Journal of the National Cancer Institute Advance Access originally published online on May 13, 2008
JNCI Journal of the National Cancer Institute 2008 100(10):721-727; doi:10.1093/jnci/djn146
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author 2008. Published by Oxford University Press.
ARTICLES |
Timing of Familial Breast Cancer in Sisters
Affiliation of authors: Department of Medical Epidemiology and Biostatistics, Karolinska Institute, Box 281, Stockholm, Sweden
Correspondence to: Marie Reilly, PhD, Department of Medical Epidemiology and Biostatistics, Karolinska Institute, Box 281, 171 77 Stockholm, Sweden (e-mail: marie.reilly{at}ki.se).
 |
ABSTRACT |
|---|
 Top Abstract Context and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
Background: Women who have had a first-degree relative diagnosed with breast cancer (ie, a positive family history) have a rate of breast cancer that is approximately twice that of all women their age, but it is unclear how they should perceive this risk at different ages or if they should be considered at higher risk for the remainder of their lifetime.
Methods: We used Swedish population–based data to assess the risk of breast cancer in 23654 sisters of women diagnosed with breast cancer and in 1 732 775 sisters of unaffected women from 1958 through 2001. Poisson models were used to express the rate of breast cancer as a function of current age, whether a woman had an affected sister, time since the first diagnosis in the family, and family size (number of sisters). The effect of the age of the index case (the first sister diagnosed in the family) at diagnosis and whether her "at-risk" sisters had achieved this age were examined in stratified analyses. Incidence rate ratios of breast cancer in exposed compared with unexposed sisters were calculated with 95% confidence intervals. All estimates were adjusted for calendar time.
Results: Sisters of breast cancer patients had higher breast cancer incidence than unexposed sisters at all ages. The association of exposure (ie, a diagnosis of breast cancer in a sister) with the risk of breast cancer was most pronounced in young women (age 20–39; incidence rate ratio = 6.64, 95% confidence interval = 4.66 to 9.48), and the relative risk decreased to approximately 2 in women older than 50 years. The risk associated with having a sister diagnosed with breast cancer was not modified substantially by the age of the index case at diagnosis (
45 years vs >45 years). The risk was similar for women who were approaching the age at which the first sister was diagnosed in their family and those who had already attained it. The incidence rate ratio of breast cancer in exposed sisters compared with unexposed sisters was constant over time for all age categories of at-risk women.
Conclusions: Women who have a sister diagnosed with breast cancer have an increased risk of breast cancer throughout much of their lifetimes.
Prior knowledge Having a sister diagnosed with breast cancer increases a woman’s risk of breast cancer, but how this increase in risk depends on the time elapsed since the sister's diagnosis, the age of the at-risk sister, and other factors was unclear. Study design Using data on birth dates and parentage that were linked to a national cancer register, Poisson models were constructed to determine the risk of breast cancer associated with having a sister diagnosed with the disease and how it varied according to the age of the at-risk woman, the time elapsed since the diagnosis of her sister, and other parameters. Contribution The increased risk of breast cancer associated with having a sister diagnosed with the disease was most pronounced in younger women, and for all women the increased risk was constant over the time elapsed since the sister's diagnosis. Implications Women with a sister diagnosed with breast cancer remain at increased risk for much of their lifetime. Limitations The study could not assess the risk associated with having a sister diagnosed with breast cancer in women older than 70 years.
|
There is a large volume of published work on the risk of breast cancer in women with a family history, with first-degree relatives of breast cancer patients having a risk approximately twice that of all women their age in the population (1,2). As a result, women are perceived to be at increased risk from the time of diagnosis in a first-degree relative, but less is known about how they should assess this risk at different ages. Some studies have shown that the familial breast cancer risk in first-degree relatives of breast cancer patients decreases with increasing age at diagnosis of the breast cancer patient and age of the relative (3,4). The association of young age at onset of breast cancer with familial risk has been previously emphasized (5), but it is still unclear whether the age at onset is genetically determined (6,7).
