Rational Design of Metal-Organic Frameworks for Pancreatic Cancer Therapy: from Machine Learning Screening to In Vivo Efficacy.

Abstract

Despite improvements in cancer survival rates, metastatic and surgery-resistant cancers, such as pancreatic cancer, remain challenging, with poor prognoses and limited treatment options. Enhancing drug bioavailability in tumors, while minimizing off-target effects, is crucial. Metal-organic frameworks (MOFs) have emerged as promising drug delivery vehicles owing to their high loading capacity, biocompatibility, and functional tunability. However, the vast chemical diversity of MOFs complicates the rational design of biocompatible materials. This study employed machine learning and molecular simulations to identify MOFs suitable for encapsulating gemcitabine, paclitaxel, and SN-38, and identified PCN-222 as an optimal candidate. Following drug loading, MOF formulations are improved for colloidal stability and biocompatibility. In vitro studies on pancreatic cancer cell lines have shown high biocompatibility, cellular internalization, and delayed drug release. Long-term stability tests demonstrated a consistent performance over 12 months. In vivo studies in pancreatic tumor-bearing mice revealed that paclitaxel-loaded PCN-222, particularly with a hydrogel for local administration, significantly reduced metastatic spread and tumor growth compared to the free drug. These findings underscore the potential of PCN-222 as an effective drug delivery system for the treatment of hard-to-treat cancers.