HIF-1α Overexpression in Ductal Carcinoma In Situ of the Breast in BRCA1 and BRCA2 Mutation Carriers

Recent studies have revealed that BRCA1 and BRCA2 germline mutation-related breast cancers show frequent overexpression of hypoxia inducible factor-1α (HIF-1α), the key regulator of the hypoxia response. However, the question remained whether hypoxia is a late stage bystander or a true carcinogenetic event in patients with hereditary predisposition. We therefore studied HIF-1α overexpression in ductal carcinoma in situ (DCIS), an established precursor of invasive breast cancer. We used immunohistochemistry to examine the expression of the hypoxia markers HIF-1α, CAIX and Glut-1 in DCIS and available invasive carcinoma lesions of 32 BRCA1, 16 BRCA2 and 77 non-BRCA mutation-related cases. HIF-1α expression was detected in 63% of BRCA1 and 62% of BRCA2 as compared to 34% of non-BRCA mutation-related DCIS cases (p = 0.005). CAIX overexpression was present in 56% of BRCA1 and 44% of BRCA2 as compared to 6% of non-BRCA mutation-related DCIS cases (p = 0.000). Glut-1 overexpression was observed in 59% of BRCA1, 75% of BRCA2 and 67% of non-BRCA mutation-related DCIS cases (p = 0.527). Overall, HIF-1α, CAIX and Glut-1 expression in BRCA mutation-related DCIS matched the expression in the accompanying invasive cancers in 60% or more of cases. In non-BRCA mutation-related cases the expression of the hypoxia markers in DCIS matched the expression in the invasive part in 46% or more of the cases. Although BRCA1 and BRCA2 germline mutation-related invasive breast cancers are different in many ways, the hypoxia-related proteins HIF-1α, CAIX and Glut-1 are expressed in both DCIS and invasive lesions of BRCA1 and BRCA2 mutation carriers. This suggests that hypoxia may already play a role in the DCIS stage of BRCA1 and BRCA2 germline mutation related breast carcinogenesis, and may also drive cancer progression. Hypoxia-related proteins are therefore putative targets for therapy and molecular imaging for early detection and monitoring therapy response in BRCA mutation patients.


Introduction
Hereditary breast cancer accounts for about 5% of all breast cancers in women and is primarily caused by a germline mutation in one of the BRCA genes. Several studies have indicated that the genetic makeup of BRCA1 and BRCA2 mutation-related breast cancer is different from that of non-BRCA mutation-related breast cancer. These differences comprise gains and losses of specific parts of chromosomes, as well as differences in protein expression [1][2][3][4][5][6][7]. Consistent with this, the morphological and immunohistochemical phenotype of BRCA1 mutation-related breast cancer is also different from that of non-BRCA mutation-related breast [8][9][10][11][12][13]. However, the phenotype of BRCA2 mutation-related breast cancer is still difficult to distinguish from non-BRCA mutationrelated breast cancers [14,15].
However, studies in pre-invasive lesions are required to address the question whether hypoxia is a late stage bystander or a true carcinogenetic event.
There is both clinical and experimental evidence to suggest that ductal carcinoma in situ (DCIS) is a precursor lesion to most, if not all, non-BRCA mutation-related invasive breast cancers [34][35][36][37][38]. DCIS and other premalignant lesions such as lobular neoplasia, fibroadenoma, and ductal hyperplasia seems to be more common in prophylactic mastectomy (PM) specimens of BRCA1 and BRCA2 mutation carriers than in control mammoplasty specimens [10,[39][40][41][42]. Furthermore, DCIS lesions adjacent to invasive cancers in BRCA mutation carriers have been described [43,44]. DCIS in BRCA mutation carriers is often high grade [43] and shows a similar morphology and immunophenotype as the accompanying invasive cancer [45]. High grade DCIS of non-BRCA-related cases often shows central necrosis [46] indicative of hypoxia. Indeed, overexpression of hypoxia-related proteins HIF-1a, CAIX and Glut-1 DCIS of non-BRCA mutation carriers has been described [22]. To find clues whether changes in hypoxia related proteins also is an early event in BRCA mutation-related carcinogenesis, we evaluated HIF-1a expression in BRCA1 and BRCA2 mutationrelated DCIS in relation with the accompanying invasive cancers.

