Use of immunohistochemical evaluation of DNA repair proteins to demonstrate cisplatin response prediction in resected NSCLC squamous cell carcinoma.

10579 Background: Platinum-based chemotherapy remains the standard of care for many solid tumor malignancies including non-small cell lung cancer (NSCLC). The mechanism of action is related to the accumulation of DNA adducts that in turn activate pro-apoptotic pathways. An intense area of focus is the identification of molecular biomarkers of DNA adduct repair that may be predictive/prognostic for treatment of NSCLC patients. A complex network of compensating DNA repair pathways exists so interrogation of multiple DNA repair markers may be more informative in predicting cisplatin efficacy than individual markers. Additionally, biomarker expression may be utilized to define treatment of histological subclasses of NSCLC. Methods: International Adjuvant Lung Cancer Trial (IALT) NSCLC FFPE patient specimens constructed on TMAs were stained by IHC for DNA repair biomarkers: ATM, MSH2, ERCC1, p53, pMK2, PARP1, BRCA1, XPF. An average of 603 patients were analyzed for each. Tumor biomarker nuclear or cytoplasmic levels were determined using digital image user defined macros. Scores were generated based on weighted intensity and quantity of stained cells. Cox PH models adjusted for relevant clinical and stratification variables were used in the univariate analyses of DFS. Results: The univariate biomarker analyses yielded significant prognostic and predictive values using disease-free survival (DFS) as the primary endpoint. This exploratory data support predictive modeling of DNA repair enzyme expression levels that cosegregate with SCC but not adenocarcinoma, and include MSH2 (predictive p = 0.012, HR=1.218, 95%CI=[1.044-1.42]), p53 (p=0.005, HR=1.120, 95%CI=[1.035-1.212],) and ATM (p=0.010, HR=1.212, 95%CI=[1.046-1.405]). Partition models for pMK2, p53, ERCC1, ATM, and PARP1 were statistically significant for prediction in SCC but not adenocarcinoma. Conclusions: These data suggest that expression of DNA repair enzymes represent a distinct molecular difference between SCC and adenocarcinoma and that application of biomarker modeling may be predictive for specific tumor histological subclasses.