Units

Keywords

Contact

---

Citation

Type

Conference Name

Publisher Proceedings

Publication Date

Pages

Publisher

Brief description/objective

Rationale
We investigated predictors of treatment outcomes for response to BE, based on data from a recent phase II trial testing
first-line BE in unselected, advanced, non-squamous NSCLC patients (SAKK 19/05). We analyzed expression variations
at the exon-level of 3 genes potentially playing a key role in modulating treatment response: epidermal growth factor
(EGFR), V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) and vascular endothelial growth factor (VEGFA).
Methods
Treatment-naive bronchoscopic biopsies were obtained in 49 out of 101 evaluable patients, from whom mRNA material
was extracted. After quality assessment, 42 samples were hybridized on Affymetrix exon arrays. Blood samples from 75
patients were investigated to compare and validate our tumor biopsy findings. Endpoints included tumor shrinkage at week
12 (TS12) after BE, time-to-progression under BE and overall survival. Correlation between exon-level expression and
continuous endpoints was assessed using Spearman's rho statistics. Receiver operating characteristic (ROC) curves
estimated the sensitivity, specificity and accuracy of classifiers based on exonic probesets intensities. Gene expression
information included in several exonic probesets was summarized using principal component analysis.
Results
Overall, most exonic EGFR-probesets (86%) showed significant correlation with TS12 (p<0.05). The score on the first
principal component, summarizing all exon-level intensity information within EGFR, showed significant correlation with
TS12 (Spearman's rho=0.50; p=0.006). A relevant heterogeneity existed among different EGFR exons. Exon 18
expression showed the highest association with TS12, which remained significant after Bonferroni correction (Spearman's
rho=0.69, p<0.0001) (Figure 1, left panel), and which was independent of EGFR- and KRAS-mutational status. This
marker provided 75% classification accuracy of BE responders at a cut-off value fixed to the median expression intensity
of exon 18. The area under ROC curve was 93% (Figure 1, central panel). A waterfall plot shows the change in tumor size
at week 12 (Figure 1, right panel). Exon 18-EGFR signal was validated using peripheral blood samples of 75 patients. No
association was found between exonic expression of KRAS and VEGFA and response to BE.
Conclusions
EGFR-expression analyses on an exon-level showed heterogeneity among different exons. Exonic EGFR expression -,
particularly in exon 18 -, was found to be a relevant predictive biomarker for response to BE and identified an EGFR
wild-type or KRAS-mutated subgroup who benefits from BE. The association of exon 18-EGFR expression and TS12 was
significant in peripheral blood as well, although with some signal "dilution". The current results warrant further testing in an
independent prospective randomized controlled trial.