Professional Documents
Culture Documents
JURNAL MATERNITAS
FAMILY HISTORY AND THE RISK OF BREAST
CANCER: A SYSTEMATIC REVIEW AND META-
ANALYSIS
OLEH
NIRSAM
NIM : 201401117
PRODI S1 KEPERAWATAN
STIKes WIDYA NUSANTARA
PALU
2016
Int. J. Cancer: 71, 800–809 (1997) Publication of the International Union Against Cancer
Publication de l’Union Internationale Contre le Cancer
r 1997 Wiley-Liss, Inc.
FAMILY HISTORY AND THE RISK OF BREAST CANCER:
A SYSTEMATIC REVIEW AND META-ANALYSIS
Paul D.P. PHAROAH1,2,*, Nicholas E. DAY1, Stephen DUFFY1, Douglas F. EASTON1 and Bruce A.J. PONDER2
1Institute of Public Health, Robinson Way, Cambridge, UK
2CRC Human Cancer Genetics Group, Addenbrooke’s Hospital, Cambridge, UK
An increased risk of breast cancer in women with a family then read to abstract any reported family history data. The
history of breast cancer has been demonstrated by many computer-aided searches were complemented by hand-searching
studies using a variety of study designs. However, the extent the references of all studies identified. This method was also used
of this risk varies according to the nature of the family history to identify the studies reported before 1966.
(type of relative affected, age at which relative developed
breast cancer and number of relatives affected) and may also Design of the meta-analysis
vary according to age of the individual. The aim of our study
was to identify all the published studies which have quantified The results of the studies are presented as relative risks (RR):
the risk of breast cancer associated with a family history of equivalent to the odds ratio estimate from a case-control study, and
the disease, and to summarise the evidence from these the observed/expected ratio in a retrospective cohort design. Most
studies, with particular emphasis on age-specific risks accord- studies reported only the crude risk estimate, but 14 studies
ing to subject and relative age. Seventy-four published studies reported both crude risk and risk adjusted for other risk factors in a
were identified. The pooled estimate of relative risk (RR) multivariate analysis (Bain et al., 1980; Brinton et al., 1995; Byrne
associated with various family histories was as follows: any et al., 1991; Calle et al., 1993; Carter et al., 1989; Chen et al., 1994;
relative, RR 5 1.9 (95% CI, 1.7–2.0); a first-degree relative, Colditz et al., 1993; Gomes et al., 1995; Harris et al., 1990;
RR 5 2.1 (CI 5 2.0, 2.2); mother, RR 5 2.0 (CI 5 1.8, 2.1); Helmrich et al., 1983; Parazzini et al., 1992; Rookus et al., 1994;
sister, RR 5 2.3 (CI 5 2.1, 2.4); daughter, RR 5 1.8 (CI 5 1.6,
2.0); mother and sister, RR 5 3.6 (CI 5 2.5, 5.0); and a Toti et al., 1986; UK National Case-Control Study Group, 1989).
second-degree relative, RR 5 1.5 (CI 5 1.4, 1.6). Risks were For this analysis, the adjusted values have been used. In addition,
increased in subjects under age 50 and when the relative had 12 studies report only adjusted relative risks (Clavel et al., 1991;
been diagnosed before age 50. Int. J. Cancer 71:800–809, 1997. Jick et al., 1989; Le et al., 1984; Lee H.P. et al., 1992; Luporsi,
r 1997 Wiley-Liss, Inc. 1988; Negri et al., personal communication; Richardson et al.,
1991; Rohan et al., 1988; Sattin et al., 1985; Siskind et al., 1989;
A familial component to breast cancer has been recognised for White et al., 1994; Zaridze and Andrieu et al., 1995). 95%
many years. Indeed, Paul Broca (1866) described a family with confidence intervals (CI) described are either those published, or
multiple cases of breast cancer over 130 years ago. Since then, a have been estimated from available published data. Where insuffi-
large number of epidemiological studies have been carried out to cient data were given to calculate a confidence interval, only the
attempt to quantify the risks of breast cancer associated with a point estimate of RR is given. Pooled estimates of risk were
positive family history, and nearly all of these have found elevated obtained by calculating a weighted average of the log relative risk
relative risks to female relatives of breast cancer patients. However, estimates (Breslow and Day, 1980). Studies were weighted accord-
the extent of this risk varies according to the nature of the family ing to the inverse of the variance of the log relative risk, which was
history (type of relative affected, age at which the relative estimated from the confidence interval by the formula:
developed breast cancer and number of relatives affected) and may
also vary according to the age of the individual (Kelsey and study variance (v1) 5 [ln (upper 95% confidence limit)
Gammon, 1990; Goldgar et al., 1996). 2 ln (lower 95% confidence limit)]2/(3.92)2.
Recent rapid increases in our understanding of the genetic basis
of breast cancer have created a resurgent interest in identifying For the few studies where no confidence interval was given or
those at increased risk because of family history. There is thus a could be estimated and the relative risk was reported as significant
need for age-specific risks to be quantified more precisely to enable (Adami et al., 1981; Burns et al., 1981; Nomura et al., 1986), the
appropriate counselling to be provided for at-risk women and to variance was calculated based on the assumption that the lower
determine the most appropriate treatment for those at risk. Al- 95% confidence limit was 1.0. If the risk was reported as not
though reviews of family history and risk of breast cancer have significant, a nominal weight of 5 was assigned (Bucalossi et al.,
been published, there has been no systematic review which has 1954; Murphy and Abbey, 1959; Ross et al., 1980; Schildkraut
identified all the relevant studies. The purpose of our study was to et al., 1990; Smithers et al., 1952). Confidence intervals for the
identify all the published studies which have quantified the risk of summary estimate of log RR were calculated using the formula
breast cancer associated with a family history of the disease, and to
summarise the evidence from these studies, with particular empha- 95% CI ln (RR) 5 ln (RR) 6 1.96 3 SE ln (RR),
sis on the age-specific risks according to subject and relative age. where SE ln (RR) 5 1/Œowi .
