Proteomic Analysis of Barrett's Associated Neoplasia Identifies Protein Markers Correlating with Neoplastic Progression
JM Davison, MS Flint, BL Hood, BA Jobe, TP Conrads. U of Pittsburgh Med Ctr, Pittsburgh, PA
Background: Barrett's esophagus (BE) is a major risk factor for the development of esophageal adenocarcinoma (AdCA), a clinically aggressive malignancy which has dramatically increased in incidence during the past 20-30 years. The fact that >90% of patients with adenocarcinoma are discovered when the adenocarcinoma is advanced underscores the need for more effective identification of people who are at risk. Protein biomarkers associated with neoplastic progression from BE to AdCA may facilitate diagnosis.
Design: We harvested formalin-fixed paraffin-embedded tissues from 14 esophagectomy specimens and analyzed 10 samples of BE, 11 samples of high grade columnar dysplasia (HGD) and 10 samples AdCA. Epithelium and surrounding stroma were obtained by laser capture microdissection. Approximately 6x106 um2 of tissue (approximately 20,000 cells estimated) was collected for each sample. Tissue was then and analyzed in duplicate by nanoflow LC-MS/MS. The primary tandem MS data were searched against the human proteome database (UniProt) for peptide identification and subsequently analyzed to identify unique peptides. A protein's spectral count (i.e., the number of sequenced peptides per protein) served as a quantitative estimate of the abundance of proteins in a sample.
Results: 1. Over 2000 unique peptides were identified per LC-MS/MS analysis of 20,000 cells on column with an average relative standard deviation (RSD) of peptide identification rate of 15% across all injections, and an average of nearly 700 proteins identified per tissue sample. 2. Over 350 proteins showed significant differences in abundance between BE, HGD and AdenoCA tissue sample sets (Kruskal-Wallace non-parametric ANOVA with p < 0.05). 3. A supervised hierarchical cluster analysis incorporating proteins with significant differences in abundance correctly classified 10/10 BE samples, 10/11 HGD and 9/10 AdCA samples and showed patterns of relative protein abundance to correlate with tissue diagnosis. 4. This approach identified expected declines in abundance of CK20, MUC-5AC and MUC-2 between BE, HGD and AdenoCA. Proteins involved in signal transduction, cell cycle regulation, cell growth and differentiation, angiogenesis, matrix remodeling, cell motility, among other functions showed gradients of abundance between sample sets.
Conclusions: Mass spectrometric analysis of small samples of formalin-fixed tissue is capable of identifying gradients of protein abundance between BE, HGD and AdenoCA which are of likely biologic relevance in neoplastic progression.
Category: Pan-genomic/Pan-proteomic Approaches to Diseases
Tuesday, March 23, 2010 9:15 AM
Platform Session: Section H 1, Tuesday Morning