Familial Breast/Ovarian Cancer and BRCA1/2 Genetic Screening: The Role of Immunohistochemistry as an Additional Method in the Selection of Patients

Abstract

Only 20–25% of families screened for BRCA1/2 mutations are found positive. Because only a positive result is informative, we studied the role of BRCA1/2 immunohistochemistry as an additional method for patient selection. From 53 high-risk-affected probands, 18 (34%) had available paraffin blocks of their tumors and were selected for this study. Mutation screening was done by conformation-sensitive gel electrophoresis and multiplex ligation-dependent probe amplification. For immunohistochemistry, 21 neoplastic specimens (15 breast carcinomas, 5 ovary neoplasms, and 1 rectal adenocarcinoma) were analyzed with BRCA1 (monoclonal antibody, Ab-1, oncogene) and BRCA2 (polyclonal antibody, Ab-2, oncogene) antibodies. Absence of the BRCA1 protein was confirmed in negative tumors by Western blotting. Seven patients were positive for BRCA1/2 mutations: 5 for BRCA1 and 2 for BRCA2. Four out of five positive patients had tumors negative for BRCA1 immunostaining, and the remaining 13 BRCA1-negative patients had positive BRCA1 immunostaining in all tumor samples. Sensitivity to predict for BRCA1 mutation carriers was 80%, and specificity was 100%, with a positive predictive value of 100% and a negative predictive value of 93%. This correlation was statistically significant (p=0.001). No correlation was observed for BRCA2. If larger studies confirm these results, high-risk patients with BRCA1-negative tumors should be screened first for this gene.

Breast cancer is hereditary in 10% of cases, the majority related to mutations in the BRCA1 and BRCA2 genes (Ford et al. 1998). BRCA1 and BRCA2 are tumor suppressor genes (Crook et al. 1997; Scully et al. 1997) that span over 100 kb of genomic DNA each and encode proteins of 1863 and 3418 amino acids, respectively. Mutations in these genes predispose carriers mainly to breast cancer but also to other cancers (Easton et al. 1995; Breast Cancer Linkage Consortium 1999; Liede et al. 2004).

BRCA1/2 mutation detection is complex because of the large size of both genes and the absence of hot spots. Besides positive and negative tests, indeterminate results of this screening pose particular problems in the management of high-risk families. Selection of patients is then crucial and relies mainly on family history and phenotype ascertainment (Frank et al. 1998; Parmigiani et al. 1998; American Society of Clinical Oncology 2003), but only 20–25% of screened families have positive mutation results (Shih et al. 2002). This is due to the disparity of selection criteria used in different breast evaluation clinics, to the different molecular methodologies available, and also because other genes may be involved in hereditary breast cancer (Walsh et al. 2006). Screening is also complicated by the fact that mutation detection in one family must start in an affected relative, and in many families, all affected relatives are deceased.

More specific selection for genetic screening, taking into account characteristics of BRCA1/2 tumors, is essential. Histopathology of BRCA1/2 hereditary breast cancer may be useful, because a basal epithelial phenotype appears to be associated with germline BRCA1 mutations (Foulkes et al. 2003) and other pathological features distinguish BRCA1 tumors from BRCA2 and sporadic breast cancers (Lakhani et al. 2002). Immunohistochemistry of BRCA1/2 proteins in tumor cells could be useful as an additional method in patient selection for genetic screening. If the cancer tissue from the genetic proband of a high-risk breast/ovarian cancer family (the first person to be tested is a cancer-affected individual) shows an absence of labeling for either BRCA1 or BRCA2 protein in tumor cells, that could predict a BRCA mutation carrier. Tumorigenesis in individuals with germline BRCA mutations requires somatic inactivation of the wild-type allele, and BRCA deficiency is critical to the development of disease (Welcsh and King 2001). With a reliable immunohistochemical method, in most cases, only the mutated gene would be screened. This methodology has recently been demonstrated to be useful in the identification of carriers of mutations in the mismatch repair genes associated with hereditary non-polyposis colorectal cancer (Stormorken et al. 2005).

