This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Porter, D. Articles by Polyak, K. Search for Related Content PubMed PubMed Citation Articles by Porter, D. Articles by Polyak, K.

---

Cancer Genes and Genomics

Molecular Markers in Ductal Carcinoma in Situ of the Breast¹

¹ Departments of Medical Oncology and ² Biostatistics, Dana-Farber Cancer Institute, Boston, MA;
³ Department of Pathology, Brigham and Women's Hospital, Boston, MA;
⁴ Department of Pathology, Faulkner and Brigham and Women's Hospital, Boston, MA;
⁵ Department of Pathology, Beth-Israel Deaconess Medical Center, Boston, MA;
⁶ Harvard Medical School and ⁷ Harvard School of Public Health, Boston, MA;
⁸ Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, MD;
⁹ National Cancer Institute, Bethesda, MD; and
¹⁰ Department of Pathology, Duke University Medical Center, Durham, NC

Requests for reprints: Kornelia Polyak, Dana-Farber Cancer Institute, 44 Binney St. D740C, Boston, MA 02115. Phone: (617) 632-2106; Fax: (617) 632-4005. E-mail: Kornelia_Polyak{at}dfci.harvard.edu

	Abstract

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Gene expression patterns in ductal carcinoma in situ (DCIS), and in invasive, and metastatic breast tumors were determined using serial analysis of gene expression (SAGE). We used mRNA in situ hybridization to examine gene expression at the cellular level and immunohistochemistry on tissue microarrays to determine association between gene expression patterns and histopathologic characteristics of the tumors. We found that that the most dramatic transcriptome change occurs at the normal to DCIS transition, while there is no clear universal "in situ" or "invasive" tumor molecular signature. From the 16,430 transcripts analyzed, we identified only 5 and 11 that were preferentially up-regulated in DCIS and invasive tumors, respectively. The majority of invasive cancer specific SAGE tags correspond to novel genes. The genes we identified may define biologically and clinically meaningful subgroups of DCIS with a high risk of progression to invasive disease.

	Introduction

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Ductal carcinoma in situ (DCIS) of the breast includes a heterogeneous group of preinvasive breast tumors with a wide range of malignant potential. Some DCIS, if untreated, will rapidly progress to invasive cancer, while others will change very little in 5–20 years. DCIS now represents a significant (up to 25%) fraction of newly diagnosed breast cancer cases, yet the clinical management of DCIS patients is still controversial. The major challenge is to reliably determine the risk of local recurrence and progression to invasive disease, and to treat patients accordingly. Understanding the pathobiology of DCIS may provide insight into their initiation and progression, but very little is known about DCIS at the molecular level. Current classification of DCIS is based on cytological and architectural features. Some studies have demonstrated a correlation between these features, particularly nuclear grade and the presence of comedo necrosis, and clinical behavior (1–3). In general, high nuclear grade and comedo DCIS are associated with high rate of local recurrence, while low-grade non-comedo tumors have low recurrence rates (4). However, classification based on histological features can be determined only with moderate consistency and is complicated by intratumoral heterogeneity. Moreover, none of these features is able to predict the risk of progression to invasive disease accurately. Therefore, the molecular characterization of DCIS with the aim of identifying biologically and potentially clinically meaningful subgroups is of utmost importance.

Comprehensive gene expression profiling has proven useful for the molecular classification of invasive breast tumors, but no such extensive study has been performed in DCIS largely due to technical difficulties with obtaining DCIS samples suitable for RNA-based analysis (5, 6). Our laboratory uses serial analysis of gene expression (SAGE) to identify genes implicated in breast tumorigenesis (7–9). SAGE requires a relatively small amount of tissue (50,000 cells or less) for the generation of comprehensive expression profiles without the requirement for RNA/cDNA amplification steps; thus, it is particularly well suited for the analysis of small, preinvasive tumors (10). Although SAGE allows the quantitative measurement of the mRNA levels of thousands of genes simultaneously in one specimen, to establish how frequently an emerging candidate gene is differentially expressed, it is necessary to examine hundreds of breast specimens. The recently developed tissue microarray technology allows rapid profiling of hundreds of specimens on one slide and is therefore ideally suited for the further evaluation of candidate genes emerging from SAGE analysis (11, 12).

Here we report that based on SAGE analyses of breast tumors of different histopathologic stages, we identified several transcripts that belong to one of the following groups: up- or down-regulated in breast cancer regardless of stage; preferentially up-regulated in DCIS; or in invasive carcinomas. The expression of 14 genes was confirmed at the cellular level by mRNA in situ hybridization using a panel of frozen DCIS and invasive breast tumors. Tissue microarrays composed of primary breast cancers of different pathological stages were used to explore the potential clinical usefulness of 10 genes by investigating their relationship to histopathologic features of the tumors and patient outcome.