This analysis should prove useful in clarifying directions for future
research.
In order to calculate summary age-specific estimates, published
age-specific risks were grouped. Two age-group categories were
METHODS
Search methods
Contract grant sponsor: Cancer Research Campaign.
Published studies were identified using the Medline (National
Library of Medicine, Washington, DC) databases for 1966–present
using the terms ‘‘breast neoplasms’’ and ‘‘family’’. In order to *Correspondence to: Institute of Public Health, Robinson Way, Cam-
identify studies in which family history-associated risk was not a bridge CB2 2SR, UK. Fax: 144 1223 330330.
E-mail: pp10001@medschl.cam.ac.uk
primary aim, but in which family history had been recorded, a
search on the terms ‘‘breast neoplasms’’ and ‘‘cohort’’ or ‘‘case-
control’’ was also performed. Studies identified in this way were Received 18 December 1996; revised 24 January 1997
FAMILY HISTORY AND BREAST CANCER RISK 801
used for both subject and relative age: under 50 years and 50 years controls selected from the general population, 22 used hospital
and over. However, some studies reported age-specific risks using controls and the remaining 3 had both population and hospital
different age categories. For these, 2 categories were used with an controls.
age cutoff as near to 50 as possible. For example, if a cutoff of 45 Twenty-two cohort studies were identified and are summarised
was published, the 45 and over age group would be categorised in in Table II. Two types of design have been used, i.e., the
the 50 and over group for the summary estimates. Similarly, for
retrospective cohort and the prospective cohort. Many of the early
studies using a cutoff at 55 years, the under 55 year group would be
categorised in the under 50 year group for the purposes of obtaining studies utilised the retrospective cohort design, in which breast
summary estimates. Where a cutoff based on menopausal status cancer incidence or mortality in relatives of breast cancer probands
was used, the pre- and peri-menopausal groups were included in was retrospectively ascertained. The observed number of cases/
the under 50 group, and the post-menopausal group in the 501 deaths was then compared with the number expected if population
group. age-specific incidences were applied to the cohort of relatives. Four
studies used a prospective design in which family history of breast
cancer was ascertained in a cohort of women initially free of breast
RESULTS
cancer. Breast cancer incidence or mortality was then compared in
Fifty-two case control studies were identified. These are sum- women in different family history categories. Two studies utilised a
marised in Table I. Twenty-seven of the case-control studies used nested case-control design in which incident breast cancer cases in
a prospective cohort were matched to controls free of breast cancer al., 1989; Slattery and Kerber, 1993; Teare et al., 1994; Toti et al.,
from the same cohort. 1986; Tao et al., 1988; UK National Case-Control Study Group,
1989; Wassink, 1935; White et al., 1994; Yuan et al., 1988; Hiatt et
al., 1984; Brownson et al., 1988) which have quantified the risk of
History of breast cancer in any relative
breast cancer in women with a history of breast cancer in at least
Twenty-six studies report the risk of breast cancer associated one first-degree relative are given in Figure 2. This shows that the
with having a history of breast cancer in any relative (unspecified) relative risk estimates ranged from 1.2–8.8, though in most studies
(Adami et al., 1981; Burns et al., 1981; Byrne et al., 1991; La the relative risk was between 2 and 3. The pooled relative risk
Vecchia et al., 1987; Clavel et al., 1991; Harris et al., 1990; Holzel estimate was 2.1 (CI 5 2.0, 2.2). The risk estimates varied
et al., 1986; Le et al., 1984; Lee N.C. et al., 1987; Luporsi, 1988;
according to the study design: the pooled risk estimates for the
Marubini et al., 1989; McPherson et al., 1987; Morabia et al.,
prospective cohort studies was 1.6 (CI 5 1.4, 1.9) compared to 2.2
1993; Newcomb et al., 1994; Ngelangel et al., 1994; Olsson et al.,
1989; Primic-Zakelj et al., 1995; Richardson et al., 1991; Roseman (CI 5 2.1, 2.3) for the retrospective study designs. The pooled risk
et al., 1990; Rossing et al., 1996; Sattin et al., 1985; Schildkraut et estimate for the case-control studies with population controls was
al., 1990; Toti et al., 1986; WHO Collaborative Study of Neoplasia 2.2 (CI 5 2.1, 2.3) compared to 2.5 (CI 5 2.2, 2.7) for the
and Steroid Hormones, 1990; Yang et al., 1992; Andrieu et al., case-control studies using hospital controls.
1995). These are summarised in Figure 1, which shows that all but Five studies reported risks according to age at diagnosis of the
one study found a positive association, with relative risks ranging relative. Four of these showed the risk of breast cancer to be
from 0.9–3.6. The pooled relative risk estimate was 1.9 (CI 5 1.7, slightly higher for women whose relatives were diagnosed at a
2.0). younger age (Mettlin et al., 1990; Byrne et al., 1991; Sattin et al.,
Age-specific effects were described in four studies. McPherson 1985; Houlston et al., 1992), and one study found age at diagnosis
et al. (1987) found that the risk was higher if the relative had been of the relative to have no effect (Toti et al., 1986).
diagnosed under age 45, RR 5 2.2, compared to a relative risk of Eleven studies reported risks according to age of the subject
1.6, if the relative had been diagnosed aged 45 or over. The other (Mettlin et al., 1990; Calle et al., 1993; Adami et al., 1981;
three studies report risks according to the age of the subject. Ambrosone et al., 1995; Byrne et al., 1991; Meirik et al., 1986;
Roseman et al. (1990) found the risk to women under 30 was Sattin et al., 1985; Parazzini et al., 1992; Roseman et al., 1990;
slightly higher than for older women, Brignone et al. (1987) found Anderson K.E. et al., 1992; Houlston et al., 1992), with conflicting
the risk in pre- and post-menopausal women to be similar and results. Three studies showed an increase in risk in older women
Clavel et al. (1991)/Andrieu et al. (1993) found little difference in (Mettlin et al., 1990; Ambrosone et al., 1995; Parazzini et al.,
risk in women under 40 compared to those over 50. 1992), 5 studies showed an increased risk in younger women (Calle
et al., 1993; Adami et al., 1981; Meirik et al., 1986; Sattin et al.,
History of breast cancer in a first-degree relative 1985; Roseman et al., 1990) and 3 studies showed no effect of
The results of the 38 studies (Adami et al., 1981; Ambrosone et subject age on risk (Byrne et al., 1991; K.E. Anderson et al., 1992;
al., 1995; Anderson K.E. et al., 1992; Bain et al., 1980; Brinton et Houlston et al., 1992).