In addition to age, there is another timescale that may be important for counseling purposes, namely, the time since the "index" (ie, first) diagnosis in a first-degree relative. Bermejo et al. (8) noted that just after the first diagnosis of breast cancer in a family, increased surveillance of the other at-risk family members can lead to earlier detection of cancers, thus biasing the estimate of the risk of familial cancers. The contributions to familial breast cancer risk of the age of the "at-risk" woman, the age at diagnosis of the first case in her family, and the time since the index diagnosis are still unclear. To our knowledge, no work has been undertaken to investigate whether women in affected families should be considered at higher risk for the rest of their lifetimes or whether those who have survived free of breast cancer for a given time after a diagnosis in a relative have a risk similar to that of women their age in the general population.
The main aim of our study was to clarify the associations of breast cancer risk with the current age of the woman at risk, age at diagnosis of her index sister (the first sister to be diagnosed in the family), and time since the index diagnosis. Our goal was to better understand the timing of familial breast cancer.
|
Methods |
|---|
|
TopAbstractContext and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
Study Design
We analyzed the timing of familial breast cancer in Swedish sisters in terms of both absolute and relative risks. We estimated the absolute risk of breast cancer as a function of age adjusting for calendar year in exposed (ie, a sister had been diagnosed with breast cancer) and unexposed women. We then proceeded to study the risk in exposed women relative to the risk in unexposed women, stratifying by age to accommodate age-specific differences in incidence rates. Third, we investigated whether the risk in exposed women (relative to the risk in unexposed women) changed with time after the diagnosis of the index case. Finally, to combine the time elements, we stratified at-risk sisters by their current age and calculated the relative risk for different time intervals after exposure.
Data Sources
Statistics Sweden maintains a "Multi-Generation Register" that registers children born in Sweden since 1932 along with their biologic or adopted parents. Children who died before 1960 are not included. Using the unique national registration numbers, we linked the individuals in this register to the Swedish Cancer Register to obtain all malignancies that were recorded between 1958 and 2001. Cancers were coded according to the seventh revision of the International Classification of Diseases (ICD-7), and the information in the cancer register enables first and subsequent primary cancers and cancers in situ to be identified.
Due to the inclusion criteria of the Multi-Generation Register, we could not identify any sisters born before 1932. Furthermore, approximately 40% of offspring who died before 1991 did not have links to both parents, so we could not identify their full siblings (9). By linking to the nationwide census, death notification, and migration databases, we obtained the vital status and migration information necessary for censoring and some additional demographic information. The study was approved by the Stockholm regional ethics committee.
Study Design and Study Subjects
We identified all women born in Sweden between 1932 and 2001 and with at least one sister recorded in the Multi-Generation Register (1739186 individuals in 766483 families). After the exclusion of 4063 women (7 who emigrated before 1958, 2302 who were censored in the same month as their birth, and 1754 for whom there was an inconsistent registration of first immigration), the remaining 1735123 women included 2348 who no longer had any sister in the dataset. Thus, the final dataset for analysis included 1732775 women from 763963 families who were followed from the start of the Cancer Register (January 1, 1958) or from their birth until they experienced a diagnosis of malignant breast cancer (ICD-7 code 170) or until December 31, 2001. Women were censored if they died (1%), emigrated (4%), were diagnosed with another cancer (2%), or reached the end of the study on December 31, 2001 (92%). The first breast cancer diagnosis in a family (among sisters) was defined as the index diagnosis, and we considered all sisters of this woman to be exposed from the date of the index diagnosis in their family. Thus, women were unexposed until one of their sisters was diagnosed with breast cancer, but from the date of this index diagnosis they were considered as exposed until they experienced breast cancer themselves or were censored. Index cases contributed just one record (unexposed) to the data, whereas the at-risk sisters of index cases could contribute two, one unexposed and one exposed.