	Results

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Normal and Cancerous Breast Transcriptomes
Genes differentially expressed between normal and cancerous breast tissues were identified using SAGE. Confirming our previous report using a limited set of SAGE libraries (8), we found that the most dramatic difference in gene expression patterns occurs at the normal to in situ carcinoma transition and it involves the uniform down-regulation of 34 genes (Table 1). Because many of these genes encode secreted proteins and genes related to epithelial cell differentiation, loss of the differentiated epithelial phenotype and abnormal autocrine/paracrine interactions appear to play an essential role in the initiation of breast tumorigenesis. We also identified 144 genes up-regulated in a fraction of in situ and invasive or metastatic tumors (Table 2). Nearly one-fourth of these SAGE tags currently have no database match, indicating that many transcripts specifically expressed in certain breast carcinomas remain to be identified. To delineate overall similarities and differences among samples, the 19 SAGE libraries were analyzed by hierarchical clustering (Fig. 1A). A dendogram created using this program revealed that the two normal samples (N1 and N2) and primary invasive tumors and lymph node metastases (I1 and LN1, and I2 and LN2) derived from the same patients were most similar to each other. In situ and invasive tumors and metastases did not form distinct clusters, suggesting that in this tumor set, there is not a pronounced and common "in situ," "invasive," or "metastasis" signature. Correlating with this observation using clustering and other statistical analyses, we were not able to identify any gene that was universally and specifically up- or down-regulated in DCIS or invasive tumors (Fig. 1A). This result confirms previous studies performed in invasive breast carcinomas and highlights that DCIS tumors are just as heterogeneous at the molecular level as their invasive counterparts (6).

View larger version (31K): [in this window] [in a new window]   FIGURE 1. Hierarchical clustering analysis of breast SAGE libraries. A. Dendogram depicting the relatedness of SAGE libraries generated from normal mammary epithelial cells (N1 and N2, green), DCIS (D1–7 and T18, blue), primary invasive breast tumors (I1–6, red), and lymph node (LN1 and LN2, orange) and distant lung metastases (M1, orange). Examples for SAGE tags preferentially present in normal breast tissue, or invasive or in situ carcinomas. B. Dendogram demonstrating similarities among DCIS tumors using 515 genes for hierarchical clustering. C. Dendogram of hierarchical clustering performed using the 26 most highly differentially expressed genes selected for further validation studies (genes listed in Table 4).

Putative Molecular Markers in DCIS
To determine if there are genes that are statistically significantly more likely to be expressed in DCIS or invasive tumors, we performed various statistical tests (see "Materials and Methods"). On the basis of these analyses, we found that the expression of CD74 and a SAGE tag (CTGGGCGCCC) with no database match were significantly more abundant in invasive or metastatic tumors than in DCIS (P = 0.02 and P = 0.05, respectively, Table 3). The expression of MGC2328, DCD, and eight other genes was also more likely to occur in invasive/metastatic tumors than in DCIS, but none of these reached statistical significance (Table 3). Similarly, the expression of S100A7 and keratin 19 (KRT19) was more frequent and at higher levels in DCIS than in invasive/metastatic tumors, but this was also only marginally statistically significant. In a second statistical analysis, receiver operating characteristic (ROC) curve analysis was used to choose the "best" cutoff for values that result in the most samples being correctly classified as DCIS or invasive, weighing both kinds of misclassification equally (Table 3). Tags that do not include 0.50 in the CI might be potentially useful for the differential diagnosis of in situ and invasive carcinomas. This includes all the tags that had P 0.13 using the higher of two normal cutoffs as well as three other high in DCIS tags and three other high in invasive tags (Table 3). Using the best cutoff values, several of the SAGE tags correctly classified most of the DCIS and invasive SAGE libraries. For example, KRT19 classified 75% of the DCIS and 0% of the invasive libraries as DCIS, while MGC23280 identified 78% of the invasive cancer and 0% of the DCIS libraries as invasive. Thus, MGC23280 had 78% sensitivity and 100% specificity to correctly categorize breast tumors as DCIS or invasive/metastatic in this data set.

Confirming Gene Expression by Messenger RNA in Situ Hybridization
mRNA in situ hybridization determines gene expression at the cellular level, which is particularly useful in solid tumors heterogeneous in their cellular composition. We used 18 frozen DCIS and invasive breast cancer samples for this purpose. Whenever possible, tumors were selected to include normal, DCIS, and invasive components on the same slide to obtain expression data in these three stages of breast tumorigenesis. Examples of in situ hybridization results are depicted in Fig. 2A. Interestingly, we found that the expression of several genes up-regulated in DCIS was localized mostly or exclusively to non-epithelial cells. Specifically, CTGF (connective tissue growth factor) and RGS5 (regulator of G protein signaling) were highly expressed in DCIS myoepithelial cells and stromal fibroblasts, although in certain tumors, we detected expression in DCIS epithelial cells as well (Fig. 2A). Cumulative scores for in situ hybridization were used for hierarchical clustering analysis and statistical tests. A dendogram of the 18 different tumors and 5 normal breast tissues determined that using 14 genes, we were able to differentiate between normal and cancer samples and group the tumors into subclasses (Fig. 2B). Confirming our SAGE results, we were not able to differentiate DCIS and invasive tumors based on gene expression profiles. Surprisingly, in the majority of cases within the same tissue sample, the in situ and invasive areas of the tumor did not show the highest similarity to each other (Fig. 2B). Although this result could be due to the use of mRNA in situ hybridization and the selected genes, it may suggest that gene expression profiles are not necessarily maintained during tumor progression. Fisher's exact test revealed significant positive correlation between the expression of TFF3 and IFI-6-16 (P = 0.01), LOC51235, and BEX1 (P = 0.05), while inverse correlation was found between the expression of S100A7 and RGS5Tu (tumor cell specific expression) (P = 0.04), S100A7 and TFF3 (P = 0.04), and CTGF and TM4SF1 (P = 0.01). No statistically significant associations were found between the expression of the genes and histopathologic features of the tumors.