al., 1995; Bucalossi et al., 1954; Byrne et al., 1991; Calle et al., Four studies reported possible interactions between age of the
1993; Carter et al., 1989; Chen et al., 1994; Colditz et al., 1993; proband and age of the relative at diagnosis. These all reported that
Gomes et al., 1995; Helmrich et al., 1983; Houlston et al., 1992; the subgroup of women with the highest risk are young women
Meirik et al., 1986; Mettlin et al., 1990; Negri et al., 1988; Nomura whose relative was also young at the time of diagnosis (Mettlin et
et al., 1986; Paul et al., 1990; Peto et al., 1996; Rohan et al., 1988; al., 1990; Calle et al., 1993; Houlston et al., 1992; Peto et al.,
Rookus et al., 1994; Roseman et al., 1990; Ross et al., 1980; Sattin 1996). One of these studies had found the overall risk to be higher
et al., 1985; Parazzini et al., 1992; Sellers et al., 1993; Siskind et in older women (Mettlin et al., 1990).
FAMILY HISTORY AND BREAST CANCER RISK 803
no effect (Byrne et al., 1991; Colditz et al., 1993), but in the other 3
the risk was found to be higher if the mother developed cancer at an
early age. Chen et al. (1994) found a relative risk of 4.4 if the
mother was affected before age 50 years compared to a relative risk
of 1.2 if the mother was aged 50 years or over at diagnosis; Tulinius
et al. (1982) found a relative risk of 2.7 for women whose mother
was under age 45 years at diagnosis compared with a RR of 1.1 for
women whose mothers were diagnosed aged 55–64 years; and
Andrieu et al. (1993)/Clavel et al. (1991) found a relative risk of
2.3 if the mother was affected before age 50 years compared to a
relative risk of 1.1 if the mother was aged 50 years or over at
diagnosis.
Eight studies reported the effect of age of the subject on risk in
women whose mother had breast cancer. In 5 of these, age was
found to have no effect (Andrieu et al., 1993; Clavel et al., 1991;
Byrne et al., 1991; Colditz et al., 1993; Paffenbarger et al., 1980;
Tulinius et al., 1982), and in 3 the risks were higher in younger
women (Bain et al., 1980; Calle et al., 1993; Jick et al., 1989). Bain
et al. (1980) found a higher risk only in the youngest age group, i.e.,
RR of 3.1 in women aged 30–39 compared to 1.9 in those aged
50–55.
Two studies described the effect of an interaction between
subject and relative age (Andrieu et al., 1993/Clavel et al., 1991;
Colditz et al., 1993). In both these studies, the risk was highest in
young women whose mother was also affected at a young age.
FIGURE 2 – Risk of breast cancer in women with a history of breast cancer in a first-degree relative. See legend to Figure 1 for explanation of
symbols.
FAMILY HISTORY AND BREAST CANCER RISK 805
History of breast cancer in mother and sister Two studies found a constant age-specific risk for women with a
The effect of having two affected first-degree relatives was history of breast cancer in a second-degree relative (Mettlin et al.,
reported by 5 studies (Adami et al., 1981; Bain et al., 1980; Colditz 1990; Roseman et al., 1990).
et al., 1993; Negri, personal communication; Sattin et al., 1985), Summary estimates of age-specific risk
with a range of risk estimates from 2.5–13.6. The pooled risk
estimate was 3.6 (CI 5 2.5, 5.0). As described above, those studies that reported on the effect of
age at diagnosis of the relative suggest a general trend for the risk to
be higher if the relative had been diagnosed at younger age, though
History of breast cancer in a second-degree relative this effect has not been shown in all studies for all types of family
Figure 6 shows the results from 10 studies that reported on the history categories. A similar trend is apparent for age of the subject.
risk conferred by a family history of breast cancer in a second- However, a more consistent pattern begins to emerge from the
degree relative (Adami et al., 1980, 1981; Teare et al., 1994; White summary estimates of age-specific risk. These are shown in Table
et al., 1994; Siskind et al., 1989; Paul et al., 1990; Rookus et al., III. For all classes of family history except a positive maternal
1994; Sattin et al., 1985; Roseman et al., 1990; Toti et al., 1986; history, risk is greater in younger subjects, and when the relative
Mettlin et al., 1990). These were generally lower than the risk was affected before age 50, with the greatest risk in women under
associated with having an affected first-degree relative, with risk age 50 who have a relative with early-onset breast cancer. The only
estimates ranging from 1.2–1.9. The summary risk estimate was 1.5 exceptions to this were the risks to women who have a mother
(CI 5 1.4, 1.6). diagnosed with breast cancer aged 50 or more. In this group the risk
806 PHAROAH ET AL.
FIGURE 5 – Risk of breast cancer in women with a history of breast FIGURE 6 – Risk of breast cancer in women with a history of breast
cancer in a daughter. See legend to Figure 1 for explanation of symbols. cancer in a second-degree relative. See legend to Figure 1 for
explanation of symbols.
in women aged less than 50 were found to be lower then the risks to
older women. However, these risks are based on only 2 studies and where the overlap from study populations was large, or where one
the confidence intervals are wide. publication superseded another as a more recent report of an
ongoing study: as in reports from the Breast Cancer Detection
DISCUSSION
Demonstration Project (Brinton et al., 1982; Byrne et al., 1991),
studies from Sweden (Adami et al., 1980, 1981) and studies from
We identified 74 studies between 1935–1995 reporting the northern Italy (La Vecchia et al., 1987; Parazzini et al., 1992).
