Breast cancers arising in women with and without a germline mutation in the BRCA1 or BRCA2 gene display different histological features, which suggests unique mechanisms of molecular pathogenesis: We used a molecular pathological analysis to define the genetic abnormalities relevant to these specific pathogeneses. Tumor material was studied from 40 women with breast cancer diagnosed before 40 years of age, sampled from a population-based study and stratified by BRCA1 and BRCA2 germline mutation status. Cases were not selected for family history or ethnic origin, and none were known to be genetically related. Thus, germline mutation itself is likely to impact on the molecular pathogenesis of these tumors, with no substantial influence due to modifying genetic or environmental factors. Breast cancers occurring in BRCA1 mutation carriers had significantly higher levels of p53 expression, including the preinvasive (carcinoma in situ) stage of disease, compared with cancers occurring in BRCA2 mutation carriers or women with no detectable germline mutation. These cancers also had a higher proliferation rate as measured by Ki-67 antibody. Expression of the prognostic factors c-erbB-2, cyclin D1, and estrogen receptor was significantly less common in BRCA1 mutation carriers. Lower levels of cyclin D1 were also found in cancers from BRCA2 mutation carriers compared with non-mutation carriers. Direct p53 mutation analysis revealed mutations in 18% of all of the early-onset breast cancers within the study and included rare insertion and deletional mutations in cancers from BRCA1 mutation carriers. Our data indicate that a BRCA1 breast cancer phenotype may be recognized by an exceptionally high proliferation rate and early and frequent p53 overexpression but infrequent selection for overexpression of several other prognostic factor proteins known to be involved in breast oncogenesis. In contrast, breast cancers arising in BRCA2 mutation carriers have a more heterogeneous phenotypic profile.
There is increasing evidence that specific histological phenotypes can be recognized in breast cancers occurring in women with germline mutations in the breast cancer susceptibility genes, BRCA1 and BRCA2 (1, 2, 3, 4, 5, 6, 7) . Typically, breast cancers occurring in BRCA1 mutation-carrying heterozygotes are high-grade and have a high proliferation rate, with medullary or atypical medullary cancers being over-represented. In contrast, lobular cancers and extensive intraduct cancers are more frequent in women with germline BRCA2 mutations. Certain histological subtypes of invasive breast cancers have been associated with altered somatic expression of tumor suppressor genes and oncogenes, such as the reduced E-cadherin expression and increased cyclin D1 expression reported in infiltrating lobular cancers (8 , 9) . Thus, the distinct histological phenotypes identified in breast cancers occurring in BRCA1 and BRCA2 mutation carriers may be a consequence of both the germline mutation and subsequent specific somatic mutations occurring in breast tissue.
Recent studies on breast cancers diagnosed in all age groups have indicated that those in BRCA1 mutation carriers are typically ER-, 3 PR-, and c-erbB-2-negative (10, 11, 12, 13) . In addition, high rates of mutation in the tumor suppressor gene p53 have been identified in breast cancers from BRCA1 and BRCA2 mutation carriers (14, 15, 16) . However, one study of Jewish women found no difference in p53 protein expression in BRCA1 and BRCA2-associated breast cancers diagnosed before age 42 years when compared with breast cancers in noncarriers (13) . A further study demonstrated that p53 over-expression in cancers occurring in BRCA1 mutation carriers was predominantly associated with germline mutations leading to protein truncation within the ring finger domain of the BRCA1 gene (17) . The majority of these molecular data have been accrued from families with multiple individuals affected by breast cancer or from Jewish or Icelandic women, who are known to have a high prevalence (because of founder effects within their populations) of a germline mutation at specific sites within these two genes. Thus it is possible that the pathogenesis of breast cancers occurring in these defined groups is influenced not only by the specific site of the germline BRCA1 or BRCA2 mutation but also by other modifying genetic or environmental factors specific to these groups of women. Conversely, it is not known whether the somatic molecular alterations described in the cancers of these selected population subgroups resemble those seen in the general population.
