BC200 RNA, a small functional RNA that operates as a translational modulator, has been implicated in the regulation of local synaptodendritic protein synthesis in neurons. Cell type-specific expression of BC200 RNA is tightly controlled such that the RNA is not normally detected in somatic cells other than neurons. However, the neuron-specific control of BC200 expression is deregulated in a number of tumors. We here report that BC200 RNA is expressed at high levels in invasive carcinomas of the breast. In normal breast tissue or in benign tumors such as fibroadenomas, in contrast, we found that the RNA is not detectable at significant levels. The difference in expression levels between invasive carcinomas and normal/benign tissue was statistically highly significant. Receiver Operating Characteristics analysis of sensitivity and specificity confirmed the diagnostic power of BC200 RNA as a molecular marker of invasive breast cancer. In ductal carcinomas in situ , furthermore, significant BC200 expression was associated with high nuclear grade, suggesting that the presence of BC200 RNA in such tumors may be used as a prognostic indicator of tumor progression. The combined results demonstrate the potential of BC200 expression to serve as a molecular tool in the diagnosis and/or prognosis of breast cancer.
Key to enhancing survival chances in breast cancer is early detection. However, no reliable molecular marker is currently available that could be used to complement mammography in the detection of breast cancer. This situation contrasts with prostate cancer, for example, where prostate-specific antigen (PSA) status can be established by simply analyzing a blood sample. PSA is a molecular marker for prostate cancer (reviewed in 1) and although PSA status is not a very reliable tumor indicator, due to low sensitivity and specificity, it is routinely used in the clinical diagnosis of prostate malignancies. No such marker is currently available for breast cancer.
We have previously reported that the normally neuron-specific transcript BC200 RNA ( 2 ) is atypically expressed in various tumors, including mammary carcinomas ( 3 ). BC200 RNA is a small untranslated RNA polymerase III transcript that is expressed in neurons of the human nervous system ( 2 ). BC200 RNA, and its rodent counterpart BC1 RNA ( 4 ), do not contain open reading frames to encode protein amino acid sequences. Collectively called BC RNAs, BC1 RNA and BC200 RNA are thus members of the growing family of untranslated functional RNAs that have been identified to date (reviewed in 5). BC RNAs have been suggested to operate as modulators of translation in the activity-dependent synthesis of new proteins at the synapse ( 6 ). This hypothesis is now supported by several lines of evidence.
Subject to activity-dependent regulation ( 7 ), BC RNAs are specifically transported to dendrites where they are located in synaptodendritic domains ( 4 , 8 , 9 ). BC1 RNA and BC200 RNA have recently been shown to be functionally active RNAs that repress translation by interacting with components of the translational machinery ( 10 ). The combined results therefore identify BC RNAs as modulators of protein synthesis that operate in postsynaptic dendritic microdomains to regulate on-site translation of local mRNAs.
Translational control, a key element in the modulation of gene expression, is frequently deregulated in cancer cells (reviewed in 11 ). We have previously reported that expression of BC RNAs is under tight tissue-specific control. Thus, BC RNAs are exclusively expressed in nerve cells and germ cells and not in glial cells or somatic cells in non-neural organs such as kidney, liver, lung and others ( 2 , 4 , 12 , 13 ). As a major exception to this rule, however, BC RNAs were found to be expressed at high levels in a number of tumors of non-neuronal origin ( 3 , 14 ). In view of the functional role of BC RNAs as modulators of translation in neurons and germ cells, the combined evidence thus leads us to hypothesize that neurons and germ cells have developed unique mechanisms of translational control and that certain types of tumor cells misappropriate such mechanisms so as to be able to sustain their increased growth requirements.
Can the deregulation of BC expression in tumor cells be used to develop a molecular marker for malignancies? At which point during tumorigenesis does BC expression become manifest and detectable? Here we address these questions by examining BC200 expression in various benign and malignant lesions of the breast. Our results establish the potential of BC200 expression as a molecular tumor indicator of diagnostic and prognostic value in breast cancer.
Materials and methods
Breast tissue samples were obtained through surgery, biopsy or reduction mammoplasty. Tissue was collected at the University of Münster School of Medicine and State University of New York Health Science Center at Brooklyn (SUNY HSCB) in accordance with applicable local, federal or European Union regulations. Specimens were identified and classified by pathologists at the Gerhard-Domagk-Institute of Pathology, University of Münster School of Medicine or at the Department of Pathology, SUNY HSCB. Samples from a total of 41 patients were analyzed (10 normal/benign, 11 ductal carcinoma in situ , 20 invasive carcinoma). Tissue samples were quick-frozen in liquid nitrogen and stored at −80°C until processing.