relationship between family history and breast cancer. Four major We found consistent results for the various study designs across
study designs were used, i.e., the retrospective cohort, the prospec- different family history types, despite the potential for different
tive cohort and case-control studies using either hospital or kinds of bias that may apply. This consistency may in part be the
population controls. In addition, various categories of family result of publication bias. However, the results of several early
history were investigated, encompassing both types of relative studies of inheritance and cancer were interpreted as evidence that
affected, (i.e., mother or sister) and the age of the affected relative. inheritance plays no part in the development of cancer (Lane-
Our study has attempted to be as inclusive as possible, and we Claypon, 1926), and subsequent site-specific studies were carried
have avoided introducing any subjective biases of our own such as out to confirm or refute these findings (Wassink, 1935; Penrose et
study quality. This does, however, introduce the potential for bias al., 1948; Anderson V.E. et al., 1950; Woolf, 1955). It would thus
by repetition of data due to overlap of study populations. We have seem likely that any early studies which confirmed these findings
thus taken great care to include only one data set from studies for site-specific cancers would have been published, but we failed
FAMILY HISTORY AND BREAST CANCER RISK 807
TABLE III – POOLED ESTIMATES OF RISK (95% CI) BY AGE OF SUBJECT et al., 1992; Gomes et al., 1995; Harris et al., 1990; Helmrich et al.,
AND AGE OF AFFECTED RELATIVE
1983; Rookus et al., 1994; Toti et al., 1986; UK National
Family history Subject age Case-Control Study Group, 1989; Calle et al., 1993; Carter et al.,
category ,50 years $50 years All ages 1989; Colditz et al., 1993), and in all of these the unadjusted
relative risk was similar to the adjusted relative risk.
Any first-degree
relative The risk associated with a positive mother history tended to be
,50 years 3.3 (2.8, 3.9)1 1.8 (1.6, 2.0)1 2.3 (2.2, 2.5)2 less than the risk associated with a positive sister history, though
$50 years 1.8 (1.5, 2.2)1 1.7 (1.5, 2.0)1 1.8 (1.6, 2.0)2 the difference fails to reach statistical significance. This could be
All ages 2.4 (2.2, 2.7)3 1.9 (1.8, 2.0)3 2.1 (2.0, 2.2) the result of differential bias in the risk estimates, in that recall of
Mother maternal history is likely to be less complete than for sister history,
,50 years 2.5 (1.6, 3.8)4 1.7 (1.1, 2.6)4 2.0 (1.7, 2.4)5 though this effect would probably be small. There is also some bias
$50 years 1.6 (1.1, 2.3)4 1.7 (1.2, 2.4)4 1.7 (1.5, 2.0)5 induced by the confounding effects of parity; parous women are at
All ages 2.2 (1.9, 2.6)6 1.8 (1.6, 2.1)6 2.0 (1.8, 2.1)
Sister lower risk of breast cancer compared to nulliparous women (Kelsey
,50 years 3.3 (2.1, 4.5)7 1.8 (1.2, 2.4)7 2.7 (2.4, 3.2)8 and Gammon, 1990). Mothers are by definition parous, whereas
$50 years 3.0 (1.4, 4.6)7 1.9 (1.1, 2.7)7 2.0 (1.7, 2.4)8 sisters may be nulliparous, and this will explain some of the
All ages 3.0 (2.5, 3.5)9 2.0 (1.8, 2.3)9 2.3 (2.1, 2.4) difference risk. Easton et al. (1996) estimated that the risk to
Second-degree mothers or daughters of cases is reduced by a factor of about 0.8 as
relative a result of this effect. This would, for example, increase the true
All ages 1.7 (1.4, 2.0)10 1.6 (1.3, 2.0)10 1.5 (1.4, 1.6) relative risk to a woman under 50 with a mother affected before the
1(Calle et al., 1993; Houlston et al., 1992; Mettlin et al., 1990; Sattin age of 50 from 2.5 to 2.8. This compares with a relative risk of 3.3
et al., 1985; Peto et al., 1996).–2(Byrne et al., 1991; Calle et al., 1993; for the same woman with a sister history of breast cancer before age
Houlston et al., 1992; Mettlin et al., 1990; Sattin et al., 1985; Peto et 50. The remaining difference could be caused by other environmen-
al., 1996; Toti et al., 1986).–3(Calle et al., 1993; Houlston et al., 1992; tal factors that predispose to breast cancer, which are more likely to
Mettlin et al., 1990; Sattin et al., 1985; Parazzini et al., 1992; Peto et be shared by sisters, or by an unidentified genetic predisposition to
al., 1996; Adami et al., 1981; Ambrosone et al., 1995; Anderson K.E. et cancer which behaves in a recessive manner.
al., 1992; Byrne et al., 1991; Meirik et al., 1986; Roseman et al.,
1990).–4(Colditz et al., 1993; Andrieu et al., 1993/Clavel et al., It is now known that a high proportion of high-risk breast cancer
1991).–5(Byrne et al., 1991; Chen et al., 1994; Colditz et al., 1993; families are due to either of the genes BRCA1 (Miki et al., 1994) or
Tulinius et al., 1982; Andrieu et al., 1993/Clavel et al., 1991).–6(Bain et BRCA2 (Wooster et al., 1995). However, the extent to which the
al., 1980; Byrne et al., 1991; Calle et al., 1993; Colditz et al., 1993; familial risks estimated in our report can be explained by these
Paffenbarger et al., 1980; Tulinius et al., 1982; Jick et al., 1989; genes remains unclear, and population-based studies of BRCA1 and
Andrieu et al., 1993/Clavel et al., 1991).–7(Peto et al., 1996).–8(Byrne BRCA2 mutations are required to address this point. In particular, it
et al., 1991; Chen et al., 1994; Tulinius et al., 1982; Peto et al., seems unlikely that the excess familial risk seen above age 50 is due to
1996).–9(Bain et al., 1980; Byrne et al., 1991; Calle et al., 1993;
Paffenbarger et al., 1980; Tulinius et al., 1982; Peto et al., 1996).– these genes, which confer a risk of 20–30% between ages 50–70. This is
10(Mettlin et al., 1990; Roseman et al., 1990). only 6–9 times the population risk. Assuming a carrier frequency of 1 in
500 for BRCA1 and BRCA2 mutations, the familial relative risk in this
age group would be predicted to be around 1.05, much lower than the
to identify any such study. Further evidence for the lack of 1.7 estimated in our report. Thus other genes, perhaps with more
publication bias comes from the plots of relative risk by study. If common alterations, conferring lower risks, are likely to be important in
publication bias is operating, one would expect that, of published explaining familial breast cancer at higher ages.
