Bertelsen, Lisbeth; Mellemkjær, Lene; Christensen, Jane; Rawal, Rajesh; Olsen, Jørgen H.
It had been suggested that women with breast cancer have a high, roughly constant incidence of new cancer in the contralateral breast of approximately 700 cases per 100,000 survivors per year or 0.7%.1 This annual risk is surprisingly similar to the per-breast risk for breast cancer in monozygotic twins of women with breast cancer, who seem to have a fairly constant annual risk of about 1.3%–1.4% (or 0.7% per breast).1,2 Recent studies, however, have suggested a more complex pattern for contralateral breast cancer, with an even higher rate in very young women followed by a decline in the rate from around age 30 years until around the age of menopause.3,4 This pattern of risk for contralateral breast cancer is in contrast to that for primary breast cancer seen in the Danish female population in recent years with incidence rates that have increased steeply up to the age of around 60 years and flattened out on a rate of about 300 per 100,000 women (0.3%) per year thereafter.5
According to Peto and Mack,1 a constant breast cancer rate might also be present in first-degree relatives of breast cancer patients when they have passed the age of the family proband at diagnosis of breast cancer. On the basis of a pooled analysis of more than 12,000 mothers and sisters of patients in the United Kingdom and United State,1,6,7 they estimated that the incidence rate of breast cancer in female relatives was constant at 300–400 per 100,000 women (0.3%–0.4%) per year, that is, about half the rate of contralateral breast cancer in breast cancer survivors and one-fourth of that in monozygotic twins. According to Peto and Mack, this pattern of a high constant rate of contralateral breast cancer among survivors of breast cancer and their female relatives indicates that a major proportion of breast cancers arise in a susceptible minority of women in the population,1,8,9 a possibility that has implications for the organization of breast cancer screening.
We tested the hypothesis of a high constant rate of contralateral breast cancer in breast cancer survivors and of breast cancer in their first-degree female relatives by using a national dataset containing information on the families of women in Denmark in whom breast cancer was diagnosed at an early age.
From the Danish population born in 1935 or later, we identified 8868 women in whom early-onset breast cancer had been detected, defined here as breast cancer (ICD-7 code 170) diagnosed before 50 years of age (index patients). Of these, 4096 cases were diagnosed among women younger than 40 years of age during the period 1943–1999 and 4772 were diagnosed among women aged 40–49 years during the period 1991–1999 (Table 1). Index women were identified from the files of the Danish Cancer Registry, which also provided the name of the patient, date of birth, personal identification number, date of diagnosis, and treatment initiated within 4 months of the diagnosis (eg, chemotherapy and/or hormonal treatment). The completeness of registration of breast cancer in the Danish Cancer Registry has been found to be virtually 100%.10 The personal identification number, which incorporates sex and date of birth and permits accurate linkage of information among registries, was available for 8859 women alive on 1 April 1968, when the Central Population Registry was established in Denmark, or were born thereafter. Linkage to the files of this registry also provided the date of death or date of emigration. For 9 women who died before 1 April 1968, the date of death was obtained manually from the death certificate file at the National Board of Health.
Identification of Female Relatives
Mothers of index women were identified by a computerized search of the Central Population Registry. For index women in whom breast cancer was diagnosed when they were younger than 40 years, this search was supplemented by manual searches in relevant local population registers. Sisters of index women were traced through the mothers. Details of these procedures are given elsewhere.11 Daughters of the breast cancer patients were located through the Central Population Registry by use of the personal identification numbers of the patients; this tracing was regarded as almost complete. Linkage of mothers, sisters, and daughters to the Danish mortality files and the Central Population Registry provided dates of death or dates of emigration whichever was applicable.
Probands and First-Degree Relatives
To define unambiguously first-degree relatives of women with early-onset breast cancer, we had to define the family proband. If the index patient was the only female member of a nuclear family who was affected by early-onset breast cancer, she became the proband of the family. We found that 49 pairs of index women were sisters and 12 were mother-and-daughter pairs. In these 61 families, the woman whose breast cancer was diagnosed earliest in calendar time was regarded as the family proband, and the other woman was defined as having “familial” breast cancer. Thus, the 8868 early-onset breast cancer patients were distributed in 8807 different families, in which a total of 15,311 female relatives (4351 mothers, 3184 sisters, and 7776 daughters) were identified.