Several antibodies for BRCA proteins are available, but conflicting results have been published in the literature concerning their diagnostic usefulness. Concerning BRCA1 protein, different types of staining have been described: nuclear, cytoplasmic, or both (Chen et al. 1995; Scully et al. 1996; Coene et al. 1997; Perez-Valles et al. 2001). Although these different observations could be explained by differences in the specificity of the antibodies (Perez-Valles et al. 2001), different fixation methods (Scully et al. 1996; Coene et al. 1997), or by the existence of splice variant isoforms of the BRCA1 protein (Wilson et al. 1997), the use of monoclonal antibodies and fixation in neutral buffered formalin after antigen exposure in a microwave demonstrates a predominantly nuclear labeling (Scully et al. 1996). This localization is consistent with the role of BRCA1 in the maintenance of genome integrity: cell cycle control, apoptosis, and DNA repair. BRCA2 is also found in the nucleus, as expected for a key component of cell cycle control and DNA repair pathways (Marmorstein et al. 1998), although truncated forms have been observed in the cytoplasm (Spain et al. 1999). Because nuclear entry for both BRCA1 and BRCA2 is dependent on nuclear localization signals (Spain et al. 1999; Henderson 2005), it is expected that mutations in these domains may affect subcellular localization of BRCA proteins (Henderson 2005).

The use of BRCA1 and BRCA2 immunohistochemistry in the prediction of BRCA mutation carriers has been more extensively studied for BRCA1, but with contradictory results. The absence of BRCA1 in 4% of 50 tissue sections was proposed to represent familial cases (Chen et al. 1995), and more recent studies have considered the use of BRCA1 staining to be a good indicator for the presence of mutations in this gene (Kashima et al. 2000; Schofield et al. 2000). However, the opposite has also been reported (Perez-Valles et al. 2001). In this study, we analyzed the correlation between the immunoexpression for BRCA1/2 and the results of genetic screening for BRCA1/2 mutations in a group of high-risk women with available tumor paraffin blocks. We also established the sensitivity, specificity, and predictive values of the method.

Materials and Methods

Case Selection

Between July 2000 and July 2002, we identified 18 women selected for BRCA1/2 mutation screening with available paraffin-embedded tissue blocks of their tumors. These women belonged to a group of 53 affected individuals at high risk for BRCA mutations based on personal and family history [combined probability of BRCA1/2 mutation over 25% (Frank et al. 1998; Parmigiani et al. 1998) or female breast cancer under 30 years of age] included in a research study for the analysis of these genetic events in the Portuguese population. All patients were counseled and consented to genetic screening, according to procedures approved by the ethics committee of our center. In one case, BRCA1 genetic screening results were correlated with immunochemistry data from an obligate carrier in the same family.

Clinicopathological Data

Information about age and other neoplasias whose specimens were not available for review was obtained from patients' medical records. A pathologist blinded to the results of BRCA1/2 mutation testing reviewed the cases. Histological grade of differentiation was estimated using the Elston and Ellis system (Elston and Ellis 1991).

BRCA1/2 Mutation Screening

DNA was extracted from whole blood using the QIAmp blood Midi Kit (Qiagen; Hilden, Germany) and aliquots of patients' DNA were each subjected to PCR amplification using primers designed to amplify whole exonic sequences and intronic/exonic boundaries of BRCA1 and BRCA2 (The Breast Cancer Information Core Database 1999; http://research.nhgri.nih.gov/bic/). The amplified products were analyzed by conformation-sensitive gel electrophoresis (CSGE) (Ganguly et al. 1993) and positive samples were sequenced using an automated fluorescence-based cycle sequencer (ABI Prism 310; Applied Biosystems, Foster City, CA). Mutations were classified according to current guidelines (den Dunnen and Antonarakis 2000).