studies, the larger ones report the smaller effects, as small positive Although the risk of breast cancer associated with a family
trials are more likely to be published than negative ones (Egger and history is higher in younger women, some studies suggest that the
Smith, 1995). This can be examined using the funnel plot (a modifica- risk is particularly high in women under 40 and then falls off
tion of Fig. 1) in which the effect size is plotted against sample size. If rapidly to a background level of elevated risk. The recent increase
this is done, the plot resembles an inverted funnel, with the results of the in our understanding of the genetic basis of breast cancer has
smaller studies being more widely scattered than those of the larger created a renewed interest in how those at increased risk because of
studies, as would be expected if there is no publication bias. family history can be identified, and how those risks might be
A second potentially important form of bias in the case-control modified. There is thus a need for age-specific risks to be more
and some of the retrospective cohort studies is recall bias. Because precisely quantified to enable appropriate counselling to be pro-
women with breast cancer are more likely to remember a positive vided for at-risk women and for use in determining treatment for
family history than control subjects, these study designs are likely those at risk. Our report has provided estimates for broad categories
to overestimate the effect size. This bias is avoided by prospective of age and family history, but meta-analysis based on individual
studies, and indeed there the risk estimates from the four prospec- data will be required to provide more precise age-specific risks.
tive cohorts are less than those for the other types of study.
However, this difference may also be partly accounted for by age
differences in the study populations.
ACKNOWLEDGEMENTS
Confounding by other risk factors for breast cancer (e.g., age at
menarche, parity, age at first birth, age at menopause) may also bias We thank Miss A. Matthews for help in obtaining copies of
results. However, 14 studies reported crude risk estimates and risk publications referred to in this study. B.A.J.P. is a Gibb Fellow of
estimates after adjustment for other risk factors (Bain et al., 1980; the Cancer Research Campaign (CRC), and P.D.P.P. and D.F.E. are
Brinton et al., 1995; Byrne et al., 1991; Chen et al., 1994; Parazzini supported by the CRC.
REFERENCES
ADAMI, H.O., HANSEN, J., JUNG, B. and RIMSTEN, A., Familiality in breast ALBERT, S. and CHILD, M., Familial cancer in the general population.
cancer: a case-control study in a Swedish population. Brit. J. Cancer, 42, Cancer, 40, 1674–1679 (1977).
71–77 (1980).
ALEXANDER, F.E., ROBERTS, M.M. and HUGGINS, A., Risk factors for breast
cancer with applications to selection for the prevalence screen. J. Epide-
ADAMI, H.O., HANSEN, J., JUNG, B. and RIMSTEN, A., Characteristics of miol. Comm. Health, 41, 101–106 (1987).
familial breast cancer in Sweden: absence of relation to age and unilateral
versus bilateral disease. Cancer, 48, 1688–1695 (1981). AMBROSONE, C.B., MARSHALL, J.R., VENA, J.E., LAUGHLIN, R., GRAHAM, S.,
808 PHAROAH ET AL.
NEMOTO, T. and FREUDENHEIM, J.L., Interaction of family history of breast CAMARGOS, A.F., A case-control study of risk factors for breast cancer in
cancer and dietary antioxidants with breast cancer risk. Cancer Causes Brazil, 1978–1987. Int. J. Epidemiol., 24, 292–299 (1995).
Control, 6, 407–415 (1995). HARRIS, R.E., ZANG, E.A. and WYNDER, E.L., Oral contraceptives and breast
ANDERSON, K.E., EASTON, D.F., MATTHEWS, F.E. and PETO, J., Cancer cancer: a case-control study. Int. J. Epidemiol., 19, 240–246 (1990).
mortality in the first degree relatives of young breast cancer patients. Brit. J. HELMRICH, S.P. and 13 OTHERS, Risk factors for breast cancer. Amer. J.
Cancer, 66, 599–602 (1992). Epidemiol., 117, 35–45 (1983).
ANDERSON, V.E., GOODMAN, H.O. and REED, S.C., Variables related to HIATT, R.A., BAWOL, R., FRIEDMAN, G.D. and HOOVER, R., Exogenous
human breast cancer, University of Minnesota Press, Minneapolis (1950). estrogen and breast cancer after bilateral oophorectomy. Cancer, 54,
ANDRIEU, N., CLAVEL, F., AUQUIER, A., LE, M.G., GAIRARD, B., PIANA, L., 139–144 (1984).
BREMOND, A., LANSAC, J., FLAMANT, R. and RENAUD, R., Variations in the HOLZEL, D., BOLLSCHWEILER, E. and EIERMANN, W., Risikofactoren zum
risk of breast cancer associated with a family history of breast cancer Mammakarzinom und ihre Nutzung fur die Fruherkennung (Risk factors for
according to age at onset and reproductive factors. J. clin. Epidemiol., 46, breast cancer and their use in early detection. Results of a case control
973–980 (1993). study). Geburtshilfe Frauenheilkd., 46, 573–580 (1986).