Statistical Methods
Absolute Risk of Breast Cancer by Age. We plotted the smoothed hazard of breast cancer by age, adjusted for calendar year, using an algorithm developed by Pawitan (10) that was modified to account for late entry (ie, entry in the study after birth) and empty bins (ie, age categories with no individuals) and to adjust for linear covariates (eg, calendar time). The algorithm performs a likelihood-based smoothing in a mixed model framework using the Poisson structure of the data. The number of events in each increment of time can be considered to follow a Poisson distribution whose mean is the expected number of events (ie, the product of person-time at risk and instantaneous hazard rate). A smoothed hazard function is constructed from a linear combination of B-splines (11), where the coefficients can be regarded as random effects in a mixed model (12). This method enables one to obtain confidence intervals (CIs). The algorithm is written in R and is available from the authors.
Relative Risk of Breast Cancer by Age.
We compared the incidence of breast cancer in exposed and unexposed sisters using a Poisson model that accommodates more than one timescale. To do so, we split the records into unexposed and exposed time windows, age categories spanning the range covered by the Multi-Generation Register (20–39, 40–49, 50–59, and 60–69 years), and calendar periods (1958–1962, 1963–1967, 1968–1972, 1973–1977, 1978–1982, 1983–1987, 1988–1992, 1993–1997, and 1998–2001). We then aggregated the records with the same covariate patterns (exposure status, age, calendar time, and family size [ie, number of sisters]), recording the total number of breast cancer events and the total time that each sister spent in each of these intervals. We modeled the number of events (yi) in interval i as a Poisson variable:
 |
where Xi represents the vector of covariates in interval i, β represents the vector of regression coefficients, and pti is the total person-time that sisters spent in interval i. We conducted a stratified analysis to examine whether the risk in exposed women was modified by diagnosis of a younger vs older index case (45 years vs >45 years), where the cut point was chosen to have sufficient events in the young sisters at risk.
To investigate whether the relative risk of breast cancer is different when the woman is approaching or has already passed the age at which her index sister was diagnosed, we split the exposure time of sisters into two periods depending on whether the current age was younger or older than the age of the index sister when she was diagnosed, creating a new dichotomous variable (younger or older). We then used a Poisson model similar to that used before, with age categorized as 25–34, 35–44, 45–54, and 55–69 years, and an interaction term between exposure and the younger/older variable. Again, we repeated the analysis stratified by age at diagnosis of the index case (45 years vs >45 years).
Relative Risk of Breast Cancer by Time Since Diagnosis of the Index Case. Finally, we studied how the relative risk changed with time from the first breast cancer diagnosis among the sisters in a family, adjusting for age, calendar time, and family size (number of sisters). The number of events was modeled as Poisson with the following covariates: time from exposure in 5-year intervals (reference category is the unexposed), calendar time (in 5-year intervals), age (in 5-year intervals), and family size. We repeated this analysis stratifying by the current age of the at-risk sisters.
To assess the potential underestimation of standard errors due to correlation between sisters in the same family, we repeated our analyses restricted to the two oldest sisters in each family. We also performed analyses restricted to calendar years before 1988 to minimize potential bias from population mammography screening.
All data preparation and analyses were performed using SAS 9.1 (SAS Institute, Cary, NC), and estimates of the parameter β and its 95% confidence intervals were obtained using PROC GENMOD. All of the figures were produced with R (www.r-project.org).
|
Results |
|---|
|
TopAbstractContext and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
We followed 1732775 sisters from 1958 until they attained the age of 70 years, and among them we identified 16505 (0.95%) index cases of breast cancer (ie, the first registered case among the sisters in the family) who had 25294 sisters, with 23654 of these at-risk sisters having follow-up time after the diagnosis of the index sister. Among the 23654 women exposed (ie, with an affected sister), 714 (3.02%) developed breast cancer (Table 1). Most (92%) of the women analyzed had two or fewer sisters. The number of exposed sisters was usually somewhat less than the total number of sisters for whom there was an index case because some of these women were censored before the index sister's diagnosis.