Detection of BRCA1 Rearrangements

All samples negative for BRCA1/2 mutations by CSGE analysis were tested for the presence of large rearrangements in the BRCA1 gene with the use of the multiplex ligation-dependent probe amplification (MLPA) (MRC Holland; Amsterdam, The Netherlands), according to the instructions provided by the manufacturer. Separation and relative quantification of the amplification products were obtained with an ABI Prism 310 automatic sequencer (Applied Biosystems) using the Genescan software.

Immunohistochemistry

Formalin-fixed, paraffin-embedded blocks of the 21 specimens available were analyzed using a monoclonal antibody for the BRCA1 protein [anti-BRCA1 (Ab-1), human (mouse); epitope: 1–304 N-terminal amino acids of BRCA1] and a polyclonal antibody for the BRCA2 protein [anti-BRCA2 (Ab-2), rabbit polyclonal IgG; epitope: 3245–3418 of human BRCA2] (Oncogene Research Products/Calbiochem; Darmstadt, Germany) at a dilution of 1:300 and 1:100, respectively. Antigen retrieval was performed by heating the specimens in a pressure cooker for 6 min in citrate buffer (pH 6). The Envision system (Dako; Copenhagen, Denmark) was used for detection. Sections from positive breast invasive ductal carcinoma were used as positive controls, and negative controls were obtained by omitting the primary antibodies. The reaction was considered positive if more than 10% of the cells showed distinctive nuclear staining. Specific staining was evaluated by two investigators who had no knowledge of the patients' family histories or the results of mutation analysis.

Protein Extraction

Total protein extracts were obtained from formalin-fixed paraffin-embedded tissue sections. Three 50-μm-thick sections were deparaffinized, and exclusively cancer tissues were collected and further cut into small pieces. Proteins were extracted from tissue sections with 50 μl of radioimmunoprecipitation (RIPA) buffer, pH 7.6 [50 mM Tris, 5 mM EDTA, 1% Triton X-100, 0.4% sodium deoxycholate, 150 mM NaCl, and protease inhibitors (Complete Mini, EDTA-free; Roche, Mannheim, Germany)], followed by incubation at 100C for 20 min and incubation at 60C for 2 hr. Next, the tissue lysates were centrifuged at 14,000 × g for 20 min at 4C, and the supernatants were collected (Ikeda et al. 1998). Protein concentration of the lysates was determined using the Bio-Rad protein assay kit (Bio-Rad Laboratories; Munich, Germany). This reaction is similar to the well-documented Bradford assay (Bradford 1976).

Immunoprecipitation and Western Blot Analysis

Total protein extracts were immunoprecipitated with the Ab-1 monoclonal antibody. Samples were precleared with protein G-Sepharose beads (Sigma-Aldrich Chemie; Steinheim, Germany) for 1 hr at 4C in RIPA buffer, and supernatants were recovered and incubated overnight at 4C with the antibody at 2 μg/ml. The binding reactions were incubated for 1 hr 30 min at 4C with protein G-Sepharose beads. The immunoprecipitated proteins were obtained after three washes in RIPA buffer and boiling at 100C for 5 min in sample loading buffer (2-fold concentrate: 125 mM Tris, 20% glycerol, 4% SDS, 2% 2-mercaptoethanol, and 0.001% bromophenol blue). Proteins were analyzed by 8% SDS-PAGE gels and transferred onto nitrocellulose membranes (Hybond-C extra; Amersham Life Sciences, Buckinghamshire, UK). After incubation with the primary antibody (Ab-1) overnight at 2 μg/ml, the membranes were incubated for 2 hr with a goat anti-mouse IgG-horseradish peroxidase (Santa Cruz Biotechnology; Santa Cruz, CA) at a dilution of 1:5000. An enhanced chemiluminescence detection system and Kodak film (Amersham Pharmacia Biotech; Buckinghamshire, UK) were used to visualize the presence of proteins on the nitrocellulose blots.

Statistical Analysis

Sensitivity, specificity, and positive and negative predictive values were calculated for the results of immunohistochemistry to predict for BRCA1/2 mutations. The association between negative immunohistochemistry and mutation carrier status was analyzed using the two-tailed Fisher's exact test.