In situ hybridization
We decided to identify and quantify BC200 RNA by in situ hybridization in tissue sections, rather than by solution or filter hybridization using homogenized tissue. In the latter case, information would be lost as to what percentage of the tissue was in fact tumor tissue and tumor tissue would thus be ‘diluted’ with normal tissue, in the extreme case such that a signal from tumor cells would no longer be detectable. For this work all tissue samples and sections were treated identically and hybridization experiments were performed under identical conditions to allow direct comparisons.
Tissue samples were cryo-embedded in Tissue-Tek OCT embedding medium (Sakura FineTek, Torrance, CA) and sectioned in a Bright Microtome Cryostat at 10 µm thickness. Sections were thaw-mounted onto microscope slides and stored at −80°C. RNA probes specific for BC200 RNA were generated from plasmid pVL450-1 ( 3 ). 35 S-labeled probes (‘sense’ and ‘antisense’ strands) were transcribed from linearized templates, using T3 or T7 RNA polymerase, and used for in situ hybridization as described ( 3 ).
Microscopy, image analysis and quantification
Dark and bright field images of tissue sections were acquired with a Photometrics CoolSnap HQ cooled CCD camera (Roper Scientific, Tucson, AZ) mounted on a Nikon Microphot-FXA microscope. Quantification of autoradiographic silver grain densities in regions of interest (ROIs) was performed using MetaMorph software (Universal Imaging Corporation, Downingtown, PA). Briefly, all images were sharpened, without altering the average gray value, using the same convolution kernel. Gray scale images were thresholded to distinguish silver grains from tissue. The area above threshold (number of pixels) was measured for each ROI. A thresholded area of identical dimensions was also measured in a nearby region outside the tissue. This value was considered background and was subtracted from the tissue ROI signal for the same slide. The resulting value (the relative signal intensity, RSI) was taken as a measure for relative BC200 expression levels. Numerical RSI values are expressed as means and medians in the text and are presented in the format mean ± SEM in the diagrams.
To examine the significance of expression differences between groups of tissues, we applied the non-parametric Kruskal–Wallis one-way ANOVA test. The non-parametric Mann–Whitney U -test was used for direct comparisons of two groups. The level of significance was set at P < 0.05. To establish the discriminative efficacy of BC200 RNA as a diagnostic or prognostic marker, we compiled the data in receiver operating characteristics (ROC) curves ( 15 ). In this analysis, sensitivity is plotted against (1 − specificity). The Area Under the Curve (AUC) is used as a measure for discriminative diagnostic power. The AUC of poor markers will be close to 0.5 (and the ROC curve close to the diagonal), whereas the AUC of markers with high discriminative value will be close to 1.0 (and the ROC curve close to the upper left-hand corner). AUC–ROC analysis calculates discriminative efficacy as a single parameter without setting artificial cut-off values. A Bootstrap confidence interval was calculated for the AUC of invasive breast cancer. GB STAT (Dynamic Microsystems, Silver Spring, MD), SAS (SAS Institute, Cary, NC) and SPSS (SPSS, Chicago, IL) software were used for statistical analyses.
We have previously reported that BC200 RNA is expressed in several types of malignancies ( 3 ). To ascertain the potential utility of BC200 expression as a diagnostic and/or prognostic molecular indicator, we have in the present work decided to focus on breast cancer.
Normal/benign breast tissue
We first established BC200 expression levels in normal breast tissue. Little, if any, specific labeling signal was observed and no focal or clustered BC200 expression was apparent in stromal or glandular tissue ( Figure 1A and B ). Analogous observations were made in normal breast tissue undergoing lactational change ( Figure 1C and D ). The BC200 signal was quantified and expressed as RSI (see Materials and methods for details). In normal breast tissue the RSI was established as 208 (mean) or 203 (median). No significant difference in labeling intensities was detected between normal breast tissue and ‘sense strand’ controls (RSI for the latter: mean 261, median 226) (see also Figure 2 ). The results indicate that labeling levels observed in normal tissue are no different from background labeling. They confirm previous data obtained by northern hybridization showing that normal human breast tissue does not express BC200 RNA ( 2 ). Similarly low BC200 background levels were also observed in a number of other benign conditions, e.g. in fibrocystic change (not shown).
