Journal of Genetic Counseling, Vol. 1, No. 2, 1992
Breast Cancer Risk Analysis: A Genetic Epidemiology Service for Families Patricia T. Kelly 1,2
Breast Cancer Risk Analysis, a service for women with breast cancer and their families, makes use of information drawn from fields such as genetics, epidemiology, and counseling, lts aim is to provide individuals with background information and information regarding the etiology and risks of breast cancer. Based on this information, individuals are helped to make appropriate decisions pertaining to treatment and follow-up. While making use of some of the information and processes used in genetic counseling~ Breast Cancer Risk Analysis differs from most types of genetic counseling for reproductive decisions. Some of these differences are discussed here. KEY WORDS: breast cancer; risk, genetics; epidemiology; family history.
INTRODUCTION In 1978, I was asked to interview participants in a study of women's beliefs and health practices pertaining to breast cancer, all of whom had a mother or sister diagnosed with breast cancer. As a medical geneticist, I had previously counseled individuals concerned about a n u m b e r of different diseases, and so did not anticipate a great difference counseling those with questions about breast cancer risk. I soon learned that individuals and families concerned about breast cancer risk had special needs for information and support. Those initial interviews were the start of a continuing involvement with counseling for breast cancer r i s k - - a n involvement 1Director, Medical Genetics and Cancer Risk Counseling, Salick Health Care, Inc., Los Angeles, California 90048. 2Correspondence should be directed to Patricia T. Kelly, Ph.D., Atta Bates Comprehensive Cancer Center, 5730 Telegraph Avenue, Oakland, California 94609. 155 1059-7700/92/0600-0155506.50/1 9 1992 NationalSocietyof Genetic Counselors,Inc.
156
Kelly
that has evolved and expanded over time. This paper recounts some of my experiences in Breast Cancer Risk Analysis, and presents an overview of the approach that has been developed. The women I interviewed in 1978 were profoundly affected by their relative's diagnosis, even 20 or more years later. For example, one woman told me that as a 9-year-old child she would lay in bed and say each night, "Please God, I know I am going to die of breast cancer, but not this year." W o m e n who were teenagers when their mothers were diagnosed told of acts of rebellion that included attempts to burn down the house, theft, and excessive alcohol consumption. Some remembered intense anger directed toward their father and other family members. Even those who were adults at the time of their mother's diagnosis spoke of how lost they felt during her illness and subsequently. In fact, several were convinced that the disturbance produced by their mother's disease led them to make inappropriate marriage choices. The conviction that she would die young of breast cancer remained strong for many of these women. For example, one woman said she had chosen her boring husband 40 years previously because she knew that when she died of breast cancer, he would be there to take care of her. Now some 40 years later, she asked rather plaintively where her breast cancer was. Another woman said she had hoped to become a missionary, but gave up her plans because she feared dying of breast cancer in a foreign country. Many of the women said they experience difficulty achieving closeness with others. Many expressed regret for not paying more attention to their relative when she was ill, or for not expressing more clearly the concern they felt about her. Those who were teenagers or young adults at the time of their mother's or sister's diagnosis said they still felt guilty about their response at that time. As one explained, "I was so afraid of saying the wrong thing, I didn't say anything." All of the women I interviewed had questions about their own risk of breast cancer. Many said they had been waiting over the years for a breast cancer diagnosis, living as if there were a sword hanging over their heads, or they were a walking time bomb. Frankly, I had been unaware of the emotional impact breast cancer could have on family members. Since I was trained as a medical geneticist, I immediately responded to questions about risks, saying I would do some research and be back in touch soon. I was used to dealing with many different diseases, and so anticipated few problems with my research. I soon learned that there were many contradictory findings, and that information was often presented in a form that was not clinically relevant. It meant little to a woman, for example, to learn that her lifetime risk was 2 to 3 times higher than average. She wanted not to compare herself to a
Breast Cancer Risk Analysis
157
hypothetical average woman, but to know her actual lifetime risk as well as her risk from age 35 to 45 and so on. I then began, with the help of colleagues across the country, to collect such information. Some of these scientists graciously shared information with me before it was published, and agreed to re-analyze their data or even to do new studies to provide clinically useful information. After learning what I could about the risk of breast cancer to sisters and daughters of affected women, I called study participants to tell them what I had learned. To my amazement, several contacted me weeks later with questions about their risks due to other factors such as benign breast disease. In time, friends and relatives of this initial group began to call and ask to come and talk to me. Physicians, also, began to ask about risks to their patients or to ask me to help patients better understand their risks. I will never forget the day that a well-known plastic surgeon asked me about risk to the contralateral breast following a diagnosis of breast cancer. When I responded that I didn't know, he firmly but courteously requested me to find out "before next Thursday" when he said he had to decide whether or not the risk was sufficient to warrant a prophylactic mastectomy of his patient's contralateral breast. Obviously, this physician and his patient were not interested in relative risk, but were seeking information about the actual percent risk at present and in the future. In this manner, my practice of breast cancer risk analysis began. Women consulting me about their own risks of breast cancer or the risks to their sisters and daughters soon let me know that they wanted more than a risk figure, no matter how carefully analyzed and clinically relevant. In addition, patients were clear that they also needed: (1) background information, (2) information about other risk factors such as benign breast disease and exogenous hormone use, and (3) help making use of the information they received to obtain adequate breast health care, and to make appropriate decisions. This included assistance in: (a) finding a physician, (b) obtaining second medical opinions, (c) learning breast self-examination, and (d) making decisions about early detection techniques such as mammography and treatments ranging from biopsy to prophylactic mastectomy. From these patients' questions and concerns arose the initial breast cancer risk analysis service. Over the years, as patients told me their needs, and asked further questions, I adapted the service to be responsive to their questions and concerns. Physicians, also, have consulted me to ask how to communicate with difficult patients, and to ask about the risk to patients with various possible risk factors. This service does much to meet the needs of women and their families who have questions about etiology and risk of breast cancer, as well as the most prudent course to follow given this risk. The service has in large measure been created by patients who were
158
Kelly
kind enough and brave enough to let me know what they needed, and by physician colleagues who forthrightly communicated the help they felt would be of value to them and their patients.