Patients
The study group comprised DCIS lesions of 32 patients with pathogenic germline BRCA1 mutations, 16 patients with pathogenic germline BRCA2 mutations and 77 patients unselected for family history (further denoted ''non-BRCA mutation-related''). A synchronous invasive tumor was also present in 28 BRCA1, 17 BRCA2 and 50 non-BRCA mutation-related cases. Tissue from these patients was available from our own archives, and from different pathology laboratories in The Netherlands (St Antonius Hospital Nieuwegein, Diakonessenhuis Utrecht, Gelre Ziekenhuizen Apeldoorn, Rijnstate Arnhem, Stichting Pathologisch en Cytologisch laboratorium West Brabant Bergen op Zoom, Ziekenhuis Gelderse Vallei Ede, Deventer Ziekenhuis Deventer, Meander medisch centrum Amersfoort, Onze Lieve Vrouwe Gasthuis Amsterdam, the VU University Medical Center, Amsterdam and the University Medical Center Groningen). Since we used archival pathology material which does not interfere with patient care and does not involve the physical involvement of the patient, no ethical approval is required according to Dutch legislation [the Medical Research Involving Human Subjects Act (Wet medisch-wetenschappelijk onderzoek met mensen, WMO [47])]. Use of anonymous or coded left over material for scientific purposes is part of the standard treatment contract with patients and therefore informed consent procedure was not required according to our institutional medical ethical review board. This has also been described by van Diest et al. [48].

Histopathology
Tumor size was measured in the fresh resection specimens, and tumor samples were subsequently fixed in neutral buffered formaldehyde, and processed to paraffin blocks according to standard procedures. Four mm thick sections were cut and stained with H&E for histopathology. Tumor type was assessed according to the WHO 2003, and tumors were graded according to the Nottingham grading system. Mitoses counting was performed as previously described [49]. Scoring was performed by one observer (PJvD) who was blinded to the origin of the tumors.

Immunohistochemistry
After deparaffinization and rehydration, antigen retrieval was performed using EDTA buffer at boiling temperature for 20 minutes for ER, HER2 and HIF-1a. A cooling period of 30 minutes preceded the incubation of the slides for HIF-1a with protein block (Novolink Max Polymer detection system, ready to use, Novocastra Laboratories Ltd, Newcastle Upon Tyne, UK) for 5 minutes at room temperature. Incubation of the slides with the HIF-1a mouse monoclonal (BD Biosciences, Pharmingen, Lexington, MA, USA), was done at a dilution of 1:50 overnight at 4uC. For detection, a polymer (Novolink Max Polymer detection system, ready to use) was used. For ER and HER2, the slides were incubated with primary antibodies for ER (1:100, Dako) and HER2 (1:100, Neomarkers) 60 minutes at room temperature.
For PR, Glut-1 and CAIX, antigen retrieval was performed in citrate buffer, pH = 6.0, for 20 minutes at 100uC. A cooling period of 30 minutes preceded the incubation (60 minutes at room temperature) with the primary antibodies. Polyclonal primary antibodies used were: PR (1:100, Dako), Glut-1 (1:200, DAKO) and CAIX (1:1000, Abcam, Cambridge Science Park, Cambridge, UK). For detection of the primary antibodies against ER, PR, HER2, CAIX and Glut-1, a poly HRP anti-Mouse/Rabbit/Rat IgG (ready to use, ImmunoLogic, Duiven, Netherlands) was used. All slides were developed with diaminobenzidine (10 minutes) followed by hematoxylin counterstaining. Before the slides were mounted all sections were dehydrated in alcohol and xylene. Positive controls were used throughout, negative controls were obtained by omission of the primary antibodies from the staining procedure. Representative pictures of positive and negative controls for HIF-1a, CAIX and Glut-1 have been provided as Figure S1.
Scoring of immunohistochemistry was performed by one observer (PJvD). HIF-1a was regarded overexpressed when .1% of nuclei were positive as described before [26]. ER and PR expression was regarded positive when 10% or more of the tumor nuclei stained positive. HER2 was scored positive when a 3+ membrane staining was observed according to the Dako system. CAIX and Glut-1 stainings were scored positive when a clear membrane staining pattern was seen. Associations between stainings were tested by Chi-square analysis. P-values,0.05 were considered to be statistically significant.

