a matter of seconds and download the relevant data (Panel C). With these data in hand, experiments, such as those involving the polymerase chain reaction (PCR), can be designed to analyze the genes for mutations (Panel D). The final step in the identification of genes is to compare the sequence from patients with the disease of interest with the normal reference sequence to discover the mutations (Panel E).
BRCA3 and beyond#lt@, 百拇医药
Several candidate regions for BRCA3 have been proposed, including chromosome 13q2121 and chromosome 8p12–22,22 but both have been strongly refuted by analysis of data from independent families.23,24 The search for BRCA3 has been difficult for several reasons. First, ovarian cancer and male breast cancer were recognized as components of syndromes of breast-cancer susceptibility before either BRCA1 or BRCA2 was isolated, allowing targeted identification of affected families. Since no such phenotype has been associated with the putative BRCA3 gene or genes, families in current studies are selected only on the basis of a young age at the diagnosis of breast cancer and the absence of ovarian and male breast cancer. Ideally, these families should have multiple members with early-onset breast cancer and strong evidence against the involvement of either BRCA1 or BRCA2. However, the breast cancers in most such families are in fact due to germ-line mutations in BRCA1 or BRCA2,25 and those that are not may represent the effects of multiple susceptibility alleles (genetic heterogeneity), reducing the power of linkage analysis. What is needed to further this effort are larger families to increase the statistical power of such studies, as well as novel means of clustering families into subgroups most likely to represent single-gene disorders.
One approach is to classify families with breast cancer according to the molecular profile of the associated tumors. These analyses could be based either on expression profiling or on array-based comparative genomic hybridization, both of which provide unique molecular signatures. Nonetheless, hundreds of small pedigrees may be needed to identify the BRCA3 locus.)x!r, 百拇医药
fig.ommtted)x!r, 百拇医药
Array-Based Comparative Genomic Hybridization.)x!r, 百拇医药
In Panel A, bacterial artificial chromosome clones or complementary DNAs are placed on glass slides at high density; tumor and normal DNA are labeled with CY3 and CY5, respectively; and the combined probe is hybridized to the array. The array is analyzed with use of a laser scanner that reads each color channel individually and then calculates an intensity ratio for each spot. In Panel B, spots with intensity ratios greater than 1.25 (green spots) represent increases in copy number (amplification), and those with intensity ratios of less than 0.75 (red spots) represent decreases in copy number (deletion). Each spot is a DNA segment that can be linked directly to the human genome sequence (Panel C), thus defining changes in the number of copies of a specific gene. In Panel D, the plotting of intensity ratios for the chromosome 9 bacterial artificial chromosome clones on the array in linear order identifies a homozygous loss of CDKN2A in a melanoma cell line.
Use of the Human Genome Sequence to Identify Low-Penetrance Genes2, 百拇医药
As noted, the susceptibility genes identified to date are not responsible for most breast and ovarian cancers, leaving a considerable potential contribution from less penetrant genes. One of the implicit problems in isolating low-penetrance genes is that such genes will rarely produce striking familial patterns involving multiple cases that can be used in traditional linkage studies. An additional concern is that very large studies, with statistical power to evaluate multiple interactions between genes, may be needed before genetic profiles involving this class of genes can be used for risk prediction.14 As the computational methods for finding coding sequences embedded in a sequence of genomic DNA become increasingly powerful, the value of the human genome sequence as a tool for identifying unknown genes also increases. These algorithms for finding genes have largely replaced laborious experimental techniques to identify potential coding sequences of unknown genes for mutation analysis within linkage regions. These methods are another illustration of the fact that it is the annotation of the genomic sequence (i.e., the identification of genes and their function) that brings the sequence to life. Annotation parses the sequence into genes and noncoding regions. By including genomic features such as CpG islands, which mark the promoter regions of many genes, annotations produce a complete rendering of each sequence. Annotated sequences are publicly available in several data bases (http://www.ensembl.org, http://www.ncbi.nlm.nih.gov, and http://www.genome.ucsc.edu) with associated genome browsers.
The depth and value of annotation have also grown through the addition of millions of single-nucleotide polymorphisms, which are invaluable in the search for susceptibility genes.26 One such example is the recent demonstration that a silent single-nucleotide polymorphism in LIG4, a gene encoding a DNA ligase important in the repair of breaks in double-stranded DNA, is associated with survival among patients with breast cancer.27 This effect was demonstrated in a British population-based study that included 2430 cases of breast cancer. DNA from these patients was genotyped for polymorphisms in 22 DNA-repair, hormone-metabolism, carcinogen-metabolism, and other genes, and the effect of each single-nucleotide polymorphism on the outcome was assessed by Cox regression analysis. The silent polymorphism D501D (t>c) in LIG4 had the largest effect. The estimated hazard ratio for death among patients homozygous for the polymorphism, as compared with those homozygous for the wild-type sequence, was 4.0 (95 percent confidence interval, 2.1 to 7.7; P=0.002), and this effect remained significant after stratification according to tumor stage, grade, and type (hazard ratio, 4.2; 95 percent confidence interval, 1.8 to 9.4; P=0.01). The inclusion of these single-nucleotide polymorphisms in the annotation of the human genome sequence greatly facilitated this analysis, which would otherwise have had to have been preceded by an extensive sequence-based effort to identify single-nucleotide polymorphisms.
Whole-Genome Approaches to the Analysis of Breast and Ovarian Cancerj, 百拇医药
Many genomic approaches to the identification of cancer genes are based on microarray techniques. For gene-expression profiling, each element usually represents one gene and is created with the use of a complementary DNA (cDNA) or oligonucleotide for the gene in question. Similar arrays have been produced with large genomic clones for array-based comparative genomic hybridization to identify changes in the number of copies of DNA. This approach replaced lower-resolution comparative genomic hybridization of cells in metaphase and provides a direct link to genes in the altered region .j, 百拇医药
A comparative genomic hybridization can be used to identify the loss of one or both copies of a given gene as well as regions of amplification. Arrays made with cDNAs can be used for expression profiling and comparative genomic hybridization simultaneously.28 This approach allows a direct comparison between the number of copies of a gene and the level of expression of that gene, but the results of comparative genomic hybridization may be variable, presumably because the cDNA sequence and the genomic sequence are not collinear. However, the alternative approach of using large cloned segments of genomic DNA in the bacterial artificial chromosomes consistently provides excellent data.29 The genomic clones can be spaced evenly across the genome, and the array set can be enriched with selected clones that contain candidate cancer genes to enhance resolution. The use of DNA microarrays has been suggested for other applications; however, epigenetic changes such as changes in DNA methylation, which are likely to be a critical component in the development of cancer, have been notoriously difficult to assay regardless of the format.
The power of a comparative genomic hybridization was recently demonstrated by Albertson and colleagues, who used this approach to map the recurrent breast-cancer amplicon at chromosome 20q12.3.30 This approach clearly demonstrated that what had previously been described as a single amplicon was, in fact, two distinct amplicons, one containing the putative oncogene ZNF21731 and the other containing CYP24, which encodes vitamin D24-hydroxylase.32 The overexpression of this enzyme alters the control of growth mediated by vitamin D. There were two distinct peaks of high copy numbers within this 2-Mb region, with a gene at the peak of each amplicon. The ability of comparative genomic hybridization to show peaks in increases in copy numbers across regions of recurrent abnormality at high resolution is very useful for locating oncogenes in many human cancers.i#;$!t, 百拇医药
Much less advanced, but critically important, are techniques involving proteomics, which examine the entire complement of proteins expressed in a specific tissue or cell. The information supplied complements that provided by a comparative genomic hybridization, expression profiling, and screening for mutations in cancer research,33 since the genetic code does not indicate which proteins are expressed, in what quantity, and in what form. For example, post-translational modifications, such as phosphorylation or glycosylation, may determine the function or stability of a protein and are not detected by transcriptional analyses. Many differences between normal tissue and malignant tumors are due to post-translational modifications, and a complete analysis of the cancer phenotype will require a whole-proteome approach.
