Background: BR.21 is a double-blind, placebo-controlled trial of second-/third-line erlotinib in stage IIIB/IV non-small cell lung cancer patients. Predictive and prognostic analyses of epidermal growth factor receptor (EGFR), ABCG2, and AKT1 genetic polymorphisms were performed.
Methods: Two hundred forty-two patients were genotyped for EGFR-216G>T (EGFR216), EGFR-191C>A (EGFR191), EGFR intron 1 CA-dinucleotide-repeat (CADR), ABCG2+421C>A (ABCG2), and AKT1-SNP4G>A (AKT1). Cox proportional hazard and logistic regression models compared genotypes with overall survival (OS), progression-free survival (PFS), and presence/absence of skin toxicity.
Results: Prognostic evaluation was based on the placebo arm: patients carrying at least one CADR long allele (>16 repeats) had a trend toward worse PFS: the adjusted hazard ratio was 1.7 (95% confidence interval [CI]: 1.0-3.0; p = 0.07). EGFR216, EGFR191, ABCG2, and AKT1 were not prognostic. Polymorphisms were not predictive for erlotinib effect (OS/PFS): no treatment-polymorphism interactions were demonstrated. Individuals carrying the rare T/T genotype of EGFR216 had an adjusted odds ratio of 8.8 (95% CI: 1.1-72; p = 0.04) of developing skin toxicity; no other significant polymorphic relationships with skin toxicity were found.
Conclusions: In contrast to previous publications, carrying shorter alleles of the EGFR CADR polymorphism was not predictive of OS or PFS. EGFR216 homozygous variants were associated with greater skin toxicity from erlotinib.