The histological and biological characteristics of breast cancers occurring in premenopausal women are said to differ from those diagnosed in older women. Compared with later-onset breast cancers, early-onset breast cancers are commonly high-grade with a high proliferation rate, and medullary or atypical medullary cancers are more common (18, 19, 20) . Early-onset breast cancers also commonly lack ER expression and show more frequent p53 overexpression (20 , 21) . All of the above characteristics have also been reported in cancers arising in BRCA1 mutation carriers. Hence, there is uncertainty as to the extent to which breast cancers occurring in women with germline BRCA1 mutations can be distinguished from early-onset cancers arising in women without germline mutations in this gene. In addition, little is known about the molecular pathogenesis of early-onset breast cancer in general during its development and progression to invasive cancer. In this study, we investigate the molecular and immunophenotypic events that occur during the preinvasive and invasive stages of early-onset breast cancers occurring in a population-based sample, either with or without a BRCA1 or BRCA2 germline mutation.
MATERIALS AND METHODS
We analyzed tissue from 40 women with breast cancer diagnosed before the age of 40 years, selected from the population-based ABCFS (22 , 23) . Participants in the ABCFS were recruited from incident, histologically confirmed, first primary cases of invasive breast cancer as recorded on the Victorian and New South Wales State Cancer Registries. Verification of every cancer reported in a relative was sought through cancer registries, pathology reports, hospital records, treating clinicians, and death certificates. All of the women gave informed consent to participate, and the study was carried out with the approval of all of the relevant institutional ethics committees. A total of 467 breast cancer cases, diagnosed between 1992 and 1995, were studied, and blood samples were available from 388 (83%) of these. Mutation detection for BRCA1 and BRCA2 identified 10 cases with a germline BRCA1 mutation and 9 cases with a germline BRCA2 mutation (1) . In addition, 21 cases were randomly selected, who did not carry a mutation in either BRCA1 or BRCA2, and were designated as “control” cancers for comparisons with mutation carriers.
BRCA1 and BRCA2 Mutation Detection.
BRCA1 and BRCA2 mutation detection was performed on DNA extracted from peripheral blood-derived buffy coat using a Progenome II DNA extraction kit (Progen, Ipswich, Australia) and stored in TE buffer. The PTT (24) was used to detect truncating mutations in exon 11 of BRCA1, using primers described by Hogervorst et al. (25) and Friedman et al. (26) . Exon-flanking primers based on the BRCA2 genomic sequence (27) were used for the PTT of BRCA 2 exons 10 and 27, and primers to three overlapping exon fragments were used for exon 11. Mutations detected by PTT were confirmed by cycle sequencing (Perkin-Elmer). If no mutation was found by these methods, sequencing of the remaining exons of BRCA1 and BRCA2 was performed except in 4 cases for whom no further peripheral blood was available. The PCR primers used for cycle sequencing were based on those described by Simard et al. (28) for BRCA1, and Wooster et al. (27) for BRCA2.
Immunohistochemical studies were performed on sections obtained from paraffin blocks of tissues fixed in 10% neutral buffered formalin. Three-μm sections cut from paraffin wax blocks were placed onto silane-coated slides and dried at 60°C for 30 min. The sections were dewaxed in histolene and rehydrated through graded alcohols. Antigen retrieval was used for the antibodies indicated in Table 1 ⇓ . The antigen retrieval methods used were 2 min at pressure in a pressure cooker or 20 min on low setting in a microwave, both in 10 mm sodium citrate (pH 6.0), or 5 min enzyme digestion with pepsin at 37°C. The DAKO Autostainer was used for the immunohistochemistry staining. All of the wash steps used 50 mm Tris-HCl (pH 7.6) and 0.05% Tween 20. The sections were treated with 3% hydrogen peroxide for 10 min. Some of the antibodies had a blocking step for 5 min using Protein Blocking Agent (Immunon, Pittsburgh, PA) as indicated in Table 1 ⇓ . The sections were incubated in primary antibody at the dilutions listed in Table 1 ⇓ for 30 min at room temperature with the exception of EGFR, which was incubated at 37°C for 60 min. Sections were incubated with biotinylated secondary antibody, followed by peroxidase-conjugate Streptavidin using DAKO LSAB2 or LSAB+ kits as indicated in Table 1 ⇓ . LSAB2 incubation times were 10 min for each step; LSAB+ incubation times were 15 min for each step. Staining was visualized using 3-amino-9-ethyl-carbazole, washed in water, and counterstained with hematoxylin. Crystal Mount (Biomedia, Foster City, CA) was applied to sections and dried on a 60°C hotplate. The sections were then coverslipped with DPX mountant.