ANDRIEU, N., DUFFY, S.W., ROHAN, T.E., LE, M.G., LUPORSI, E., GERBER, HOULSTON, R.S., MCCARTER, E., PARBHOO, S., SCURR, J.H. and SLACK, J.,
M., RENAUD, R., ZARIDZE, D.G., LIFANOVA, Y. and DAY, N.E., Familial risk, Family history and risk of breast cancer. J. med. Genet., 29, 154–157
abortion and their interactive effect on the risk of breast cancer; a combined (1992).
analysis of six case-control studies. Brit. J. Cancer, 72, 744–751 (1995).
JACOBSEN, O., Heredity in breast cancer: a genetic and clinical study of two
BAIN, C., SPEIZER, F.E., ROSNER, B., BELANGER, C. and HENNEKENS, C.H., hundred probands, H.K. Lewis, London (1946).
Family history of breast cancer as a risk indicator for the disease. Amer. J.
Epidemiol., 111, 301–308 (1980). JICK, J.S., WALKER, A.M., STERGACHIS, A. and JICK, H., Oral contraceptives
and breast cancer. Brit. J. Cancer, 59, 618–621 (1989).
BRESLOW, N.E. and DAY, N.E., Statistical methods in cancer research,
Volume 1. The analysis of case-control studies, International Agency for KELSEY, J.L. and GAMMON, M.D., Epidemiology of breast cancer. Epide-
Research on Cancer, Lyon (1980). miol. Rev., 12, 228–240 (1990).
BRIGNONE, G., CUSIMANO, R., DARDANONI, G., GUGLIUZZA, M., LANZARONE, LA VECCHIA, C., DECARLI, A., PARAZZINI, F., GENTILE, A., NEGRI, E.,
F., SCIBILIA, V. and DARDANONI, L., A case-control study on breast cancer CECCHETTI, G. and FRANCESCHI, S., General epidemiology of breast cancer
risk factors in a southern European population. Int. J. Epidemiol., 16, in northern Italy. Int. J. Epidemiol., 16, 347–355 (1987).
356–361 (1987). LANE-CLAYPON, J.E., A further report on cancer of the breast with special
BRINTON, L.A., HOOVER, R. and FRAUMENI, J.F., Interaction of familial and reference to its associated conditions, Ministry of Health, London (1926).
hormonal risk factors for breast cancer. J. nat. Cancer Inst., 69, 817–822 LE, M.G., BACHELOT, A., DOYON, F., KRAMAR, A. and HILL, C., Oral
(1982). contraceptive use and breast cancer or cervical cancer: preliminary results
BRINTON, L.A., DALING, J.R., LIFF, J.M., SCHOENBERG, J.B., MALONE, K.E., of a French case-control study. In: J.P.U. Wolff and J.D. Scott (eds.),
STANFORD, J.L., COATES, R.J., GAMMON, M.D., HANSON, L. and HOOVER, Hormones and sexual factors in human cancer aetiology, pp. 139–147,
R.N., Oral contraceptives and breast cancer risk among younger women. J. Elsevier, Amsterdam (1984).
nat. Cancer Inst., 87, 827–835 (1995). LEE, H.P., GOURLEY, L., DUFFY, S.W., ESTÈVE, J., LEE, J. and DAY, N.E.,
Risk factors for breast cancer by age and menopausal status. Cancer Causes
BROCA, P.P., Traité des tumeurs, Asselin, Paris (1866). Control, 3, 313–322 (1992).
BROWNSON, R.C., BLACKWELL, C.W., PEARSON, D.K., REYNOLDS, R.D., LEE, N.C., ROSERO-BIXBY, L., OBERLE, M.W., GRIMALDO, C., WHATLEY,
RICHENS, J.W. and PAPERMASTER, B.W., Risk of breast cancer in relation to A.S. and ROVIRA, E.Z., A case-control study of breast cancer and hormonal
cigarette smoking. Ann. intern. Med., 148, 140–144 (1988). contraception in Costa Rica. J. nat. Cancer Inst., 79, 1247–1254 (1987).
BUCALOSSI, P., VERONESI, U. and PANDOLFI, A., Il problema dell’ereditarieta LUPORSI, E., Breast cancer and alcohol, Ph.D. Thesis, University of
nell’uomo. II. Il cancro della mamella. Tumori, 40, 365–402 (1954). Paris-Sud (1988).
BURNS, P.E., LEES, A.W., HURLBURT, M.E., MAY, C.L. and GRACE, M., MACKLIN, M.T., Comparison of the number of breast cancer deaths
Reproductive events and family history as risk factors for breast cancer in observed in relatives of breast-cancer patients and the numbers expected on
northern Alberta. Canad. med Assoc. J., 124, 1451–1457 (1981). the basis of mortality rates. J. nat. Cancer Inst., 22, 927–951 (1959).
BYRNE, C., BRINTON, L.A., HAILE, R.W. and SCHAIRER, C., Heterogeneity of MARUBINI, E., DECARLI, A., COSTA, A., MAZZOLENI, C., ANDREOLI, C.,
the effect of family history on breast cancer risk. Epidemiology, 2, 276–284 BARBIERI, A., CAPITELLI, E., CARLUCCI, M., CAVALLO, F., MONFERRONI, N.,
(1991). PASTORINO, U. and SALVINI, S., The relationship of dietary intake and serum
CALLE, E.E., MARTIN, L.M., THUN, M.J., MIRACLE, H.L. and HEATH, C.W., levels of retinol and beta-carotene with breast cancer: the results of a
Family history, age and risk of fatal breast cancer. Amer. J. Epidemiol., 138, case-control study. Cancer, 61, 173–180 (1989).
