Functional personalized complex combination nano therapy for osteosarcoma

Osteosarcoma (OS) remains a challenging malignancy, particularly for metastatic cases, due to its heterogeneous genetic landscape characterized by lack of actionable oncogenic drivers and thus lack of personalized therapy. Though personalization of therapy based on functional assays using patient derived cells is emerging as a promising approach, it was not explored as a workflow for identifying patient specific drug cocktails. In this study we investigated a sequential drug screening approach for single and pair drugs which can be rationally used to identify 4 drug cocktails. We started with a systematic screening of 17 drugs from multiple classes across four osteosarcoma cell lines (U2OS, MG-63, SaOS-2, and K7M2), and observed differential drug responses among the cell lines. Interestingly, leading large language models (LLMs) failed to predict cell specific efficacy in OS cells while they were successful in KRAS mutation driven cells. Both SaOS-2 and K7M2 showed sensitivity to kinase inhibitors, particularly ponatinib, and we further explored their combinatorial therapeutic potential. We identified promising non-chemotherapeutic drug pairs, including trametinib-ponatinib and rapamycin-ganetespib and demonstrated potent synergistic effects. To mitigate dose limiting toxicities, drug pairs were formulated in polydopamine-stabilized nanoparticles and K7M2 murine models in vivo studies revealed that they preferably accumulated in tumors and were highly superior to the chemotherapeutic standard of care. Surprisingly, alternating administration of nanoparticle drug pairs was superior to concomitant regimen in both efficacy, survival and toxicity profiles. These findings strengthen a functional approach for combinatorial personalized treatment, potentially overcoming the limitations of current therapeutic strategies.

Drug combinations; Large language models; Nanomedicine; Osteosarcoma; Personalized medicine; Polydopamine.