Results

Clinicopathological Data

Nine out of 18 women had been diagnosed with unilateral breast cancer, 2 with bilateral breast cancer, 3 with breast and ovarian cancer, 2 with ovarian cancer, 1 with breast and rectal cancer, and 1 with breast and lung cancer. Twenty-one specimens were available: 15 with breast cancer tissue, 5 with ovarian cancer, and 1 with rectal cancer. Two of the five ovarian cancer specimens were of surgically excised metastasis (epiploon and hepatic metastasis). The other three specimens included the primary tumor. Clinicopathological characteristics are described in Table 1.

Table 1.

Clinopathological data

Case	Age (years)	Tumor specimen	Histological diagnosis	Histological grade	Pathological stage	Estrogen receptors
1	28	Breast	Invasive ductal carcinoma with intraductal component (<10%)	G2	IIB	+
2	69	Breast	Invasive ductal carcinoma, NOS	G2	IIA	+
3	28	Breast	Invasive ductal carcinoma, NOS	G1	IIA	+
4	46	Breast	Invasive ductal carcinoma, NOS	G2	IIB	−
5	35	Breast	(1) Invasive ductal carcinoma with predominant intraductal component	(1) G2	(1) IIB	+
			(2) Invasive ductal carcinoma, NOS	(2) G2	(2) IIB	+
6	41	Ovary	Serous adenocarcinoma ∗	G2	IIIB	NA
7	43	Breast	Invasive ductal carcinoma, NOS	G3	IIA	+
8	49	Breast	Osseous metastasis of invasive ductal carcinoma	G3	IV	+
9	52	Breast	Invasive ductal carcinoma	G3	I	−
10	42	Breast	Intraductal carcinoma	G2	0	+
11	58	Ovary	Serous (60%) and endometrioid (40%) carcinoma	G3	IB	NA
12	34	Breast	Invasive ductal carcinoma with intraductal component (<20%)	G2	IIB	+
13	67	Breast	Invasive ductal carcinoma with predominant intraductal component	G3	I	+
14	37	(1) Breast	(1) Invasive ductal carcinoma	(1) G3	(1) IIB	(1)−
		(2) Rectum	(2) Adenocarcinoma	(2) G2	(2) IIIB	(2) NA
15	32	Breast	Invasive ductal carcinoma	G3	IIA	+
16	33	Breast	Invasive ductal carcinoma with predominant intraductal component	G2	I	+
17	62	Ovary	Liver metastasis of ovary serous carcinoma	G3	IV	NA
18	55	Ovary	Clear cell carcinoma	G3	IIIC	NA

^∗ Two samples reviewed, one of a surgically excised metastasis and the other including the primitive tumor.

Only specimens available for review are shown. In case 3, the patient had contralateral breast cancer at age 35, and in case 5 a first diagnosis of invasive lobular carcinoma of the other breast had been made at age 33; Cases 9, 17, and 18 had both breast and ovarian cancer, and case 13 was treated for lung cancer before her breast cancer diagnosis. NA, not applicable; NOS, not otherwise specified.

Genetic Screening

Of the 18 women with available paraffin-embedded tumor blocks, 7 were found to be positive for BRCA1/2 mutations: 5 with BRCA1 mutations (c.536delA, g.Ex13ins6Kb, c.211A>G, and g.Ex11_Ex15del) and 2 with BRCA2 mutations (c.1369_1370ins2 and c.7208_ 7211del4). The remaining 11 women were negative for BRCA1/2 mutations, by CSGE and MLPA screening of BRCA1 rearrangements (Table 2).

Table 2.