HOW DOES BREAST CANCER RISK ANALYSIS WORK?
Patients are generally seen for three or more visits initially, during which time a pedigree is taken, records on affected relatives are ordered, risk assessed, and information regarding risk conveyed (Kelly, 1991). Many return once a year to keep up with new developments, or come in for a visit when they read or hear something that appears to contradict information discussed during their initial visits. All patients are encouraged to bring a friend or relative with them. For example, I recently saw two sisters diagnosed with breast cancer within 6 months of each other. They came with three as yet unaffected sisters and their unaffected mother. More often, husband and wife come together, two sisters, or a mother and daughter or several friends will come together. Four major areas are covered in breast cancer risk analysis, and will be discussed in turn. They are: background information, risk of cancer, perception of risk, and early detection and treatment.
BACKGROUND INFORMATION Background information includes whatever patients need to learn about biology, medicine, statistics, and individuals' reactions to cancer to help them understand the situation in which they find themselves. For some, it will be as basic as information about cells, nuclei in cells, as well as a quick overview of chromosomes, genes, and DNA. Others will require help in more clearly understanding specifics of their pathology report, oncogenes, or the reactions of friends and relatives.
RISK OF CANCER Risk of cancer, the second major area covered in cancer risk analysis, includes a discussion of the following concepts:
Breast Cancer Risk Analysis
159
Time
In discussing breast cancer risk, one is dealing with risk over time, not the probability at a given time, such as at conception. Most individuals are not aware that risk information without time is uninformative. Patients need to know that in the United States, the risk of breast cancer increases with age. Many are aware of the American Cancer Society estimate that 1 in 9 women (11%) will be diagnosed with breast cancer. However, most don't realize that this is the risk from birth to age 85 (Boring, C., Personal Communication, 1991). The same American Cancer Society calculations estimate a risk of 2% from birth to age 50 and 7% from birth to age 70. The elements of age and time are most important concepts when dealing with a disease that can be diagnosed from early adult years to the end of a woman's life. For health professionals used to dealing with an all or nothing effect, such as the chance of inheriting a particular gene at conception, this change in perspective can come as a surprise.
Relative Risk
Many patients mistakenly think that one can multiply a relative risk by the average woman's risk of 11% to obtain an idea of the percent risk. Such an approach can produce highly inaccurate results, as the following example of benign breast disease demonstrates. Older studies reported that women with benign breast disease had a threefold increase in risk of breast cancer. Recent investigations have shown that benign breast disease is a catch-all category for a heterogenous collection of cell types. Most types of benign breast disease are not associated with an increased risk of breast cancer. Instead, it is primarily those whose benign breast disease is associated with atypia (a cell type defined by pathologists), who are found to be at increased risk~ For example, in one excellent study (Page et al., 1985) women with atypia had a threefold increase in risk of breast cancer. Patients who multiply the threefold increase in risk by the average woman's lifetime risk of 11% may be led to think the risk of invasive breast cancer following a diagnosis of atypia is 33%. Fortunately, the Page et al. study used the same data to present risk in percent and over time. This analysis produced a risk of 10-12% spread over the 15 years following a diagnosis of atypia. That is, the risk of developing breast cancer is less than 1% a year in the initial 15-year period. Why does the multiplication of relative and average risks produce such a distorted figure? First, the average risk includes women with atypia
160
Kelly
so they would be included twice in such a multiplication. Second, most studies have few women over the age of 70, when the risk of developing breast cancer is particularly high. Remember, the average risk is 7% to age 70, not 11%. Therefore, the use of 11% as a base figure is too high. Finally, one needs to know the amount of time women in the study were followed, since time as well as age will influence the risk obtained.
Statistical Significance Most of my patients have not heard of statistical significance and so tend to accept a relative risk figure as more definitive than it may be. Individuals need help in understanding that disease occurrence in two groups can differ as a result of chance. Many benefit by learning the role confidence intervals play in helping to evaluate the likelihood that a difference between two groups is likely to be due to chance.