For all of the antibodies except Ki-67, the intensity of staining and proportion of positive cells were determined for the normal breast epithelium, DCIS, and invasive carcinoma components from each case, modified from the method described by Allred et al. (29) The proportion score represented the estimated fraction of positive staining cells (0, ≤10%; 1, 11–25%; 2, 26–50%; 3, 51–75%; 4, 76–90%; 5, ≥91%). The intensity score represented the estimated average staining intensity of positive cells (0, none; 1, weak; 2, moderate; 3, strong). The results were then analyzed by two different methods. The first was a simple estimate of whether the cellular component was positive or negative for each antibody, positivity being defined as greater than 10% of cells positive. The second method took into account both the intensity and the proportion of positive cells to give a semiquantitative estimate of the expression levels of antigen in the tissue. Samples with intensity scores of 0 or 1 were designated as negative-to-weak expression. For intensity scores of 2 and 3, a combined score was derived by adding the intensity and proportion scores. Combined scores of 2 and 3 were designated as negative-to-weak expression, 4–6 as moderate expression, and 7 or 8 as strong expression. For each antibody, only the cellular compartment that was expected to express the antigen of interest was scored (i.e., either nuclear, cytoplasmic, or membranous staining). The expression level was not calculated for Ki-67 antibody staining because this antibody was used simply as an indication of the number of cells in the cell cycle. Instead the proportion score was calculated at both 10% and 50% positive cell cutoff points.
Microdissection and DNA Extraction.
Microdissection of target tissue was performed on dewaxed, methyl green-stained slides using a 21-gauge needle. Microdissected material (approximately 500 cells per μl of buffer) was placed directly into a 0.5-ml tube containing Proteinase K buffer (10 mm Tris-KCl (pH 8.0), 10 mm MgCl, 0.5% Tween 20, and 0.4 mg/ml Proteinase K). Tubes were then incubated for 72 h at 55°C with Proteinase K (0.3 μl of 20 mg/ml Proteinase K per 15 μl of buffer) added after 24 and 48 h. After 72 h, Proteinase K was inactivated by boiling for 15 min.
Table 2 ⇓ summarizes the primers used for PCR amplification of DNA fragments spanning microsatellite sites at different chromosome loci. The PCR amplification reaction contained 2 μl of microdissected DNA sample, 10 mm Tris, 50 mm KCl (pH 8.3), 1.5–4.0 mm MgCl, 1 μm of each primer, 0.1 mm dTTP, 0.1 mm dCTP, 0.1 mm dGTP, 4.0 μm dATP, reaction buffer (1X; Perkin-Elmer), 1 unit of AmpliTaq Gold (Perkin-Elmer) and 0.5 μCi [33P]dATP (2000 Ci/mmol). DNA was amplified in a 9600 Thermocycler (Perkin-Elmer) as follows: 94°C for 5 min, 35 cycles of 55°C for 1 min, 72°C for 1 min, and 94°C for 1 min with a final 72°C hold for 10 min. The amplified products were separated in a 6% denaturing polyacrylamide gel and visualized by autoradiography. LOH was considered to be present if the intensity of alleles varied by more than 50%.
p53 Mutation Detection.
Mutations in exons 5–10 of the p53 gene were detected by cycle sequencing, SSCP, or sub-cloning. Mutations that were detected at least twice from different PCR reactions by the above methods were designated p53 mutation-positive.
Microdissected DNA (1–2 μl) was amplified with primers and MgCl2 (at concentrations described in Table 3 ⇓ ), in a total reaction volume of 25 μl containing 2 mm dNTPs, 10 mm Tris (pH 8.3), 50 mm KCl, and 0.5 units of AmpliTaq Gold DNA polymerase (Perkin-Elmer). Touch-down PCRs were performed in a 9600 Thermocycler (Perkin-Elmer) with the initial annealing temperature 8°C higher than the final temperature. DNA was denatured for 10 min at 94°C initially, followed by cycles of 94°C for 1 min, variable annealing temperatures for 1 min, 72°C for 1 min, followed by a final 72°C hold for 10 min. The annealing temperature was decreased every 2 cycles by 2°C, touching down at the final annealing temperature (see Table 3 ⇓ ). An additional 30 cycles were performed, and amplification was confirmed by agarose gel electrophoresis.
For cycle sequencing, PCR products were purified using a Sephaglas Bandprep kit (Pharmacia Biotech) and eluted in TE buffer. Purified products (6 μl) were sequenced using AmpliCycle Sequencing Kits (Perkin-Elmer) incorporating [33P]dATP (Amrad Pharmacia Biotech) and the same primers used for the amplification reaction. Fragments were analyzed using standard 6% polyacrylamide/urea sequencing gels, run for 1–2 h at 60 W, then dried, and exposed to X-ray film (Biomax-MR, Kodak) overnight.
