Abstract
Tumor heterogeneity greatly limits personalized treatment of cancer. Patient-derived tumor cell (PDC) models precisely recapitulate the molecular properties and biology of the disease, making them effective preclinical tools for assessing anti-cancer drug activities. Accurate estimation of tumor purity is essential for performing high-throughput drug screening (HTS). In the present study, we measured and predicted the tumor population index in PDC models for two-drug combinational strategies using HTS system. Gastric cancer cell-lines and PDCs were subjected to multi-color immunofluorescence analysis against EpCAM and vimentin to evaluate the tumor cell index based on EpCAM expression levels. We generated a tumor purity prediction model using five different gastric cancer cell-lines (AGS, KATO-III, MKN-45, NCI-N87, SNU-216) with fluorescence intensity-based techniques. Afterwards, stage IV gastric cancer PDC models were evaluated using a micropillar/microwell chip-based HTS system. HER2/CCNE1-amplified PDCs were considerably resistant to an HER2 inhibitor, while combinational treatment consisting of an HER2 inhibitor with anti-WEE1 compound substantially suppressed tumor cellular growth. Moreover, PDCs with BRCA1/2 mutations were synergistically sensitive to HER2 and PARP inhibition therapy. Finally, somatic mutations in TP53 and CDKN2A with MYC amplification rendered PDCs susceptible to the drug combination of WEE1 and HER2. Collectively, our systematic method of high-throughput drug sensitivity screening is an integral pre-clinical platform for evaluating potential two-drug combinational approaches for personalized treatment of cancer.