Life Table Analysis This concept is very important, since it helps avoid the problem of a shrinking denominator over time, as in determining the risk of cancer to sisters of breast cancer patients. In an initial group of 100 women with an affected sister, if one develops invasive disease during the first year of the study, and risk is calculated by percent, a 1% risk is obtained (1 in 100 = 1% risk). As women die from any cause, or develop breast cancer, they are removed from the population at risk. Therefore, at 15 years, there might be only 50 women remaining in the group. If one woman develops cancer at 15 years the risk would be 2% (1 in 50 = 2%). Life table analysis is used to overcome the greater effect of a later case, compared to one diagnosed earlier. As I hope I have made clear, an appreciation of risk requires that the patient have some understanding of concepts such as relative risk and life table analysis as they apply to the most commonly discussed risk factors listed in Table I. One of these factors, family history, is of particular concern to many patients. For the most part cancer is thought to result from an interaction of an individual's genes and the environment. In deference to Knudson's two hit theory, I have called the environmental factors insults (Knudson, 1971). There are families in which breast cancer appears to be inherited in simple, dominant Mendelian fashion (Lynch et al., 1984). However, in view of recent work on retinoblastoma and the growing information on the role
161
Breast Cancer Risk Analysis
Table I. Risk Factors Discussed in Breast Cancer Risk Analysis Age Diet/alcohol Benign breast disease In situ breast cancers Oral contraceptives Replacement hormones
Trauma Family history Reproductive history Breast feeding Stress Personality DES
played by oncogenes in cancer etiology, we now know that what appears to be dominant inheritance clinically may be recessive on the molecular level (Knudson, 1986). Among patients referred to an oncology clinic, nearly one-third were found to have two or more first or second degree affected relatives (Lynch and Lynch, 1986). In an unselected group of patients, which might be more representative of the population as a whole, 18% had either an affected first or second degree relative; 14% had only an affected first degree relative (Anderson, D.E., Personal Communication, 1990). Breast cancer is a frequent disease in the United States, so an individual may have two or more affected relatives due to chance alone, and not to genetic factors. Or, if the family is small and there are few adult women, an individual might have few or no affected relatives and be at high risk genetically. Therefore, families that appear to be similar clinically are probably a heterogenous group. Until biochemical tests are available to distinguish genetic susceptibilities in similar appearing families, empiric risk will be most useful in providing information to concerned individuals. A number of valuable studies have investigated risks to relatives of breast cancer patients (Claus et aL, 1990; Mettlin et aL, 1990; Ottman et aL, 1983, Schwartz et al., 1985). In an attempt to limit heterogeneity, Dr. David Anderson has included in his studies only families in which at least two first degree relatives are affected. Risks have been calculated taking age at diagnosis, laterality (unilateral or bilateral), and occurrence in one, two, or more generations into account. In addition, Dr. Anderson has verified the presence of cancer in all families included in his studies. This is very important, since 20% of first degree and even more of second degree relatives' diagnoses may not be known with accuracy (Love et al., 1985). Dr. Anderson's most recent studies find no difference in risk to sisters of patients diagnosed before and after age 50, as shown in Table II (Anderson and Badzioch, 1985). This finding differs from some of his earlier results and could be due to heterogeneity in the families, larger sample
162
Kelly Table IL Risk of Breast Cancer to Sisters of Patients with Unilateral Disease Patient's age at diagnosis Sister's age 30-39 40-49 50-59 60-69+ Total
_50 b
2% 5% 6% 2%
1% 5% 4% 4%
15%
14%
a276 sisters at risk. b322 sisters at risk.
size, and the use of age 50 in the analysis. As Claus et al. (1990) have recently shown in their case control study, risk to mothers and sisters of affected women increases as age at diagnosis decreases. In their study, a patient's age at diagnosis by decade, starting at 20, was used in analyzing risk to their mothers and sisters (Table III). The number of generations affected seems to play an important role in determining a woman's risk, as shown in Table IV. As you can see, Anderson and Badzioch (1985) obtained a 14% lifetime risk for individuals from families in which a single generation was affected and 20% for those in which women in two generations are affected. Claus et al. (1990) found a lifetime risk of 25% for those with an affected mother and sister compared to an 11% risk for those with an affected sister. These results are surprisingly close, given the different study designs. Laterality appears to strongly influence risk, with bilateral disease leading to a 30% lifetime risk to relatives in some studies (Anderson and Badzioch, 1985) and 50% in others (Ottman et al., 1983). Paternal history must also be evaluated in determining risk due to family history. As Macklin (1959) has so elegantly shown, the risk from
Table IlL Risk of Breast Cancer to Age 59 in Mothers and Sisters of Breast Cancer Patients Age at patient's diagnosis
Risk to mother or sister
20-29 30-39 40-49 50-59
16% 10% 6% 4%
Breast Cancer Risk Analysis
163 Table IV. Risk of Breast Cancer to Sisters of Patients with Unilateral Disease--Family Type Family type Sister's age
30-39 40-49 50-59 60-69+ Total
Ia
IIb
1% 6% 3% 4%
1% 7% 8% 4%
14%
20%
a226 sisters at risk. '5215 sisters at risk.
paternal history is just as great as from maternal. An adequate assessment of risk also includes an evaluation of other cancers in the family, since in some families, first degree relatives of affected individuals have a lifetime risk approaching 50%. So-called cancer families identified to date include those with breast and ovarian cancer, breast cancer, colon cancer, uterine cancer, and others. Perhaps one of the most well-known family types includes breast cancer, sarcoma, leukemia, and brain tumors. Typically, two or more generations are affected with three or more different site-specific tumors. In these families, cancers tend to occur at younger ages than in the general population.
PERCEPTION OF RISK Genetic counselors are, of course, very accomplished in helping patients to deal with perception of risk, and are aware that the effects of a disease on an individual or the family can influence how risk is perceived by them. Some of the more important factors influencing patients' perceptions of cancer risk are listed in Table V and briefly discussed below. A more complete discussion and examples are included in Understanding Breast Cancer Risk (Kelly, 1991).
Table V. Factors Influencing Patients' Perceptions of Risk Information Beliefs about cancer etiology Beliefs about cancer prognosis Anger, guilt, fear Comparison with affected relative(s)
164
Kelly Etiology
Unless the patient's beliefs about etiology are taken into account, relevant information given to her may be discounted. For example, if an individual believes that her risk is increased because she drank from her affected relative's cup, discussions of repressor genes are going to do little to enlighten her.
Prognosis If, when I discuss breast cancer, I mean minimal breast cancer with breast conservation and a 95-99% survival to 10 years, while the patient thinks breast cancer is equivalent to death after long suffering, our communication will suffer. No matter how good a job I do in analyzing risk or in explaining its meaning to the patient, she may interpret the risk of developing breast cancer as the risk of dying from it.
Anger, Fear, and Guilt As discussed earlier, individuals are often quite emotionally and socially affected by a diagnosis of breast cancer in their families. Many are angry that their lives were disrupted by their relatives' illness, pain, and suffering. They may fear they will die or be ill as their relative was, not taking into account advances in early detection or treatment. Some feel guilt about their anger or failure to do more to help their affected relative. All of these factors can act as barriers to the patient's understanding.