PCR products were ligated into pGEM-T Easy vectors (Promega) according to the manufacturer’s instructions and transformed into JM109 High Efficiency Competent Cells (Promega). Recombinant plasmid DNA was cycle-sequenced as described above.
SSCP analysis was performed using 1 μl of microdissected DNA, amplified in a total volume of 25 μl containing 10 mm Tris-HCl (pH 8.3); 50 mm KCl; 2 mm MgCl; 200 μm (each) dGTP, dTTP, and dCTP; 20 μm dATP; 1 μCi [α-32P]dATP; 400 nm appropriate primers (see Table 3 ⇓ ), and 0.5 unit of AmpliTaq Gold (Perkin-Elmer). PCR was performed using the touch-down protocol described above.
Five μl of PCR product was diluted in SDS/EDTA to give a final concentration of 0.1% SDS-10 mm EDTA. Equal volumes of diluted product were mixed with loading dye comprising; 95% formamide, 20 mm EDTA, 0.05% bromphenol blue, and 0.02% xylene cyanole FF. Samples were denatured for 5 min at 95°C and immediately placed on ice, and 5 μl was loaded onto a 6% polyacrylamide gel containing 90 mm Tris-borate, 2 mm EDTA, and 10% glycerol and run at 60 W for 6 h at 4°C. Gels were dried and exposed to overnight autoradiography. Shifted bands were cut out of the gel, DNA was eluted in 20 μl of H20 at 60°C and 5 μl was used to amplify the appropriate exon as described above. PCR products were cleaned and sequenced as described.
Statistical comparisons between groups were assessed by Fisher’s exact test. Nominal Ps are given without adjustment for multiple comparisons. Because of the small numbers of cases in each category, raw data are provided together with the nominal Ps.
Tumor material from 40 women selected from the ABCFS were analyzed: 19 cases with a germline BRCA1 or BRCA2 mutation (10 BRCA1 and 9 BRCA2) and 21 cases for whom there were no detectable germline mutations in these genes. The 19 cases with germline mutations comprised all but one of the 20 mutations detected in the ABCFS (1) . One woman with a germline protein-truncating mutation in both BRCA1 and BRCA2 was excluded from the statistical analysis because the tumor could not be classified into either the BRCA1 or BRCA2 category. Ten different BRCA1 protein-truncating mutations and 7 different BRCA2 protein-truncating mutations were represented (1) . None of the 40 women were known to be genetically related. There was no difference in the proportion with a family history of breast cancer between carriers of a germline BRCA1 or BRCA2 mutation and noncarriers (P > 0.1; Ref. 1 ). The mean ages (SD) of diagnosis of breast cancer for the BRCA1 and BRCA2 mutation carriers and the control cancers were 32.3 (5.3), 34.9 (3.7) and 36.5 (2.6) years, respectively. Immunohistochemical and molecular information could not be obtained on the invasive component of one control cancer because it was microinvasive and the invasive component was insufficient for study. A detailed histological analysis of these breast cancers has been previously reported (1) .
LOH at the BRCA1 and BRCA2 Loci.
LOH at one or more intragenic BRCA1 loci was common in cancers arising in BRCA1 mutation carriers, being present in seven of eight informative cancers. Two cases were homozygous for all of the markers studied, and in one case, LOH could not be consistently demonstrated. A focus of DCIS was also available for analysis in three of the informative cases. LOH was again documented in this preinvasive stage of disease
Similarly, LOH was commonly seen in BRCA2 germline mutation carriers, with the loss of one or more microsatellite markers approximately 0.5 cM and 2.1 cM from the BRCA2 locus demonstrated in all of the seven informative cancers. LOH was analyzed in the DCIS component of five of these. Loss of the same allele was demonstrated in four cases, and in three of these, multiple sites of DCIS were dissected, all showing similar loss. LOH was not demonstrated in the DCIS accompanying an invasive carcinoma in one case, despite the DCIS having an homogeneous cellular morphology with no apparent contamination by other cell types.