Diffuse large -cell lymphoma was the first human cancer to undergo gene-expression profiling, and a microarray containing 17,800 cDNAs was used.34 Breast and ovarian cancer have now been subjected to molecular profiling as well. In the first such study, Perou and colleagues used a cDNA microarray containing 8000 genes to assay 65 breast-biopsy specimens, primarily invasive breast cancers.35 Perhaps not surprisingly, estrogen-receptor status was a key predictor of the outcome and treatment response, with estrogen-receptor and coregulated genes being the primary elements needed for these tumors to cluster. In addition, a profile of tumors that overexpress ERBB2 was easily identifiable. Thus, the primary clusters recognized were tumors that expressed estrogen receptor and had a luminal-cell pattern of gene expression, tumors that did not express estrogen receptor and had a myoepithelial-cell pattern of expression, tumors that overexpressed ERBB2, and a fourth group of tumors that clustered with normal breast tissue. More recently, Hedenfalk and colleagues suggested that transcriptional profiling can also accurately differentiate breast cancers with underlying germ-line mutations in BRCA1 or BRCA2 from those without such mutations, an advance that could facilitate the identification of high-risk families on the basis of molecular phenotyping, as well as identify characteristic molecular differences that may be useful clinical targets for directed therapy.36
Ovarian cancers have been subjected to transcriptional profiling with similar results. In one series, 27 serous papillary ovarian cancers and 3 samples of normal ovarian tissue underwent gene-expression profiling with oligonucleotide-based arrays representing more than 6000 human genes.37 Normal ovarian tissue was clearly distinguishable from malignant tissues, and three types of tumors were identified. The first subtype clustered with normal tissue and was well differentiated on conventional histologic analyses. The second group was characterized by the expression of genes from admixed stromal cells and infiltrating lymphocytes. Although this profile could represent a random admixture of cell types, it could also represent an immune response to the tumor, as suggested recently by Zhang and colleagues.38 The third group of tumors had a high level of expression of cell-cycle–associated genes, most likely reflecting a high proliferative rate. More than half the tumors in this cluster were poorly differentiated on histologic analysis. Which of the differentially expressed genes in this and other series represent the root causes of malignant transformation rather than markers of progression is not known but must be determined in order to distinguish diagnostic markers from therapeutic targets.
Microarrays have also been used to show how an expression profile can change as cancer cells develop resistance to doxorubicin-based therapy.39 In these experiments, a set of genes that were transiently overexpressed after initial exposure to doxorubicin included a subgroup of genes that became constitutively overexpressed as resistance to doxorubicin developed. These experiments, although just the beginning of what will be a fundamental change in molecular oncology owing to the deciphering of the human genome sequence, demonstrate the power of this approach.q]i'2p, 百拇医药
In perhaps the most extensive and informative study to date, the expression profiles of 117 primary breast cancers were compared with known prognostic markers and the clinical outcome at least five years after diagnosis.40 Expression profiling with the use of 25,000 genes separated the tumors into two groups, one in which distant metastases developed in 34 percent at five years and one in which metastases developed in 70 percent at five years. From the original 25,000 genes in the array, 70 were identified as having the greatest accuracy in predicting recurrent disease. When the tumors were sorted on the basis of this smaller set of genes, fewer than 10 percent of the tumors in the poor-prognosis group were misclassified. A comparative multivariate analysis using clinical prognostic factors that included tumor grade, tumor size, the presence or absence of angiolymphatic invasion, patients' age, and tumor estrogen-receptor status demonstrated that as compared with the good-prognosis gene-expression signature, the poor-prognosis microarray profile was an independent predictor of recurrence, with an odds ratio of 18 (95 percent confidence interval, 3 to 94). This approach has now been tested in 295 consecutive patients with stage I or II breast cancer.41 Of these, 180 had the poor-prognosis profile and 115 had the good-prognosis profile. Ten years after the diagnosis of breast cancer, the probability of remaining free of metastases was 51 percent among women with a poor-prognosis profile and 85 percent among those with a good-prognosis profile. These data provide compelling evidence that the genetic program of a cancer cell at diagnosis defines its biologic behavior many years later, refuting a competing hypothesis that the genetic changes driving the development of metastatic disease are acquired in residual cells after adjuvant treatment.
Clinical Management of Inherited Susceptibility to Breast and Ovarian Cancer??:}%w%, http://www.100md.com
Several computational models have been developed to predict an individual woman's risk of breast cancer, including one in which family history is the predominant risk factor. This model, developed by Claus and colleagues and published as a series of tables clinicians can use,42 is based on the number and degree of relatedness of family members with breast cancer and their age at diagnosis. However, this model does not provide estimates of the likelihood that an individual woman will have a germ-line mutation in BRCA1 or BRCA2. Several studies have identified factors that are associated with an increased likelihood that a BRCA1 or BRCA2 mutation will be identified, including early-onset breast cancer, the occurrence of breast and ovarian cancer in the same woman, a history of male family members with breast cancer, and Ashkenazi Jewish ancestry. These characteristics have also been included in predictive models designed for use by clinicians.43,44,45
Testing for germ-line mutations in BRCA1 and BRCA2 is an important tool for predicting the risk of breast cancer and developing management strategies. Once such mutations are identified, we recommend that the woman choose between annual screening mammography and prophylactic mastectomy, which significantly reduced the risk of breast cancer in a small, retrospective study of mutation carriers.46 We recommend that women who choose surveillance also investigate the possibility of participating in a clinical trial evaluating the utility of magnetic resonance imaging for screening high-risk women. Several studies have shown that in women with germ-line BRCA1 and BRCA2 mutations, breast cancers are likely to occur as interval cancers47 and that standard mammograms are more likely to be negative than in women at low or moderate risk.48,49,50|;'-, http://www.100md.com
With respect to the risk of ovarian cancer among carriers of BRCA1 and BRCA2 mutations, we strongly recommend that such women undergo prophylactic oophorectomy as soon as they have completed childbearing, since no surveillance regimen to date has been shown to decrease the percentage of women who receive a diagnosis of advanced disease. Among mutation carriers, this procedure has been shown to reduce the risk of breast and ovarian cancer by more than 60 percent and 95 percent, respectively.51,52 Again, although no prospective data are available, we recommend that these women receive hormone-replacement therapy until the age of 50 years, approximately the time of natural menopause. Although extending exposure to estrogens beyond the age of 50 years has been associated with a small increase in the risk of breast cancer, these younger women would be producing endogenous estrogens in the absence of prophylactic oophorectomy. The addition of hormone-replacement therapy makes this choice acceptable to women who would otherwise refuse it because of concern about premature menopause, and the risk–benefit ratio is strongly in favor of oophorectomy, with or without hormone-replacement therapy.
