Pharmacogenomic (PG) analysis of Japan-SWOG common arm study in advanced stage non-small cell lung cancer (NSCLC): A model for testing population-related pharmacogenomics

Abstract

7500 Background: We have previously reported differences in outcomes (increased survival, neutropenia & febrile neutropenia) in Japanese versus US patients (pts) treated with a paclitaxel-carboplatin “common arm” in a prospectively designed fashion in 3 phase III trials (FACS, LC03, S0003) in advanced NSCLC (Gandara, ASCO 05 & Crowley, ASCO 06). We hypothesized that these findings were due in part to PG alterations in paclitaxel disposition. Methods: Genomic DNA was prospectively collected from pts in 2 of these phase III trials (LC03 [N=78, 37 on common arm] & S0003 [N=78]) with identical eligibility, staging, treatment plan (paclitaxel 225 mg/m2 & carboplatin AUC 6), response & toxicity criteria. Analysis for genotypic variants of CYP3A4, CYP3A5, CYP2C8, NR1I2–206, ABCB1, ERCC2 was performed by pyrosequencing, and results assessed by Cox model for survival/PFS & logistics regression for response/toxicity. Results: There was a significant difference between Japan & US pts in genotypes: CYP3A4*1b (p=0.01), CYP3A5*3c (p=0.03), ERCC2 k751q (p <0.001), and CYP2C8 r139k (p=.01). Genotypic correlations were observed between CYP3A4*1b for PFS (HR 2.75, 1.06–7.08, p=0.04) & ERCC2 k751q for response (HR 0.33, 0.13–0.84, p=0.02). There were no other statistically significant associations, although for grade 4 neutropenia, the HR for ABCB1 3425c->T was 1.84 (0.77–4.48), p=0.19. The low number of events for febrile neutropenia within this data set precluded assessment of this parameter. Additional PG testing is ongoing & will be presented. Conclusions: 1) Differences in allelic distribution for genes involved in paclitaxel disposition or DNA repair were observed between Japanese & US pts. 2) Statistically significant genotype-associated correlations were present for PFS (CYP3A4*1b) & response (ERCC2 k751q), but not for neutropenia (p=0.19). 3) The small sample size limits interpretation of these data. Further studies based on this common arm approach are warranted for the prospective study of population-related PGs where ethnic/racial differences in anti-neoplastic drug disposition are anticipated. [Table: see text]