Results
The clinicopathological characteristics and expression of ER, PR, HER2, HIF-1a, CAIX and Glut-1 of BRCA1, BRCA2 and non-BRCA mutation-related DCIS cases are described in Table 1. The age of onset is lower in BRCA compared to non-BRCA mutation carriers (p = 0.000). BRCA1 mutation-related DCIS cases often are ER, PR and HER2-negative as compared to the BRCA2 and non-BRCA mutation-related DCIS (see Table 1 for correlations).
Furthermore, in the BRCA1 and BRCA2 mutation-related DCIS no correlations between HIF-1a expression and grade, ER, PR and HER2 expression were found. For the non-BRCA mutationrelated DCIS cases, a positive trend was observed with grade, and a negative trend with ER ( Table 2).
In summary, these non-significant differences indicate that HIF-1a positivity was similar in DCIS and the accompanying invasive lesions. Differences in HIF-1a expression between BRCA1 and BRCA2 and non-BRCA mutation related DCIS were borderline significant (p = 0.062). A significant difference in HIF-1a expression was seen between BRCA1 and BRCA2 as compared to non-BRCA mutation-related invasive cancer (p = 0.000). Table 4 shows the expression of HIF-1a, CAIX and Glut-1 in paired, DCIS and concomitant invasive cancer, for BRCA mutation and non-BRCA mutation carriers.
In the BRCA2 mutation-related cases with DCIS and concomitant invasive cancer, 38% (6/16) of the cases HIF-1a expression was observed and was absent in 38% (6/16) of the cases (Table 4). Thus, in 75% (12/16) of the BRCA2 mutation-related cases, the DCIS and invasive lesions of the same patient showed similar expression levels of HIF-1a. Expression of HIF-1a in only the DCIS lesion was seen in 25% (4/16) of the BRCA2 mutationrelated cases. CAIX was expressed in both lesions in 31% (5/16) of BRCA2 mutation-related cases and in 44% (7/16) of the cases both lesions lacked expression (total match 75%). CAIX was expressed in the invasive, but not in the DCIS part in 13% (2/16) of the cases, and CAIX was expressed in the DCIS, but not in the invasive part of 13% (2/16) of the cases. Glut-1 was expressed or absent in both lesions in 50% (8/16) and 19% (3/16) of cases, respectively (total match 69%). Further, Glut-1 expression was confined to the invasive part in 6% (1/16) of cases and the DCIS part in 25% (4/16) of the cases.
When BRCA1 and BRCA2 mutation-related cases were examined together, HIF-1a expression in DCIS matched the expression in the accompanying invasive cancers in 68% (31/45) of cases, as compared to in 68% (34/50) of the non-BRCA mutation carrier cases. The expression of CAIX matched in 64% (29/45) of BRCA1 and BRCA2 mutation-related cases, as compared to in 88% (44/ 50) of non-BRCA mutation carrier cases. For Glut-1, the expression in DCIS matched the expression in the accompanying invasive cancers in 60% (27/45) of BRCA1 and BRCA2 mutationrelated cases as compared to 46% (23/50) for non-BRCA mutation carrier cases.

