Background: Epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs), developed for patients with advanced non-small cell lung cancer (NSCLC), give modest results similar to those with chemotherapy. There is evidence of a greater survival benefit from TKIs in patients with certain molecular and clinical features, but results are conflicting. To assess the role of these factors in predicting TKI efficacy, a pooled analysis was performed on data from randomized trials in NSCLC.
Materials and methods: An electronic search of all randomized trials comparing the efficacy or activity of TKIs and a pooled analysis were performed. The hazard ratio (HR) with 95% confidence interval (CI) was calculated for each level of the factors and an interaction test was used to detect differences in treatment effect related to the different levels.
Results: Of ten randomized trials identified, five were considered for analysis. Smoking was shown to be the only clinical factor to have a predictive effect (non smokers: overall survival (OS) HR 0.53, 95% CI 0.42-0.67; smokers: HR 0.91, 95% CI 0.81-1.02; p-value for interaction <0.001). A negative predictive value was suggested for K-ras mutations (K-ras(+): HR 1.97 95% CI 1.16-3.33; K-ras(-): HR 0.79, 95% CI 0.59-1.05; p-value for interaction 0.003).
Conclusion: At the present time, none of the biological features which have been evaluated in patients who have undergone therapy using TKIs is proven to be of predictive value; only K-ras mutations and smoking habits can be considered as a possible criteria for selection. Results of prospective randomized trials on biological markers are awaited.