The use of tamoxifen to prevent breast cancer in carriers of BRCA1 and BRCA2 mutations remains controversial. One retrospective study suggested that adjuvant tamoxifen therapy in carriers of BRCA1 and BRCA2 mutations with estrogen-receptor–positive breast cancer reduced the risk of contralateral breast cancers by the same amount as that in unselected patients with breast cancer.53 However, data showing that most BRCA1-associated breast cancers are negative for estrogen receptors54 and recent data from the Breast Cancer Prevention Trial (BCPT) of the National Surgical Adjuvant Breast and Bowel Project have led to widespread speculation that tamoxifen will not prevent breast cancer in women with germ-line BRCA1 mutations.55 It is important to consider that all available data suggest that endogenous exposure to hormones has a central role in defining the risk of cancer among carriers of BRCA1 mutations, that breast cancer developed in only eight carriers of BRCA1 mutations in the BCPT, and that the lack of a preventive effect of tamoxifen was statistically insignificant (odds ratio, 1.67; 95 percent confidence interval, 0.41 to 8.00). The length of treatment in the BCPT is also consistent with an early treatment effect, rather than true prevention, which, given the preponderance of estrogen-receptor–negative tumors among carriers of BRCA1 mutations, would produce the data seen in this study. Thus, we recommend that carriers of BRCA1 mutations consider taking tamoxifen once they discontinue hormone-replacement therapy at about the age of 50 years.
Summary5h}{&q, 百拇医药
The past decade has been a period of unparalleled discovery in the field of the genetics and genomics of breast and ovarian cancer. Two major susceptibility genes have been isolated, and subsequent work provided sufficient management information to allow genetic testing for BRCA1 and BRCA2 mutations to become a part of routine practice in many clinical centers. In addition, work has begun on the characterization of genetic variants that, although associated with a lower risk of cancer than germ-line BRCA1 and BRCA2 mutations, are far more common in the population and thus may have a substantial role in defining the risk of cancer. Finally, gene-expression profiling, coupled with the sequencing of most or all of the genes in the human genome, is revolutionizing the study of the biology and the molecular classification of breast and ovarian cancer. Combined with data from projects conducting a genome-wide mutation analysis of all genes implicated in the development of cancer, the importance of which has just been illustrated with the discovery that more than 60 percent of melanomas have mutations in BRAF (v-raf murine sarcoma viral oncogene homologue B1),56 and progress in developing effective preventive measures, a marked reduction in mortality from breast and ovarian cancer is a realistic goal for the next decade.5h}{&q, 百拇医药
Source Information5h}{&q, 百拇医药
From the Wellcome Trust Sanger Institute, Hinxton, Cambridge, United Kingdom (R.W.); and the University of Pennsylvania, Philadelphia(Richard Wooster Ph.D. and Barbara L. Weber M.D.)
PDF of this article+, http://www.100md.com
PDA version of this article+, http://www.100md.com
Editor's Summary+, http://www.100md.com
CME Exam+, http://www.100md.com
Related Letters to the Editor+, http://www.100md.com
Find Similar Articles in the Journal+, http://www.100md.com
Notify a friend about this article+, http://www.100md.com
Articles citing this article+, http://www.100md.com
Ethical, Legal, and Social Implications of Genomic Medicine+, http://www.100md.com
Ellen Wright Clayton, M.D., J.D.+, http://www.100md.com
Table of Contents+, http://www.100md.com
PDF of this article+, http://www.100md.com
PDA version of this article+, http://www.100md.com
Editor's Summary+, http://www.100md.com
CME Exam+, http://www.100md.com
Related Letters to the Editor+, http://www.100md.com
Find Similar Articles in the Journal
Notify a friend about this article5h}{&q, 百拇医药
Articles citing this article5h}{&q, 百拇医药
Services 5h}{&q, 百拇医药
Add to Personal Archive5h}{&q, 百拇医药
Download to Citation Manager5h}{&q, 百拇医药
Alert me when letters appear5h}{&q, 百拇医药
Alert me when this article is cited5h}{&q, 百拇医药
ISI Web of Science5h}{&q, 百拇医药
Related Articles5h}{&q, 百拇医药
Citing Articles (8)5h}{&q, 百拇医药
Medline 5h}{&q, 百拇医药
Related Articles in Medline5h}{&q, 百拇医药
Articles in Medline by Author:5h}{&q, 百拇医药
Clayton, E. W.5h}{&q, 百拇医药
Medline Citation5h}{&q, 百拇医药
Genetics5h}{&q, 百拇医药
Legal Issues in Medicine5h}{&q, 百拇医药
Medical Ethics5h}{&q, 百拇医药
Related Chapters at Harrison's Online
As detailed in other articles inthe Genomic Medicine series, genomics has contributegreatly to our understanding of the molecular basis of disease and, to a lesser but growing extent, to the development of effective interventions. Clinicians and society at largewever, are concerned about the effect geneticknowle will have on the well-being of individual persons and groups. Much effort is being devoted to trying to anticipate, understand, and address the ethical, legal, social, and political implications of genetics and genomics..x?@^, http://www.100md.com
The inquiry is complex. Understanding the social effects of genomics requires an analysis of the ways in which genetic information and a genetic approach to disease affect people individually, within their families and communities, and in their social and working lives. Genomics presents particular challenges with respect to clinicians' ethical and professional responsibilities, including the appropriate use of genomic information in the health care setting. In this article, I examine public concerns about genetic information and discuss a few recent cases in some depth to highlight a few of the dilemmas presented by genomics and emerging solutions.
What Genetic Information Is and What People Are Worried About|85w}pw, 百拇医药
Genes affect virtually all human characteristics and diseases. These influences can be ascertained in individual patients through a review of the family history, physical examination, and the use of medical diagnostics. In some conditions, such as cystic fibrosis and sickle cell disease, the specific molecular mechanisms are largely understood, but in many, including such common chronic diseases as diabetes mellitus and hypertension, the relevant genes — and there are often many — are only beginning to be identified. Given the variety of these effects and the limits of our knowledge, it is not surprising that the term "genetic information" is used in different ways at different times. Sometimes it is used to mean the influence of the entire genome, but more often it is used to refer to recognized, single-gene disorders or, even more narrowly, the results of DNA-based tests. These various meanings may make sense in context, but confusion can occur unless the speaker and listener are defining the term in the same way.
The most commonly expressed fear is that genetic information will be used in ways that could harm people — for example, to deny them access to health insurance, employment, education, and even loans. Part of that concern is fueled by the growing recognition that health information is not entirely private, despite people's expectations and desires to the contrary. In fact, both federal and state governments have been actively engaged in discussions about who ought to have access to health information and under what conditions.1,2 This debate is informed appropriately by the recognition that limiting access to the medical record to the patient and the treating clinician is neither possible nor unequivocally desirable.31rs$\(, http://www.100md.com
People tend to see genetic information as more definitive and predictive than other types of data, in the sense that "you cannot change your genes" and that "genes tell all about your future." This notion of genetic determinism, however, includes an unwarranted sense of inevitability, because it reflects a fundamental failure to understand the nature of biologic systems. The DNA sequence is not the Book of Life. Human characteristics are the product of complex interactions over time between genes — both a person's own and those of other organisms — and the environment. Both germ-line and somatic cells undergo mutations, the latter being a primary way in which cancer develops. Moreover, a pathogenic mutation does not doom one to ill health; many diseases can be treated. As is true for so many conditions in medicine, clinicians have a variable but usually limited ability to predict when, how severely, and even whether a person with a genetic predisposition to a certain illness is going to become ill.