Correlation between mutation testing and immunohistochemistry

Case	BRCA1/2 mutation (designation in BIC database)	Codon	Exon	Coding effect	BRCA1 nuclear immunostaining	BRCA2 nuclear immunostaining
1	Negative	−	−	−	+	+
2	Negative	−	−	−	+	+
3	c.536delA	179	8/1	STOP 233	−	−
4	Negative	−	−	−	+	−
5	c.1369_1370ins2	457	10/2	STOP 460	+	+ ∗
6	g.Ex13ins6Kb (exon13ins6Kb)	NA	13/1	STOP 1460	−	−
7	c.7208_7211del4 (7436del4)	2403	14/2	STOP 2467	+	−
8	Negative	−	−	−	+	−
9	Negative	−	−	−	+	−
10	Negative	−	−	−	+	+
11	g.Ex11_Ex15del	NA	11–15/1	ID	+	−
12	Negative	−	−	−	+	−
13	Negative	−	−	−	+	−
14	c.211A>G (R71G)	71	5/1	STOP 64	−	−
15	Negative	−	−	−	+	−
16	Negative	−	−	−	+	+
17	Negative	−	−	−	+	+
18	c.211A>G (R71G)	71	5/1	STOP 64	−	+

^∗ This patient relapsed in the breast 3 years latter and the specimen of invasive ductal carcinoma was negative for BRCA2 immunohistochemistry. BIC, Breast Cancer Information Core; ID, indeterminate; NA, not applicable.

Immunohistochemistry

Tumor samples from four of five women with BRCA1 mutations were BRCA1 negative, with absence of nuclear or cytoplasmic staining (Figure 1B). These cases included two breast cancer specimens (from cases 3 and 14), three ovarian cancer specimens (two from case 6 and one from case 18), and one rectal cancer specimen (case 14). In contrast, the other patient with a BRCA1 mutation (g.Ex11_Ex15del) had her ovarian cancer specimen stain positive for BRCA1 immunohistochemistry, with a clear nuclear immunoreactivity in tumor cells. All other tumor samples, from the 2 BRCA2-positive women and from the 11 patients negative for mutations in both genes, were positive for BRCA1 immunostaining (Table 2).

Figure 1.

BRCA1 immunohistochemistry in cancer specimens. (A) Breast cancer specimen with clear nuclear BRCA1 labeling in cancer (bcc), epithelial (ebc), and stromal (sc) cells contrasts with an ovarian cancer sample (B), where cancer cells (occ) are negative for BRCA1, even if inflammatory (ic), and stromal cells (sc) keep the expected nuclear labeling.

BRCA2 staining was also optimized and nuclear staining observed in tumor samples. Cytoplasmic staining was not observed. One BRCA2-positive woman had her invasive breast cancer sample stain negative for BRCA2, whereas the opposite was observed in a sample of invasive ductal carcinoma with large areas of intraductal carcinoma of the other BRCA2 mutation carrier. In this last case, a relapse in the same location of the right breast was observed 3 years later, and the corresponding tumor sample, of invasive ductal carcinoma, was negative for BRCA2 staining. Most of the tumor samples from the 18 women were negative for BRCA2 labeling (Table 2).

Protein Analysis

For BRCA1, the absence of the 220-kDa BRCA1 protein in negative immunohistochemistry tumor samples was further confirmed with total protein extracts from paraffin tumor specimens, by Western blot. Normal breast tissue was used as control, and adequate total protein lysates were obtained from seven of our tissue-embeded samples. Total proteins were concentrated by immunoprecipitation, and in the experiment shown in Figure 2, the band corresponding to the BRCA1 protein was observed by Western blotting. BRCA1 protein was observed in normal breast tissue (positive control) and in a BRCA1-negative mutation case with positive stained BRCA1 specimens (case 1) but not in two BRCA1 mutation—positive cases with negative BRCA1 immunostaining (cases 6 and 14) and in the negative control (protein G-Sepharose beads with Ab-1 monoclonal antibody).

Figure 2.

Absence of BRCA1 protein in negative breast cancer specimens. After Western blotting of tumor protein extracts with the Ab-1 antibody, BRCA1 protein was identified in normal breast, positive control (1) and in breast cancer from a BRCA1 mutation-negative woman but with positive Ab-1 immunostaining (2). In BRCA1 mutation-positive cases with negative immunostaining, the 220-kDa band corresponding to the BRCA1 protein was not observed (3 and 4). Negative control, protein G-Sepharose beads with Ab-1 monoclonal antibody (5).