Comparison with Affected Relative(s) Patients often believe that their risk is influenced by their resemblance to an affected relative. Some attempt to have a more positive disposition or outlook than their relative's, hoping that such an approach to life's problems will protect them. Many, regardless of resemblance, feel doomed. For this reason, breast cancer risk analysis, to be effective in helping individuals improve the quantity and quality of their lives, must go bey o n d a clear e x p o s i t i o n of risk. In addition, individuals also n e e d information about early detection, breast health care, and treatment.
Breast Cancer Risk Analysis
165
EARLY DETECTION AND TREATMENT Many patients are unaware of the benefits and limitations of early breast cancer detection. For example, many are surprised to learn that mammograms can detect 85-90% of breast cancers, but not 100%. Patients therefore need to learn the value of physical examinations as well as mammograms. Generally speaking, most patients need help in devising a breast health program that is not only safe but which helps them to feel safe. Many are so afraid of the disease or so afraid of being called hypochondriacs, that they don't have a regular follow-up by a breast health center or a breast specialist, and instead rely on scattered examinations throughout the year (Kelly, 1980). As a result, the anxiety suffered by many is tremend o u s - a n d many do not receive sufficient follow-up. Misconceptions about modern treatments may be present and diminish a patient's ability to make the best use of today's methods.
CONCLUSIONS The aim of breast cancer risk analysis is to act as a conduit between relevant medical, social, and scientific information and patients. As I have tried to show, much of the information presented is complex, and requires both time and skill to transmit successfully. When different studies report conflicting results, the construction of the studies themselves must be analyzed. Breast cancer risk analysis is designed to answer questions such as those listed in Table VI. To do so, it employs approaches and information familiar to genetic counselors who help families to make reproductive decisions. For example, genetic counselors realize that beliefs and life experiences can influence patients' perceptions of risk. Genetic counselors are aware that patients must be given appropriate background information if they are to understand sophisticated medical and scientific information. Cancer risk analysis differs from genetic counseling for reproductive decisions in the following ways: (1) emphasis is usually the counseled individual's own health, (2) focus is not the risk at a given time or event, such as conception, but in helping individuals to understand risk over time, and (3) an important aim is to help individuals establish and maintain effective health practices or to choose an appropriate treatment. Patients have great needs for relevant information about all factors that might influence their risk of breast cancer, not just family history. In my own practice, I regularly see women and their families from all over the country. Of course, only those who can afford to travel are able to come to the California service.
166
Kelly Table VI. Q u e s t i o n s Asked by Relatives of W o m e n with Breast Cancer W h e r e does breast cancer come from? Why did it h a p p e n to my relative? W h a t are my risks? W h a t can I do to be safe? How can I help my relative? W h a t can or should I say to her? A m I doing enough? How can I make sense of the conflicting information about breast cancer?
Breast cancer is a very frequent disease in our society, and one which is unique in its emotional and social impact on the patient and the patient's family. New information about breast cancer and breast cancer risk is announced almost weekly, leading to new concerns and questions by patients and their loved ones. Over the years patients have shown us how great their needs are as they have traveled across the country to receive information. The question now is whether we will follow their lead, in providing the genetic epidemiology service they so clearly need, or, in the name of economy, will fail to do so. The provision of breast cancer risk analysis to families at the time of diagnosis would actually save money by instilling in female relatives of breast cancer patients the need for early detection, and providing the peace of mind that would lead to fewer medical utilizations of all types by all family members. In addition to increased quantity of life, these individuals also receive an increased quality of life by having their questions answered, and by being assured that as new advances occur they will learn about them. Genetic counselors are uniquely educated and trained to provide breast cancer risk analysis to women and their families. Not only are they experts in translating technical, difficult, and confusing scientific and medical information to individuals in a manner that can be understood, but they have long appreciated the effects of disease on family structure and family life. It is my hope that in the years ahead genetic counselors will be able to join with me in providing this much needed service to women throughout the country.
REFERENCES A n d e r s o n DE, Badzioch M D (1985) Risk of familial breast cancer. Cancer 56:383-387. Claus EB, Risch NJ, T h o m p s o n , W D (1990) Age at onset as an indicator of familial risk of breast cancer. A m J Epidemiol 131:961-972. Kelly, PT (1980) Counselling needs of w o m e n with a maternal history of breast cancer. Patient Counsel Health Educ 2:118-124.
Breast Cancer Risk Analysis
167
Kelly PT (1991) Understanding Breast Cancer Risk. Philadelphia: Temple University Press. Knudson A G (1971) Mutation and cancer: Statistical study of retinoblastoma. PNAS 68:820-823. Knudson AG (1986) Genetics of human cancer. Ann Rev Genet 20:231-251. Love RR, Evans AM, Josten DM (1985) The accuracy of patient reports of a family history of cancer. J Chron Dis 38:289-293. Lynch HT, and Lynch JF (1986) Breast cancer genetics in an oncology clinic: 328 consecutive patients. Cancer Genet Cytogenet 22:369-371. Lynch HT, Albano WA, Danes S, Layton MA, Kimberling W J, Lynch JF, Cheng SC, Costello K_A, Mulcahy GM, Wagner CA, Tindall SL (1984) Genetic predisposition to breast cancer. Cancer 53:612-622. Macklin MT (1959) Comparison of the number of breast-cancer deaths observed in relatives of breast-cancer patients, and the number expected on the basis of mortality rates. JNCI 22:927-951. Mettlin C, Croghan I, Natarajan N, Lane W (1990) The association of age and familial risk in a case-control study of breast cancer. Am J Epidemiol 131:973-983. Ottman R, Pike MC, King M-C, Henderson BE (1983) Practical guide for estimating risk for familial breast cancer. Lancet ii:556-558. Page DL, Dupont WD, Rogers LW, Rados MS (1985) Atypical hyperplastic lesions of the female breast. Cancer 55:2698-2708. Schwartz AG, King MC, Belle SH, Satariano WA, Swanson GM (1985) Risk of breast cancer to relatives of young breast cancer patients. JNCI 75:665-668.
