Genetic epidemiology of breast cancer

  • Published on
    11-Jun-2016

  • View
    212

  • Download
    0

Embed Size (px)

Transcript

  • 279

    Genetic Epidemiology of Breast Cancer W. Douglas Thompson, Ph.D.

    It has been recognized for some time that a family history of breast cancer is associated rather strongly with a wo- mans own risk of developing the disease. Recent segrega- tion analyses of population-based data on familial pat- terns provide evidence for a rare autosomal dominant al- lele that increases a carriers susceptibility to breast can- cer. The estimated proportion of breast cancer patients who carry this allele declines sharply with age at diagno- sis. Empirical estimates of the risk associated with par- ticular patterns of family history of breast cancer indi- cate the following: (1) having any first-degree relative with breast cancer increases a womans risk of breast cancer 1.5-3-fold, depending on age, (2) having multiple first degree relatives affected is associated with particu- larly elevated risks, (3) having a second-degree relative affected increases the risk by approximately 50%, (4) affected family members on the maternal side and the pa- ternal side contribute similarly to the risk, (5) a family history of breast cancer is associated with bilateral dis- ease, and (6) breast cancer in males is associated with breast cancer in female relatives in much the same way as is breast cancer in women.

    Ovarian cancer clearly has been shown to be associ- ated with breast cancer in families, and genetic linkage has provided strong evidence for a breast-ovarian cancer gene located somewhere on chromosome 17q. At the pop- ulation level, having a first degree relative with ovarian cancer may be at least as predictive of a womans risk for developing breast cancer as is having a second-degree relative with breast cancer. Considerably weaker evi- dence points to a possible familial relationship between breast and endometrial cancer and between breast cancer in women and prostatic cancer in males.

    The clinical applications of the genetic epidemiology of breast cancer are complicated by uncertainty as to the efficacy of mammographic screening in women under the age of 50. For the vast majority of women with a positive family history, the epidemiologic evidence does provide the basis for offering considerable reassurance in that

    Presented at the National Conference on Breast Cancer, Boston, Massachusetts, August 26-28, 1993.

    From the Department of Applied Medical Sciences, School of Applied Science, University of Southern Maine, Portland, Maine.

    Address for reprints: W. Douglas Thompson, Ph.D., Department of Applied Medical Sciences, School of Applied Science, University of Southern Maine, 96 Falmouth Street, Portland, ME 04103.

    Accepted for publication January 31, 1994.

    risks are not extremely high. For that rather small sub- group at exceptionally high risk, realistic estimates of the magnitude of absolute risk over the next 10-20 years may be more informative and less alarming than lifetime probabilities. Cancer 1994; 74979-87.

    Key words: breast neoplasms, genetics, ovarian neo- plasms, prostatic neoplasms, endometrial neoplasms.

    Breast cancer is by far the most frequently diagnosed type of malignant neoplasm among women, and it claims tens of thousands of lives annually in the United States. Despite extensive investigation into the etiology of breast cancer over the past several decades, the risk factors that have been identified to date account for only a minority of the disease in the population. Breast cancer in a first degree relative generally has been found to increase the risk of breast cancer two- to threefold. Having multiple affected first degree relatives may in- crease the risk sub~tant ia l ly .~ ,~

    Awareness of family history as a risk factor leads to considerable concern on the part of many women who have affected family members, and some even seek pro- phylactic mastectomy. Clinicians also recognize the im- portance of family history as an indicator of risk and often query their female patients about the occurrence of breast cancer in family members. The use to which such information most appropriately should be put, however, is generally unclear.

    This paper is not intended as an exhaustive review of the literature on the genetic epidemiology of breast cancer. Instead, its purpose is to present and synthesize some of the more recent data available on the subject and to evaluate those data from the perspective of pub- lic health and clinical decision making for individual patients.