The immunophenotypes of invasive breast cancers in BRCA1, BRCA2, and non-mutation carriers are shown and contrasted in Table 4 ⇓ . Compared with the other two groups, p53 positivity was more common in tumors arising in BRCA1 mutation carriers. In addition, when both the intensity and extent of staining was scored, all of the positive breast cancers in the BRCA1 mutation carriers showed strong p53 expression (Fig. 1A) ⇓ . In contrast, none of the cancers in the other two groups showed strong expression (P < 0.002). Positive p53 expression was not related to the site of germline mutation in BRCA1 mutation carriers (data not shown). Strong p53 expression was also seen in the DCIS component of all of the cancers from the BRCA1 mutation carriers with positive p53 staining in the invasive component (Fig. 1B) ⇓ . However, strong p53 expression was not present in the morphologically normal breast epithelium. Overall, 44% of all of the invasive breast cancers studied were p53-positive at the 10% cutoff level.
Positive cyclin D1 staining was rare in cancers arising in BRCA1 or BRCA2 mutation carriers compared with control cancers (P < 0.01, Table 4 ⇓ ). The normal breast epithelium showed variable staining. The accompanying DCIS component was cyclin D1-negative in all but one of these cancers in BRCA1 mutation carriers. Infrequent cyclin D1 staining in cancers arising in BRCA1 mutation carriers was further accentuated when the degree of antigen expression was assessed as strong or moderate versus weak or negative. Only 11% (1 of 9) of cancers in BRCA1 mutation carriers showed strong or moderate staining, compared with 63% of control cancers (P = 0.02).
Fifty-seven per cent of the cancers in BRCA1 mutation carriers, compared with none of the cancers in BRCA2 mutation carriers and only 5% of the control cancers, showed exceptionally high Ki-67 staining (defined as >50% of cells positive; Table 4 ⇓ ). Ki-67 staining was also exceptionally high in the DCIS component of disease in BRCA1 mutation carriers (data not shown).
c-erbB-2 positivity was less frequent (40%) in BRCA1 mutation carriers compared with 78 and 90% of cancers in BRCA2 mutation carriers and control cancers, respectively (Table 4) ⇓ . In addition, none of the cancers in BRCA1 mutation carriers showed strong expression of c-erbB-2, whereas 44% (4 of 9) of cancers in BRCA2 mutation carriers and 45% (9 of 20) of control cancers showed strong membranous expression (BRCA1versusBRCA2, P = 0.03; BRCA1versus control cancers, P = 0.01).
ER-positive cancers were rare in the cancers from BRCA1 mutation carriers (10%, Table 4 ⇓ ). The DCIS component was also negative in all of the ER-negative BRCA1 mutation carriers. These nine ER-negative cases were also negative for PR and pS2. In contrast, 67% of invasive cancers in BRCA2 mutation carriers and 55% of control cancers were ER-positive. PR and pS2 staining were variable in these cancers. Overall, 44% of all of the breast cancers studied were ER-positive. The ER status of the normal epithelial component was variable in all of the cancers examined. There was no difference in Cathepsin D positivity when cancers from BRCA1 mutation carriers were compared with cancers from BRCA2 mutation carriers and control cancers. Nonetheless, when the antigen expression level was assessed, only 13% (one of eight) of cancers in BRCA1 mutation carriers showed strong or moderate staining, compared with 67% (six of nine) of cancers in BRCA2 mutation carriers (P = 0.05). No obvious differences between the three categories of breast cancer were noted in the immunophenotype of the other antibodies studied.
Mutation Detection and LOH at the p53 Locus.
p53 mutation was detected in 7 of the 39 invasive cancers (18%). The mutations were scattered over four of the six exons examined (Table 5) ⇓ . Three of the mutations were detected in cancers from germline BRCA1 mutation carriers, and three from control cancers. No functional mutations were detected in the BRCA2 mutation carriers, although one silent mutation (pro>pro) was detected. Five mutations were missense; all of them were transitions with 4 being G>A transitions. Five of the six cases with functional mutations showed positive immunohistochemical staining for p53, whereas one—with a mutation in exon 9—was negative by immunohistochemistry. Conversely, no p53 mutations were identified in 12 other carcinomas in which p53 antigen was detected by immunohistochemistry. DCIS was available for study in two cases with p53 mutations in exon 7 of the invasive component (two control cancers). Identical mutations were documented in the DCIS component of each case.