One might be tempted to conclude that the way to allay people's fears about genetics is simply to give them a more realistic understanding of the informative power of these tools. Given the optimistic predictions about genetics that pervade the media and public opinion today, that path is unlikely to succeed in the short term. A more promising approach to addressing the social implications of genetics requires us to consider both how genes are perceived in the real world and what is actually known about their function.m(@p1cd, http://www.100md.com
The Problem of Discriminationm(@p1cd, http://www.100md.com
The question of whether genetic information should ever be used to affect one's access to health and other forms of insurance has been a dominant issue of public concern in the past decade. People cite fear of losing insurance as a major reason to avoid genetic testing.4 Others argue that discrimination by insurance companies is not a problem, often pointing out that few of these cases, which are difficult for employees to win, have been filed.5 Insurers assert that they do not perform tests to obtain genetic information but argue that they should be free to use such information if it is available, citing the need to avoid "moral hazard" — the risk that people who know they will become ill or die soon will try to obtain insurance at regular rates.6 In response to consumer pressure, many states have passed laws in this area .7,8 In passing the Health Insurance Portability and Accountability Act (HIPAA),9 Congress specifically banned certain uses of genetic information in determining insurance eligibility, but it placed no limits on rate setting.10 Vigorous debate about optimal solutions is ongoing,11 and bills have been introduced in every recent session of Congress.12
fig.ommtted:]y/|6c, http://www.100md.com
Summary of Statutes Regarding Discrimination on the Basis of Genetic Information and the Privacy of Such Information.:]y/|6c, http://www.100md.com
The complexity of the issues surrounding discrimination can be illustrated more generally by examining a case involving Burlington Northern Santa Fe Railroad (BNSF). Allegedly relying on the advice of its company physician, who in turn had apparently relied on the representations of a diagnostic company, BNSF began obtaining blood for DNA testing from employees who were seeking disability compensation as a result of carpal tunnel syndrome that occurred on the job. The employees were reportedly not told the purpose of the tests, which was to detect a mutation associated with hereditary neuropathy with liability to pressure palsies.13 The company's motive for pursuing testing was never made clear, but it seems reasonable to suspect that BNSF would have tried to deny disability benefits to any employee who had such a mutation, arguing that the mutation, and not the job, caused the carpal tunnel syndrome. When the company's practice came to light, it was almost immediately stopped by the federal Equal Employment Opportunity Commission,14 and shortly thereafter, the company settled claims brought by its employees for an undisclosed amount of money.15
What lessons can be learned here? One is that the company's effort to find mutations for hereditary neuropathy with liability to pressure palsies made little sense. This disorder is very rare, affecting about 3 to 10 persons per 100,000, and more important, although carpal tunnel syndrome can be a part of hereditary neuropathy with liability to pressure palsies, it has not been reported as the sole symptom. The injuries these employees sustained were not the result of an epidemic of hereditary neuropathy with liability to pressure palsies. Getting the biologic process correct is a critical step in making decisions about genetic testing.:&o3, http://www.100md.com
Another important lesson is that identifying a genetic predisposition to carpal tunnel syndrome would not have been the end of the discussion in the eyes of the law. The company got in trouble because its practice violated numerous laws forbidding discrimination in the workplace. In particular, the Americans with Disabilities Act permits employers to require a medical evaluation only under clearly specified circumstances.16 Testing employees after they were disabled without their informed consent clearly fell outside the bounds of this and other antidiscrimination laws.
The actions of BNSF led to widespread criticism and, not surprisingly, to calls to ban genetic discrimination in the workplace.17 Although some states have enacted laws the need for federal action has grown as the Supreme Court has progressively narrowed the protection provided under the Americans with Disabilities Act.18,19 The answer, however, is not simply to forbid employers to use genetic information or to require genetic testing.7, 百拇医药
The first step in developing an appropriate response is to determine how the use of genetic information fits within the broader framework of antidiscrimination laws, which were passed to create a certain kind of society, one in which people must be included regardless of race, sex, or disability, even at some cost to employers. Biology alone does not determine the social outcome. To use an analogy, an employer cannot exclude women from the workplace, even if he or she believes, with some justification, that women are more likely than men to take time off to care for family members. At the same time, employers are not required to bear unlimited costs to promote these social goals — the employee, male or female, who misses months of work at a time to care for sick relatives can still be fired.
A similar debate about social goals and the limits of our pursuit of them must occur with regard to genetic discrimination. The Equal Employment Opportunity Commission recently awarded damages to Terri Sergeant, who was fired from her job as an office manager for an insurance broker because she required extremely expensive medication to treat her at-worst mildly symptomatic alpha1-antitrypsin deficiency.20 A person's need for expensive health care is not sufficient reason to fire that person or to refuse to hire him or her in the first place. The fact that the costs may cause the employer to go under or to decide not to provide health insurance simply underlines the inherent weakness of employment-based health insurance.g, http://www.100md.com
At the same time, one can imagine a genetic condition that might affect a person's ability to perform a job in ways that could not be accommodated with reasonable efforts. Suppose a person with a recurrent and untreatable cardiac arrhythmia that leads to loss of consciousness, owing to an inherited ion-channel defect, is seeking employment as a long-distance truck driver. Because of the risk to third parties, such a person would not even be able to get a driver's license in many jurisdictions. The more difficult question — and the one posed particularly with respect to genetics — would arise if an asymptomatic person had a predisposing, but incompletely penetrant, mutation for the same disorder. Deciding what to do about such predispositions will require close attention both to the true, as opposed to the feared, likelihood that symptoms will develop and to the complex weighing of the interests of the individual, the employer, and society.
A similar calculus must be applied to every question regarding who can obtain and use genetic information to distinguish, or discriminate, among people in ways that affect their ability to obtain social goods, such as health insurance and education .21 If, as is likely, some uses are deemed to be appropriate, the challenge for clinicians will be to discuss with their patients the potential adverse social consequences of testing so that the patients can make informed choices about whether or not to proceed with testing.&^, http://www.100md.com
fig.ommtted&^, http://www.100md.com
Elements to Be Considered in Decisions about the Use of Genetic Information.&^, http://www.100md.com
The Challenge of Genomic Medicine with Respect to the Physician–Patient Relationship&^, http://www.100md.com
Consider the case of a man who died of colon cancer in the 1960s. When the same disease developed in his daughter approximately 25 years later, she obtained her father's pathology slides, discovered that he had had diffuse adenomatous polyposis coli, and sued the estate of her father's surgeon, alleging that the physician should have warned her about her 50 percent risk of having the disorder. An intermediate appellate court in New Jersey ruled that the physician had a duty to warn the daughter directly (she would have been a child at the time of her father's death), perhaps even over her father's objections.22
This is only one court's view in one case, but given how much attention it received, it is important to ask whether this was a good result. Two central tenets of Western medicine are that physicians should focus on the interests of their patients and that they should protect the confidentiality of their patients' medical information. Yet the tools of genomic medicine often reveal information about health risks faced not only by patients but also by their relatives. What should clinicians do? It seems clear that they should tell their patients about the risks faced by family members. The harder questions are whether physicians are ethically permitted to contact the relatives themselves, in contravention of traditional patient-centered norms, and whether they should be legally required to do so.'0, 百拇医药
This issue must be viewed in the light of the fact that the duty to protect confidentiality is not absolute. Physicians are required to report numerous infectious diseases,23 and they have been held liable for failing to warn people whom their patients have specifically threatened with violence.24 The question then becomes more complex: are genetic risks sufficiently similar to these existing exceptions to the requirement of confidentiality that they warrant an exception as well? Over the years, numerous prominent advisory bodies have said no, opining that physicians should be permitted to breach confidentiality in order to warn third parties of genetic risks only as a last resort to avert serious harm.25,26,27
These learned opinions, however, are not the end of the matter, in part because they lack the force of law. In fact, as the case above illustrates, relatives have sued the primary patients' physicians for failing to warn them of their own genetic risks — and won limited victories, although none have been awarded monetary damages. The decisions in the colon-cancer case and a similar one in Florida28 have been criticized for both their legal reasoning and their deviation from ethical guidelines, but they have not been overturned and, in the tradition of the common law, may be persuasive to other courts. Physicians who breach their patients' confidentiality and warn family members are not likely to incur substantial liability, even under HIPAA.29 As a result, physicians might understandably conclude that warning relatives is the least risky option.6s, 百拇医药
The existing directives are thus in conflict: "expert consensus," ethical analysis, and the HIPAA regulations argue for honoring confidentiality, whereas at least one legal opinion holds that physicians fail to warn a patient's relatives at their peril. Given the press of other business, legislators are not likely to resolve this conflict soon. In this setting, clinicians should inform their patients about the risks their relatives face, discuss the appropriateness of sharing this information and offer assistance, trust — usually realistically — that patients will in turn tell their relatives who are at risk, and hope that the courts will get it right in the future.
