Background: Erlotinib shows large inter-patient pharmacokinetic variability, but the impact of early drug exposure and genetic variations on the clinical outcomes of erlotinib remains fully investigated. The primary objective of this study was to clarify the population pharmacokinetics/pharmacodynamics of erlotinib in Japanese patients with non-small cell lung cancer (NSCLC). The secondary objective was to identify genetic determinant(s) for the cerebrospinal fluid (CSF) permeability of erlotinib and its active metabolite OSI-420.
Methods: A total of 88 patients treated with erlotinib (150 mg/day) were enrolled, and CSF samples were available from 23 of these patients with leptomeningeal metastases. Plasma and CSF concentrations of erlotinib and OSI-420 were measured by high-performance liquid chromatography with UV detection. Population pharmacokinetic analysis was performed with the nonlinear mixed-effects modelling program NONMEM. Germline mutations including ABCB1 (1236C>T, 2677G>T/A, 3435C>T), ABCG2 (421C>A), and CYP3A5 (6986A>G) polymorphisms, as well as somatic EGFR activating mutations if available, were examined. Early exposure to erlotinib and its safety/efficacy relationship were evaluated.
Results: The apparent clearance of erlotinib and OSI-420 were significantly decreased by 24 and 35 % in patients with the ABCG2 421A allele, respectively (p < 0.001), while ABCB1 and CYP3A5 polymorphisms did not affect their apparent clearance. The ABCG2 421A allele was significantly associated with increased CSF penetration for both erlotinib and OSI-420 (p < 0.05). Furthermore, the incidence of grade ≥2 diarrhea was significantly higher in patients harboring this mutant allele (p = 0.035). A multivariate logistic regression model showed that erlotinib trough (C0) levels on day 8 were an independent risk factor for the development of grade ≥2 diarrhea (p = 0.037) and skin rash (p = 0.031). Interstitial lung disease (ILD)-like events occurred in 3 patients (3.4 %), and the median value of erlotinib C0 levels adjacent to these events was approximately 3 times higher than that in patients who did not develop ILD (3253 versus 1107 ng/mL; p = 0.014). The objective response rate in the EGFR wild-type group was marginally higher in patients achieving higher erlotinib C0 levels (≥1711 ng/mL) than that in patients having lower erlotinib C0 levels (38 versus 5 %; p = 0.058), whereas no greater response was observed in the higher group (67 %) versus the lower group (77 %) within EGFR mutation-positive patients (p = 0.62).
Conclusions: ABCG2 can influence the apparent clearance of erlotinib and OSI-420, and their CSF permeabilities in patients with NSCLC. Our preliminary findings indicate that early exposure to erlotinib may be associated with the development of adverse events and that increased erlotinib exposure may be relevant to the antitumor effects in EGFR wild-type patients while having less of an impact on the tumor response in EGFR mutation-positive patients.