|
Absolute Risk of Breast Cancer by Age
We derived the smoothed hazard function adjusted for calendar year for breast cancer at ages 20–70 years in all sisters of breast cancer cases and in unexposed sisters (Figure 1). The widening confidence intervals with age reflect the relatively young age of the cohort: approximately 76% of the sisters were younger than 50 years at the end of follow-up. Sisters of breast cancer patients had higher breast cancer incidence than unexposed sisters at all ages: at age 40 years the rate of breast cancer in unexposed sisters was 5.94 per 10000 person-years (95% CI = 5.70 to 6.18) whereas in exposed sisters the rate was 19.83 per 10000 person-years (95% CI = 16.11 to 24.40); at age 60 years the absolute rate in unexposed sisters was 28.87 per 10000 person-years (95% CI = 27.49 to 30.31), compared with 59.51 per 10000 person-years (95% CI = 52.79 to 67.08) in exposed sisters.
|
Relative Risk of Breast Cancer by Age
The overall incidence of breast cancer in sisters of case subjects was more than double the incidence in the unexposed sisters (Table 2). The increased risk associated with exposure (ie, having a sister diagnosed with breast cancer) was greatest for women aged 20–39 years (incidence rate ratio [IRR] = 6.64, 95% CI = 4.66 to 9.48). The relative risk was approximately 2.5 for sisters in their forties, and the risk then decreased further in older women so that there was an increased risk of approximately twofold in women older than 50 years. To gain some insight into possible differences in risks associated with having a sister diagnosed with breast cancer at a younger or older age, we repeated the analysis after stratifying by the age at which the index case was diagnosed (45 and >45 years, Table 2). We found no evidence of a statistically significant difference in the risk for sisters of younger and older index cases (
2 = 0.15, P = .70 from Wald test). All risk estimates according to age were adjusted for calendar time and family size. The calendar time was statistically significant in all analyses, as expected from Figure 1, and failure to adjust resulted in overestimation of familial risk.
|
The relative risk of breast cancer had a similar pattern in sisters who were approaching and sisters who had already attained the age at which their index sister was diagnosed (Figure 2). The incidence rate ratio of breast cancer in exposed sisters was highest at age 25–35 years, whether they were younger (IRR = 7.6, 95% CI = 3.6 to 15.9) or older (IRR = 8.3, 95% CI = 2.7 to 25.9) than the age at diagnosis of the index case. After age 35 years, the relative risk was approximately 2.5, whether the woman was younger or older than the age of the index case at the time of diagnosis. We repeated these analyses stratified by the age at which the index case was diagnosed and the pattern was similar (data not shown).
|
In analyses restricted to calendar years before nationwide population mammography screening in Sweden, the pattern of risk with age (data not shown) was similar to that reported in Table 2. Restricting our cohort to the two oldest sisters for each family, the 555 exposed women had very similar IRRs and confidence intervals (age 20–39: IRR = 6.05, 95% CI = 3.64 to 10.05; age 40–49: IRR = 2.64, 95% CI = 2.22 to 3.15; age 50–59: IRR = 2.16, 95% CI = 1.91 to 2.44; age 60–69: IRR = 2.15, 95% CI = 1.80 to 2.57), and all the associations remained statistically significant (data not shown), except for one stratum with very limited data (sisters aged 20–39 years with an index case diagnosed after age 45 years), for which only one event remained in the restricted dataset.
