Lung cancer is the leading cause of cancer deaths in the developed world with a 5-year survival rate of 15%; thus, there is a critical need for better treatment options, as well as better prognostic and classifying biomarkers. Here we report objective, quantitative expression analysis of EGFR (HER1), HER2, HER3, HER4, Erk1/2, Akt1, and Stat3 on a tissue microarray of NSCLC using automated quantitative analysis (AQUATM). The array is comprised of 213 NSCLC tumors at two-fold redundancy [median survival: 26.8 months; 17.8% Stage 0; 43.7% Stage I; 13.1% Stage II; 15.5% Stage III; 9.9% Stage IV], 20 normal lung samples, and 7 cell line controls. We observed variable expression of all markers, but did not observe significant correlation between marker expression level and tumor stage. For survival analysis of each marker, optimal cut points were generated using X-tile1, and significance was determined using Monte Carlo simulations, a robust statistical analysis that is appropriate for generation of optimal cut points for continuous data. HER1, HER2, and Stat3 expression were associated with a statistically significant decrease in 3-year disease specific survival [HER1 expression (top 20%): 25% decrease in survival, Monte Carlo p = 0.02; HER2 expression (top 15%): 22% decrease in survival, Monte Carlo p < 0.0001; Stat3 expression (top 15%): 22% decrease in survival, Monte Carlo p = 0.03]. Furthermore, optimal cut-points were generated on training sets and subsequently applied to a validation set within the cohort with significance for HER1 (p=0.02), HER2 (p=0.005), and Stat3 (p = 0.03), thus adding to the statistical rigor of these analyses. In contrast, Erk1/2 expression was associated with increased survival [top 15%: 24% increase in survival, Monte Carlo p = 0.04; training/validation p = 0.03). HER3, HER4, and Akt1 expression did not have a significant effect on survival. Taken together, this objective quantitative analysis confirms HER1 and HER2 expression to be of prognostic significance in NSCLC but adds the reproducibility of objective analy sis. Furthermore, expression of Stat3 led to decreased survival while expression of Erk1/2 led to increased survival, suggesting the importance of differential pathway activation in lung tumorigenesis. Although, expression of HER3, HER4, and Akt1 did not show an effect on overall survival, we anticipate that future studies multiplexing these markers with other HER family pathway members may demonstrate their possible involvement in NSCLC oncogenesis and allow for the development of sets of predictive and prognostic biomarkers. 1Camp et al. X-Tile: a new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. 2004 Clinical Cancer Research. 10(21) 7252-9.
[Proc Amer Assoc Cancer Res, Volume 47, 2006]