While several prognostic factors have been identified in breast carcinoma, the clinical outcome remains hard to predict for individual patients. Better predictive markers are needed to help guide difficult treatment decisions. In a previous study of 78 breast carcinoma specimens, we noted an association between poor clinical outcome and the expression of cytokeratin 17 and/or cytokeratin 5 mRNAs. Here we describe the results of immunohistochemistry studies using monoclonal antibodies against these markers to analyze more than 600 paraffin-embedded breast tumors in tissue microarrays. We found that expression of cytokeratin 17 and/or cytokeratin 5/6 in tumor cells was associated with a poor clinical outcome. Moreover, multivariate analysis showed that in node-negative breast carcinoma, expression of these cytokeratins was a prognostic factor independent of tumor size and tumor grade.

Advances in genomics and proteomics are dramatically increasing the need to evaluate large numbers of molecular targets for their diagnostic, predictive or prognostic value in clinical oncology. Conventional molecular pathology techniques are often tedious, time-consuming, and require a lot of tissue, thereby limiting both the number of tissues and the number of targets that can be evaluated. Here, we demonstrate the power of our recently described tissue microarray (TMA) technology in analyzing prognostic markers in a series of 553 breast carcinomas. Four independent TMAs were constructed by acquiring 0.6 mm biopsies from one central and from three peripheral regions of each of the formalin-fixed paraffin embedded tumors. Immunostaining of TMA sections and conventional "large" sections were performed for two well- established prognostic markers, estrogen receptor (ER) and progesterone receptor (PR), as well as for p53, another frequently examined protein for which the data on prognostic utility in breast cancer are less unequivocal. Compared with conventional large section analysis, a single sample from each tumor identified about 95% of the information for ER, 75 to 81% for PR, and 70 to 74% for p53. However, all 12 TMA analyses (three antibodies on four different arrays) yielded as significant or more significant associations with tumor-specific survival than large section analyses (p< 0.0015 for each of the 12 comparisons). A single sample from each tumor was sufficient to identify associations between molecular alterations and clinical outcome. It is concluded that, contrary to expectations, tissue heterogeneity did not negatively influence the predictive power of the TMA results. TMA technology will be of substantial value in rapidly translating genomic and proteomics information to clinical applications.

Multiple different oncogenes have been described previously to be amplified in breast cancer including HER2, EGFR, MYC, CCND1, and MDM2. Gene amplification results in oncogene overexpression but may also serve as an indicator of genomic instability. As such, presence of one or several gene amplifications may have prognostic significance. To assess the prognostic importance of amplifications and coamplifications of HER2, EGFR, MYC, CCND1, and MDM2 in breast cancer, we analyzed a breast cancer tissue microarray containing samples from 2197 cancers with follow-up information. Fluorescence in situ hybridizations revealed amplifications of CCND1 in 20.1%, HER2 in 17.3%, MDM2 in 5.7%, MYC in 5.3%, and EGFR in 0.8% of the tumors. All gene amplifications were significantly associated with high grade. HER2 (P < 0.001) and MYC amplification (P < 0.001) were also linked to shortened survival. In case of HER2, this was independent of grade, pT, and pN categories. MYC amplification was almost 3 times more frequent in medullary cancer (15.9%), than in the histologic subtype with the second highest frequency (ductal; 5.6%; P ‫؍‬ 0.0046). HER2 and MYC amplification were associated with estrogen receptor/progesterone receptor negativity (P < 0.001) whereas CCND1 amplification was linked to estrogen receptor/progesterone receptor positivity (P < 0.001). Coamplifications were more prevalent than expected based on the individual frequencies. Coamplifications of one or several other oncogenes occurred in 29.6% of CCND1, 43% of HER2, 55.7% of MDM2, 65% of MYC, and 72.8% of EGFR-amplified cancers. HER2/MYC-coamplified cancers had a worse prognosis than tumors with only one of these amplifications. Furthermore, a gradual decrease of survival was observed with increasing number of amplifications. In conclusion, these data support a major prognostic impact of genomic instability as determined by a broad gene amplification survey in breast cancer.

Previous studies in small series of patients with invasive breast cancer suggested a prognostic value of Ep-CAM overexpression in primary tumor tissue. To corroborate these findings, we performed a retrospective analysis of Ep-CAM expression using a tissue microarray containing tissue specimens from a large patient set. Ep-CAM expression was evaluated by immunohistochemistry in breast cancer tissue from 1715 patients with documented raw survival data. High level Ep-CAM expression (overexpression) was found in 41.7% of tumor samples, low level expression was found in 48.0% and no expression in 10.3% of tumor samples. Ep-CAM expression predicted poor overall survival in this patient cohort (p < 0.0001). Overall survival decreased significantly with increasing Ep-CAM expression. However, in this patient sample Ep-CAM expression was not an independent prognostic marker by multivariate analysis. Subgroup analysis revealed that Ep-CAM expression was a prognostic marker in node-positive (p < 0.0001) but not in node-negative (p = 0.58) breast cancer patients. Intriguingly, Ep-CAM expression was predictive for a dismal prognosis in patients receiving adjuvant cytotoxic (p = 0.03) or hormonal therapy (p < 0.0001) but not in untreated patients (p = 0.41). In summary, this study provides strong evidence that expression of Ep-CAM is a powerful marker of poor prognosis in node-positive invasive breast carcinoma and a potential predictive marker of sensitivity to adjuvant hormonal and/or cytotoxic treatment modalities.

The aim of this study was to compare prospectively the accuracy of whole-body positron emission tomography (PET), CT and MRI in diagnosing primary and recurrent ovarian cancer. Nineteen patients (age range 23-76 years) were recruited with suspicious ovarian lesions at presentation (n = 8) or follow-up for recurrence (n = 11). All patients were scheduled for laparotomy and histological confirmation. Whole-body PET with FDG, contrast-enhanced spiral CT of the abdomen, including the pelvis, and MRI of the entire abdomen were performed. Each imaging study was evaluated separately. Imaging findings were correlated with histopathological diagnosis. The sensitivity, specificity and accuracy for lesion characterization in patients with suspicious ovarian lesions (n = 7) were, respectively: 100, 67 and 86% for PET; 100, 67 and 86% for CT; and 100, 100 and 100% for MRI. For the diagnosis of recurrent disease (n = 10), PET had a sensitivity of 100%, specificity of 50% and accuracy of 90%. The PET technique was the only technique which correctly identified a single transverse colon metastasis. Results for CT were 40, 50 and 43%, and for MRI 86, 100 and 89%, respectively. No statistically significant difference was seen. Neither FDG PET nor CT nor MRI can replace surgery in the detection of microscopic peritoneal disease. No statistically significant difference was observed for the investigated imaging modalities with regard to lesion characterization or detection of recurrent disease; thus, the methods are permissible alternatives. The PET technique, however, has the drawback of less accurate spatial assignment of small lesions compared with CT and MRI.