Discussion
Non-BRCA mutation-related DCIS lesions, especially high grade ones, are known to become centrally deprived of oxygen resulting in activation of the hypoxia pathway, as shown in several studies by the presence of HIF-1a and its downstream targets. The aim of the present study was to examine the expression of HIF-1a in DCIS lesions of BRCA1 and BRCA2 mutation carriers in comparison with their invasive counterparts. Activation of HIF-1a in the DCIS stage of BRCA1 or BRCA2 germline mutated patients would indicate that hypoxia is an early driver of BRCA mutationrelated carcinogenesis. HIF-1a overexpression was indeed frequently observed in BRCA1 and BRCA2 mutation-related DCIS cases, in association with expression of its downstream genes, indicating that HIF-1a is active.
Overall, 63% (30/48) of BRCA mutation-related DCIS lesions were HIF-1a-positive, which was significantly different compared to non-BRCA mutation carriers (34%, 26/77). The latter figure is  lower compared to our earlier observations where 67% of sporadic DCIS lesions were HIF-1a positive [22]. Nevertheless, the current study suggests that hypoxia and HIF-1a already play a similar role in the DCIS stage of BRCA mutation-related carcinogenesis as in non-BRCA mutation-related DCIS. BRCA mutation-related invasive cancers (especially BRCA1 mutation-related ones) more frequently show HIF-1a overexpression than non-BRCA mutation-related ones [33,34]. This suggests that hypoxia plays a more important role in cancer progression in BRCA mutation carriers than in non-BRCA mutation carriers. HIF-1a, CAIX and Glut-1 expression in BRCA mutation-related DCIS was usually similar in the accompanying invasive lesions. This implies that next to being involved in early BRCA mutationrelated carcinogenesis, hypoxia and HIF-1a overexpression may also be a driver of cancer progression, especially in BRCA1 mutation carriers. Although the number of BRCA2 mutationrelated cases with DCIS and invasive lesions was small, there was a trend towards higher expression of the hypoxia-related markers in BRCA2 mutation-related DCIS as compared to the invasive lesions. We can speculate that progression to the invasive state in these BRCA2 mutation carriers might be due to the switch of the HIF-1a to HIF-2a expression under prolonged hypoxia [50]. HIF-2a expression has been observed in sporadic breast cancer [51] and should be analysed in BRCA mutation-related breast cancer and pre-invasive lesions. As HIF-1a already plays a role in the preinvasive lesions of BRCA mutation carriers, hypoxia proteins would therefore be putative therapeutic targets for prevention of invasive disease. HIF-1a signalling inhibitors like PX-478 [52], farnesyltransferase inhibitor R115777 or trans-farnesylthiosalicyclic acid [53,54], Cetuximab [55] and other antibodies with the same structural motif [56], 2-methoxyestradiol (2ME2) [57,58], and inhibitors of the RNA binding protein Hur [59,60] are some of the therapeutics currently available.
We conclude that BRCA1 and BRCA2 germline mutationrelated DCIS show a high frequency of overexpression of HIF-1a and its downstream proteins CAIX and Glut-1, as compared to non-BRCA mutation-related DCIS. This suggests that hypoxia may already play a role at the DCIS stage of BRCA1 and BRCA2 germline mutation-related breast carcinogenesis, and may also drive cancer progression. The current findings could be clinically relevant for BRCA mutation-related breast cancer treatment in several ways. First, HIF-1a and its downstream effectors may be used as molecular imaging targets for early detection and monitoring of therapy response. Second, HIF-1a is an interesting therapeutic target at the pre-invasive stage of BRCA mutationrelated breast disease to prevent invasive disease. Figure S1 Positive controls: Immunohistochemical staining of HIF-1a and CAIX in renal clear cell carcinoma (B and D) and for Glut-1 in placental tissue (F). In A, C and E the primary antibody was omitted to provide negative controls. (TIF)