Correlation Between Genetic Screening and Immunohistochemistry

Sensitivity of immunohistochemistry to detect BRCA1 mutation carriers was 80% and specificity 100%. Negativity for BRCA1 labeling in a tumor paraffin specimen was significantly correlated with BRCA1 carrier status (p=0.001); the positive predictive value of this test was 100%, and the negative predictive value 93%. For BRCA2, sensitivity of immunochemistry was 50% and specificity 38%. The results of immunohistochemistry for BRCA2 were not significantly correlated with BRCA2 carrier status (p=1).

Discussion

In this study, we observed that immunostaining with a monoclonal antibody against the N-terminal amino acids of the BRCA1 protein has a high specificity for the prediction of BRCA1 mutation carriers. These results suggest that BRCA1 immunohistochemistry, a rapid and easy test, can be used before the expensive mutation screening, to select which high-risk cases should be submitted to analysis of this gene; when BRCA1 staining shows integrity of the protein, BRCA2 screening should be done first. This methodology is helpful in case selection for subsequent mutation analysis and it can also be the only method to demonstrate inherited breast or ovarian cancer in deceased individuals belonging to high-risk breast/ovarian cancer families. However, the inclusion of BRCA1 immunohistochemistry in algorithms concerning patient selection for BRCA1/2 mutation screening is not recommended at this time, because the results of our study need confirmation in a larger, preferentially prospective sample. Confirmations of familiar cancers are not easy to obtain in breast cancer risk evaluation clinics, and cancer specimens are even more difficult to obtain. Collaborative studies are needed to overcome these difficulties.

Specificity of BRCA1 immunohistochemistry to predict for BRCA1 mutations was 100% and sensitivity 80%. In only one case was BRCA1 immunohistochemistry not concordant with the final mutation results: a woman negative for BRCA1/2 screening by CSGE was found to test positive for a large BRCA1 deletion after MLPA screening. This rearrangement included the deletion of exons 11–15 of BRCA1, and immunohistochemistry of this patient's ovarian tumor sample showed positive labeling of the BRCA1 protein. The most reasonable explanation for this observation is that besides this large deletion, the N-terminal portion of the BRCA1 protein is transcribed and is accessible to the Ab-1 antibody. A previous study had already shown that BRCA1 exon 11 mutations may not affect the immunostaining of both C-terminal and N-terminal antibodies (Kashima et al. 2000) (Table 3). In the present study, most of the BRCA1 mutations occurred downstream of exon 11; the only two exceptions were the large deletion encompassing exons 11–15 and Ex13ins6Kb, another large rearrangement occurring upstream of exon 11. In this last case, complete negativity of BRCA1 was observed, as in all other cases with mutations downstream of exon 11. These observations reinforce the notion that BRCA1 immunohistochemistry can be used as a prescreen for mutation testing but, because sensitivity is 80%, it cannot replace mutation testing. However, its inclusion in a decision process also considering phenotypic and pathological factors can rationalize the approach and the cost of BRCA1/2 mutations.

Table 3.

Prediction of BRCA1 carrier status with immunohistochemistry for the BRCA1 protein

In cases negative for immunostaining, this finding was striking in that almost all tumor cells were negative. Although we indicated a cutoff at less than 10% of labeled tumor cells to consider the labeling negative, in only one of the tumors analyzed (one of the ovarian cancer specimens) very few cells (much less than 10%) were scarcely positive for BRCA1. This is in contrast with the results of another study in which only “islands” of negativity for BRCA1 immunohistochemistry were observed in the presence of BRCA1 mutations (Schofield et al. 2000) (Table 3). The results we present here, however, were confirmed by Western blotting: negative samples after BRCA1 immunohistochemistry were negative for the detection of the BRCA1 protein, and the opposite was observed for positive samples. No other study had previously done this type of assay, probably because of the technical difficulties in obtaining protein lysates from paraffin-embedded specimens. A recent study whose aim was to assess the specificity and sensitivity of immunohistochemistry as a screening method for demonstrating BRCA1 expression concluded that Ab-1 was the only BRCA1 antibody whose staining correlated significantly with loss of BRCA1 expression as determined by RT-PCR (Al-Mulla et al. 2005). Table 3 summarizes prior studies attempting to analyze the sensitivity and specificity of several BRCA1 antibodies and the prediction of BRCA1 mutation carrier status with immunohistochemistry.

