Machine-Learning-Accelerated Design of Ternary Carrier-Free Nanomedicine for Intranasal Therapy of Brain Metastatic Non-small-cell Lung Cancer

TL;DR

A machine learning-guided nanomedicine was developed for intranasal treatment of brain metastatic lung cancer, showing improved delivery and immune response.

Contribution

A machine learning platform was used to design a ternary, carrier-free nanodrug for targeted cancer therapy.

Findings

DTL nanodrug achieved efficient delivery to both primary and metastatic brain tumors via intranasal administration.

DTL induced multiple forms of tumor cell death and remodeled the immune microenvironment.

DTL inhibited tumor growth in vivo with reduced toxicity compared to conventional therapies.

Abstract

Non-small-cell lung cancer (NSCLC) with brain metastases poses formidable therapeutic challenges due to acquired resistance and the inherent pharmacokinetic defects of traditional delivery. We developed an innovative lipoic acid-based self-assembled nanodrug (dabrafenib, trametinib, and lipoic acid self-assembly [DTL]) system, whose rational design was guided by a novel machine learning platform to overcome high-cost, empirical screening bottlenecks. Multifunctional lipoic acid, serving as a universal self-assembling molecule, enabled DTL’s robust assembly and enhanced penetration across mucosal and solid tumor barriers via its unique thiol-mediated exchange mechanism while simultaneously exerting distinct antitumor efficacy. Intranasal administration of DTL achieved efficient dual-targeted delivery to both primary NSCLC and established intracranial metastases. Furthermore, compared to conventional targeted combination therapies, DTL induced diverse, multimodal tumor cell death (apoptosis, pyroptosis, and ferroptosis) and profoundly remodeled the immune microenvironment. In vivo, DTL markedly inhibited tumor growth with reduced toxicity, offering a clinically translatable strategy for advanced NSCLC.

Figures