A phase II study for selection of clinical and molecular predictors for survival in patients (pts) with advanced NSCLC treated with erlotinib

17066 Background: The primary objective of the ongoing phase II study was to find clinical and molecular predictors for survival. Consequently, we prospectively examined gene and protein expression and EGFR mutation profiles of tumor cells, and clinical variables. Methods: Pts with advanced NSCLC (stage IIIb/IV) previously treated with platinum-based chemotherapy received 150mg/day erlotinib (Tarceva an EGFR tyrosine-kinase inhibitor) orally until disease progression. For molecular analyses, biopsies were taken from the primary lung tumor prior to erlotinib treatment. Pure tumor cells were isolated using laser-capture microdissection. For expression analysis, a microarray of 8793 genes (Human Focus Array, Affymetrix) was used. Data were analyzed using VSN method and significance analysis of microarrays. Results: In an interim analysis, clinical data were available for 36 of 43 so far included pts in the ongoing phase II study. For 36 pts, the tumor responses were: 1 CR (3%), 4 PR (11%), 11 SD (30%) and 19 PD (53%), and the median survival was 20 wks. Overall survival was significantly better in pts with rash than without rash: 9.2 vs 4.3 months. In a 14 months follow up, all of the responding pts survived while the median survival for pts with SD or PD was 11.4 and 4.0 months, respectively. Comparing gene expression profiles of tumors from 6 pts with a response or SD with those from 13 pts with tumor progression, we could not identify genes that were differentially expressed with sufficient significance. In 20 tumor samples evaluated for EGFR protein expression and mutation analysis, 19 expressed EGFR (IHC). A base exchange in exon 18 (2166G>A; A722A) and 20 (2547A>G; Q787Q) of the EGFR already described as a single nucleotide polymorphism was seen in 4 samples performing SSCP gel electrophoresis and sequencing. Conclusions: Rash, a side effect with erlotinib, was significantly associated with better survival. No activating EGFR mutations were found indicating that the response in 14% of pts is independent of mutations. For identification of genes predictive for response to erlotinib, a larger sample number is necessary. No significant financial relationships to disclose.