Breast Cancer Risk Analysis: A Genetic Epidemiology Service for Families Patricia T. Kelly 1,2
Breast Cancer Risk Analysis, a service for women with breast cancer and their families, makes use of information drawn from fields such as genetics, epidemiology, and counseling, lts aim is to provide individuals with background information and information regarding the etiology and risks of breast cancer. Based on this information, individuals are helped to make appropriate decisions pertaining to treatment and follow-up. While making use of some of the information and processes used in genetic counseling~ Breast Cancer Risk Analysis differs from most types of genetic counseling for reproductive decisions. Some of these differences are discussed here. KEY WORDS: breast cancer; risk, genetics; epidemiology; family history.
INTRODUCTION In 1978, I was asked to interview participants in a study of women's beliefs and health practices pertaining to breast cancer, all of whom had a mother or sister diagnosed with breast cancer. As a medical geneticist, I had previously counseled individuals concerned about a n u m b e r of different diseases, and so did not anticipate a great difference counseling those with questions about breast cancer risk. I soon learned that individuals and families concerned about breast cancer risk had special needs for information and support. Those initial interviews were the start of a continuing involvement with counseling for breast cancer r i s k - - a n involvement 1Director, Medical Genetics and Cancer Risk Counseling, Salick Health Care, Inc., Los Angeles, California 90048. 2Correspondence should be directed to Patricia T. Kelly, Ph.D., Atta Bates Comprehensive Cancer Center, 5730 Telegraph Avenue, Oakland, California 94609. 155 1059-7700/92/0600-0155506.50/1 9 1992 NationalSocietyof Genetic Counselors,Inc.
156
Kelly
that has evolved and expanded over time. This paper recounts some of my experiences in Breast Cancer Risk Analysis, and presents an overview of the approach that has been developed. The women I interviewed in 1978 were profoundly affected by their relative's diagnosis, even 20 or more years later. For example, one woman told me that as a 9-year-old child she would lay in bed and say each night, "Please God, I know I am going to die of breast cancer, but not this year." W o m e n who were teenagers when their mothers were diagnosed told of acts of rebellion that included attempts to burn down the house, theft, and excessive alcohol consumption. Some remembered intense anger directed toward their father and other family members. Even those who were adults at the time of their mother's diagnosis spoke of how lost they felt during her illness and subsequently. In fact, several were convinced that the disturbance produced by their mother's disease led them to make inappropriate marriage choices. The conviction that she would die young of breast cancer remained strong for many of these women. For example, one woman said she had chosen her boring husband 40 years previously because she knew that when she died of breast cancer, he would be there to take care of her. Now some 40 years later, she asked rather plaintively where her breast cancer was. Another woman said she had hoped to become a missionary, but gave up her plans because she feared dying of breast cancer in a foreign country. Many of the women said they experience difficulty achieving closeness with others. Many expressed regret for not paying more attention to their relative when she was ill, or for not expressing more clearly the concern they felt about her. Those who were teenagers or young adults at the time of their mother's or sister's diagnosis said they still felt guilty about their response at that time. As one explained, "I was so afraid of saying the wrong thing, I didn't say anything." All of the women I interviewed had questions about their own risk of breast cancer. Many said they had been waiting over the years for a breast cancer diagnosis, living as if there were a sword hanging over their heads, or they were a walking time bomb. Frankly, I had been unaware of the emotional impact breast cancer could have on family members. Since I was trained as a medical geneticist, I immediately responded to questions about risks, saying I would do some research and be back in touch soon. I was used to dealing with many different diseases, and so anticipated few problems with my research. I soon learned that there were many contradictory findings, and that information was often presented in a form that was not clinically relevant. It meant little to a woman, for example, to learn that her lifetime risk was 2 to 3 times higher than average. She wanted not to compare herself to a
Breast Cancer Risk Analysis
157
hypothetical average woman, but to know her actual lifetime risk as well as her risk from age 35 to 45 and so on. I then began, with the help of colleagues across the country, to collect such information. Some of these scientists graciously shared information with me before it was published, and agreed to re-analyze their data or even to do new studies to provide clinically useful information. After learning what I could about the risk of breast cancer to sisters and daughters of affected women, I called study participants to tell them what I had learned. To my amazement, several contacted me weeks later with questions about their risks due to other factors such as benign breast disease. In time, friends and relatives of this initial group began to call and ask to come and talk to me. Physicians, also, began to ask about risks to their patients or to ask me to help patients better understand their risks. I will never forget the day that a well-known plastic surgeon asked me about risk to the contralateral breast following a diagnosis of breast cancer. When I responded that I didn't know, he firmly but courteously requested me to find out "before next Thursday" when he said he had to decide whether or not the risk was sufficient to warrant a prophylactic mastectomy of his patient's contralateral breast. Obviously, this physician and his patient were not interested in relative risk, but were seeking information about the actual percent risk at present and in the future. In this manner, my practice of breast cancer risk analysis began. Women consulting me about their own risks of breast cancer or the risks to their sisters and daughters soon let me know that they wanted more than a risk figure, no matter how carefully analyzed and clinically relevant. In addition, patients were clear that they also needed: (1) background information, (2) information about other risk factors such as benign breast disease and exogenous hormone use, and (3) help making use of the information they received to obtain adequate breast health care, and to make appropriate decisions. This included assistance in: (a) finding a physician, (b) obtaining second medical opinions, (c) learning breast self-examination, and (d) making decisions about early detection techniques such as mammography and treatments ranging from biopsy to prophylactic mastectomy. From these patients' questions and concerns arose the initial breast cancer risk analysis service. Over the years, as patients told me their needs, and asked further questions, I adapted the service to be responsive to their questions and concerns. Physicians, also, have consulted me to ask how to communicate with difficult patients, and to ask about the risk to patients with various possible risk factors. This service does much to meet the needs of women and their families who have questions about etiology and risk of breast cancer, as well as the most prudent course to follow given this risk. The service has in large measure been created by patients who were
158
Kelly
kind enough and brave enough to let me know what they needed, and by physician colleagues who forthrightly communicated the help they felt would be of value to them and their patients.
