OBJECTIVE:

The overall goal of this study is to develop independent and significant predictive measures of tumor behavior for head and neck squamous cell carcinoma (HNSCC).

METHODS:

In our systems biology approach to study HNSCC, TRIzol® extracted tumors yield DNA, RNA and protein from the same piece of tissue. The proteins were purified and peptides resulting from chemical and enzymatic digestion of the proteins were analyzed with two dimensional liquid chromatography-mass spectrometry (2D LC-MS). The methodology utilized has been validated for its reproducibility and high stringency (Sudha et al. Laboratory Investigation, 2007). Acquired data were used to generate arrays for statistical analysis through an extensive optimization process. This procedure has resulted in peptides that discriminate HNSCC patients and are potential prognostic for HNSCC.

RESULTS:
We have performed 2D LC-MS analysis of 125 tumor samples from each of three sites of HNSCC including, 37 oropharynx, 43 oral cavity and 45 larynx human specimens. Once acquired, arrays were generated through optimized preprocessing steps including peak extraction and feature alignment. Resulting global proteomic profiles have been used to discriminate/classify HNSCC with different clinical outcomes. Initially, the entire data set was analyzed irrespective of tumor sites and 15 peptides were found to be correlated with clinical outcome such as node+ status, local-regional recurrence, late-stage disease (III/IV), time to disease progression or distant metastasis, and disease-specific survival.
However, site-specific prognostic signatures were found to be stronger than prognostic signatures derived from a heterogeneous mix of anatomical sites during our studies on global RNA expression (Belbin et al. Head Neck Pathology, 2008). Accordingly, we chose to re-analyze the oral cavity data set first. Discriminators were generated from the statistical modeling and are being identified with tandem mass spectrometry. Once, identified, a clinical mass spectrometry assay will be developed in collaboration with our GCRC clinical research laboratory.
In the meantime, we have found that loss of miR-375 in HNSCC is significantly associated with poor cancer specific survival (Harris et al. Amer. J Pathol, 2011 in press). 90 potential targets have been compiled from the following: global RNA expression studies comparing miRNA expressing cells and control cells, a prediction database, and the literature. The targets have been searched against our global proteomic profile for oral cavity and produced several biologically relevant proteins as matches. We are now conducting studies to confirm the expression of these invasion associated proteins in the primary tumor specimens.

CONCLUSION:

The results strongly suggest that a small number of peptides can be used to develop a predictive model that can discriminate disease severity at initial diagnosis as well as clinical outcome prospectively. RNA expression profiling of primary HNSCC specimens has also been performed. We will integrate the proteomic data with the RNA expression data to identify the best potential prognostic biomarkers to develop clinical diagnostics. In addition, the miR-375 target study integrated with global proteomic data demonstrates the significant potential of the proteomic profiles as a database which can be utilized for the study of proteins involved in HNSCC.