Establishment of a large-scale drug sensitivity screening platform on gastric cancer organoid cultures with potential to guide treatment

Abstract

Gastric cancer (GC) is the 3rd leading cause of cancer death worldwide and has an exceptionally high incidence in Asia, including China and Hong Kong. Currently, other than complete surgical resection of the tumour during early stage disease, there is no effective treatment for GC. There are a few approved chemotherapeutic agents, including 5-FU, cisplatin, epirubicin, paclitaxel, irinotecan or their combinations, along with several targeted drugs such as Trastuzumab and Ramucirumab. Unfortunately, most patients are diagnosed at late stages with dismal prognosis despite these treatments. There are also no good biomarkers guiding the efficacy of chemotherapy. Therefore, there is an urgent need to develop a reliable pre-clinical model for drug screening for precision medicine development. With recent advances in 3D organoid culture techniques, we have successfully generated a GC living biobank from 34 patients encompassing major molecular subtypes with detailed whole exome and transcriptome analysis. This living biobank serves as a valuable tool for drug sensitivity screening to guide patient treatment and identify new therapeutic opportunities. We aim to establish a large-scale drug sensitivity screening platform on our gastric organoid biobank. 37 Anticancer drugs were selected to treat 9 GC organoids derived from 7 patients. Each drug was screened at 7 drug concentrations in triplicate, and each run was repeated once using GC organoids from different passages as biological replicates, generating 13,986 datapoints. There was good reproducibility (r>0.96) between biological replicate runs of all organoids in terms of the area under the curve (AUC). The relative sensitivity against individual drugs is expressed as log2-transformed IC50 values and mean-centered against the therapeutic plasma steady state concentration. Our organoid cohort showed differential drug responses to standard chemo-regimens, and these results can effectively serve as a first line to guide patient treatment and improve patient prognosis. We also identified good response to several new drugs approved for the treatment of gastric or other cancer types, such as Napabucasin, a stemness STAT3-target inhibitor, or Abemaciclib, which was recently approved for breast cancer, suggesting that a clinical trial for GC patients may yield beneficial results. Finally, we observed genotype and drug response interactions for organoids carrying p53 and ARID1a mutations to nutlin3a and ATR inhibitors (VE821 and VE822), respectively. We further tested an additional panel of GC organoids for sensitivity to VE821 and VE822. The IC50 of VE822 was significantly lower in GC organoids with ARID1A mutation (n=19, p=0.043), whereas VE821 also showed a similar trend (n=19, p=0.113) Overall, we demonstrated that large-scale drug sensitivity screening is feasible in our gastric organoid biobank which facilitates the development of precision cancer therapy.