HOW DOES BREAST CANCER RISK ANALYSIS WORK?
Patients are generally seen for three or more visits initially, during which time a pedigree is taken, records on affected relatives are ordered, risk assessed, and information regarding risk conveyed (Kelly, 1991). Many return once a year to keep up with new developments, or come in for a visit when they read or hear something that appears to contradict information discussed during their initial visits. All patients are encouraged to bring a friend or relative with them. For example, I recently saw two sisters diagnosed with breast cancer within 6 months of each other. They came with three as yet unaffected sisters and their unaffected mother. More often, husband and wife come together, two sisters, or a mother and daughter or several friends will come together. Four major areas are covered in breast cancer risk analysis, and will be discussed in turn. They are: background information, risk of cancer, perception of risk, and early detection and treatment.
BACKGROUND INFORMATION Background information includes whatever patients need to learn about biology, medicine, statistics, and individuals' reactions to cancer to help them understand the situation in which they find themselves. For some, it will be as basic as information about cells, nuclei in cells, as well as a quick overview of chromosomes, genes, and DNA. Others will require help in more clearly understanding specifics of their pathology report, oncogenes, or the reactions of friends and relatives.
RISK OF CANCER Risk of cancer, the second major area covered in cancer risk analysis, includes a discussion of the following concepts:
Breast Cancer Risk Analysis
159
Time
In discussing breast cancer risk, one is dealing with risk over time, not the probability at a given time, such as at conception. Most individuals are not aware that risk information without time is uninformative. Patients need to know that in the United States, the risk of breast cancer increases with age. Many are aware of the American Cancer Society estimate that 1 in 9 women (11%) will be diagnosed with breast cancer. However, most don't realize that this is the risk from birth to age 85 (Boring, C., Personal Communication, 1991). The same American Cancer Society calculations estimate a risk of 2% from birth to age 50 and 7% from birth to age 70. The elements of age and time are most important concepts when dealing with a disease that can be diagnosed from early adult years to the end of a woman's life. For health professionals used to dealing with an all or nothing effect, such as the chance of inheriting a particular gene at conception, this change in perspective can come as a surprise.
Relative Risk
Many patients mistakenly think that one can multiply a relative risk by the average woman's risk of 11% to obtain an idea of the percent risk. Such an approach can produce highly inaccurate results, as the following example of benign breast disease demonstrates. Older studies reported that women with benign breast disease had a threefold increase in risk of breast cancer. Recent investigations have shown that benign breast disease is a catch-all category for a heterogenous collection of cell types. Most types of benign breast disease are not associated with an increased risk of breast cancer. Instead, it is primarily those whose benign breast disease is associated with atypia (a cell type defined by pathologists), who are found to be at increased risk~ For example, in one excellent study (Page et al., 1985) women with atypia had a threefold increase in risk of breast cancer. Patients who multiply the threefold increase in risk by the average woman's lifetime risk of 11% may be led to think the risk of invasive breast cancer following a diagnosis of atypia is 33%. Fortunately, the Page et al. study used the same data to present risk in percent and over time. This analysis produced a risk of 10-12% spread over the 15 years following a diagnosis of atypia. That is, the risk of developing breast cancer is less than 1% a year in the initial 15-year period. Why does the multiplication of relative and average risks produce such a distorted figure? First, the average risk includes women with atypia
160
Kelly
so they would be included twice in such a multiplication. Second, most studies have few women over the age of 70, when the risk of developing breast cancer is particularly high. Remember, the average risk is 7% to age 70, not 11%. Therefore, the use of 11% as a base figure is too high. Finally, one needs to know the amount of time women in the study were followed, since time as well as age will influence the risk obtained.
Statistical Significance Most of my patients have not heard of statistical significance and so tend to accept a relative risk figure as more definitive than it may be. Individuals need help in understanding that disease occurrence in two groups can differ as a result of chance. Many benefit by learning the role confidence intervals play in helping to evaluate the likelihood that a difference between two groups is likely to be due to chance.