We further examined BC200 expression in benign fibroadenomas ( Figure 2A and B ) and found that signal intensities and distributions in epithelial and stromal elements did not noticeably differ from either normal tissue ( Figure 1 ) or from ‘sense’ strand controls ( Figure 2C and D ) that were analyzed in parallel. Quantitative analysis showed that BC200 expression in fibroadenomas (mean RSI 501, median 416) was slightly higher than in normal breast tissue. However, this elevation was not statistically significant. The data therefore suggest that BC200 RNA is not significantly expressed in normal human breast tissue or in benign conditions such as fibroadenomas.
In contrast to normal or benign breast tissue, invasive carcinomas of the breast were found to express BC200 RNA at high levels. Invasive ductal carcinomas (IDCs) are shown in Figure 3 . The labeling signal was observed heterogeneously distributed in the tumor tissue and was specifically associated, in clustered fashion, with tumor cells. Analogous observations were made with invasive lobular carcinomas and with invasive tubulo-lobular carcinomas (an example of the latter is shown in Figure 4 ). In all these cases it was clearly the tumor cells (not surrounding stromal and glandular tissue) that were BC200-positive. For all invasive carcinomas examined, BC200 signal intensities were substantial and significantly higher than the background levels observed in normal or benign breast tissue.
This qualitative evaluation was confirmed by quantitative analysis ( Figure 5 ). For the purposes of this analysis, invasive breast carcinomas (IBCs) were treated as one group as no significant differences in BC200 expression levels were revealed between different types or grades of IBCs. BC200 expression in IBCs was scored at a mean RSI of 2525 (median 2347) ( Figure 5A ). Statistical analysis showed that BC200 expression in IBCs was significantly higher than in normal or benign tissue (Kruskal–Wallis one-way ANOVA, P < 0.0001; Mann–Whitney U -test, comparison with normal tissue, P < 0.001, compared with fibroadenomas, P < 0.01).
To ascertain the diagnostic efficacy of BC200 expression as a marker to discriminate between IBCs and normal/benign tissue, we plotted sensitivity against (1−specificity) in a ROC curve ( Figure 5B ). An AUC of 0.95 was obtained (with a 95% confidence interval of 0.86–1.0), a result that indicates high discriminating power of BC200 expression as a marker for IBCs versus normal/benign tissue.
Ductal carcinomas in situ
Ductal carcinomas in situ (DCIS) constitute a heterogeneous group of malignant epithelial cell growth that remains confined to the ductal-lobular tree and has not penetrated the basement membrane. We found that in DCIS, BC200 RNA levels were a function of tumor grade. Thus, non-high grade (NHG) DCIS showed very little specific BC200 labeling ( Figure 6A and B ). Quantitative and statistical analysis showed that BC200 signal intensities in NHG DCIS (mean RSI 456, median 543) were not significantly different from background levels observed in normal or benign breast tissue. In contrast, high grade (HG) DCIS showed substantial BC200 expression ( Figure 6C and D ). BC200 signal intensities in HG DCIS (mean RSI 3112, median 2212) were in the same range as invasive carcinomas and were significantly higher than background levels observed in normal tissue or fibroadenomas. Direct quantitative comparison showed that expression in HG DCIS was also significantly higher than in NHG DCIS (Mann–Whitney U -test, P < 0.01) ( Figure 6E ). The results indicate that in DCIS, BC200 expression is strongly dependent on tumor grade. An AUC–ROC of 1.0 was obtained in this case (not shown), indicating high discriminative power between HG DCIS and NHG DCIS.
In summary, our data show that in DCIS, BC200 expression is a function of tumor grade, with significant differences in expression levels between HG DCIS and NHG DCIS. We therefore propose that BC200 expression may be developed as a prognostic indicator in ‘preinvasive’ carcinomas in situ (see Discussion).
Progress in the diagnosis and prognosis of breast cancer has been slow because reliable and sensitive molecular markers have not been developed to date. Such indicators are urgently needed to identify and characterize breast lesions, to distinguish benign from malignant disease and to be able to prognosticate the potential of a non-invasive lesion to become invasive. The results presented in this paper indicate that BC200 expression in breast cancer cells may serve as a molecular indicator of invasive or potentially invasive malignancy.
BC200 RNA as a diagnostic indicator of invasive malignancy
BC200 RNA is expressed at high levels in invasive breast cancer but is not detectable above background in normal breast tissue and in benign tumors such as fibroadenomas. In line with these results, BC200 RNA was also not detected in other normal non-neural tissues ( 2 ). Thus, in contrast to many tumors markers (e.g. PSA) that are merely up-regulated from basal levels, BC200 RNA is strongly expressed in invasive breast tumors but is not detectable at significant levels in normal or benign breast tissue. BC200 RNA may therefore be used as a categorical indicator, for diagnostic purposes a significant advantage over traditional incremental indicators such as PSA.