675–681 (1993). MCPHERSON, K., VESSEY, M.P., NEIL, A., DOLL, R., JONES, L. and ROBERTS,
CARTER, C.L., JONES, D.Y., SCHATZKIN, A. and BRINTON, L.A., A prospective M., Early oral contraceptive use and breast cancer: results of another
study of reproductive, familial and socioeconomic risk factors for breast case-control study. Brit. J. Cancer, 56, 653–660 (1987).
cancer using NHANES I data. Publ. Health Rep., 104, 45–50 (1989). MEIRIK, O., LUND, E., ADAMI, H., BERGSTRÖM, R., CHRISTOFFERSEN, T. and
CHEN, P.L., SELLERS, T.A., RICH, S.S., POTTER, J.D. and FOLSOM, A.R., BERGSJO, P., Oral contraceptive use and breast cancer in young women.
Examination of the effect of nongenetic risk factors on the familial risk of Lancet, ii, 650–654 (1986).
breast cancer among relatives of postmenopausal breast cancer patients. METTLIN, C., CROGHAN, I., NATARAJAN, N. and LANE, W., The association of
Cancer Epidemiol. Biomarkers Prev., 3, 549–555 (1994). age and familial risk in a case-control study of breast cancer. Amer. J.
CLAVEL, F., ANDRIEU, N., GAIRARD, B., BREMOND, A., PIANA, L., LANSAC, J., Epidemiol., 131, 973–983 (1990).
BREART, G., RUMEAU ROUQUETTE, C., FLAMANT, R. and RENAUD, R., Oral MIKI, Y. and 44 OTHERS, A strong candidate for the Hq linked breast and
contraceptives and breast cancer: a French case-control study. Int. J. ovarian cancer susceptibility gene BRCA1. Science, 266, 66–71 (1994).
Epidemiol., 20, 32–38 (1991).
MORABIA, A., SZKLO, M., STEWART, W., SCHUMAN, L. and THOMAS, D.B.,
COLDITZ, G.A., WILLETT, W.C., HUNTER, D.J., STAMPFER, M.J., MANSON, Consistent lack of association between breast cancer and oral contracep-
J.E., HENNEKENS, C.H. and ROSNER, B.A., Family history, age, and risk of tives using either hospital or neighborhood controls. Prev. Med., 22,
breast cancer. Prospective data from the Nurses’ Health Study. J. amer. med. 178–186 (1993).
Ass., 270, 338–343 (1993).
MORGAN, R.W., VAKH, D.V. and CHIPMAN, M.L., Breast feeding, family
EASTON, D.F., MATTHEWS, F.E., FORD, D., SWERDLOW, A.J. and PETO, J., history and breast disease. Amer. J. Epidemiol., 99, 117–122 (1974).
Cancer mortality in relatives of women with ovarian cancer: the OPCS
study. Int. J. Cancer, 65, 284–294 (1996). MURPHY, D.P. and ABBEY, H., Cancer in families. (A study of the relatives of
200 breast cancer probands), Harvard University Press, Cambridge (1959).
EGGER, M. and SMITH, G.D., Misleading meta-analysis (editorial). Brit. NEGRI, E., LA VECCHIA, C., BRUZZI, P., DARDANONI, G., DECARLI, A., PALLI,
Med. J., 310, 752–754 (1995). D., PARAZZINI, F. and DEL TURCO, M.R., Risk factors for breast cancer:
GOLDGAR, D.E., STRATTON, M.R. and EELES, R.A., Familial breast cancer. pooled results from three Italian case-control studies. Amer. J. Epidemiol.,
In: R.A. Eeles, B.A.J. Ponder, D.F. Easton and A. Horwich (eds.), Genetic 128, 1207–1215 (1988).
predisposition to cancer, pp. 227–238, Chapman and Hall, London (1996). NEWCOMB, P.A., STORER, B.E., LONGNECKER, M.P., MITTENDORF, R.,
GOMES, A.L., GUIMARAES, M.D., GOMES, C.C., CHAVES, I.G., GOBBI, H. and GREENBERG, E.R., CLAPP, R.W., BURKE, K.P., WILLETT, W.C. and MAC
FAMILY HISTORY AND BREAST CANCER RISK 809
MAHON, B., Lactation and a reduced risk of premenopausal breast cancer. N. SCHILDKRAUT, J.M., HULKA, B.S. and WILKINSON, W.E., Oral contraceptives
Engl. J. Med., 330, 81–87 (1994). and breast cancer: a case-control study with hospital and community
NGELANGEL, C.A., LACAYA, L.B., CORDERO, C. and LAUDICO, A.V., Risk controls. Obstet. Gynaecol., 76, 395–402 (1990).
factors for breast cancer among Filipino women. Phil. J. Int. Med., 32, SELLERS, T.A., GAPSTUR, S.M., POTTER, J.D., KUSHI, L.H., BOSTICK, R.M.
231–236 (1994). and FOLSOM, A.R., Association of body fat distribution and family histories
NOMURA, A.M.Y., KOLONEL, L.N., HIROHATA, T. and LEE, J., The associa- of breast and ovarian cancer with risk of postmenopausal breast cancer.
tion of replacement estrogens with breast cancer. Int. J. Cancer, 37, 49–53 Amer. J. Epidemiol., 138, 799–803 (1993).
(1986). SISKIND, V., SCHOFIELD, F., RICE, D. and BAIN, C., Breast cancer and
OLSSON, H., MOLLER, T.R. and RANSTAM, J., Early oral contraceptive use breastfeeding: results from an Australian case-control study. Amer. J.
and breast cancer among premenopausal women: final report from a study Epidemiol., 130, 229–236 (1989).
in southern Sweden. J. nat. Cancer Inst., 81, 1000–1004 (1989). SLATTERY, M.L. and KERBER, R.A., A comprehensive evaluation of family
PAFFENBARGER, R.S.J., KAMPERT, J.B. and CHANG, H.G., Characteristics history and breast cancer risk. The Utah Population Database. J. Amer. med.
that predict risk of breast cancer before and after the menopause. Amer. J. Ass., 270, 1563–1568 (1993).