View larger version (48K): [in this window] [in a new window]   FIGURE 2. Analysis of gene expression in DCIS tumors using mRNA in situ hybridization and immunohistochemistry. A. Digitonin labeled anti-sense riboprobes corresponding to the indicated genes were hybridized to frozen DCIS tumors as described in "Materials and Methods" to give a dark purple precipitate that corresponds to the presence of the appropriate mRNA. Hybridization with the sense probes gave no signal (data not shown). Strong expression of IFI-6-16 and S100A7 is detected in a subset of DCIS (T18, 96-331, and 6164), but not normal (N24), mammary epithelial cells, while the expression of CTGF and RGS5 is mostly seen in DCIS stromal fibroblasts and myoepithelial cells, respectively. B. Dendogram depicting hierarchical clustering of 5 normal, 18 DCIS, and invasive breast tumors based on mRNA in situ hybridization results obtained using the indicated 14 genes. Numbers correspond to coded specimen identifiers, "N" denotes normal, "D" DCIS, and "I" invasive breast tissue. Coloring scheme as in Fig. 1. Tu and Str stand for expression in tumor and stromal cells, respectively. C. A representative DCIS sample from a tissue microarray composed of DCIS tumors stained with hematoxylin–eosin, or with antibodies against the indicated genes. Immunohistochemistry was performed as described in "Materials and Methods"; black/dark brown precipitate indicates the presence of S100A7, TFF3, SPARC, or CTGF protein. No DCD expression is seen in DCIS.

	Discussion

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

Although no genes were absolutely specific for DCIS tumors, two calcium binding proteins of the S100 family, S100A7 (psoriasin) and S100A9, trefoil factor 3 (TFF3), keratin 19 (KRT19), and apolipoprotein D (APOD) were preferentially more abundant in DCIS than in normal or invasive/metastatic cancerous breast tissue. S100A7 has been previously identified as a gene differentially expressed between in situ and invasive breast cancer and it has been demonstrated to bind calcium and act as a chemoattractant for T-lymphocytes and neutrophils (14). Correlating with this, S100A7 expression was particularly high in high-grade comedo DCIS that frequently demonstrate strong lymphocytic infiltration. Similarly, S100A9 is known to influence cell migration (15). However, it is likely that both S100A7 and S100A9 have additional intracellular functions such as the regulation of cell growth, differentiation, or survival. Keeping with this, the expression of S100A7 was shown to be dramatically up-regulated in response to apoptosis inducing stimuli and in psoriatic and premalignant keratinocytes that are unable to differentiate (16–18). The high expression of S100A7 in ER negative, poorly differentiated, and lymph node positive invasive breast tumors suggests that its expression may predict bad clinical outcome and high risk of recurrence or progression in DCIS. Kaplan-Meier curve analysis demonstrating a somewhat decreased overall survival of patients with S100A7 positive invasive breast tumors supports this hypothesis, but to conclusively determine if S100A7 could be useful for the prognostication of breast cancer patients requires further studies.

Another gene highly expressed in DCIS is trefoil factor 3, a secreted protein that has been implicated in tumorigenesis, wound healing, and regulation of epithelial integrity (19). Although its role in breast tumorigenesis is unclear, in colon cancer cells, it has been demonstrated to confer apoptosis resistance (20). Keratin 19 and apolipoprotein D were also more abundant in DCIS than in normal epithelial cells and invasive carcinomas. Keratin 19 is the smallest known acidic keratin that is preferentially expressed in breast tumors of ductal origin (21). Apolipoprotein D is a glycoprotein involved in lipid transport that is negatively regulated by estrogen and its expression in invasive breast cancer may correlate with shorter disease free survival (22, 23). Although the differential expression of KRT19, S100A9, TFF3, and APOD between in situ and invasive carcinomas has not been analyzed in detail, the differences we detected by SAGE would be impossible to confirm by mRNA in situ hybridization or immunohistochemistry due to the semiquantitative nature of these latter techniques. Because we believe that the most likely clinical application of molecular markers would be their evaluation by immunohistochemistry, these four genes have limited potential as molecular markers.

In addition to selecting DCIS and invasive/metastatic cancer specific genes, we also identified 26 transcripts most highly differentially expressed between normal and DCIS or between different types of DCIS and demonstrated that these genes are able to classify DCIS as accurately as over 500 genes could (Fig. 1, B and C). The expression of 15 and 10 of these genes was confirmed by mRNA in situ hybridization and immunohistochemistry, respectively (Fig. 2 and Table 5), and the expression of some of them appeared to correlate with the histopathologic features of the tumors suggesting that they may identify subgroups of DCIS.

The majority of genes that are preferentially expressed in invasive/metastatic carcinomas correspond to uncharacterized genes. Four transcripts (MGC23280, MGC14480, FLJ30428, and NUDT8) encode hypothetical proteins with no known functions, while one of the SAGE tags currently has no ESTs match (Table 3). Dermcidin (DCD) was recently identified as a small secreted protein highly expressed in sweat glands of the skin, but its role in breast cancer is unclear (24). Calmodulin-like skin protein is a calcium-binding protein with a putative role in keratinocyte differentiation (25). Similarly, cellular retinol-binding protein is a protein carrier of retinol, a compound essential for normal epithelial cell differentiation (26). The high expression of immunoglobulin and CD74 antigen may reflect leukocytic infiltration of the tumors. However, CD74, the invariant () chain of the MHC class II antigen, was also highly abundant in SAGE libraries generated from purified epithelial cells; thus, it is likely to be expressed by the cancer cells themselves although the functional relevance of this is unclear. We hypothesized that these "invasive cancer specific" genes may be used for the identification of DCIS tumors with the highest risk for recurrence and progression. As a first step toward testing this hypothesis, we confirmed by immunohistochemical analysis that the expression of DCD is very rarely detected in DCIS (data not shown). However, testing this hypothesis will require the examination of hundreds of DCIS tumors with long-term clinical follow-up data.