Relative Risk of Breast Cancer by Time Since Diagnosis of the Index Case
The overall relative risk of breast cancer in exposed sisters compared with unexposed was constant over time for 20 years, with the incidence rate ratios ranging from 2.09 to 2.16 (Table 3). To examine how the risk for exposed women depends on time since the index diagnosis while taking into consideration age effects, we studied the relative risks by time in analyses stratified by current age of at-risk sisters (Table 3). For women who were older than 40 years and had a sister diagnosed with breast cancer, the increased risk compared with that in unexposed women was approximately constant over time. Women who were younger (<40 years) had the highest relative risk for breast cancer, and there was no clear trend in the incidence rate ratio over time for these women, but the estimates were based on small numbers.
|
|
Discussion |
|---|
|
TopAbstractContext and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
Although there has been considerable effort to estimate the dependence of the risk of familial breast cancer on age (2–5,13,14), the focus of these efforts was more on the age of the breast cancer patients than on the age of their relatives (4,15). This emphasis is understandable because young age at the time of diagnosis of breast cancer is associated with the presence of BRCA1 and BRCA2 mutations (14,16,17). However, the risk of breast cancer in the general female population and in relatives of breast cancer patients depends strongly on age (2,13,14). Furthermore, there is a third timescale that may need to be considered, that is, the time that has elapsed since the index diagnosis in the family. We are unaware of any published research that has attempted to determine the dependence of the familial breast cancer risk on these different timescales. Our work offers a better understanding of the timing of familial breast cancer, which is important for providing a more clear and focused counseling message to sisters of breast cancer patients and could also help in the planning of screening programs (18), which have been shown to be beneficial for women younger than 50 years with a family history of breast cancer (19).
The age-specific incidence of breast cancer in our cohort agreed with the Swedish population rates (20,21), although the overall incidence was low (approximately 1%) due to limited follow-up. We observed a higher incidence of breast cancer in sisters of case subjects than in women from unaffected families (Figure 1), similar to results obtained by many studies (4,13,14,16,22). The inflection in risk around the time of menopause (6,7) known as "Clemmesen's Hook" (23) is clearly visible in the unexposed sisters (Figure 1). The absolute risk in affected and unaffected families increased steadily with age, and there was a statistically significant difference in the risks in these two populations at all ages.
In agreement with previous studies (2,4), we found a very high risk for women who were young (20–39 years) when the index case was diagnosed in a sister relative to the risk in women with unaffected sisters (Table 2). For women older than 50, the increased relative risk (relative to women with no affected sisters) was approximately 2; for women aged 40–49 years, the relative risk was slightly higher than 2. The high risk in young women is expected because in families where there is a breast cancer there will be a higher prevalence of BRCA1 and/or BRCA2 mutations than in the general population and thus an increased risk of cancer, especially at younger ages (24). However, this risk was modified only slightly by the age of the index case (15): comparing families with index cases of 45 years and younger with families where the index cases were older than 45 years, the relative risks in sisters aged 20–39 years were approximately 7.0 and 6.0, respectively. This similarity contrasts with many studies that have reported a higher risk in relatives of younger women diagnosed with breast cancer, but these studies used case–control designs (14) or analyzed cohorts of relatives of patients that did not distinguish the time order of the diagnoses in the family (15). Only a few studies have attempted to study the interaction between the age of the breast cancer patient and the age of her relative, with inconsistent findings (14).
To examine whether the observed high risk in young sisters of affected women could be confounded by screening behavior, we repeated the analysis for the time periods before the introduction of screening in Sweden. The pattern of risk with age was similar to that reported in Table 2, providing no evidence of preferential screening of young sisters.
Exposed sisters showed similar breast cancer risks at ages younger and older than the age at diagnosis of the index case (Figure 2). These results suggest that there is not a predetermined age at which members of a family enter a high-risk state, as proposed by Peto and Mack (6). In that paper, the authors claimed that the absolute breast cancer risk increases to a high constant level at an age that is specific to each family. Although our work focused on relative and not absolute risk, the results suggest that there is no change in the risk profile of a woman as she approaches and passes the age at which the index sister was diagnosed in her family. Recent work in which breast cancer patients were tested for BRCA1 and BRCA2 mutations also found no evidence of risk depending on age in families of patients with these mutations (15).