Cancers from 37% (13 of the 35 tested) of cases were shown to have LOH at one or more intragenic p53 microsatellite locus (4 in cancers from BRCA1 mutation carriers, 5 from BRCA2 mutation carriers, and 4 from control cancers). However, only 6 of the 13 cases with LOH showed p53 antigen expression by immunohistochemistry. Only two of the six cases with functional p53 mutations detected by molecular methods showed LOH at the p53 locus (Table 5) ⇓ . No LOH was detected in three of the p53 mutation-positive cases, and one case could not be assessed because of insufficient material.
This paper has demonstrated, for the first time on a population basis, that breast cancers occurring in young BRCA1 or BRCA2 mutation carriers show distinct molecular pathogeneses, and that these pathways are different from cancers occurring in mutation-negative women. Previous studies of multiple-case BRCA1 or BRCA2-associated breast cancer kindreds and studies performed on populations with strong founder mutations have identified reduced expression of ER, PR, and c-erbB-2 in cancers from BRCA1 mutation carriers (10, 11, 12, 13) . In addition, somatic mutations in p53, or p53 protein overexpression, have been documented in cancers occurring in BRCA1 and BRCA2 mutation carriers (14, 15, 16, 17) . Although the number of cases with BRCA1 or BRCA2 mutations was small in our sample and, therefore, some of the nominally significant findings reported herein may not be reproducible in subsequent population-based studies, our study confirms the above findings drawn from selected groups at high-risk of BRCA1 or BRCA2 germline mutations. In addition, the present study also serves to identify molecular and immunophenotypic changes that are independent of germline mutation site or presence of a family history. In this way, we hope to emphasize the molecular pathogenic consequences of the germline mutations themselves and, as far as possible, minimize the impact of modifying genetic or environmental factors.
Elevated p53 expression seems to be common in cancers from BRCA1 mutation carriers. However, data from the present series and Crook et al. (15) indicate that the type and site of mutation documented in cancers arising in BRCA1 mutation carriers are unusual compared with breast cancers in general. We were able to show that p53 overexpression consistently occurred at the preinvasive (DCIS) stage of the disease, but not in morphologically normal breast epithelium. Hence, these data suggest p53 overexpression is an important and early event in the molecular pathogenesis of cancers arising in BRCA1 mutation carriers.
We found few cancers in germline BRCA2 mutation carriers to have p53 overexpression or mutation. This is in contrast to two previous studies using cancer-dense, predominantly European-based families and Icelandic women (15 , 16) . This discrepancy is interesting in light of the two distinct histological phenotypes reported for cancers arising in BRCA2 mutation carriers. High-grade carcinomas, without an excess of lobular carcinoma, have been described in Icelandic and European-based patients (2 , 30) , whereas we (1) and Marcus et al. (6) have described lower-grade cancers in BRCA2 mutation carriers, which frequently show a lobular pattern of growth. Thus it is possible that p53 mutations in cancers arising in BRCA2 germline mutation carriers confer a high-grade phenotype to these tumors. Unfortunately, there is no comprehensive follow-up data available on cancer patients with germline BRCA2 mutations, and, therefore, the effect of high-grade histology and p53 mutation cannot be assessed.
Cyclin D1 is a G1 cyclin, active in the regulation of G1-S cell cycle progression (31 , 32) . Cyclin D1 is overexpressed in several cancer types and acts as an oncogene in some breast cancers (9 , 33) . We found an unusually high rate of cyclin D1 overexpression over our entire sample of early-onset breast cancers (73%) compared with previous studies on breast cancers occurring at all ages (30–50%; 34 , 35 ). One possible explanation for this high rate may be high circulating estrogen levels within these patients with early-onset breast cancers because estrogen increases cyclin D1 protein synthesis in vivo (36) . Similar variance according to hormonal status is well described for other proteins affected by estrogen stimulation, notably the ER. However, cyclin D1 expression was not solely a consequence of physiological estrogen stimulation because cyclin D1 expression in the normal breast epithelium in individual cases was not identical to the staining properties of the invasive cancer.
Despite the overall high rate of cyclin D1-positive tumors in our sample, the invasive and accompanying in situ cancers arising in BRCA1 mutation carriers were rarely positive for cyclin D1 compared with the control cancers. However, there was variable staining in the normal breast epithelium. These findings suggest that cyclin D1 overexpression is rarely involved in the oncogenesis of breast cancers arising in germline BRCA1 mutation carriers. One previous study has examined cyclin D1 expression in breast cancers arising in BRCA1 or BRCA2 mutation carriers (13) . In this series no difference was seen between cyclin D1 expression in Jewish women harboring one of three founding mutations in the BRCA1 and BRCA2 genes compared with women from the same population without these mutations. However, we have found that the frequency of cyclin D1 expression in BRCA2 mutation carriers is intermediate between BRCA1 mutation carriers and control cancers, and, therefore, it is likely that combining BRCA1 and BRCA2 mutation carriers into one group, as in the above study, would mask any differences between BRCA1 mutation carriers and control cancers.
