In Vitro Study on AI-PRS Enabled Precision Cocktail Drugs Design for Treating Human Colorectal Carcinoma

Colorectal cancer (CRC) is currently the third most common cancer in the world. Due to the development of treatment resistance, the efficacy of current chemotherapeutic agents against CRC has reached a plateau. Drug activity depends on the entire physiological response; therefore, drug-dose parameters cannot be designed efficiently by using conventional prediction-based methodologies. In this work, the AI-PRS (artificial intelligence-based phenotypic response surface) platform is successfully applied to find optimal drug-dose combinations in vitro from a pool of ten approved drugs. The AI-PRS platform optimizes effective drug-dose combinations without reference to molecular pathways or drug interaction data. With the aid of AI-PRS platform, efficient one, two, three, and four drug-dose combinations from in vitro studies are found. Of hundreds of combinations, regorafenib (R)/gemcitabine (G)/cetuximab (C)/5-fluorouracil (U) drug-dose combination exhibits the best activity on four CRC cell lines, two circulating tumor cells (CTCs), and one patient derived xenografts (PDX) cell lines. The three-drug combination of R/G/U shows the highest toxicity (70%) in the PDX cell line. Four-drug combination of R/G/C/U displays the best toxicity (80%) in in vitro cultured CTCs. The findings from the present derived cells reveal the prospective validation of the AI-PRS platform, which may help identify customized and highly efficient drug-dose combinations for future CRC treatment.