Genomic Medicine and Public Health(z, 百拇医药
When Sierra Creason underwent state-mandated newborn screening, she had abnormally low levels of both thyroxine and thyrotropin, findings consistent with the presence of congenital hypothyroidism. Her physician was not notified of these results, however, because the state had chosen not to divulge the actual values and, instead, to report as abnormal only results in which thyroxine levels were low and thyrotropin levels were high.30 As a result, the diagnosis of congenital hypothyroidism and subsequent treatment were seriously delayed, resulting in permanent harm. When the child's family sued the state, however, the California Supreme Court ruled that the state program could not be held liable, in part to avoid diverting funds that would have been used for other state purposes. By contrast, had a private diagnostic laboratory given the same report, especially without providing the actual results, which would have enabled the child's physician to make an independent assessment, it almost certainly would have been held responsible.
Complex questions arise when the government requires testing and interventions. State-mandated screening of newborns for metabolic and genetic disorders was described by Khoury et al. in an earlier article in this series.31 Governments undertake many activities to promote health — universal screening of newborns for phenylketonuria, for example, is generally considered a resounding success — but it is worth asking in each case whether there is sufficient justification to pursue mandatory as opposed to voluntary action or to place such activities in the public rather than the private sector. Requiring public health agencies to assume such responsibilities has advantages, such as more transparent accountability to the public and greater uniformity in access and results. Relying on public health entities in matters that directly affect the health of individual persons, however, entails certain risks as well. Physicians and patients count on receiving accurate and informative results regardless of whether a private or a public entity is doing the testing. Permitting state agencies to avoid financial responsibility when their actions harm patients like Sierra Creason is unjust and should raise questions about the wisdom of proposals that would dramatically expand newborn screening.
A public health analysis of genomics, of course, involves more than state-run testing. The broadest question is whether the public's health is actually improved by the knowledge derived. A major determinant is access to testing and to the medical interventions that may be warranted as a result. In our current multipayer system of health care, people will have widely differing levels of access to these forms of technology. One cannot assume that everyone will reap the benefits of this knowledge.93@%i, 百拇医药
From a public health perspective, it might do to go one step further and ask whether people will actually use the test results to alter their behavior in ways that improve health. Some people whom testing identifies as predisposed to cancer subsequently decline to undergo surveillance or other interventions for psychological reasons or because of other demands on their time. Some preventive or therapeutic measures are more likely to be pursued than others; most people find it difficult to take medications for a lifetime or to maintain major lifestyle changes, no matter how important such approaches are for their health.
Public health agencies exist not only to identify barriers to health but also to improve health and health care. Efforts to determine when genetic tests are reliable enough for routine clinical use are quintessential public health activities.32,33 The Secretary's Advisory Committee on Genetic Testing and its successor, the Secretary's Advisory Committee on Genetics, Health, and Society, were formed to provide such guidance.34 The development of strategies to educate health care providers and patients about genomic medicine, a long-standing goal of the Human Genome Project, and to decrease obstacles to health-promoting behavior also falls comfortably within this rubric.(lvwuz2, http://www.100md.com
Conclusions(lvwuz2, http://www.100md.com
This brief discussion illustrates public expectations and fears about the effect of genomics, challenges to the goals of antidiscrimination laws and to the nature of the physician–patient relationship, and the contrasting perspectives and legal rules that apply to personal medical care and public health. Acknowledgment and examination of these complex issues are critical for identifying the appropriate ethical principles that should be applied and for creating the necessary legislative and regulatory responses.
Supported in part by a grant from the National Institutes of Health (5R01 HG 01974-02).87quxo;, http://www.100md.com
I am indebted to William O. Cooper and Jay Clayton for their helpful comments on earlier drafts and to Vanessa M. Spencer for her research assistance.87quxo;, http://www.100md.com
Source Information87quxo;, http://www.100md.com
From the Center for Genetics and Health Policy, Vanderbilt University, Nashville.87quxo;, http://www.100md.com
Address reprint requests to Dr. Clayton at Vanderbilt University, 507A Light Hall, Nashville, TN 37232-0165.87quxo;, http://www.100md.com
References87quxo;, http://www.100md.com
Services 87quxo;, http://www.100md.com
Add to Personal Archive87quxo;, http://www.100md.com
Download to Citation Manager87quxo;, http://www.100md.com
Alert me when letters appear87quxo;, http://www.100md.com
Alert me when this article is cited87quxo;, http://www.100md.com
ISI Web of Science87quxo;, http://www.100md.com
Related Articles87quxo;, http://www.100md.com
Citing Articles (9)
Medline np/mr*, 百拇医药
Related Articles in Medlinenp/mr*, 百拇医药
Articles in Medline by Author:np/mr*, 百拇医药
Wooster, R.np/mr*, 百拇医药
Weber, B. L.np/mr*, 百拇医药
Medline Citationnp/mr*, 百拇医药
Breast Cancernp/mr*, 百拇医药
Geneticsnp/mr*, 百拇医药
Oncologynp/mr*, 百拇医药
Women抯 Healthnp/mr*, 百拇医药
Related Chapters at Harrison's Onlinenp/mr*, 百拇医药
Despite years of intensive study and substantial progress in understanding susceptibility to breast and ovarian cancer, these diseases remain important causes of death in women. However, several recent critical advances — sequencing of the human genome and the development of high-throughput techniques for identifying DNA-sequence variants, changes in copy numbers, and global expression profiles — have dramatically accelerated the pace of research aimed at preventing and curing these diseases. We review some of the important discoveries in the genetics of breast and ovarian cancer, ongoing studies to isolate additional susceptibility genes, and early work on molecular profiling involving microarrays.
Susceptibility to Breast and Ovarian Cancer44!)0, 百拇医药
In the United States, 10 to 20 percent of patients with breast cancer and patients with ovarian cancer have a first- or second-degree relative with one of these diseases.1 Two major genes associated with susceptibility to breast and ovarian cancer — breast cancer susceptibility gene 1 (BRCA1) and breast cancer susceptibility gene 2 (BRCA2) — have been identified to date.2,3 Mutations in either of these genes confer a lifetime risk of breast cancer of between 60 and 85 percent and a lifetime risk of ovarian cancer of between 15 and 40 percent.4,5 However, mutations in these genes account for only 2 to 3 percent of all breast cancers,6,7 and susceptibility alleles in other genes, such as TP53, PTEN, and STK11/LKB1, are even less common causes of breast and ovarian cancer .44!)0, 百拇医药
fig.ommtted44!)0, 百拇医药
The Genetics of Breast Cancer.