Small sample size, one of the limitations of our study, is due to the difficulty in assembling tissue specimens from all patients tested for BRCA1/2 mutations. In spite of this, one of the strengths of our assay is that BRCA1 results were concordant not only in tumors of different origins (breast, ovarian, and one case of rectal cancer) but also in primitive and metastatic lesions. Rectal cancer has been an interesting although complicated issue since it was first associated with BRCA1 (Brose et al. 2002; Thompson and Easton 2002), but more recent studies did not confirm that association (Niell et al. 2004). BRCA1 loss is expected in a subgroup of colorectal cancers (Grabsch et al. 2006), and it may be that our specimen was negative because it was a sporadic rectal cancer in a confirmed BRCA1 carrier or because the specific splicing BRCA1 mutation that this patient inherited is truly associated with colorectal and other digestive tumors (Vega et al. 2001).

Exclusive nuclear staining could be considered to represent the normal phenotype and was observed for both BRCA1 and BRCA2 antibodies in all tumor specimens analyzed in our study. Some studies (Chen et al. 1995; Coene et al. 1997; Perez-Valles et al. 2001; Al-Mulla et al. 2005) also reported a cytoplasmatic staining. Several reasons may account for the observation of BRCA1/2 in the cytoplasm: cross-reaction between the target epitopes; the presence of BRCA1 splice variants (Wilson et al. 1997); BRCA2 truncated forms (Spain et al. 1999), or loss of the nuclear localization signals of the BRCA proteins. In fact, the ability to enter the nucleus may be compromised by germline mutations in the BRCT (Rodriguez et al. 2004) domain of BRCA1 and in the carboxyl terminus of BRCA2 (Marmorstein et al. 1998).

Prediction of BRCA2 mutation carriers remains a problem. We could not find a correlation with immunohistochemistry in our study, and several studies show that pathological or microarray tumor analysis of BRCA1 and BRCA2 tumors is more distinctive for BRCA1 than for BRCA2: BRCA1 tumors tend to be of higher grade and are also more frequently negative for hormone receptors and more p53-positive than are their BRCA2 counterparts (Lakhani et al. 2002); type D cyclins and their associated CDK4 and CDKIs (p16, p21, and p27) were found to be downregulated in BRCA1 in respect to BRCA2 carcinomas. Also, over 80% of triple-negative breast tumors (negative for estrogen, progestagen, and HER2 receptors) are “basal-like,” and basal markers have been shown to be specific to a subset of BRCA1 carcinomas (Palacios et al. 2004). The apparent relative excess of BRCA1 germinal mutations in association with triple negative tumors is under study (Kandel et al. unpublished data). The characterization of BRCA2 breast tumors and their distinction from sporadic breast cancers is an area of active research, and one recent study proposed that BRCA2 tumors could be identified through the use of an array of markers that included CHEK2 and RAD51: CHEK2 was found to be more frequently expressed in BRCA1 and BRCA2 tumors than in non-BRCA1/2, and BRCA2 was found to be necessary for the translocation of RAD51 (Honrado et al. 2005).

In conclusion, we observed a high specificity for the prediction of BRCA1 carriers with immunohistochemistry using a monoclonal BRCA1 antibody. Validation of this assay, using a larger sample, will allow the use of immunohistochemistry for deciding which high-risk patients should be screened first for the BRCA1 gene. This recommendation does not exclude the relevance of other known risk factors for mutations in this gene (family history, age at cancer diagnosis, histological characteristics of the tumors, triple negativity) but is intended to contribute to a more specific patient selection. Because only positive results are informative for probands of these families, better patient selection is likely to increase the possibilities of obtaining informative genetic results.