Although no study identical to ours has been performed using DNA microarrays, several of the genes that were overexpressed in tumors (like trefoil factor 3, X-box binding protein, collagen type I, fibronectin, etc.) were also found to be up-regulated in breast cancer by other groups (5, 6, 27).

In summary, we have identified several genes that could potentially be used for the molecular differential diagnosis of DCIS and invasive breast cancer and for the classification of DCIS tumors. Determining the clinical usefulness of these genes requires further studies.

	Materials and Methods

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Tissue Samples and Tissue Microarrays
Fresh tissue specimens obtained from the Brigham and Women's Hospital, Massachusetts General Hospital, and Faulkner Hospital (all Boston, MA), Duke University (Durham, NC), University Hospital Zagreb (Zagreb, Croatia), and the National Disease Research Interchange were snap frozen on dry ice and stored at -80°C until use. All human tissue was collected following NIH guidelines and using protocols approved by the Institutional Review Boards. Tumors with significant DCIS component were identified based on pathology report and confirmed by microscopic examination of hematoxylin-eosin-stained frozen sections. Tumors used for SAGE analysis D1, D3, D4, D5, and D6 were high-grade, comedo DCIS, while D2, D7, and T18 were intermediate-grade with no necrosis. Tumors used for mRNA in situ hybridization and immunohistochemistry included DCIS cases of all three (low, intermediate, and high grade) histological types. Most of the cases used for in situ hybridization and immunohistochemistry were DCIS with concurrent invasive carcinoma and pure DCIS, respectively, while tumors D3 and D6 used for SAGE were pure DCIS. The larger representation of frozen/fresh DCIS cases with concurrent invasive disease was due to logistic issues, because it is extremely difficult to obtain frozen pure DCIS specimens, especially ones with long-term clinical follow-up data. For in situ hybridization, 5-µm-thick frozen sections were mounted on silylated slides (CEL Associates Inc., Pearland, TX), air-dried, and stored at -80°C until use. Tissue microarrays were (a) obtained from commercial sources; Imgenex, San Diego, CA (49 invasive breast tumor) and Ambion, Austin, TX (92 primary invasive tumors and 41 distant metastasis); (b) provided by Cooperative Breast Cancer Tissue Resource (40 normal breast tissue, 20 DCIS-10 pure and 10 with concurrent invasive tumor, 192 primary invasive breast tumors); and (c) generated at Johns Hopkins University (299 invasive breast tumors and 10 distant metastasis) and at Beth Israel Deaconess Medical Center (30 invasive breast cancer, and 70 pure DCIS of different histological grades and matched normal breast tissue) following published protocols (11). With the exception of the Imgenex and the DCIS arrays (1 mm punches), all other tissue microarrays (TMAs) had 0.6 mm punches and contained at least two punches/tumor to control for tumor and immunohistochemical staining heterogeneity.

Generation and Analysis of SAGE Libraries
SAGE libraries were generated and analyzed essentially as previously described as part of the National Cancer Institute Cancer Gene Anatomy Project (8, 9, 28, 29). SAGE libraries generated from normal (N1 and N2), two DCIS (D1 and D2), two invasive tumors (I1 and I2), and corresponding lymph node metastases (LN1 and LN2) were previously reported (8). Some of the DCIS tumors were pure DCIS (D3 and D6), while others were derived from patients with concurrent invasive breast carcinomas. Epithelial cells from normal breast tissue (N1 and N2) and tumors (D2, D3, D6, and D7) were purified using BerEP4-coated magnetic beads (Dynal, Oslo, Norway), while other tumors were macroscopically dissected based on adjacent hematoxylin-eosin-stained slides. Approximately 50,000 SAGE tags were obtained from each library. For further analyses, libraries were normalized to the library with the highest tag number (89,541 total tags). Hierarchical clustering was applied to data using the Cluster program developed by Eisen et al. (30). Differentially expressed genes were identified based on statistical analysis of comparisons of groups of normal (two cases), DCIS (eight samples), and invasive or metastatic (nine samples) breast cancer SAGE libraries using the SAGE2000 software (7). Similarly, for the identification of genes specific for DCIS or invasive breast cancer, the eight DCIS samples were treated as a group and the nine invasive or metastatic patients were treated as another group. First we used the SAGE tag numbers highest in two normal libraries (N1 and N2) as the cutoff and evaluated tag numbers in the DCIS and invasive libraries above this "normal" value using a two-sided Fisher's exact test without multiple comparisons (see Table 3). In a second test, ROC curve analysis was used to choose the best cutoff for values. A ROC area of 0.50 is no better than chance and a ROC area of 1.00 is the best possible. The bound 2 in the first row means that values greater than or equal to 2 are counted as predicting DCIS.