BRCA1 and BRCA2 mutations occur in approximately 20 percent of families with evidence of inherited susceptibility to breast cancer. Germ-line mutations in TP53 cause the Li–Fraumeni syndrome and account for no more than 1 percent of cases of familial breast cancer, but women who survive the childhood cancers associated with the Li–Fraumeni syndrome have as much as a 90 percent risk of breast cancer.8 Mutations in the cell-cycle–checkpoint kinase gene (CHEK2) account for about 5 percent of all cases of familial breast cancer (defined by the diagnosis of breast cancer in two or more family members before the age of 60 years), but the risk for individual mutation carriers is probably less than 20 percent.9 All other cases of breast cancer are presumed to be due to an undefined number of additional susceptibility genes with various degrees of penetrance, exposure to hormonal and environmental factors, and stochastic genetic events.[k/k, 百拇医药
The prediction that there are common DNA-sequence variants that confer a small but appreciable enhanced risk of cancer has been validated with the recent discovery of the 1100delC mutation in the cell-cycle–checkpoint kinase gene (CHEK2).9 This mutation was found in 1.1 percent of women without breast cancer, 1.4 percent of women with a personal but no family history of breast cancer, and 4.2 percent of index patients from 718 families in which two or more members had been given a diagnosis of breast cancer before the age of 60 years but in which there was no detectable BRCA1 or BRCA2 mutation. This mutation doubles the risk of breast cancer among women and increases the risk among men by a factor of 10. CHEK2, an important component of the cellular machinery that recognizes and repairs damaged DNA, is activated after phosphorylation by the checkpoint gene ATM and in turn activates BRCA1. The role of ATM mutations in the predisposition to the early onset of breast cancer remains controversial, but some missense mutations do appear to increase susceptibility to breast cancer in humans10 and mice.11
There is convincing evidence that additional high-penetrance genes that increase susceptibility to breast cancer exist. In contrast, it has been suggested that, other than BRCA1 and BRCA2, high-penetrance genes that confer susceptibility to ovarian cancer do not exist.12 An ovarian-cancer–susceptibility locus on chromosome 3p22–25 has putatively been identified, but this finding has yet to be confirmed by an independent group.13s?z+./, http://www.100md.com
Many additional genetic variants in low-penetrance susceptibility alleles may moderately increase the risk of breast cancer, ovarian cancer, or both. These genetic variants are much more common in the population than are high-penetrance gene mutations and, thus, in aggregate may make a substantially greater contribution to breast and ovarian cancer in the population than mutations in high-risk genes.14 However, genetic heterogeneity and the rarity of high-penetrance genes make both high- and low-penetrance genes difficult to identify.
Identification of Genes That Increase Susceptibility to Breast and Ovarian Cancer|\1tn, 百拇医药
Identification of High-Penetrance Genes|\1tn, 百拇医药
Genetic linkage was used to identify the BRCA1 and BRCA2 loci on chromosomes 17q and 13q, respectively.15,16 In both cases, no information was initially available on the location, structure, or function of the genes, and they were identified through positional cloning. Loss-of-heterozygosity mapping was of little assistance in either search; however, a homozygous deletion on chromosome 13 in a pancreatic adenocarcinoma helped identify the location of BRCA2.17 Finally, critical data in the search for BRCA2 came from studies of breast cancer in Iceland, whose population derives from a small group of settlers from Norway and Ireland.18,19 Such populations share more genetic information than large, admixed populations and have been used successfully many times in gene mapping.20 After the BRCA1 locus was identified, it took almost four years to isolate the gene and involved several labor-intensive strategies.2 By contrast, the BRCA2 locus was one of the first genomic intervals to be systematically sequenced as part of the Human Genome Project. These data, together with other information about the genes in the region, reduced the time it took to isolate BRCA2 to two years.3 Thus, information from the human genome sequence greatly enhances the utility of linkage analysis for gene identification .
fig.ommtted*5f2$, 百拇医药
Effect of Sequencing the Human Genome on Gene-Discovery Strategies.*5f2$, 百拇医药
The annotated DNA sequence of the human genome can be used to locate genes, repeat sequences, and other features and has revolutionized the identification of cancer genes. A sequence without annotation is of limited utility (Panel A). As shown in Panel B, an annotated sequence shows genetic markers such as CA and GT repeats along with other data, such as CpG islands, known genes, genes predicted to exist on the basis of computational models, and Basic Local Alignment Search Tool (BLAST) matches. Using publicly available data (http://www.ensembl.org, http://www.ncbi.nlm.nih.gov, and http://www.genome.ucsc.edu), it is possible to jump from a genetic region of interest to the identification of candidate genes in a matter of seconds and download the relevant data (Panel C). With these data in hand, experiments, such as those involving the polymerase chain reaction (PCR), can be designed to analyze the genes for mutations (Panel D). The final step in the identification of genes is to compare the sequence from patients with the disease of interest with the normal reference sequence to discover the mutations (Panel E).
BRCA3 and beyond#lt@, 百拇医药
Several candidate regions for BRCA3 have been proposed, including chromosome 13q2121 and chromosome 8p12–22,22 but both have been strongly refuted by analysis of data from independent families.23,24 The search for BRCA3 has been difficult for several reasons. First, ovarian cancer and male breast cancer were recognized as components of syndromes of breast-cancer susceptibility before either BRCA1 or BRCA2 was isolated, allowing targeted identification of affected families. Since no such phenotype has been associated with the putative BRCA3 gene or genes, families in current studies are selected only on the basis of a young age at the diagnosis of breast cancer and the absence of ovarian and male breast cancer. Ideally, these families should have multiple members with early-onset breast cancer and strong evidence against the involvement of either BRCA1 or BRCA2. However, the breast cancers in most such families are in fact due to germ-line mutations in BRCA1 or BRCA2,25 and those that are not may represent the effects of multiple susceptibility alleles (genetic heterogeneity), reducing the power of linkage analysis. What is needed to further this effort are larger families to increase the statistical power of such studies, as well as novel means of clustering families into subgroups most likely to represent single-gene disorders.
One approach is to classify families with breast cancer according to the molecular profile of the associated tumors. These analyses could be based either on expression profiling or on array-based comparative genomic hybridization, both of which provide unique molecular signatures. Nonetheless, hundreds of small pedigrees may be needed to identify the BRCA3 locus.)x!r, 百拇医药
fig.ommtted)x!r, 百拇医药
Array-Based Comparative Genomic Hybridization.)x!r, 百拇医药
In Panel A, bacterial artificial chromosome clones or complementary DNAs are placed on glass slides at high density; tumor and normal DNA are labeled with CY3 and CY5, respectively; and the combined probe is hybridized to the array. The array is analyzed with use of a laser scanner that reads each color channel individually and then calculates an intensity ratio for each spot. In Panel B, spots with intensity ratios greater than 1.25 (green spots) represent increases in copy number (amplification), and those with intensity ratios of less than 0.75 (red spots) represent decreases in copy number (deletion). Each spot is a DNA segment that can be linked directly to the human genome sequence (Panel C), thus defining changes in the number of copies of a specific gene. In Panel D, the plotting of intensity ratios for the chromosome 9 bacterial artificial chromosome clones on the array in linear order identifies a homozygous loss of CDKN2A in a melanoma cell line.