Identifying Cancer in Patients and Relatives
Data on women with early-onset breast cancer and their female relatives were linked to the Danish Cancer Registry with the personal identification number or, if the women had died before 1 April 1968, their date of birth, date of death, and name. Women in whom synchronous, bilateral breast cancer was diagnosed or who were registered with a contralateral breast cancer within 4 months of the primary diagnosis were excluded from the analysis of contralateral breast cancer, which left us with 8737 index patients available for follow-up. The follow-up period for contralateral breast cancer extended from 4 months after diagnosis of the initial breast cancer to the date of diagnosis of contralateral breast cancer, date of death, date of emigration, or the end of the study on 31 December 2003, whichever occurred first. The cohort of 15,311 first-degree female relatives was also followed-up for breast cancer. The follow-up period for mothers and sisters began at age 40 (1943–1990), age 50 (1991–1999), or the date of diagnosis of breast cancer in the proband, whichever occurred first. For daughters, the period of follow-up began on the date of birth of the daughter or the date of diagnosis of breast cancer in the proband, whichever occurred last. The end of follow-up of relatives was the date of death, date of emigration, or end of the study on 31 December 2003, whichever occurred first.
Age-specific incidence rates of contralateral breast cancer among the index patients were calculated per 100,000 person-years. We used national age- and period-specific incidence rates of breast cancer in which bilateral breast cancer is counted as 1 tumor (the first), except in cases for in which the histology of the 2 breast cancers is different (about 2% of all breast tumors in the Danish Cancer Registry); in those cases, both tumors are counted. These rates were applied to the appropriate person-years of observation to obtain the rates that would have occurred had the index patients experienced rates of contralateral breast cancer corresponding to breast cancer rates in the general population of Denmark.
Among the female relatives, the number of breast cancer cases and person-years of observation were calculated separately at ages younger and older than the proband's age at diagnosis. The age- and period-specific incidence rates of breast cancer in the general population of women in Denmark were applied to the appropriate person-years of observation to obtain the expected breast cancer rate. Tests for trends for the observed rates of breast cancer were performed with the χ2 test.
The analyses of contralateral breast cancer were based on Cox proportional hazards models, with follow-up time as the time axis. Attained age was included as a time-dependent variable and modeled by a linear spline12 with boundaries at each 5-year interval starting at age 25 years. The models were adjusted for calendar time of diagnosis (<1970 and 5-year calendar periods from 1970–1999). All statistical tests were based on the likelihood ratio test statistic. The SAS procedure PHREG was used for statistical analyses (release 9.1 for Unix; SAS Institute, Cary, NC).
The 8737 index women with unilateral breast cancer at a young age accrued 73,445 person-years of follow-up (mean, 8.4 years; range, >0–59 years). Overall, there were 466 patients in whom a diagnosis of contralateral breast cancer was made, yielding an overall incidence rate of 634 per 100,000 person-years (0.63% per year). The age-specific incidence rates of contralateral breast cancer were high during the age range 25–44 years (755, 613, 806, and 841 per 100,000 person-years in the 5-year age groups, respectively), consistent with a constant rate of about 0.8% per year (slope = 0.98 [95% CI = 0.84–1.14] per 5-year increase in age) (Fig. 1). Thereafter, the rate decreased by 0.77 (95% CI = 0.65–0.91) per 5-year increase in age to approach the cancer rate per breast in the general female population at about age 65 years of age (assuming that the rate per breast in the general female population is about half of the rate of breast cancer in all women). The rates of contralateral breast cancer during follow-up were higher for women in whom the first primary tumor was diagnosed when they were younger than 40 years than in those aged 40–49 years (Fig. 2), and the rate of contralateral breast cancer decreased by time since first diagnosis among women diagnosed both before 40 years and at ages 40–49 years (age <40 years; P for trend = 0.03 and age 40–49 years: P for trend = 0.26 [Fig. 2]). When we restricted the study population to women in whom breast cancer was diagnosed during 1980–1999 (n = 7634) and stratified by use of chemotherapy or hormonal treatment according to the Cancer Registry (yes = 3324, no = 4310), the rates for those treated were lower than for those not treated for the age groups 45–49, 50–54, and 55–59 years, but the rates for both treated and untreated women declined in a similar manner by age in this age interval (no cases of contralateral breast cancer were observed among women ≥60 years) (data not shown).