When women of all ages were analyzed together, the incidence rate ratio of breast cancer appeared to be constant with time after diagnosis of the index case (Table 3). We found that exposed sisters had approximately double the risk of breast cancer relative to unexposed women for at least 20 years after the index sister was diagnosed in their family. Previous work has shown the familial risk of breast cancer to be higher in the same year as the index diagnosis, suggesting increased surveillance in the family (8,25), but the lack of an increased risk in the first 5 years in our data suggests that there is no substantial surveillance bias. Stratification by the age of the at-risk sisters revealed that exposed sisters of all age categories had a constant relative risk of breast cancer over time. Women who were young (less than 40 years) had the highest relative risk for breast cancer (in agreement with Table 2), and we could discern no clear trend over time from these analyses, which were based on small numbers.
A strength of our study is the availability of the population registers of Sweden, which enabled us to plot the age-specific smoothed incidence rate of breast cancer with narrow confidence intervals for the general population and for sisters. We also developed an algorithm to adjust these hazards for the known increasing incidence with calendar time. Our focus on sisters and not mother and daughter pairs is more relevant for studying whether time since index diagnosis is important for counseling because sisters are likely to belong to the same age group. Moreover, by using sisters, we avoided the possible bias due to the different environmental exposure between mothers and daughters because sisters live in more similar calendar periods (14,26). Our results may be relevant for other familial relationships because previous studies show that the increased risk associated with having an affected sister is similar to that associated with having an affected mother (4,14). Our study did not include cancers diagnosed before 1958 (the start-up year of the Swedish cancer registry), but there was minimal bias due to this truncation: the sisters in our study were all born after 1931, so that in 1958 they were at most 27 years old and thus highly unlikely to have already developed a cancer.
A limitation of our study is that we could not study breast cancer in very old women. Because the Swedish Multi-Generation Register started with individuals born in or after 1932 and follow-up ended in 2001, we could only follow sisters to a maximum age of 70 years. This age span covers times of substantial risk in the life of a woman. However, the apparent flattening of the hazard in exposed sisters older than 60 years may be due to the small sample sizes, reflected in wide confidence intervals (Figure 1). There is also potential bias due to the available register data because some older sisters (born before1932) may not have been included in the analysis. Another limitation is the possible misclassification of exposure because full siblings cannot be identified for many women who died before 1991. However, we have shown elsewhere (27) that there is no bias from this source except where there is a large difference in the mortality of familial and nonfamilial cases. Even if there was evidence of such differential mortality for breast cancer, it would lead to only a slight underestimate of the modest familial risk, and, thus, any bias would have little impact on our conclusions.
Our analysis did not model the correlation between breast cancer risk in sisters in the same family so that the inclusion of women from families of three or more sisters could have led to some underestimation of the confidence intervals. Nevertheless, when we repeated all the analyses including only the two oldest sisters for each family, the results were similar. In our analysis, we did not consider important risk factors for breast cancer such as age at first birth or use of hormone replacement therapy. However, we do not expect this decision to have a large impact on our results because a previous study in the same population (28) showed that adjustment for reproductive factors did not change the estimates of familial risk.
Our choice of starting time of exposure (date of diagnosis in a sister) was not based on a biologic hypothesis but was motivated by the aim of providing a counseling message for women who have a family member that was diagnosed with breast cancer. The first diagnosis in a family is indeed the time where women first become aware of their increased risk of breast cancer, and we wished to estimate whether the perceived higher risk from the date of diagnosis of a sister was modified by time. We found no evidence of such effect modification or of a change in risk profile as a woman approaches and passes the age at onset of breast cancer in her affected sister. Thus, the assumption of constant risk with respect to these time axes in current breast cancer prediction models (29) is reasonable, although our observation of the confounding effect of calendar time suggests that its effect on these predictions is worthy of investigation.
