Influence of germline variations in drug transporters ABCB1 and ABCG2 on intracerebral osimertinib efficacy in patients with non-small cell lung cancer.

Patients who started treatment with osimertinib for EGFRm+ NSCLC between November 2014 and June 2021 were included in this retrospective observational multicentre cohort study. For patients with baseline CNS metastases, the primary endpoint was CNS progression-free survival (CNS-PFS; time from osimertinib start until CNS disease progression or death). For patients with no or unknown baseline CNS metastases, the primary endpoint was CNS disease-free survival (CNS-DFS; time from osimertinib start until occurrence of new CNS metastases). Relationships between SNPs and baseline characteristics with CNS-PFS and CNS-DFS were studied with competing-risks survival analysis. Secondary endpoints were relationships between SNPs and PFS, overall survival, severe toxicity, and osimertinib pharmacokinetics.

Central nervous system (CNS) metastases are present in approximately 40% of patients with metastatic epidermal growth factor receptor-mutated (EGFRm+) non-small cell lung cancer (NSCLC). The EGFR-tyrosine kinase inhibitor osimertinib is a substrate of transporters ABCB1 and ABCG2 and metabolized by CYP3A4. We investigated relationships between single nucleotide polymorphisms (SNPs) ABCB1 3435C>T, ABCG2 421C>A and 34G>A, and CYP3A4∗22 and CNS treatment efficacy of osimertinib in EGFRm+ NSCLC patients.

ABCG2 34G>A and ABCB1 3435C>T are predictors for developing new CNS metastases during osimertinib treatment, probably because of diminished drug levels in the CNS. ABCG2 421C>A was significantly related with the incidence of severe toxicity. Pre-emptive genotyping for these SNPs could individualize osimertinib therapy. Addition of ABCG2 inhibitors for patients without ABCG2 34G>A should be studied further, to prevent new CNS metastases during osimertinib treatment.

From 572 included patients, 201 had baseline CNS metastases. No SNP was associated with CNS-PFS. Genotype ABCG2 34GA/AA and/or ABCB1 3435CC --present in 35% of patients-- was significantly associated with decreased CNS-DFS (hazard ratio 0.28; 95% CI 0.11-0.73; p = 0.009) in the multivariate analysis. This remained significant after applying a Bonferroni correction and internal validation through bootstrapping. ABCG2 421CA/AA was related to more severe toxicity (27.0% versus 16.5%; p = 0.010).

No funding was received for this trial.