Use of the Human Genome Sequence to Identify Low-Penetrance Genes2, 百拇医药
As noted, the susceptibility genes identified to date are not responsible for most breast and ovarian cancers, leaving a considerable potential contribution from less penetrant genes. One of the implicit problems in isolating low-penetrance genes is that such genes will rarely produce striking familial patterns involving multiple cases that can be used in traditional linkage studies. An additional concern is that very large studies, with statistical power to evaluate multiple interactions between genes, may be needed before genetic profiles involving this class of genes can be used for risk prediction.14 As the computational methods for finding coding sequences embedded in a sequence of genomic DNA become increasingly powerful, the value of the human genome sequence as a tool for identifying unknown genes also increases. These algorithms for finding genes have largely replaced laborious experimental techniques to identify potential coding sequences of unknown genes for mutation analysis within linkage regions. These methods are another illustration of the fact that it is the annotation of the genomic sequence (i.e., the identification of genes and their function) that brings the sequence to life. Annotation parses the sequence into genes and noncoding regions. By including genomic features such as CpG islands, which mark the promoter regions of many genes, annotations produce a complete rendering of each sequence. Annotated sequences are publicly available in several data bases (http://www.ensembl.org, http://www.ncbi.nlm.nih.gov, and http://www.genome.ucsc.edu) with associated genome browsers.
The depth and value of annotation have also grown through the addition of millions of single-nucleotide polymorphisms, which are invaluable in the search for susceptibility genes.26 One such example is the recent demonstration that a silent single-nucleotide polymorphism in LIG4, a gene encoding a DNA ligase important in the repair of breaks in double-stranded DNA, is associated with survival among patients with breast cancer.27 This effect was demonstrated in a British population-based study that included 2430 cases of breast cancer. DNA from these patients was genotyped for polymorphisms in 22 DNA-repair, hormone-metabolism, carcinogen-metabolism, and other genes, and the effect of each single-nucleotide polymorphism on the outcome was assessed by Cox regression analysis. The silent polymorphism D501D (t>c) in LIG4 had the largest effect. The estimated hazard ratio for death among patients homozygous for the polymorphism, as compared with those homozygous for the wild-type sequence, was 4.0 (95 percent confidence interval, 2.1 to 7.7; P=0.002), and this effect remained significant after stratification according to tumor stage, grade, and type (hazard ratio, 4.2; 95 percent confidence interval, 1.8 to 9.4; P=0.01). The inclusion of these single-nucleotide polymorphisms in the annotation of the human genome sequence greatly facilitated this analysis, which would otherwise have had to have been preceded by an extensive sequence-based effort to identify single-nucleotide polymorphisms.
Whole-Genome Approaches to the Analysis of Breast and Ovarian Cancerj, 百拇医药
Many genomic approaches to the identification of cancer genes are based on microarray techniques. For gene-expression profiling, each element usually represents one gene and is created with the use of a complementary DNA (cDNA) or oligonucleotide for the gene in question. Similar arrays have been produced with large genomic clones for array-based comparative genomic hybridization to identify changes in the number of copies of DNA. This approach replaced lower-resolution comparative genomic hybridization of cells in metaphase and provides a direct link to genes in the altered region .j, 百拇医药
A comparative genomic hybridization can be used to identify the loss of one or both copies of a given gene as well as regions of amplification. Arrays made with cDNAs can be used for expression profiling and comparative genomic hybridization simultaneously.28 This approach allows a direct comparison between the number of copies of a gene and the level of expression of that gene, but the results of comparative genomic hybridization may be variable, presumably because the cDNA sequence and the genomic sequence are not collinear. However, the alternative approach of using large cloned segments of genomic DNA in the bacterial artificial chromosomes consistently provides excellent data.29 The genomic clones can be spaced evenly across the genome, and the array set can be enriched with selected clones that contain candidate cancer genes to enhance resolution. The use of DNA microarrays has been suggested for other applications; however, epigenetic changes such as changes in DNA methylation, which are likely to be a critical component in the development of cancer, have been notoriously difficult to assay regardless of the format.
The power of a comparative genomic hybridization was recently demonstrated by Albertson and colleagues, who used this approach to map the recurrent breast-cancer amplicon at chromosome 20q12.3.30 This approach clearly demonstrated that what had previously been described as a single amplicon was, in fact, two distinct amplicons, one containing the putative oncogene ZNF21731 and the other containing CYP24, which encodes vitamin D24-hydroxylase.32 The overexpression of this enzyme alters the control of growth mediated by vitamin D. There were two distinct peaks of high copy numbers within this 2-Mb region, with a gene at the peak of each amplicon. The ability of comparative genomic hybridization to show peaks in increases in copy numbers across regions of recurrent abnormality at high resolution is very useful for locating oncogenes in many human cancers.i#;$!t, 百拇医药
Much less advanced, but critically important, are techniques involving proteomics, which examine the entire complement of proteins expressed in a specific tissue or cell. The information supplied complements that provided by a comparative genomic hybridization, expression profiling, and screening for mutations in cancer research,33 since the genetic code does not indicate which proteins are expressed, in what quantity, and in what form. For example, post-translational modifications, such as phosphorylation or glycosylation, may determine the function or stability of a protein and are not detected by transcriptional analyses. Many differences between normal tissue and malignant tumors are due to post-translational modifications, and a complete analysis of the cancer phenotype will require a whole-proteome approach.
Diffuse large -cell lymphoma was the first human cancer to undergo gene-expression profiling, and a microarray containing 17,800 cDNAs was used.34 Breast and ovarian cancer have now been subjected to molecular profiling as well. In the first such study, Perou and colleagues used a cDNA microarray containing 8000 genes to assay 65 breast-biopsy specimens, primarily invasive breast cancers.35 Perhaps not surprisingly, estrogen-receptor status was a key predictor of the outcome and treatment response, with estrogen-receptor and coregulated genes being the primary elements needed for these tumors to cluster. In addition, a profile of tumors that overexpress ERBB2 was easily identifiable. Thus, the primary clusters recognized were tumors that expressed estrogen receptor and had a luminal-cell pattern of gene expression, tumors that did not express estrogen receptor and had a myoepithelial-cell pattern of expression, tumors that overexpressed ERBB2, and a fourth group of tumors that clustered with normal breast tissue. More recently, Hedenfalk and colleagues suggested that transcriptional profiling can also accurately differentiate breast cancers with underlying germ-line mutations in BRCA1 or BRCA2 from those without such mutations, an advance that could facilitate the identification of high-risk families on the basis of molecular phenotyping, as well as identify characteristic molecular differences that may be useful clinical targets for directed therapy.36
Ovarian cancers have been subjected to transcriptional profiling with similar results. In one series, 27 serous papillary ovarian cancers and 3 samples of normal ovarian tissue underwent gene-expression profiling with oligonucleotide-based arrays representing more than 6000 human genes.37 Normal ovarian tissue was clearly distinguishable from malignant tissues, and three types of tumors were identified. The first subtype clustered with normal tissue and was well differentiated on conventional histologic analyses. The second group was characterized by the expression of genes from admixed stromal cells and infiltrating lymphocytes. Although this profile could represent a random admixture of cell types, it could also represent an immune response to the tumor, as suggested recently by Zhang and colleagues.38 The third group of tumors had a high level of expression of cell-cycle–associated genes, most likely reflecting a high proliferative rate. More than half the tumors in this cluster were poorly differentiated on histologic analysis. Which of the differentially expressed genes in this and other series represent the root causes of malignant transformation rather than markers of progression is not known but must be determined in order to distinguish diagnostic markers from therapeutic targets.
