Large Language Model-Assisted Genotoxic Metal-Phenolic Nanoplatform for Osteosarcoma Therapy

Area of Science:

• Biomaterials Engineering
• Oncology
• Artificial Intelligence in Medicine

Background:

• Osteosarcoma is a primary bone cancer in children and adolescents with poor prognosis and fertility concerns.
• Developing effective and targeted therapies for osteosarcoma remains a significant challenge.

Purpose of the Study:

• To leverage a large language model (LLM)-assisted phenolic network (LLMPN) platform for designing novel anti-osteosarcoma therapeutics.
• To identify potent natural compounds and engineer them into targeted nanomedicines for osteosarcoma treatment.

Main Methods:

• Utilized GPT-4, a large language model, to screen 60 polyphenols for anti-osteosarcoma activity.
• Identified gossypol as a promising candidate and synthesized Cu²⁺-gossypol nanocomplexes with hyaluronic acid surface modification (CuGOS NPs).
• Evaluated the therapeutic efficacy and safety of CuGOS NPs in osteosarcoma models.

Main Results:

• GPT-4 successfully identified gossypol as a superior anti-osteosarcoma compound.
• CuGOS NPs demonstrated significant induction of genetic alterations and cell death in osteosarcoma cells.
• The treatment showed therapeutic benefits against primary tumors and reduced metastasis without systemic toxicity or impact on fertility.

Conclusions:

• An LLM-driven approach enables efficient discovery and engineering of metal-organic nanoplatforms for cancer therapy.
• Gossypol-based CuGOS NPs represent a promising, targeted therapeutic strategy for osteosarcoma with a favorable safety profile.
• This work highlights the potential of LLMs to accelerate research and development in nanomedicine and oncology.