Study of TARCEVA (Erlotinib) as Adjuvant Treatment for Locally Advanced Head and Neck Squamous Cell Carcinoma

This trial was originally designed and powered to compare biomarker modulation in the neo-adjuvant setting (erlotinib versus erlotinib plus sulindac versus placebo) with clinical response to erlotinib in the adjuvant setting. Since implementing the trial in late 2005, The investigators have encountered significant obstacles to implementing the adjuvant therapy phase of the trial. * Barriers included: 1. disease recurrence 2. patient refusal to take the agent 3. patient refusal to travel to Pittsburgh for clinical evaluations. Given the institutional challenges to implement and complete the adjuvant portion, the investigators have decided to change the primary endpoint to a biomarker modulation endpoint. To achieve this goal, the investigators determined that they needed 39 paired tissue specimens (see statistical justification below). The central hypothesis to be tested in this study is that persistent activation of parallel and/or downstream pathways contributes to tumor progression in the setting of EGFR blockade. While not all head and neck squamous cell carcinoma (HNSCC) patients will respond to EGFR targeting, the optimal strategy to identify those subjects whose tumors are sensitive to EGFR inhibition remains unknown. The primary objective is centered around the concept of tumor biomarkers which may be modulated by EGFR and Cox-2 inhibitors and may serve as future therapeutic targets for therapy. To this end patients on this trial will be randomly assigned to one of three arms to receive either Tarceva, Tarceva plus sulindac, or a placebo in the 2 week pre-operative period. A panel of biomarkers will be obtained by biopsy prior to pre-operative therapy and again at surgery. Biomarkers will be examined for modulation in the 2-week pre-operative period, for group differences, for treatment effects and for further understanding of protein signaling pathways. Sample size for the primary objective Modification of Statistical Design: The primary endpoi