The 15,311 female relatives accrued a total of 106,005 person-years at risk in the age range below the age of the proband at diagnosis of her breast cancer and 184,372 person-years at risk above that age. The observed breast cancer incidence rates among all first-degree relatives combined were clearly higher than those of the general population both at ages younger (31 per 100,000 per year) and older (384 per 100,000 per year) than that of the proband at diagnosis of breast cancer and appeared roughly constant irrespective of age at diagnosis of the proband (Table 2). As shown in Figure 3, however, the age-specific rates of breast cancer in relatives who were older than the proband had been at diagnosis increased with increasing age (P for trend <0.0001). The difference between the observed rates in relatives and those expected in the general population (the excess risk) was fairly constant (P for trend = 0.17), as illustrated by the almost parallel course of the 2 incidence curves. The rate difference was on average 187 breast cancer cases per 100,000 women per year, equivalent to an annual excess risk of about 0.2% independent of age. Figure 4 includes similar data stratified by the age of the proband at diagnosis (30–34, 35–39, and 40–49 years) extended to show the full age range of the relatives, that is, at ages below as well as above the age of the proband at diagnosis of her breast cancer. Among relatives of probands age 30–34 years, the incidence raised steeply until the proband age, whereas the increase until proband age was less steep for relatives of older probands (Fig. 4).
During this long-term follow-up of a large group of women in whom breast cancer was diagnosed when they were young, the incidence rate of contralateral breast cancer was much higher than the age-specific rate of breast cancer in the general population of women. In particular, the rate was high and roughly constant in the age range 25–44 years, after which it slowly decreased to approach the rate of cancer per breast in the general population about the age of 65 years. Previous studies have reported high incidence rates of contralateral breast cancer in the United States4 and Sweden,3,13 with a decrease in the annual risk from approximately 0.9% at about age 30 to 0.5%–0.6% at about age 50.3,4 We could not compare our data with the age-specific curves for women older than 50 years, as we restricted our study population to women younger than 50 at the time of diagnosis of a first breast cancer. Nevertheless, one of the Swedish studies13 showed that the rates for contralateral breast cancer among breast cancer survivors younger than 45 years decreased markedly by time since first breast cancer, consistent with our results. Although the rate of contralateral breast cancer does not appear constant over age and time after first diagnosis of breast cancer as proposed by Peto and Mack,1 the rate of contralateral breast cancer does seem to remain high throughout life.
Several factors may influence the rate of contralateral breast cancer. Prophylactic mastectomy reduces the occurrence of contralateral breast cancer markedly,14 but patients in whom breast cancer was diagnosed during the study period in Denmark were generally not encouraged to undergo prophylactic mastectomy, so this is not likely to have had major impact on the rates of contralateral breast cancer in our study. Breast cancer patients who receive chemotherapy or tamoxifen also seem to have a reduced rate of contralateral breast cancer.15,16 Few of the young breast cancer patients included in our study received tamoxifen. Therefore we combined chemotherapy and hormonal therapy in our stratified analysis, which suggested that adjuvant therapy had little impact on the observed pattern by age. We cannot rule out the possibility that metastases of the first breast cancer could have been misdiagnosed as new primary tumors in the opposite breast; however, in recent studies, comparisons of pairs of first primaries and contralateral breast cancers from the same patient by molecular profiling suggest that most diagnosed contralateral breast cancers are independent occurrences of the disease.17,18 Thus, the decreasing rate of contralateral breast cancer with increasing age of the survivors, observed in our study as well as other studies might be due to exhaustion of the population at risk, as women are censored at diagnosis of contralateral breast cancer leaving a study population that becomes genetically more similar to the general population during follow-up. Another factor that might contribute to the observations is that cancer in the opposite breast is diagnosed earlier because of heightened surveillance among the women with breast cancer. Finally, the observed pattern could suggest that the effect of the susceptibility genes ebbs with age, perhaps as the genes become inactivated.
