Genomic Analysis of diffuse Gastric cancer cells from malignant Ascites. Genomic Analysis of diffuse Gastric cancer cells from malignant Ascites

Diffuse-type gastric cancer (DGC) is histologically and clinically distinct from intestinal gastric cancer with poor prognosis. Preclinical models that reliably predict clinical activity of novel compounds are lacking, and molecular analyses of DGC have been hampered by difficulty in obtaining primary tumors of sufficient tumor cellularity without contaminating stromal fibroblasts. To overcome this, we isolated cancer cells from malignant ascites of DGC patients. Red blood cells and CD45+ hematopoietic cells were depleted using Ficoll-separation and antibody-conjugated magnetic beads respectively. To establish preclinical cancer models for DGC, we generated, for the first time, patient-derived xenografts (PDX) which recapitulated the histology of primary DGC tumors with characteristic poorly cohesive signet ring cells and intracellular mucin. Exome sequencing was performed on primary cancer cells and matched whole blood from the same patient to identify somatic alterations. Further analyses revealed FGFR2, MYC and KRAS amplifications as well as TP53 and BAP1 mutations. Exome sequencing of PDX tumors also revealed high concordance of mutations between primary and corresponding PDX tumors. FGFR2 amplification in 3 out of 4 PDX tumors was confirmed by fluorescence in situ hybridization (FISH), with correspondingly high levels of FGFR2 transcripts and protein expression. Drug treatments on PDX tumors using FGFR2 inhibitor AZD4547 revealed that FGFR2 amplification alone was sufficient to predict the sensitivity of PDX tumors to AZD4547 treatment. In contrast, these tumors regardless of amplification status of FGFR2 responded poorly to cisplatin, a standard of care drug for treatment of DGC. In conclusion, we report a high prevalence of FGFR2 gene amplification in DGC and the establishment of reliable PDX models that can predict tumor response to targeted therapy.