In conclusion, we confirmed that familial breast cancer risk is highly dependent on age but appears to be quite constant with time since the index diagnosis. In particular, sisters of women diagnosed with breast cancer still have an increased risk of breast cancer 20 years after diagnosis of the sister, suggesting that women live with the burden of familial breast cancer for their lifetime. This suggests a need for intense screening of sisters in affected families for the rest of their life, independent of the age-related screening recommendations.
|
Funding |
|---|
|
TopAbstractContext and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
Swedish Cancer Society (Cancerfonden, registration number 05 6837, contracts 05 0444 and 06 0224).
|
NOTES |
|---|
|
TopAbstractContext and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
The funder had no role in the design of this study, the data analysis, data interpretation, the writing of the report, or the authors’ decision to submit the paper for publication.
We thank Agus Salim for help with the computer code for smoothing adjusted hazard ratios.
|
REFERENCES |
|---|
|
TopAbstractContext and CaveatsMethodsResultsDiscussionFundingReferencesNotes |
|---|
1. Hemminki K, Czene K. Attributable risks of familial cancer from the Family-Cancer Database. Cancer Epidemiol Biomarkers Prev (2002) 11(12):1638–1644.
2. 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 (2006) 42(17):3034–3040.[CrossRef][Web of Science][Medline]
3. Tulinius H, Sigvaldason H, Olafsdóttir G, Tryggvadóttir L, Bjarnadóttir K. Breast cancer incidence and familiality in Iceland during 75 years from 1921 to 1995. J Med Genet (1999) 36(2):103–107.
4. Collaborative Group on Hormonal Factors in Breast Cancer. Familial breast cancer: collaborative reanalysis of individual data from 52 epidemiological studies including 58,209 women with breast cancer and 101,986 women without the disease. Lancet (2001) 358(9291):1389–1399.[CrossRef][Web of Science][Medline]
5. Claus EB, Risch NJ, Thompson WD. Age at onset as an indicator of familial risk of breast cancer. Am J Epidemiol (1990) 131(6):961–972.
6. Peto J, Mack TM. High constant incidence in twins and other relatives of women with breast cancer. Nat Genet (2000) 26(4):411–414.[CrossRef][Web of Science][Medline]
7. Easton D. Breast cancer—not just whether but when? Nat Genet (2000) 26(4):390–391.[CrossRef][Web of Science][Medline]
8. Bermejo JL, Hemminki K. Familial risk of cancer shortly after diagnosis of the first familial tumor. J Natl Cancer Inst (2005) 97(21):1575–1579.
9. Statistic Sweden. 2004:7 Multi-Generation Register 2003. A description of contents and quality Multi-Generation register. (2004) Örebro, Sweden: STATISTISKA CENTRALBYRÅN;. http://www.scb.se/statistik/BE/OV9999/2003A01/OV9999_2003A01_BR_BE96ST0407.pdf. [Last accessed: April 16, 2008.
10. Pawitan Y. In All Likelihood: Statistical Modelling and Inference Using Likelihood (2001) Oxford, UK: Oxford University Press. 544.
11. De Boor C. A Practical Guide to Splines (1978) New York, NY: Springer-Verlag. 392.
12. Lee Y, Nelder J, Pawitan Y. Generalized Linear Models with Random Effects: Unified Analysis via H-likelihood (2006) London, UK: Chapman and Hall. 416.
13. Lux MP, Fasching PA, Beckmann MW. Hereditary breast and ovarian cancer: review and future perspectives. J Mol Med (2006) 84(1):16–28.[CrossRef][Web of Science][Medline]
14. Pharoah PD, Day NE, Duffy S, Easton DF, Ponder BA. Family history and the risk of breast cancer: a systematic review and meta-analysis. Int J Cancer (1997) 71(5):800–809.[CrossRef][Web of Science][Medline]
15. Lee JS, John EM, McGuire V, et al. Breast and ovarian cancer in relatives of cancer patients, with and without BRCA mutations. Cancer Epidemiol Biomarkers Prev (2006) 15(2):359–363.
