Abstract 375P
Background
Rezivertinib (BPI-7711) is a novel, irreversible third-generation EGFR-TKI. Despite the efficacy of EGFR-TKIs in treating NSCLC, resistance invariably develops. This study utilized multi-omics analyses in BPI-7711-resistant cell lines and integrated plasma metabolomics to elucidate resistance mechanisms.
Methods
We established a BPI-7711-resistant cell line, HCC827R, from the HCC827 cell line. Multi-omics approaches were used to identify metabolic pathways related to resistance. Additionally, we collected baseline plasma samples from 186 locally advanced or metastatic NSCLC patients enrolled in the phase I and IIa studies of BPI-7711 (NCT03386955) for metabolic analysis.
Results
The analysis revealed a distinctive metabolic profile associated with EGFR-TKI resistance. Key metabolic pathways, such as phosphatidylcholine catabolism and lipoprotein particle receptor activity, were notably enriched in HCC827R cells. In plasma samples, glycerophospholipids, particularly lysophospholipids, were more prevalent in the non-responder (NR) group, while fatty acyls dominated the responder (R) group. NMF clustering identified two patient clusters, with Cluster 1 exhibiting higher lysophospholipid and lower fatty acid levels, correlating with poorer OS (P=0.0244). Univariate Cox analyses and the LASSO algorithm further identified prognostic metabolite biomarkers, yielding ROC-AUC values of 0.7762 and 0.7485 in the training (n=131) and test (n=55) cohorts, respectively. A multivariate model adjusted for confounding factors remained significant in predicting therapeutic efficacy. Based on median risk scores from the training cohort, patients were grouped into high- and low-score categories, with K-M survival analysis indicating significantly better OS in the low-score group for both training (P=0.0014) and test cohorts (P=0.055).
Conclusions
This study elucidated unique metabolic resistance mechanisms to BPI-7711 in EGFR-mutant NSCLC. The prognostic models developed from identified metabolites demonstrated strong predictive capabilities, underscoring the potential of integrating metabolomics with clinical parameters to enhance treatment strategies.
Legal entity responsible for the study
National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital
Funding
New National Natural Science Foundation of China (81972805)
Disclosure
All authors have declared no conflicts of interest.