Primary lung squamous cell carcinomas (SCC) form a distinct group in terms of molecular characteristics. There is no single assay that predicts outcome and selects optimal treatment for early stage SCC, especially for stage I where there is a 33-60% risk of disease-related death within 5 years after “curative” surgical treatment. The current staging system fails to select early stage SCC patients likely to recur and die of their disease after surgery. Emerging data using gene-expression profiling has shown that lung SCCs can be classified into high and low risk groups by signatures associated with down-regulation of epidermal development genes. However, the development of such molecular tools has been limited by lack of reproducibility and independent validation. Here we develop and validate a 3-protein classifier that stratifies early stage SCC by risk.

Quantitative immunofluorescence (AQUA®) was employed to measure proteins known to be part of SCC gene risk signatures; Cytokeratins 5 (CK5), 8 (CK8), 13 (CK13), 14 (CK14), 15 (CK15), 16 (CK16), 17 (CK17), 18 (CK18) and 19 (CK19) were analyzed in a multivariate Cox proportional hazards regression model following stepwise selection with backward elimination resulting in a 3-protein weighted classifier. A risk score was calculated combining binarized quantitative measurements of CK13, CK14 and CK17 in a training set of 160 SCC patients. X-tile software was employed to identify the optimal cut off risk score and classify patients into high and risk groups; this cut point was applied to an independent cohort of 76 SCC patients.

Weighted expression of CK13, CK14 and CK17 was used to build a model to classify patients in high and low risk groups. Survival analysis showed that patients in the low risk group had a longer median overall survival compared to the high risk group (46 vs. 28 months, log rank p=0.004). Multivariate analysis revealed an independent increased risk of death for all SCC patients with high scores (HR=1.76, 95% CI 1.28-2.5, p=0.001). Then the 3-biomarker classifier was tested on an independent cohort of 76 SCC patients which showed a 30.1 vs. 55 month median survival for high and low risk patients respectively (log rank p=0.001). Multivariate analysis in this cohort adjusting for age, gender and stage verified the significant prognostic role of the classifier (HR=1.88, 95% CI 1.22-2.9, p=0.004). The independent prognostic value of our model was especially valuable in stage I patients (HR=1.91, 95% CI 1.15-3.17, p=0.013) and even more compelling for stage I patients that were treated with surgery only (HR=2.92, 95% CI 1.31-6. 25, p=0.009).

We developed and validated a quantitative biomarker-based independent prognostic model that accurately stratifies SCC patients for risk of disease specific death. Our classifier could be used to determine which stage I SCC patients can be treated with surgery alone and which would benefit from adjuvant chemotherapy.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 4643.