Epidemiol., 112, 258–268 (1980). SMITHERS, D.W., RIGBY-JONES, P., GALTON, D.A.G. and PAYNE, P.M.,
PARAZZINI, F., LA VECCHIA, C., NEGRI, E., FRANCESCHI, S. and BOCCIOLONE, Cancer of the breast: a review. Brit. J. Radiol. (Suppl. 4) (1952).
L., Menstrual and reproductive factors and breast cancer in women with a TAO, S.C., YU, M.C., ROSS, R.K. and XIU, K.W., Risk factors for breast
family history of the disease. Int. J. Cancer, 51, 677–681 (1992). cancer in Chinese women of Beijing. Int. J. Cancer, 42, 495–498 (1988).
PASSEY, R.D., WAIMAN, M., ARMSTRONG, E.C. and RHODES, J., Heredity in TEARE, M.D., WALLACE, S.A., HARRIS, M., HOWELL, A. and BIRCH, J.M.,
human breast cancer. Acta Unio Int. Contra Cancrum, 8, 184–186 (1952). Cancer experience in the relatives of an unselected series of breast cancer
PAUL, C., SKEGG, D.C.G. and SPEARS, G.F.S., Oral contraceptives and risk patients. Brit. J. Cancer, 70, 102–111 (1994).
of breast cancer. Int. J. Cancer, 46, 366–373 (1990). TOTI, A., AGUGIARO, S., AMADORI, D., BUZZI, G., BRUZZI, P., BUIATTI, E.,
PENROSE, L.S., MACKENZIE, H.J. and KARN, M.N., A genetical study of CAPELLI, M.C., CIATTO, S., DELFINO, E. and FOTI, E., Breast cancer risk
human mammary cancer. Brit. J. Cancer, 2, 168–176 (1948). factors in Italian women: a multicentric case-control study. Tumori, 72,
PETO, J., EASTON, D.F., MATTHEWS, F.E., FORD, D. and SWERDLOW, A.J., 241–249 (1986).
Cancer mortality in relatives of women with breast cancer: the OPCS study. TULINIUS, H., DAY, N.E., BJARNASON, O., GEIRSSON, G., JOHANNESSON, G.,
Int. J. Cancer, 65, 273–283 (1996). LICEAGA DE GONZALEZ, M.A., SIGVALDASON, H., BJARNADOTTIR, G. and
PRIMIC-ZAKELJ, M., EVSTIFEEVA, T., RAVNIHAR, B. and BOYLE, P., Breast GRIMSDOTTIR, K., Familial breast cancer in Iceland. Int. J. Cancer, 29,
cancer risk and oral contraceptive use in Slovenian women aged 25 to 54. 365–371 (1982).
Int. J. Cancer, 62, 414–420 (1995). UK NATIONAL CASE-CONTROL STUDY GROUP, Oral contraceptive use and
RICHARDSON, S., GERBER, M. and CENEE, S., The role of fat, animal protein breast cancer risk in young women. Lancet, i, 973–982 (1989).
and some vitamin consumption in breast cancer: a case control study in WASSINK, W.F., Cancer et heredité. Genetica, 17, 103–144 (1935).
southern France. Int. J. Cancer, 48, 1–9 (1991).
WHITE, E., MALONE, K.E., WEISS, N.S. and DALING, J.R., Breast cancer
ROHAN, T., MCMICHAEL, A.J. and BAGHURST, P.A., A population based among young U.S. women in relation to oral contraceptive use. J. nat.
case-control study of diet and breast cancer in Australia. Amer. J. Cancer Inst., 86, 505–514 (1994).
Epidemiol., 128, 478–489 (1988).
WHO COLLABORATIVE STUDY OF NEOPLASIA AND STEROID HORMONES,
ROOKUS, M.A. and VAN LEEUWEN, F.E., for THE NETHERLANDS ORAL Breast cancer and combined oral contraceptives: results from a multina-
CONTRACEPTIVES AND BREAST CANCER STUDY GROUP, Oral contraceptives tional study. Brit. J. Cancer, 61, 110–119 (1990).
and risk of breast cancer in women aged 20–54 years. Lancet, 344, 844–851
(1994). WOOLF, C.M., Investigations on genetic aspects of carcinoma of the
stomach and breast, Volume 2. University of California Press, Berkeley
ROSEMAN, D.L., STRAUS, A.K. and SHOREY, W., A positive family history of (1955).
breast cancer. Does its effect diminish with age? Ann. intern. Med., 150,
191–194 (1990). WOOSTER, R. and 40 OTHERS, Identification of the breast cancer susceptibil-
ity gene BRCA2. Nature (Lond.), 378, 789–792 (1995).
ROSS, R.K., PAGANINI-HILL, A., GERKINS, V.R., MACK, T.M., PFEFFER, R.,
ARTHUR, M. and HENDERSON, B.E., A case-control study of menopausal WYNDER, E.L., MACCORNACK, F.A. and STELLMAN, S.D., The epidemiology
estrogen therapy and breast cancer. J. Amer. med. Ass., 243, 1635–1639 of breast cancer in 785 United States caucasian women. Amer. J. Epide-
(1980). miol., 41, 2341–2354 (1978).
ROSSING, M.A., STANFORD, J.L., WEISS, N.S. and HABEL, L.A., Oral YANG, C.P., DALING, J.R., BAND, P.R., GALLAGHER, R.P., WHITE, E. and
contraceptive use and risk of breast cancer in middle-aged women. Amer. J. WEISS, N.S., Non-contraceptive hormone use and risk of breast cancer.
Epidemiol., 144, 161–164 (1996). Cancer Causes Control, 3, 475–479 (1992).
SATTIN, R.W., RUBIN, G.L., WEBSTER, L.A., HUEZO, C.M., WINGO, P.A., YUAN, J., YU, M.C., ROSS, R.K., GAO, Y. and HENDERSON, B.E., Risk factors
ORY, H.W. and LAYDE, P.M., Family history and the risk of breast cancer. J. for breast cancer in Chinese women in Shanghai. Cancer Res., 48,
Amer. med. Ass., 253, 1908–1913 (1985). 1949–1953 (1988).