Microarrays have also been used to show how an expression profile can change as cancer cells develop resistance to doxorubicin-based therapy.39 In these experiments, a set of genes that were transiently overexpressed after initial exposure to doxorubicin included a subgroup of genes that became constitutively overexpressed as resistance to doxorubicin developed. These experiments, although just the beginning of what will be a fundamental change in molecular oncology owing to the deciphering of the human genome sequence, demonstrate the power of this approach.q]i'2p, 百拇医药
In perhaps the most extensive and informative study to date, the expression profiles of 117 primary breast cancers were compared with known prognostic markers and the clinical outcome at least five years after diagnosis.40 Expression profiling with the use of 25,000 genes separated the tumors into two groups, one in which distant metastases developed in 34 percent at five years and one in which metastases developed in 70 percent at five years. From the original 25,000 genes in the array, 70 were identified as having the greatest accuracy in predicting recurrent disease. When the tumors were sorted on the basis of this smaller set of genes, fewer than 10 percent of the tumors in the poor-prognosis group were misclassified. A comparative multivariate analysis using clinical prognostic factors that included tumor grade, tumor size, the presence or absence of angiolymphatic invasion, patients' age, and tumor estrogen-receptor status demonstrated that as compared with the good-prognosis gene-expression signature, the poor-prognosis microarray profile was an independent predictor of recurrence, with an odds ratio of 18 (95 percent confidence interval, 3 to 94). This approach has now been tested in 295 consecutive patients with stage I or II breast cancer.41 Of these, 180 had the poor-prognosis profile and 115 had the good-prognosis profile. Ten years after the diagnosis of breast cancer, the probability of remaining free of metastases was 51 percent among women with a poor-prognosis profile and 85 percent among those with a good-prognosis profile. These data provide compelling evidence that the genetic program of a cancer cell at diagnosis defines its biologic behavior many years later, refuting a competing hypothesis that the genetic changes driving the development of metastatic disease are acquired in residual cells after adjuvant treatment.
Clinical Management of Inherited Susceptibility to Breast and Ovarian Cancer??:}%w%, http://www.100md.com
Several computational models have been developed to predict an individual woman's risk of breast cancer, including one in which family history is the predominant risk factor. This model, developed by Claus and colleagues and published as a series of tables clinicians can use,42 is based on the number and degree of relatedness of family members with breast cancer and their age at diagnosis. However, this model does not provide estimates of the likelihood that an individual woman will have a germ-line mutation in BRCA1 or BRCA2. Several studies have identified factors that are associated with an increased likelihood that a BRCA1 or BRCA2 mutation will be identified, including early-onset breast cancer, the occurrence of breast and ovarian cancer in the same woman, a history of male family members with breast cancer, and Ashkenazi Jewish ancestry. These characteristics have also been included in predictive models designed for use by clinicians.43,44,45
Testing for germ-line mutations in BRCA1 and BRCA2 is an important tool for predicting the risk of breast cancer and developing management strategies. Once such mutations are identified, we recommend that the woman choose between annual screening mammography and prophylactic mastectomy, which significantly reduced the risk of breast cancer in a small, retrospective study of mutation carriers.46 We recommend that women who choose surveillance also investigate the possibility of participating in a clinical trial evaluating the utility of magnetic resonance imaging for screening high-risk women. Several studies have shown that in women with germ-line BRCA1 and BRCA2 mutations, breast cancers are likely to occur as interval cancers47 and that standard mammograms are more likely to be negative than in women at low or moderate risk.48,49,50|;'-, http://www.100md.com
With respect to the risk of ovarian cancer among carriers of BRCA1 and BRCA2 mutations, we strongly recommend that such women undergo prophylactic oophorectomy as soon as they have completed childbearing, since no surveillance regimen to date has been shown to decrease the percentage of women who receive a diagnosis of advanced disease. Among mutation carriers, this procedure has been shown to reduce the risk of breast and ovarian cancer by more than 60 percent and 95 percent, respectively.51,52 Again, although no prospective data are available, we recommend that these women receive hormone-replacement therapy until the age of 50 years, approximately the time of natural menopause. Although extending exposure to estrogens beyond the age of 50 years has been associated with a small increase in the risk of breast cancer, these younger women would be producing endogenous estrogens in the absence of prophylactic oophorectomy. The addition of hormone-replacement therapy makes this choice acceptable to women who would otherwise refuse it because of concern about premature menopause, and the risk–benefit ratio is strongly in favor of oophorectomy, with or without hormone-replacement therapy.
The use of tamoxifen to prevent breast cancer in carriers of BRCA1 and BRCA2 mutations remains controversial. One retrospective study suggested that adjuvant tamoxifen therapy in carriers of BRCA1 and BRCA2 mutations with estrogen-receptor–positive breast cancer reduced the risk of contralateral breast cancers by the same amount as that in unselected patients with breast cancer.53 However, data showing that most BRCA1-associated breast cancers are negative for estrogen receptors54 and recent data from the Breast Cancer Prevention Trial (BCPT) of the National Surgical Adjuvant Breast and Bowel Project have led to widespread speculation that tamoxifen will not prevent breast cancer in women with germ-line BRCA1 mutations.55 It is important to consider that all available data suggest that endogenous exposure to hormones has a central role in defining the risk of cancer among carriers of BRCA1 mutations, that breast cancer developed in only eight carriers of BRCA1 mutations in the BCPT, and that the lack of a preventive effect of tamoxifen was statistically insignificant (odds ratio, 1.67; 95 percent confidence interval, 0.41 to 8.00). The length of treatment in the BCPT is also consistent with an early treatment effect, rather than true prevention, which, given the preponderance of estrogen-receptor–negative tumors among carriers of BRCA1 mutations, would produce the data seen in this study. Thus, we recommend that carriers of BRCA1 mutations consider taking tamoxifen once they discontinue hormone-replacement therapy at about the age of 50 years.
Summary5h}{&q, 百拇医药
The past decade has been a period of unparalleled discovery in the field of the genetics and genomics of breast and ovarian cancer. Two major susceptibility genes have been isolated, and subsequent work provided sufficient management information to allow genetic testing for BRCA1 and BRCA2 mutations to become a part of routine practice in many clinical centers. In addition, work has begun on the characterization of genetic variants that, although associated with a lower risk of cancer than germ-line BRCA1 and BRCA2 mutations, are far more common in the population and thus may have a substantial role in defining the risk of cancer. Finally, gene-expression profiling, coupled with the sequencing of most or all of the genes in the human genome, is revolutionizing the study of the biology and the molecular classification of breast and ovarian cancer. Combined with data from projects conducting a genome-wide mutation analysis of all genes implicated in the development of cancer, the importance of which has just been illustrated with the discovery that more than 60 percent of melanomas have mutations in BRAF (v-raf murine sarcoma viral oncogene homologue B1),56 and progress in developing effective preventive measures, a marked reduction in mortality from breast and ovarian cancer is a realistic goal for the next decade.5h}{&q, 百拇医药
Source Information5h}{&q, 百拇医药
From the Wellcome Trust Sanger Institute, Hinxton, Cambridge, United Kingdom (R.W.); and the University of Pennsylvania, Philadelphia(Richard Wooster Ph.D. and Barbara L. Weber M.D.)