Assessing the Differential Expression of Molecular Biomarkers in Breast Cancer.

Abstract

Abstract Background: Molecular profiling of breast cancer has identified multiple biomarkers with potential to better predict clinical outcomes and response to treatment than existing clinicopathological indices. We focused on 8 biomarkers of particular relevance as suggested by the literature: CD44, methyl guanine methyltransferase (MGMT), epidermal growth factor receptor (EGFR), cyclooxygenase-2 (COX2), cytokeratin-5 (CK5), estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2). This study aims to clarify the prognostic value of these breast cancer biomarkers as well as to define their relationships with each other.Methods: Formalin fixed paraffin embedded (FFPE) breast tumor samples from 140 patients diagnosed with breast cancer from Jan 2001 to Dec 2005 at St. Michael's Hospital in Toronto, Canada were examined retrospectively using tissue microarray analysis. Samples were stained for CD44, COX2, MGMT, EGFR, CK5, ER, PR, and HER2 by immunohistochemistry. Biomarkers including histological features and staining patterns were then evaluated. A manual chart review documenting relevant clinical and pathological features was also conducted. Subsequent statistical analysis utilized Kaplan-Meier survival curves to identify biomarker-survival associations and simple linear regression analysis to determine biomarker cluster groups.Results: Median patient age was 56 (range 31-86). Median follow-up time was 62 months (range 5-102). Tumors were of various pathological types and stages (I – IV). No significant associations between biomarkers and disease-free survival (DFS) were found. DFS (median = 53.5 months) did correlate with tumor stage, nuclear grade, and lymphovascular invasion (LVI) as expected. However, only LVI was found to correlate with DFS in triple negative patients (n=24). Regression analysis yielded two cluster groups of biomarkers; Cluster group 1 includes ER, PR, Cox2, CD44; and Cluster group 2 includes HER2, EGFR, CK5. Expression of biomarkers within one cluster group relate directly with those in the same group, but inversely with biomarkers from the other cluster group. In addition, biomarkers in Cluster group 1 relate inversely to the mitotic count, tumor size, and nuclear grade, while biomarkers from Cluster group 2 relate directly with these indices. For example, CK5 positive tumors with HER2 and EGFR overexpression tend to have a high mitotic count, large tumor size, and high nuclear grade. MGMT did not show association with any histological feature or DFS.Conclusions: This retrospective observational study of 140 breast cancer patients focused on the prognostic implications and relationships of 8 biomarkers: CD44, COX2, MGMT, EGFR, CK5, ER, PR, and HER2. Two cluster groups of biomarkers have been identified. Cluster group 1 (ER, PR, COX2, CD44) correlates with a less aggressive tumor morphology, while Cluster group 2 (HER2, EGFR, CK5) correlates with a more aggressive one. No associations have yet been found between biomarkers and DFS. This study has been extended to include 373 patients with further analysis underway. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 6041.