Life Table Analysis This concept is very important, since it helps avoid the problem of a shrinking denominator over time, as in determining the risk of cancer to sisters of breast cancer patients. In an initial group of 100 women with an affected sister, if one develops invasive disease during the first year of the study, and risk is calculated by percent, a 1% risk is obtained (1 in 100 = 1% risk). As women die from any cause, or develop breast cancer, they are removed from the population at risk. Therefore, at 15 years, there might be only 50 women remaining in the group. If one woman develops cancer at 15 years the risk would be 2% (1 in 50 = 2%). Life table analysis is used to overcome the greater effect of a later case, compared to one diagnosed earlier. As I hope I have made clear, an appreciation of risk requires that the patient have some understanding of concepts such as relative risk and life table analysis as they apply to the most commonly discussed risk factors listed in Table I. One of these factors, family history, is of particular concern to many patients. For the most part cancer is thought to result from an interaction of an individual's genes and the environment. In deference to Knudson's two hit theory, I have called the environmental factors insults (Knudson, 1971). There are families in which breast cancer appears to be inherited in simple, dominant Mendelian fashion (Lynch et al., 1984). However, in view of recent work on retinoblastoma and the growing information on the role
161
Breast Cancer Risk Analysis
Table I. Risk Factors Discussed in Breast Cancer Risk Analysis Age Diet/alcohol Benign breast disease In situ breast cancers Oral contraceptives Replacement hormones
Trauma Family history Reproductive history Breast feeding Stress Personality DES
played by oncogenes in cancer etiology, we now know that what appears to be dominant inheritance clinically may be recessive on the molecular level (Knudson, 1986). Among patients referred to an oncology clinic, nearly one-third were found to have two or more first or second degree affected relatives (Lynch and Lynch, 1986). In an unselected group of patients, which might be more representative of the population as a whole, 18% had either an affected first or second degree relative; 14% had only an affected first degree relative (Anderson, D.E., Personal Communication, 1990). Breast cancer is a frequent disease in the United States, so an individual may have two or more affected relatives due to chance alone, and not to genetic factors. Or, if the family is small and there are few adult women, an individual might have few or no affected relatives and be at high risk genetically. Therefore, families that appear to be similar clinically are probably a heterogenous group. Until biochemical tests are available to distinguish genetic susceptibilities in similar appearing families, empiric risk will be most useful in providing information to concerned individuals. A number of valuable studies have investigated risks to relatives of breast cancer patients (Claus et aL, 1990; Mettlin et aL, 1990; Ottman et aL, 1983, Schwartz et al., 1985). In an attempt to limit heterogeneity, Dr. David Anderson has included in his studies only families in which at least two first degree relatives are affected. Risks have been calculated taking age at diagnosis, laterality (unilateral or bilateral), and occurrence in one, two, or more generations into account. In addition, Dr. Anderson has verified the presence of cancer in all families included in his studies. This is very important, since 20% of first degree and even more of second degree relatives' diagnoses may not be known with accuracy (Love et al., 1985). Dr. Anderson's most recent studies find no difference in risk to sisters of patients diagnosed before and after age 50, as shown in Table II (Anderson and Badzioch, 1985). This finding differs from some of his earlier results and could be due to heterogeneity in the families, larger sample
162
Kelly Table IL Risk of Breast Cancer to Sisters of Patients with Unilateral Disease Patient's age at diagnosis Sister's age 30-39 40-49 50-59 60-69+ Total
_50 b
2% 5% 6% 2%
1% 5% 4% 4%
15%
14%
a276 sisters at risk. b322 sisters at risk.
size, and the use of age 50 in the analysis. As Claus et al. (1990) have recently shown in their case control study, risk to mothers and sisters of affected women increases as age at diagnosis decreases. In their study, a patient's age at diagnosis by decade, starting at 20, was used in analyzing risk to their mothers and sisters (Table III). The number of generations affected seems to play an important role in determining a woman's risk, as shown in Table IV. As you can see, Anderson and Badzioch (1985) obtained a 14% lifetime risk for individuals from families in which a single generation was affected and 20% for those in which women in two generations are affected. Claus et al. (1990) found a lifetime risk of 25% for those with an affected mother and sister compared to an 11% risk for those with an affected sister. These results are surprisingly close, given the different study designs. Laterality appears to strongly influence risk, with bilateral disease leading to a 30% lifetime risk to relatives in some studies (Anderson and Badzioch, 1985) and 50% in others (Ottman et al., 1983). Paternal history must also be evaluated in determining risk due to family history. As Macklin (1959) has so elegantly shown, the risk from
Table IlL Risk of Breast Cancer to Age 59 in Mothers and Sisters of Breast Cancer Patients Age at patient's diagnosis
Risk to mother or sister
20-29 30-39 40-49 50-59
16% 10% 6% 4%
Breast Cancer Risk Analysis
163 Table IV. Risk of Breast Cancer to Sisters of Patients with Unilateral Disease--Family Type Family type Sister's age
30-39 40-49 50-59 60-69+ Total
Ia
IIb
1% 6% 3% 4%
1% 7% 8% 4%
14%
20%
a226 sisters at risk. '5215 sisters at risk.
paternal history is just as great as from maternal. An adequate assessment of risk also includes an evaluation of other cancers in the family, since in some families, first degree relatives of affected individuals have a lifetime risk approaching 50%. So-called cancer families identified to date include those with breast and ovarian cancer, breast cancer, colon cancer, uterine cancer, and others. Perhaps one of the most well-known family types includes breast cancer, sarcoma, leukemia, and brain tumors. Typically, two or more generations are affected with three or more different site-specific tumors. In these families, cancers tend to occur at younger ages than in the general population.
PERCEPTION OF RISK Genetic counselors are, of course, very accomplished in helping patients to deal with perception of risk, and are aware that the effects of a disease on an individual or the family can influence how risk is perceived by them. Some of the more important factors influencing patients' perceptions of cancer risk are listed in Table V and briefly discussed below. A more complete discussion and examples are included in Understanding Breast Cancer Risk (Kelly, 1991).
Table V. Factors Influencing Patients' Perceptions of Risk Information Beliefs about cancer etiology Beliefs about cancer prognosis Anger, guilt, fear Comparison with affected relative(s)
164
Kelly Etiology
Unless the patient's beliefs about etiology are taken into account, relevant information given to her may be discounted. For example, if an individual believes that her risk is increased because she drank from her affected relative's cup, discussions of repressor genes are going to do little to enlighten her.
Prognosis If, when I discuss breast cancer, I mean minimal breast cancer with breast conservation and a 95-99% survival to 10 years, while the patient thinks breast cancer is equivalent to death after long suffering, our communication will suffer. No matter how good a job I do in analyzing risk or in explaining its meaning to the patient, she may interpret the risk of developing breast cancer as the risk of dying from it.
Anger, Fear, and Guilt As discussed earlier, individuals are often quite emotionally and socially affected by a diagnosis of breast cancer in their families. Many are angry that their lives were disrupted by their relatives' illness, pain, and suffering. They may fear they will die or be ill as their relative was, not taking into account advances in early detection or treatment. Some feel guilt about their anger or failure to do more to help their affected relative. All of these factors can act as barriers to the patient's understanding.
