The anti-estrogen tamoxifen has been widely used as therapy in breast cancer. However, only half of the patients with recurrent tumors show an objective response (OR) to tamoxifen therapy, while the remaining half continues in progressive disease (PD). Eventually, even responsive tumors become resistant, which is a major cause of death and therefore a problem that needs to be tackled. Identification of proteins that associate with tamoxifen-resistance is a first step towards unraveling the molecular mechanisms involved. This information can be used to better predict therapy response, thereby reducing over-treatment of patients, and to identify new targets for therapy that may lead to tailored treatment strategies.

To identify protein biomarkers indicative of tamoxifen-resistance in breast cancer, a comparative proteomics analysis was performed using nanoscale liquid chromatography-Fourier transform ion cyclotron resonance mass spectrometry (nanoLC-FTICR MS). Two independently processed datasets (n=27 and n=28) of both OR and PD tumors were subjected to laser capture microdissection (LCM), ~ 4000 tumor cells were collected and subsequently pooled into groups of seven. Tryptic digests were prepared and analyzed in triplicate; ~550 ng of peptides per analysis were separated on a 50 μm x 80 cm reversed phase nanoLC column prior to FTICR MS analysis. Peptide mass and elution time features were matched to information in previously generated accurate mass and time (AMT) tag reference databases to identify peptide sequences and proteins, and the MS peak intensities were used to determine relative peptide abundances.

More than 20,000 unique peptides were identified that corresponded to a total of 2309 non-redundant proteins identified with two or more peptides. From this total, 1713 (74%) proteins that overlapped between the two datasets were used for further statistical analysis. The two datasets were separated into a ‘training’ and ‘validation’ set and analyzed for differential protein abundance between OR and PD groups, using a univariate t-test from BRB array-tools software package, followed by a Wilcoxon rank-sum test. In both datasets, 100 differentially abundant proteins were identified (p<0.05), of which 8 were present in both sets (p<0.015). These 8 proteins were subjected to hierarchical clustering and principal components analysis. Based upon this 8-protein profile, PD and OR samples clustered into two separate groups that correctly predicted therapy-response. The sensitivity and specificity of this protein profile will be determined in an independent validation set.

In summary, an 8-protein profile observed to predict tamoxifen-resistance in breast cancer was revealed by ultra-sensitive nanoLC-FTICR technology for comprehensive proteome analyses of LCM cells. This profile may harbor biomarkers that are useful in the clinic.

98th AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA