Validation of 5- and 3-gene expression signatures for predicting outcome in non-small cell lung cancer (NSCLC) patients (p)

Abstract

7532 Background: A number of prognostic gene expression signatures have been reported to predict survival in lung cancer, including a 5-gene signature (Chen et al, N Engl J Med 2007) and a 3-gene signature in squamous cell carcinoma (SCC) (Skrzypski et al. ASCO 2007). We have examined these two signatures in tumors from 142 NSCLC p. Methods: From 2000 to 2005, freshly frozen primary tumor specimens were obtained at the time of surgery from 142 NSCLC p (80% stage I and II; 66% SCC, 33% adenocarcinomas). Expression of genes included in the 5-gene signature (MMD, ERBB3, LCK, STAT1, DUSP6) and the 3-gene signature (CSF1, EGFR, CA IX) was assessed by quantitative RT-PCR in low density arrays, using TBP as endogenous control. Risk scores were used to classify p into high- or low-risk according to the expression of the genes in the two signatures. Results: All genes included in the 5-gene signature were more highly expressed in adenocarcinoma than in SCC, leading us to adjust the risk score for histology. According to the original risk score for the 5-gene signature, median survival (MS) for low-risk p was not reached while it was 29.7 months (m) for high-risk p (P=0.22). However, using the adjusted risk score for SCC, MS for low-risk p was 63 m vs 24 m for high-risk p (P=0.003). Using the adjusted risk score for adenocarcinoma, MS for low-risk p was not reached, while it was 46.2 m for high-risk p (P=0.006). According to the risk score for the 3-gene signature, MS for high-risk p was 41 m vs 62.4 m for low-risk p (P=0.009). For stage I-II SCC p, MS in low-risk p was not reached, while it was 34 m for high-risk p (P=0.05). For adenocarcinoma p, MS was not reached for low-risk p, while it was 61.2 m for high-risk p (P=0.04). Conclusions: The 5-gene signature is highly predictive of survival when adjusted for histology. The 3-gene signature is predictive of survival without adjustment for histology. These gene signatures can be useful for selecting high-risk p for adjuvant chemotherapy. No significant financial relationships to disclose.