    Familial Aggregation of Breast Cancer

    A number of case-control epidemiologic studies have examined the degree of familial aggregation of breast cancer. A large population-based study of this type is the Cancer and Steroid Hormone Study.4 In this study,

  • 280 CANCER Supplement July 2, 2994, Volume 74, No. 1

    Table 1. Numbers and Odds Ratios for Reported History of Breast Cancer in the Female Relatives of Breast Cancer Patients and Control Subjects, Age 20-44, Cancer and Steroid Hormone Study*

    Occurrence of breast cancer in female relatives (N = 2080) (N = 2058) Odds ratio (95% CI)

    No. of patients No. of control subjects

    No first- or second-degree reported to

    First-degree have been affected

    Mother affected Sister affected Mother or sister with onset < 45 yr 2 or more affected

    Second-degree

    1459 1665 1.00 (referent)

    191 57 93 19

    70 21 30

    1

    3.11 (2.33-4.17) 3.10 (1 32-5.30) 3.54 (2.29-5.49)

    21.68 (3.09-94.74)

    Maternal grandmother or aunt affected 235 187 1.43 (1.16-1.77) Paternal grandmother or aunt affected 178 129 1.57 (1.23-2.01) 2 or more affected 66 34 2.22 (1.43-3.44)

    CI: confidence interval. Excluded are 13 patients and 11 control subjects who did not provide information on the number of sisters they had had. The only second-degree relatives for

    whom information was collected were grandmothers and aunts. Categories for first-degree relatives included women regardless of whether second-degree relatives were also affected; categories for second-degree relatives were restricted to women with no first-degree relative affected.

    detailed family histories were collected from approxi- mately 4500 newly diagnosed breast cancer patients be- tween the ages of 20 and 54 and from a comparable number of age-matched control women from the gen- eral population. The initial findings were reported in 1 985,5 and extensive genetic and epidemiologic analy- ses have been conducted using these Here, I present some previously unpublished empirical esti- mates of the magnitude of the odds ratio (relative risk) for various patterns of family history of breast cancer among breast cancer cases versus controls in the Cancer and Steroid Hormone Study.

    Table 1 gives data for breast cancer patients diag- nosed between the ages of 20 and 44 and control women in the same age group. Table 2 presents corre-

    sponding data for women between the ages of 45 and 54. From Table 1, it can be seen that for women under the age of 45, the risk of developing breast cancer is increased approximately threefold if a mother or sister has been diagnosed with the disease (odds ratios, 3.1 1 and 3.10, respectively; 95% confidence intervals, 2.33- 4.17 and 1.82-5.30, respectively). Diagnosis of breast cancer before age 45 in either a mother or sister in- creases the risk somewhat more (odds ratio, 3.54; 95% confidence interval, 2.29-5.49). The risk is elevated par- ticularly in women who report that two or more of their first-degree female relatives have had breast cancer (odds ratio, 21.68; 95% confidence interval, 3.09- 94.74). Having an affected second-degree relative but not an affected first-degree relative increases the risk by

    Table 2. Numbers and Odds Ratios for Reported History of Breast Cancer in the Female Relatives of Breast Cancer Patients and Control Subjects, Ages 45-54, Cancer and Steroid Hormone Study*

    Occurrence of breast cancer in female relatives

    No. of patients (N = 2614) (N = 2604) Odds ratio (95% CI)

    No. of control subjects

    No first- or second-degree reported to

    First-degree have been affected

    Mother affected Sister affected Mother or sister with onset < 45 yr 2 or more affected

    Maternal grandmother or aunt affected Paternal grandmother or aunt affected 2 or more affected

    Second-degree

    1865

    220 125 97 30

    257 188

    69

    2100

    132 64 39 5

    186 134 35

    1.00 (referent)

    1.88 (1.49-2.36) 2.20 (1.60-3.07) 2.80 (1.92-4.08) 6.76 (2.49-19.84)

    1.56 (1.27-1.9 1) 1.58 (1.25-2.00) 2.22 (1.44-3.42)

    CI: confidence interval. * Excluded are 23 patients and 15 control subjects who did not provide information on the number of sisters they had had. The only second-degree relatives for whom information was collected were grandmothers and aunts. Categories for first-degree relatives included women regardless of whether second-degree relatives were also affected; cateeories for second-deeree relatives were restricted to women with no first-deeree relative affected.

  • Genetic Epidemiology of Breast Cancer/Thompson 281

    approximately 50%, and the increase is of similar mag- nitude, regardless of whether the relative is on the ma- ternal or paternal side of the family.

    For women between the ages of 45 and 54, Table 2 indicates an overall pattern similar to that observed for women between the ages of 20 and 44, except that the magnitude of the increase in risk associated with having an affected first-degree relative is somewhat smaller.