Messenger RNA in Situ Hybridization
To generate templates for in vitro transcription reactions, 300- to 500-bp fragments derived from the 3' untranslated region of the selected genes were PCR amplified and subcloned into pZERO 1.0 (Invitrogen, Carlsbad, CA). This pZERO 1.0 vector contains a multiple cloning site surrounded by SP6 and T7 RNA polymerase promoters; therefore, the same plasmid was used for the generation of sense and anti-sense riboprobes for mRNA in situ hybridizations. Digitonin-labeled sense and anti-sense riboprobes were generated and mRNA in situ hybridization was performed as previously described (31). The hybridized sections were observed with a NIKON microscope, images were obtained using a SPOT CCD camera, and processed with Adobe Photoshop. Hybridizations were considered successful if the sense probe gave no significant signal. The intensity and distribution of the hybridization signal were scored (0–3 for intensity and 0–3 for distribution using a scoring scheme described below for immunohistochemistry) independently by three investigators, and scores in Fig. 2 reflect a consensus of the three independent summary scores.

Immunohistochemistry
The expression of the indicated genes in primary breast tumors was analyzed by the use of immunohistochemistry to eight tissue microarrays that contained evaluatable paraffin-embedded specimens derived from 80 DCIS, 675 primary invasive breast cancer, and 33 distant metastases. Antigen Retrieval Citra solution (Research Genetics, San Ramon, CA) and boiling in microwave (5 min at high power) was used to enhance staining. Isotype control antiserum was used as negative control. A standard indirect immunoperoxidase protocol with 3,3'-diaminobenzidine as a chromogen was used for the visualization of antibody binding (ABC-Elite; Vector Laboratories, Burlingame, CA). Primary antibodies used were as follows: mouse monoclonal anti-psoriasin antibody (17); affinity-purified rabbit polyclonal anti-CTGF (a generous gift of Dr. D. Brigstock, Childrens' Research Institute, Columbus, OH); affinity-purified rabbit polyclonal anti-TFF3 (a kind gift of Prof. Hoffman, Universitaetsklinikum, Magdeburg, Germany); mouse monoclonal anti-IL-8, GRO-1, and GRO-2 were purchased from R&D Systems (Minneapolis, MN); anti-SPARC antibody was from Hematologic Technologies (Essex Junction, VT); and anti-FASN antibody from Transduction Labs. (San Diego, CA). Antibodies were used at 1:100 dilution in PBS containing 10% heat-inactivated goat serum. Antibody staining was subjectively scored by three investigators independently in a scale of 0–3 for intensity (0 = no staining, 1 = faint signal, 2 = moderate, and 3 = intense staining) and 0–3 for extent (0 = no, 1 = 30%, 2 = 30–70%, and 3 = 70% positive cells) of staining. Cumulative scores were calculated based on the average scores assigned by the three independent observers. For statistical analyses, a cumulative score at or above 3 was considered positive. Relationships between the expression of genes determined by mRNA in situ hybridization or immunohistochemistry were analyzed by Fisher's exact test without correction for multiple comparisons.

Statistical Analyses of Clinical Correlates
The relationship of gene expression to clinicopathologic parameters and the association between the expression of different genes determined by immunohistochemistry were analyzed by the following statistical methods. The eight separate tissue microarray data sets and one combined data set were analyzed for association of gene positivity and prognostic factors using a logistic regression model (with gene positivity as the outcome), and a forward, or step-up, selection procedure to determine the best fitting model. Clinicopathologic factors analyzed were: expression of the estrogen and progesterone receptors and HER2 by immunohistochemistry, histological grade, TNM stage, tumor size, number of positive lymph nodes, patient age, and overall and distant metastasis free survival. If all patients or no patients with a particular level of a covariate had gene positivity, then the logistic regression did not converge and a significance level was obtained using Fisher's exact test. If, however, there remained some patients with and without gene positivity after deleting patients with the particular level of the covariate, then a step-up logistic regression was done on them. The significance of the variables in the logistic regression models was tested using likelihood ratio tests. The cutoff used for entry into the model was = 0.05. In addition to the analyses described above, Kaplan-Meier curves were generated and Cox models were run for two data sets that contained survival information. Calculated times to distant failure and times to survival were used and were based on the failure/death and accession dates.

	Acknowledgements

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Special thanks to the National Cancer Institute Cooperative Breast Cancer Tissue Resource for breast tissue microarrays, the National Disease Research Interchange for frozen DCIS specimens, Dr. D. Brigstock (Childrens' Research Institute, Colombus, OH) for anti-CTGF, and Prof. Hoffman (Universitaetsklinikum, Magdeburg, Germany) for anti-TFF3 antibodies provided for this study.

	Notes

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
¹ National Cancer Institute Cancer Genome Anatomy Project and Specialized Program in Research Excellence in Breast Cancer at Dana-Farber/Harvard Cancer Center (CA89393) and Johns Hopkins University (CA88843); Department of Defense Breast Cancer Center of Excellence Grants.

Received December 2, 2002; revised February 10, 2003; accepted February 13, 2003.