ROC analysis confirmed that BC200 expression provides high sensitivity and specificity and thus qualifies as a discriminative marker of invasive breast malignancies. It should be noted in this context that BC200 RNA cannot be classified as a proliferation marker, for two reasons. (i) Of all normal somatic cells, only neurons express BC200 RNA. Most neurons are post-mitotic and do not proliferate. (ii) Proliferating somatic cells, with the exception of cancer cells such as mammary carcinoma cells, do not express BC200 RNA. Thus, BC200 expression is not associated with proliferation per se , but rather with invasive or potentially invasive malignancy.
The substantial expression of BC200 RNA in invasive breast cancer indicates potential diagnostic utility in a clinical setting. On the basis of the evidence presented here, future work will be directed at developing a methodology for the standardized, routine detection of BC200 RNA in small tissue samples. Such methodology, which may employ RT–PCR, is expected to be valuable in establishing BC200 status in biopsy and other preoperative clinical samples.
BC200 RNA as a potential prognostic indicator
DCIS is a non-uniform group of neoplastic diseases. In only a fraction of such cases does the tumor progress to become invasive, and thus clinically important, if left untreated (reviewed in 16 , 17 ). An unresolved dilemma lies in our current inability to distinguish potentially life-threatening DCIS as there is to date no reliable indicator, molecular or otherwise, to prognosticate invasive potential in DCIS and thus to identify those cases that are likely to advance to invasive disease. Most women with mammographically detectable DCIS therefore opt for surgery, which has often been viewed as amounting to overtreatment ( 16 , 18 ). Clearly, a reliable prognostic indicator would be most valuable in assisting physicians and patients in making informed treatment decisions.
The clinical observations are mirrored by cytogenetic evidence. Genetic diversity in intraductal carcinoma development suggests that the step from DCIS to IDC cannot be seen as a simple linear progression. Rather, DCIS is characterized by clonal proliferation and diversification of different cytogenetic subclones, one of which may possibly become invasive ( 19 , 20 ). Comparative genomic hybridization data support the notion that the transition from DCIS to IDC follows specific cytogenetic pathways that appear to be associated with differentiation status and grade ( 21 , 22 ). HG DCIS appears to be associated with increased genomic instability as genetic aberrations are observed at higher frequency in HG DCIS than in NHG DCIS ( 21 ). In addition, there is strong evidence that nuclear grade is an indicator of aggressive behavior in DCIS ( 23 – 26 ) [but note that subgroups of NHG DCIS may share other genetic homologies with invasive carcinomas, e.g. lobular invasive carcinomas (see 20)]. Nuclear grade in DCIS can often only be determined with fair to moderate consistency ( 25 ). For these reasons, we decided to explore whether BC200 expression was associated with nuclear grade in DCIS.
Our results show that BC200 RNA is expressed at significant levels in HG DCIS, but not in NHG DCIS. High expression of BC200 RNA in carcinoma in situ is thus indicative of high grade. The data suggest that BC200 expression in DCIS is likely to be associated with a high degree of genomic instability and with an elevated degree of invasive potential of that tumor. While the prognostic power of BC200 expression will obviously have to be substantiated by longitudinal analysis in prospective follow-up studies, our results represent a significant step towards establishing the utility of BC200 expression as a prognostic indicator for DCIS.
In summary, the data presented here suggest that BC200 expression can be developed as a specific molecular indicator of invasive breast carcinomas and of invasive potential in carcinomas in situ . Our future work will be directed at establishing the utility of BC200 expression in breast cancer to a point where it can be successfully applied as a routine diagnostic and prognostic tool in a clinical setting.
We thank Jürgen Brosius for advice and discussions. This work was supported in part by fellowships from the CNR and Istituto Pasteur–Fondazione Cenci Bolognetti (A.I.), by a New York City Council Speaker's Fund for Biomedical Research Grant (I.A.M.), by National Institutes of Health grant NS034158 (H.T.) and by Department of Defense Breast Cancer Research Program grant DAMD 17-02-1-0520 (H.T.).
1Department of Physiology and Pharmacology, 2Department of Surgery and 3Scientific Computing Center, State University of New York, Health Science Center at Brooklyn, Brooklyn, NY 11203, USA, 4Institute of Pathology, Heinrich-Heine-University, D-40225 Düsseldorf, Germany, 5Gerhard-Domagk-Institute of Pathology, University of Münster School of Medicine, D-48129 Münster, Germany and 6Department of Neurology, State University of New York, Health Science Center at Brooklyn, Brooklyn, NY 11203, USA