    Other epidemiologic studies generally have found similar magnitudes of risk associated with family his- tory of breast c a n ~ e r . ~ , ~ , ' ~ - ' ~ Women with an affected first-degree relative are estimated to have a 1.5-3.0-fold increase in risk, with early age at onset in relatives often associated with a greater increase than late age at onset. An important qualification on the interpretation of many of these estimates, however, is the retrospective nature of the information collected in case-control in- vestigations. Because of their own recent diagnosis of breast cancer, case patients may be more knowledge- able than controls about cancer in their family mem- bers, and as a consequence, the relative risks reported in some studies may be inflated. Results presented in a recent report from the Nurses' Health Study" are not subject to this sort of distortion, because the informa- tion on family history was collected prospectively, be- fore the diagnosis of breast cancer in cohort members. In that study, a maternal history of breast cancer again was found to increase a woman's risk of breast cancer, with the magnitude of the increase related to age at di- agnosis of the mother's cancer. For women whose mother developed breast cancer before age 40, the mul- tivariate relative risk was 2.00 (95% confidence interval, 1.49-2.70), and for women whose mother developed the disease a t age 70 or later, the value was 1.46 (95% confidence interval, 1.01-2.11). Women with one or more affected sisters were found to be more than twice as likely as women with no affected sister to develop breast cancer. These estimates of relative risk, although of smaller magnitude than some of those obtained from case-control studies, indicate that the results from ret- rospective investigations of the familial aggregation of breast cancer are generally valid. Prospective data from a cohort of women in Iowa yielded a relative risk of 1.53 for any affected first-degree relative, but that study was restricted to women who were older than age 54 at the beginning of the follow-up period."

    Application of segregation analysis to population- based data on breast cancer in families indicates a pat- tern consistent with inheritance of a rare autosomal do- minant allele.' The estimated proportion of breast can- cer patients who carry this allele declines sharply with increasing age a t diagnosis.

    Results from several studies indicate that having a relative with bilateral breast cancer may put a woman

    at a particularly high risk of developing breast can- cer, 14~17-25 although other studies do not.3,5,26,27 Retro- spective collection of family history data may be partic- ularly problematic when the women providing the his- tory are asked to differentiate between unilateral and bilateral disease. Recently, prospective population- based interview data addressing this issue have been published.2' Selected results are summarized in Table 3. For that study, the breast cancer cases from the Cancer and Steroid Hormone Study, who were diagnosed with breast cancer between 1980 and 1982 and interviewed about their family history shortly thereafter, were fol- lowed through 1986 via the Surveillance, Epidemiology and End Results Program of the National Cancer Insti- tute to monitor the occurrence of a second primary breast cancer in the contralateral breast. As the results in the table indicate, women who have had a first pri- mary breast cancer are more likely to develop a contra- lateral second primary breast cancer if they have a fam- ily history of breast cancer than if they do not. In addi- tion, the magnitude of the rate ratio (relative risk) is greater if a first-degree relative was diagnosed with breast cancer before age 45 than if she was diagnosed with breast cancer at a later age.

    Several of the pedigrees reported in the literature as illustrative of families with an extremely high rate of occurrence of breast cancer have been remarkable in that they include instances of male breast cancer, which in the general population is extremely rare. Therefore, it would seem that the presence of male breast cancer in a family may be a particularly powerful indicator that female family members are at high risk for developing breast cancer. A recent population-based study, how- ever, indicates that the familial association between male and female breast cancer is not strikingly stronger than the corresponding association for female breast cancer within families.29 Table 4 presents selected re- sults from that study. Having a mother affected with breast cancer is found to increase the risk of male breast cancer 2.33-fold, and having an affected sister is found to increase the risk 2.23-fold. These estimates are close to the corresponding figures of 1.97 and 2.25 obtained from a metaanalysis of five studies of female breast can- cer. The table also indicates that male breast cancer may be related particularly strongly to early-onset breast cancer in female first-degree relatives, but the confi- dence interval around the estimate of 10.47 is wide. The finding of a 4.08-fold increase in risk associated with having an affected second-degree female relative makes little genetic or epidemiologic sense in the context of the estimates for first-degree relatives; it may reflect insta- bility due to small numbers or systematic bias due to case-control differences in the knowledge that men

  • 282 CANCER Supplement ju ly I , 1994, Volume 74, No. 1

    Table 3. Numbers and Rate Ratios for Reported History of Breast Cancer in Female Relatives as It Relates to the Incidence of Sec...