	References

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 

1. Walker, R. A., Jones, J. L., Chappell, S., Walsh, T., and Shaw, J. A. Molecular pathology of breast cancer and its application to clinical management. Cancer Metastasis Rev., 16: 5–27, 1997.[Medline]
2. Gupta, S. K., Douglas-Jones, A. G., Fenn, N., Morgan, J. M., and Mansel, R. E. The clinical behavior of breast carcinoma is probably determined at the preinvasive stage (ductal carcinoma in situ). Cancer, 80: 1740–1745, 1997.[Medline]
3. Badve, S., A'Hern, R. P., Ward, A. M., Millis, R. R., Pinder, S. E., Ellis, I. O., Gusterson, B. A., and Sloane, J. P. Prediction of local recurrence of ductal carcinoma in situ of the breast using five histological classifications: a comparative study with long follow-up. Hum. Pathol., 29: 915–923, 1998.[Medline]
4. Silverstein, M. (ed.). Ductal Carcinoma in Situ of the Breast. Baltimore, MD: Williams & Wilkins, 1997.
5. Luzzi, V., Holtschlag, V., and Watson, M. A. Expression profiling of ductal carcinoma in situ by laser capture microdissection and high-density oligonucleotide arrays. Am. J. Pathol., 158: 2005–2010, 2001.[Abstract/Free Full Text]
6. Perou, C. M., Sorlie, T., Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Rees, C. A., Pollack, J. R., Ross, D. T., Johnsen, H., Akslen, L. A., Fluge, O., Pergamenschikov, A., Williams, C., Zhu, S. X., Lonning, P. E., Borresen-Dale, A. L., Brown, P. O., and Botstein, D. Molecular portraits of human breast tumours. Nature, 406: 747–752, 2000.[Medline]
7. Velculescu, V. E., Zhang, L., Vogelstein, B., and Kinzler, K. W. Serial analysis of gene expression. Science, 270: 484–487, 1995.[Abstract/Free Full Text]
8. Porter, D. A., Krop, I. E., Nasser, S., Sgroi, D., Kaelin, C. M., Marks, J. R., Riggins, G., and Polyak, K. A SAGE (serial analysis of gene expression) view of breast tumor progression. Cancer Res., 61: 5697–5702, 2001.[Abstract/Free Full Text]
9. Krop, I. E., Sgroi, D., Porter, D. A., Lunetta, K. L., LeVangie, R., Seth, P., Kaelin, C. M., Rhei, E., Bosenberg, M., Schnitt, S., Marks, J. R., Pagon, Z., Belina, D., Razumovic, J., and Polyak, K. HIN-1, a putative cytokine highly expressed in normal but not cancerous mammary epithelial cells. Proc. Natl. Acad. Sci. USA, 98: 9796–9801, 2001.[Abstract/Free Full Text]
10. St Croix, B., Rago, C., Velculescu, V., Traverso, G., Romans, K. E., Montgomery, E., Lal, A., Riggins, G. J., Lengauer, C., Vogelstein, B., and Kinzler, K. W. Genes expressed in human tumor endothelium. Science, 289: 1197–1202, 2000.[Abstract/Free Full Text]
11. Kononen, J., Bubendorf, L., Kallioniemi, A., Barlund, M., Schraml, P., Leighton, S., Torhorst, J., Mihatsch, M. J., Sauter, G., and Kallioniemi, O. P. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat. Med., 4: 844–847, 1998.[Medline]
12. Moch, H., Schraml, P., Bubendorf, L., Mirlacher, M., Kononen, J., Gasser, T., Mihatsch, M. J., Kallioniemi, O. P., and Sauter, G. High-throughput tissue microarray analysis to evaluate genes uncovered by cDNA microarray screening in renal cell carcinoma [see comments]. Am. J. Pathol., 154: 981–986, 1999.[Abstract/Free Full Text]
13. Ramaswamy, S., Ross, K. N., Lander, E. S., and Golub, T. R. A molecular signature of metastasis in primary solid tumors. Nat. Genet., 33: 49–54, 2003.[Medline]
14. Watson, P. H., Leygue, E. R., and Murphy, L. C. Psoriasin (S100A7). Int. J. Biochem. Cell Biol. 30: 567–571, 1998.[Medline]
15. Donato, R. Functional roles of S100 proteins, calcium-binding proteins of the EF- hand type. Biochim. Biophys. Acta, 1450: 191–231, 1999.[Medline]
16. Madsen, P., Rasmussen, H. H., Leffers, H., Honore, B., Dejgaard, K., Olsen, E., Kiil, J., Walbum, E., Andersen, A. H., Basse, B., Lauridsen, J. B., Ratz, G. P., Celis, A., Vandekerckhove, J., and Celis, J. E. Molecular cloning, occurrence, and expression of a novel partially secreted protein "psoriasin" that is highly up-regulated in psoriatic skin. J. Invest. Dermatol., 97: 701–712, 1991.[Medline]
17. Enerback, C., Porter, D. A., Seth, P., Sgroi, D., Gaudet, J., Weremowicz, S., Morton, C. C., Schnitt, S., Pitts, R. L., Stampl, J., Barnhart, K., and Polyak, K. Psoriasin expression in mammary epithelial cells in vitro and in vivo. Cancer Res., 62: 43–47, 2002.[Abstract/Free Full Text]
18. van Ruissen, F., Jansen, B. J., de Jongh, G. J., van Vlijmen-Willems, I. M., and Schalkwijk, J. Differential gene expression in premalignant human epidermis revealed by cluster analysis of serial analysis of gene expression (SAGE) libraries. FASEB J., 16: 246–248, 2002.[Abstract/Free Full Text]
19. Sands, B. E. and Podolsky, D. K. The trefoil peptide family. Annu. Rev. Physiol., 58: 253–273, 1996.[Medline]
20. Taupin, D. R., Kinoshita, K., and Podolsky, D. K. Intestinal trefoil factor confers colonic epithelial resistance to apoptosis. Proc. Natl. Acad. Sci. USA, 97:799–804, 2000.[Abstract/Free Full Text]
21. Bratthauer, G. L., Moinfar, F., Stamatakos, M. D., Mezzetti, T. P., Shekitka, K. M., Man, Y. G., and Tavassoli, F. A. Combined E-cadherin and high molecular weight cytokeratin immunoprofile differentiates lobular, ductal, and hybrid mammary intraepithelial neoplasias. Hum. Pathol., 33: 620–627, 2002.[Medline]
22. Diez-Itza, I., Vizoso, F., Merino, A. M., Sanchez, L. M., Tolivia, J., Fernandez, J., Ruibal, A., and Lopez-Otin, C. Expression and prognostic significance of apolipoprotein D in breast cancer. Am. J. Pathol., 144: 310–320, 1994.[Abstract]
23. Harding, C., Osundeko, O., Tetlow, L., Faragher, E. B., Howell, A., and Bundred, N. J. Hormonally-regulated proteins in breast secretions are markers of target organ sensitivity. Br. J. Cancer, 82: 354–360, 2000.[Medline]
24. Schittek, B., Hipfel, R., Sauer, B., Bauer, J., Kalbacher, H., Stevanovic, S., Schirle, M., Schroeder, K., Blin, N., Meier, F., Rassner, G., and Garbe, C. Dermcidin: a novel human antibiotic peptide secreted by sweat glands. Nat. Immunol., 2: 1133–1137, 2001.[Medline]
25. Mehul, B., Bernard, D., Simonetti, L., Bernard, M. A., and Schmidt, R. Identification and cloning of a new calmodulin-like protein from human epidermis. J. Biol. Chem., 275: 12841–12847, 2000.[Abstract/Free Full Text]
26. Colantuoni, V., Cortese, R., Nilsson, M., Lundvall, J., Bavik, C. O., Eriksson, U., Peterson, P. A., and Sundelin, J. Cloning and sequencing of a full length cDNA corresponding to human cellular retinol-binding protein. Biochem. Biophys. Res. Commun., 130: 431–439, 1985.[Medline]
27. Sgroi, D. C., Teng, S., Robinson, G., LeVangie, R., Hudson, J. R., and Elkahloun, A. G. In vivo gene expression profile analysis of human breast cancer progression. Cancer Res., 59: 5656–5661, 1999.[Abstract/Free Full Text]
28. Lal, A., Lash, A. E., Altschul, S. F., Velculescu, V., Zhang, L., McLendon, R. E., Marra, M. A., Prange, C., Morin, P. J., Polyak, K., Papadopoulos, N., Vogelstein, B., Kinzler, K. W., Strausberg, R. L., and Riggins, G. J. A public database for gene expression in human cancers. Cancer Res., 59: 5403–5407, 1999.[Abstract/Free Full Text]
29. Boon, K., Osorio, E. C., Greenhut, S. F., Schaefer, C. F., Shoemaker, J., Polyak, K., Morin, P. J., Buetow, K. H., Strausberg, R. L., De Souza, S. J., and Riggins, G. J. An anatomy of normal and malignant gene expression. Proc. Natl. Acad. Sci. USA, 99: 11287–11292, 2002.[Abstract/Free Full Text]
30. Eisen, M. B., Spellman, P. T., Brown, P. O., and Botstein, D. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA, 95: 14863–14868, 1998.[Abstract/Free Full Text]
31. Qian, Y., Fritzsch, B., Shirasawa, S., Chen, C. L., Choi, Y., and Ma, Q. Formation of brainstem (nor)adrenergic centers and first-order relay visceral sensory neurons is dependent on homeodomain protein Rnx/Tlx3. Genes Dev., 15: 2533–2545, 2001.[Abstract/Free Full Text]