16. Dite GS, Jenkins MA, Southey MC, et al. Familial risks, early-onset breast cancer, and BRCA1 and BRCA2 germline mutations. J Natl Cancer Inst (2003) 95(6):448–457.
17. Suthers GK. Cancer risks for Australian women with a BRCA1 or a BRCA2 mutation. ANZ J Surg (2007) 77(5):314–319.[CrossRef][Web of Science][Medline]
18. Robson M, Offit K. Clinical practice. Management of an inherited predisposition to breast cancer. N Engl J Med (2007) 357(2):154–162.
19. Maurice A, Evans DG, Shenton A, et al. Screening younger women with a family history of breast cancer—does early detection improve outcome? Eur J Cancer (2006) 42(10):1385–1390.[CrossRef][Web of Science][Medline]
20. The National Board of Health and Welfare. www.socialstyrelsen.se. [Last accessed: April 16, 2008.].
21. Hemminki K, Li X, Plna K, Granström C, Vaittinen P. The nation-wide Swedish family-cancer database—updated structure and familial rates. Acta Oncol (2001) 40(6):772–777.[CrossRef][Web of Science][Medline]
22. Hemminki K, Granström C, Czene K. Attributable risks for familial breast cancer by proband status and morphology: a nationwide epidemiologic study from Sweden. Int J Cancer (2002) 100(2):214–219.[CrossRef][Web of Science][Medline]
23. Hakama M. The peculiar age specific incidence curve for cancer of the breast—Clemmesen's hook. Acta Pathol Microbiol Scand (1969) 75(3):370–374.[Web of Science][Medline]
24. Begg CB, Haile RW, Borg A, et al. Variation of breast cancer risk among BRCA1/2 carriers. JAMA (2008) 299(2):194–201.
25. Hemminki K, Bermejo JL. Effects of screening for breast cancer on its age-incidence relationships and familial risk. Int J Cancer (2005) 117(1):145.[CrossRef][Web of Science][Medline]
26. Paltiel O, Friedlander Y, Deutsch L, et al. The interval between cancer diagnosis among mothers and offspring in a population-based cohort. Fam Cancer (2007) 6(1):121–129.[CrossRef][Web of Science][Medline]
27. Leu M, Czene K, Reilly M. The impact of truncation and missing family links in population-based registers on familial risk estimates. Am J Epidemiol (2007) 166(12):1461–1467.
28. Hemminki K, Granström C. Familial breast cancer: scope for more susceptibility genes? Breast Cancer Res Treat (2003) 82(1):17–22.[CrossRef][Web of Science][Medline]
29. Claus EB. Risk models used to counsel women for breast and ovarian cancer: a guide for clinicians. Fam Cancer (2001) 1(3–4):197–206.[CrossRef][Medline]
Manuscript received November 9, 2007; revised March 17, 2008; accepted April 7, 2008.
Related Article in JNCI
CiteULike 
Connotea 
Del.icio.us What's this?
J Natl Cancer Inst 2008 100: 687.
This article has been cited by other articles:
|
|
 |
|
  J. Zipprich, M. B. Terry, Y. Liao, M. Agrawal, I. Gurvich, R. Senie, and R. M. Santella Plasma Protein Carbonyls and Breast Cancer Risk in Sisters Discordant for Breast Cancer from the New York Site of the Breast Cancer Family Registry Cancer Res., April 1, 2009; 69(7): 2966 - 2972. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Leu, M. Reilly, and K. Czene Evaluation of Bias in Familial Risk Estimates: A Study of Common Cancers Using Swedish Population-based Registers J Natl Cancer Inst, September 17, 2008; 100(18): 1318 - 1325. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||