In situ proteomic classification of breast cancer

An important clinical need in breast cancer research is to accurately distinguish patients who would benefit from aggressive, adjuvant treatment following initial diagnosis of primary breast cancer (high risk of recurrence) from those for whom adjuvant treatment may be unnecessary (low risk of recurrence). The identification of novel prognostic markers can not only identify subsets of patients at high or low risk, but can also identify new targets for potential future therapeutic development or decisions about treatment of metastatic disease. This dissertation summarizes the results of a discovery-based, hypothesis-generating approach to analyzing biomarkers by objective quantitation and molecular classification of breast cancer tumors.; The progression, validation, and the application of technologies for analyzing breast cancer cohort protein expression levels in a quantitative way are demonstrated in this dissertation. It begins with assessment of breast tumors by the current field standard with a study of Stat3 and PhosphoStat3 as an example of semi-quantitative immunohistochemical analysis of markers. Other previously published works on which I have been a collaborator using this methodology are included in the appendix. Automated quantitative analysis (AQUAtm) was developed in our laboratory for protein expression level analysis, and was validated with beta-catenin by showing that AQUA tm protein measurements are equivalent to ELISA quantitation. In addition, we have shown that the broad dynamic range of some biomarkers (such as HER2 and p53) are not easily distinguished by conventional methods.; The combination of tissue microarrays with AQUAtm can provide the follow-up technology for nucleic acid expression profiling-based target discoveries. The expense and difficulty of using RNA microarray gene expression analysis as a clinical test highlights the importance of quantitative protein expression assessment on tumors. Based on the demonstrated validation of AQUAtm from these studies, a large-scale examination of markers culled from microarray studies for their relationship to estrogen receptor and estrogen responsiveness were assessed by AQUAtm. By use of unsupervised hierarchical clustering and modeling by a genetic algorithm, we distilled down the data to a small group of three keystone markers, ER, NAT1 and GATA3, which could be inexpensively assessed in routine histologic breast cancer specimens to provide maximal classifying and predictive information.