In our study, cancers arising in BRCA1 mutation carriers were commonly ER-negative, which contrasted with the majority of cancers in BRCA2 mutation carriers and control cancers, which were ER-positive. These data, which suggested reduced or absent ER pathway function in BRCA1-associated cancers, were confirmed by the absence of expression of the estrogen/ER-inducible proteins pS2 and PR in these cancers. A similar lack of ER positivity in BRCA1-associated cancers has been previously reported in studies involving Jewish patients and patients from cancer-dense families (10, 11, 12, 13) . Few studies have specifically addressed ER status in confirmed BRCA2 mutation carriers. However, high rates of ER positivity have been demonstrated in two studies in Jewish and Icelandic women with specific sites of germline mutation in BRCA2 (11 , 37) .
The infrequent c-erbB-2 staining in cancers of BRCA1 mutation carriers within our sample is interesting in light of the high histological grade of these cancers inasmuch as c-erbB-2 overexpression is an adverse prognostic indicator that has been previously associated with high-grade cancers (38) . This report supports the previously reported lack of c-erbB-2 expression in cancer-dense families (10) . Taken together, these data suggest that neither hormone receptor nor c-erbB-2 stimulation seem to be important in the pathogenesis of cancers arising in women with BRCA1 germline mutations.
Data defining the molecular and immunophenotypic profiles of breast cancers arising in BRCA1 and BRCA2 mutation carriers can be used in conjunction with other experimental approaches to elucidate the mechanisms of carcinogenesis induced by aberrant function of these genes. Both BRCA1 and BRCA2 are thought to be involved in the maintenance of genomic integrity because both proteins are known to associate with RAD51 (39 , 40) . There is also mounting evidence that BRCA1 is involved in regulation of cellular proliferation: Over-expression of wild-type BRCA1 leads to growth retardation of breast and ovarian cancer cells in vitro and tumor inhibition in nude mice (41) , and decreased expression of BRCA1 leads to increased cellular proliferation (42) . In addition, recent data suggest that BRCA1 transactivates the expression of the cyclin-dependent kinase inhibitor, p21 and inhibits cell cycle progression to the S phase (43) . BRCA1 is also known to bind p53 and stimulate p53-dependent transcriptional pathways, including the p53-dependent induction of p21 (44 , 45) . Importantly, we and others have previously shown that breast cancers arising in BRCA1 mutation carriers have an exceptionally high proliferation rate measured by their cellular mitotic count and, as described in this study, a high Ki-67 index (1, 2, 3, 4, 5) . Thus, it is reasonable to postulate that once early inactivation of the normal BRCA1 allele and p53 stabilization has occurred in the breast tissue of germline BRCA1 mutation carriers, there is little selection pressure for the recruitment of additional oncogenic mechanisms of cell growth. In contrast, the molecular pathogenesis of cancers arising in BRCA2 mutation carriers would indicate that additional somatic mutations are advantageous for tumor growth, perhaps implying a less direct effect of BRCA2 on control of cellular proliferation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This research was supported by The Victorian Breast Cancer Research Consortium, The Australian National Health and Medical Research Council, The Peter MacCallum Cancer Institute, The Hancock Family Foundation, and The Victorian Health Promotion Foundation.
↵2 To whom requests for reprints should be addressed, at Department of Pathology, Peter MacCallum Cancer Institute, St. Andrew’s Place, East Melbourne, Victoria 3002, Australia.
↵3 The abbreviations used are: ER, estrogen receptor; PR, progesterone receptor; ABCFS, Australian Breast Cancer Family Study; TE buffer, 10 mm Tris-HCl (pH 8.0) and 1 mm EDTA; PTT, protein truncation test; EGFR, epidermal growth factor receptor; LSAB, labeled streptavidin-biotin; LOH, loss of heterozygosity; DCIS, ductal carcinoma in situ; SSCP, single strand confirmation polymorphism.
- Received November 23, 1998.
- Accepted February 18, 1999.
- ©1999 American Association for Cancer Research.