This article has been cited by other articles:

				I. Kijima, J. Ye, C. Glackin, and S. Chen CCAAT/Enhancer Binding Protein {delta} Up-regulates Aromatase Promoters I.3/II in Breast Cancer Epithelial Cells Cancer Res., June 1, 2008; 68(11): 4455 - 4464. [Abstract] [Full Text] [PDF]

				A. Vincent-Salomon, C. Lucchesi, N. Gruel, V. Raynal, G. Pierron, R. Goudefroye, F. Reyal, F. Radvanyi, R. Salmon, J.-P. Thiery, et al. Integrated Genomic and Transcriptomic Analysis of Ductal Carcinoma In situ of the Breast Clin. Cancer Res., April 1, 2008; 14(7): 1956 - 1965. [Abstract] [Full Text] [PDF]

				D. C. Allred, Y. Wu, S. Mao, I. D. Nagtegaal, S. Lee, C. M. Perou, S. K. Mohsin, P. O'Connell, A. Tsimelzon, and D. Medina Ductal Carcinoma In situ and the Emergence of Diversity during Breast Cancer Evolution Clin. Cancer Res., January 15, 2008; 14(2): 370 - 378. [Abstract] [Full Text] [PDF]

				G. W. Pearson and T. Hunter Real-time imaging reveals that noninvasive mammary epithelial acini can contain motile cells J. Cell Biol., December 31, 2007; 179(7): 1555 - 1567. [Abstract] [Full Text] [PDF]