The interpretation of inactivation of susceptibility genes at older ages is, however, inconsistent with our observation of a large and constant excess risk of breast cancer in first-degree relatives. The age-specific rates of breast cancer among first-degree relatives seemed to increase steeply until the women reached the age at breast cancer diagnosis in the proband, particularly among relatives of the youngest probands, followed by a slower increasing rate at ages above the age at breast cancer diagnosis in the proband. At the latter ages, the age-specific incidence curve appeared to run parallel to the breast cancer rate of the general population, with about 200 extra breast cancer cases per 100,000 women per year. This observation is not consistent with the hypothesis of a constant rate irrespective of the age of the relative proposed by Peto and Mack.1 Instead it seems to be the rate difference (or excess risk) that is constant. A Swedish study also could not confirm a constant incidence by age among daughters of breast cancer patients.19
Unlike most family studies, which are based on interview data and thus vulnerable to biased selection of relatives and differential ascertainment of cases of breast cancer,20,21 we used unbiased identification of relatives through linkage with local and central population registries and unbiased ascertainment and validation of cancer through cancer registry linkage. Furthermore we had nearly complete follow-up of index women and their relatives. In addition, we based the study on a national population of women with breast cancer in whom genetic susceptibility was more likely because of their young age at diagnosis.7 This approach makes the material especially powerful for studying the familial pattern of breast cancer in predisposed families. The incompleteness in tracing of relatives in the age range 40–49 years, due to the temporal coverage of the Central Population Registry in Denmark is unlikely to have caused bias. One basis for concern might be more intensive breast cancer surveillance of women who have a close relative with early-onset breast cancer. For example, these women might attend screening programs more assiduously than other women. There were mammography screening programs in only 4 of the 12 counties of Denmark, however, and only from the mid-1990s, so that effect of screening would be limited.22
About 5% of women in whom breast cancer is diagnosed before the age of 50 years have genetic susceptibility due to deleterious germline mutations in the BRCA1 or 2 gene,7 and perhaps other, as-yet-unidentified breast cancer susceptibility genes. We assume that the excess risk of both contralateral breast cancers and breast cancers in relatives are caused by the same genes shared by probands and relatives. The observations among close relatives may best reflect this genetic risk because this group is composed of women who had no prior breast cancer and are therefore, in principle, comparable to the general population. The population of women with breast cancer differs in several aspects that might affect the rate of a new primary breast tumor (apart from susceptibility genes) as mentioned previously. The finding of a constant rate difference (excess risk) throughout life in close relatives implies that the familial risk is not due to factors that accumulate with age. Mack et al2 has suggested that the mechanism of heritable breast cancer could be due to genetically determined high cellular sensitivity to reproductive hormones rather than to genetically determined high levels of hormones. However, the finding of a consistent rate difference (excess risk) throughout life in close relatives implies that the familial risk is not due to factors that accumulate with age.
Andrea Bautz (Danish Cancer Society) assisted with the data management.
1. Peto J, Mack TM. High constant incidence in twins and other relatives of women with breast cancer. Nat Genet
2. Mack TM, Hamilton AS, Press MF, et al. Heritable breast cancer in twins. Br J Cancer
3. Vaittinen P, Hemminki K. Risk factors and age-incidence relationships for contralateral breast cancer. Int J Cancer
4. Bernstein JL, Lapinski RH, Thakore SS, et al. The descriptive epidemiology of second primary breast cancer. Epidemiology
5. Fuglede N, Langballe O, Svendsen AL, et al. Development in incidence of breast cancer in non-screened Danish women, 1973–2002–a population-based study. Int J Cancer
6. UK National Case-Control Study Group. Oral contraceptive use and breast cancer risk in young women. Lancet
7. Peto J, Collins N, Barfoot R, et al. Prevalence of BRCA1 and BRCA2 gene mutations in patients with early- onset breast cancer. J Natl Cancer Inst
8. Peto J. Breast cancer susceptibility-A new look at an old model. Cancer Cell
9. Easton D. Breast cancer–not just whether but when? Nat Genet
10. Jensen Ar, Overgaard J, Storm HH. Validity of breast cancer in the Danish cancer registry. A study based on clinical records from one county in Denmark. Eur J Cancer.
11. Rawal R, Bertelsen L, Olsen JH. Cancer incidence in first-degree relatives of a population-based set of cases of early-onset breast cancer. Eur J Cancer
12. Greenland S. Dose-response and trend analysis in epidemiology: alternatives to categorical analysis. Epidemiology
13. Hartman M, Czene K, Reilly M, et al. Genetic implications of bilateral breast cancer: a population based cohort study. Lancet Oncol
14. Hartmann LC, Schaid DJ, Woods JE, et al. Efficacy of bilateral prophylactic mastectomy in women with a family history of breast cancer. N Engl J Med
15. Early Breast Cancer Trialists’ Collaborative Group. Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet
16. Bertelsen L, Bernstein L, Olsen JH, et al. Effect of systemic adjuvant treatmetn on risk of contralateral breast cancer in the Women's Environment, Cancer and Radiation Epidemiology Study. J Natl Cancer Inst
17. Kollias J, Man S, Marafie M, et al. Loss of heterozygosity in bilateral breast cancer. Breast Cancer Res Treat
18. Imyanitov EN, Suspitsin EN, Grigoriev MY, et al. Concordance of allelic imbalance profiles in synchronous and metachronous bilateral breast carcinomas. Int J Cancer
19. Hemminki K, Granstrom C. Risk for familial breast cancer increases with age. Nat Genet
20. Guo SW. Inflation of sibling recurrence-risk ratio, due to ascertainment bias and/or overreporting. Am J Hum Genet
21. Olson JM, Cordell HJ. Ascertainment bias in the estimation of sibling genetic risk parameters. Genet Epidemiol
22. Svendsen AL, Olsen AH, von Euler-Chelpin M, et al. Breast cancer incidence after the introduction of mammography screening: what should be expected? Cancer.
© 2009 Lippincott Williams & Wilkins, Inc.