Comparison with Affected Relative(s) Patients often believe that their risk is influenced by their resemblance to an affected relative. Some attempt to have a more positive disposition or outlook than their relative's, hoping that such an approach to life's problems will protect them. Many, regardless of resemblance, feel doomed. For this reason, breast cancer risk analysis, to be effective in helping individuals improve the quantity and quality of their lives, must go bey o n d a clear e x p o s i t i o n of risk. In addition, individuals also n e e d information about early detection, breast health care, and treatment.
Breast Cancer Risk Analysis
165
EARLY DETECTION AND TREATMENT Many patients are unaware of the benefits and limitations of early breast cancer detection. For example, many are surprised to learn that mammograms can detect 85-90% of breast cancers, but not 100%. Patients therefore need to learn the value of physical examinations as well as mammograms. Generally speaking, most patients need help in devising a breast health program that is not only safe but which helps them to feel safe. Many are so afraid of the disease or so afraid of being called hypochondriacs, that they don't have a regular follow-up by a breast health center or a breast specialist, and instead rely on scattered examinations throughout the year (Kelly, 1980). As a result, the anxiety suffered by many is tremend o u s - a n d many do not receive sufficient follow-up. Misconceptions about modern treatments may be present and diminish a patient's ability to make the best use of today's methods.
CONCLUSIONS The aim of breast cancer risk analysis is to act as a conduit between relevant medical, social, and scientific information and patients. As I have tried to show, much of the information presented is complex, and requires both time and skill to transmit successfully. When different studies report conflicting results, the construction of the studies themselves must be analyzed. Breast cancer risk analysis is designed to answer questions such as those listed in Table VI. To do so, it employs approaches and information familiar to genetic counselors who help families to make reproductive decisions. For example, genetic counselors realize that beliefs and life experiences can influence patients' perceptions of risk. Genetic counselors are aware that patients must be given appropriate background information if they are to understand sophisticated medical and scientific information. Cancer risk analysis differs from genetic counseling for reproductive decisions in the following ways: (1) emphasis is usually the counseled individual's own health, (2) focus is not the risk at a given time or event, such as conception, but in helping individuals to understand risk over time, and (3) an important aim is to help individuals establish and maintain effective health practices or to choose an appropriate treatment. Patients have great needs for relevant information about all factors that might influence their risk of breast cancer, not just family history. In my own practice, I regularly see women and their families from all over the country. Of course, only those who can afford to travel are able to come to the California service.
166
Kelly Table VI. Q u e s t i o n s Asked by Relatives of W o m e n with Breast Cancer W h e r e does breast cancer come from? Why did it h a p p e n to my relative? W h a t are my risks? W h a t can I do to be safe? How can I help my relative? W h a t can or should I say to her? A m I doing enough? How can I make sense of the conflicting information about breast cancer?
Breast cancer is a very frequent disease in our society, and one which is unique in its emotional and social impact on the patient and the patient's family. New information about breast cancer and breast cancer risk is announced almost weekly, leading to new concerns and questions by patients and their loved ones. Over the years patients have shown us how great their needs are as they have traveled across the country to receive information. The question now is whether we will follow their lead, in providing the genetic epidemiology service they so clearly need, or, in the name of economy, will fail to do so. The provision of breast cancer risk analysis to families at the time of diagnosis would actually save money by instilling in female relatives of breast cancer patients the need for early detection, and providing the peace of mind that would lead to fewer medical utilizations of all types by all family members. In addition to increased quantity of life, these individuals also receive an increased quality of life by having their questions answered, and by being assured that as new advances occur they will learn about them. Genetic counselors are uniquely educated and trained to provide breast cancer risk analysis to women and their families. Not only are they experts in translating technical, difficult, and confusing scientific and medical information to individuals in a manner that can be understood, but they have long appreciated the effects of disease on family structure and family life. It is my hope that in the years ahead genetic counselors will be able to join with me in providing this much needed service to women throughout the country.
REFERENCES A n d e r s o n DE, Badzioch M D (1985) Risk of familial breast cancer. Cancer 56:383-387. Claus EB, Risch NJ, T h o m p s o n , W D (1990) Age at onset as an indicator of familial risk of breast cancer. A m J Epidemiol 131:961-972. Kelly, PT (1980) Counselling needs of w o m e n with a maternal history of breast cancer. Patient Counsel Health Educ 2:118-124.
Breast Cancer Risk Analysis
167
Kelly PT (1991) Understanding Breast Cancer Risk. Philadelphia: Temple University Press. Knudson A G (1971) Mutation and cancer: Statistical study of retinoblastoma. PNAS 68:820-823. Knudson AG (1986) Genetics of human cancer. Ann Rev Genet 20:231-251. Love RR, Evans AM, Josten DM (1985) The accuracy of patient reports of a family history of cancer. J Chron Dis 38:289-293. Lynch HT, and Lynch JF (1986) Breast cancer genetics in an oncology clinic: 328 consecutive patients. Cancer Genet Cytogenet 22:369-371. Lynch HT, Albano WA, Danes S, Layton MA, Kimberling W J, Lynch JF, Cheng SC, Costello K_A, Mulcahy GM, Wagner CA, Tindall SL (1984) Genetic predisposition to breast cancer. Cancer 53:612-622. Macklin MT (1959) Comparison of the number of breast-cancer deaths observed in relatives of breast-cancer patients, and the number expected on the basis of mortality rates. JNCI 22:927-951. Mettlin C, Croghan I, Natarajan N, Lane W (1990) The association of age and familial risk in a case-control study of breast cancer. Am J Epidemiol 131:973-983. Ottman R, Pike MC, King M-C, Henderson BE (1983) Practical guide for estimating risk for familial breast cancer. Lancet ii:556-558. Page DL, Dupont WD, Rogers LW, Rados MS (1985) Atypical hyperplastic lesions of the female breast. Cancer 55:2698-2708. Schwartz AG, King MC, Belle SH, Satariano WA, Swanson GM (1985) Risk of breast cancer to relatives of young breast cancer patients. JNCI 75:665-668.