				C. Fabian Tamoxifen or Raloxifene in Postmenopausal Women for Prevention of Breast Cancer: A Tale of Two Choices Counterpoint Cancer Epidemiol. Biomarkers Prev., November 1, 2007; 16(11): 2210 - 2212. [Full Text] [PDF]

				S. Esseghir, A. Kennedy, P. Seedhar, A. Nerurkar, R. Poulsom, J. S. Reis-Filho, and C. M. Isacke Identification of NTN4, TRA1, and STC2 as Prognostic Markers in Breast Cancer in a Screen for Signal Sequence Encoding Proteins Clin. Cancer Res., June 1, 2007; 13(11): 3164 - 3173. [Abstract] [Full Text] [PDF]

				K. A. Hartwell, B. Muir, F. Reinhardt, A. E. Carpenter, D. C. Sgroi, and R. A. Weinberg The Spemann organizer gene, Goosecoid, promotes tumor metastasis PNAS, December 12, 2006; 103(50): 18969 - 18974. [Abstract] [Full Text] [PDF]

				M. Dolled-Filhart, L. Ryden, M. Cregger, K. Jirstrom, M. Harigopal, R. L. Camp, and D. L. Rimm Classification of breast cancer using genetic algorithms and tissue microarrays. Clin. Cancer Res., November 1, 2006; 12(21): 6459 - 6468. [Abstract] [Full Text] [PDF]

				R. Lupu and J. A. Menendez Targeting Fatty Acid Synthase in Breast and Endometrial Cancer: An Alternative to Selective Estrogen Receptor Modulators? Endocrinology, September 1, 2006; 147(9): 4056 - 4066. [Abstract] [Full Text] [PDF]

				F. Okada, M. Kobayashi, H. Tanaka, T. Kobayashi, H. Tazawa, Y. Iuchi, K. Onuma, M. Hosokawa, M. C. Dinauer, and N. H. Hunt The Role of Nicotinamide Adenine Dinucleotide Phosphate Oxidase-Derived Reactive Oxygen Species in the Acquisition of Metastatic Ability of Tumor Cells Am. J. Pathol., July 1, 2006; 169(1): 294 - 302. [Abstract] [Full Text] [PDF]

				C. S. Schuetz, M. Bonin, S. E. Clare, K. Nieselt, K. Sotlar, M. Walter, T. Fehm, E. Solomayer, O. Riess, D. Wallwiener, et al. Progression-Specific Genes Identified by Expression Profiling of Matched Ductal Carcinomas In situ and Invasive Breast Tumors, Combining Laser Capture Microdissection and Oligonucleotide Microarray Analysis. Cancer Res., May 15, 2006; 66(10): 5278 - 5286. [Abstract] [Full Text] [PDF]

				J. Yao, S. Weremowicz, B. Feng, R. C. Gentleman, J. R. Marks, R. Gelman, C. Brennan, and K. Polyak Combined cDNA array comparative genomic hybridization and serial analysis of gene expression analysis of breast tumor progression. Cancer Res., April 15, 2006; 66(8): 4065 - 4078. [Abstract] [Full Text] [PDF]

				I. Krop, A. Marz, H. Carlsson, X. Li, N. Bloushtain-Qimron, M. Hu, R. Gelman, M. S. Sabel, S. Schnitt, S. Ramaswamy, et al. A Putative Role for Psoriasin in Breast Tumor Progression Cancer Res., December 15, 2005; 65(24): 11326 - 11334. [Abstract] [Full Text] [PDF]

				M. Thangaraju, M. Rudelius, B. Bierie, M. Raffeld, S. Sharan, L. Hennighausen, A-M. Huang, and E. Sterneck C/EBP{delta} is a crucial regulator of pro-apoptotic gene expression during mammary gland involution Development, November 1, 2005; 132(21): 4675 - 4685. [Abstract] [Full Text] [PDF]

				I. Zucchi, E. Mento, V. A. Kuznetsov, M. Scotti, V. Valsecchi, B. Simionati, E. Vicinanza, G. Valle, S. Pilotti, R. Reinbold, et al. Gene expression profiles of epithelial cells microscopically isolated from a breast-invasive ductal carcinoma and a nodal metastasis PNAS, December 28, 2004; 101(52): 18147 - 18152. [Abstract] [Full Text] [PDF]

				H. Alexander, A. L. Stegner, C. Wagner-Mann, G. C. Du Bois, S. Alexander, and E. R. Sauter Proteomic Analysis to Identify Breast Cancer Biomarkers in Nipple Aspirate Fluid Clin. Cancer Res., November 15, 2004; 10(22): 7500 - 7510. [Abstract] [Full Text] [PDF]

				Y. Hu, H. Sun, J. Drake, F. Kittrell, M. C. Abba, L. Deng, S. Gaddis, A. Sahin, K. Baggerly, D. Medina, et al. From Mice to Humans: Identification of Commonly Deregulated Genes in Mammary Cancer via Comparative SAGE Studies Cancer Res., November 1, 2004; 64(21): 7748 - 7755. [Abstract] [Full Text] [PDF]

R. Namba, J. E. Maglione, L. J.T. Young, A. D. Borowsky, R. D. Cardiff, C. L. MacLeod, and J. P. Gregg Molecular Characterization of the Transition to Malignancy in a Genetically Engineered Mouse-Based Model of Ductal Carcinoma In situ Mol. Cancer Res., August 1, 2004; 2(8): 453 - 463.