Gastric cancer is the world's third leading cause of cancer mortality. In spite of significant therapeutic improvements, the clinical outcome for patients with advanced gastric cancer is poor; thus, the identification and validation of novel targets is extremely important from a clinical point of view. We generated a wide, multilevel platform of gastric cancer models, comprising 100 patient-derived xenografts (PDX), primary cell lines, and organoids. Samples were classified according to their histology, microsatellite stability, Epstein-Barr virus status, and molecular profile. This PDX platform is the widest in an academic institution, and it includes all the gastric cancer histologic and molecular types identified by The Cancer Genome Atlas. PDX histopathologic features were consistent with those of patients' primary tumors and were maintained throughout passages in mice. Factors modulating grafting rate were histology, TNM stage, copy number gain of tyrosine kinases/KRAS genes, and microsatellite stability status. PDX and PDX-derived cells/organoids demonstrated potential use-fulness to study targeted therapy response. Finally, PDX transcriptomic analysis identified a cancer cell-intrinsic microsatellite instability (MSI) signature, which was efficiently exported to gastric cancer, allowing the identification, among microsatellite stable (MSS) patients, of a subset of MSI-like tumors with common molecular aspects and significant better prognosis. In conclusion, we generated a wide gastric cancer PDX platform, whose exploitation will help identify and validate novel "druggable" targets and optimize therapeutic strategies. Moreover, transcriptomic analysis of gastric cancer PDXs allowed the identification of a cancer cell-intrinsic MSI signature, recognizing a subset of MSS patients with MSI transcriptional traits, endowed with better prognosis.Significance: This study reports a multilevel platform of gastric cancer PDXs and identifies a MSI gastric signature that could contribute to the advancement of precision medicine in gastric cancer.

Patient-Derived Xenografts (PDXs), entailing implantation of cancer specimens in immunocompromised mice, are emerging as a valuable translational model that could help validate biologically relevant targets and assist the clinical development of novel therapeutic strategies for gastric cancer.More than 30% of PDXs generated from gastric carcinoma samples developed human B-cell lymphomas instead of gastric cancer. These lymphomas were monoclonal, Epstein Barr Virus (EBV) positive, originated tumorigenic cell cultures and displayed a mutational burden and an expression profile distinct from gastric adenocarcinomas. The ability of grafted samples to develop lymphomas did not correlate with patient outcome, nor with the histotype, the lymphocyte infiltration level, or the EBV status of the original gastric tumor, impeding from foreseeing lymphoma onset. Interestingly, lymphoma development was significantly more frequent when primary rather than metastatic samples were grafted.Notably, the development of such lympho-proliferative disease could be prevented by a short rituximab treatment upon mice implant, without negatively affecting gastric carcinoma engraftment.Due to the high frequency of human lymphoma onset, our data show that a careful histologic analysis is mandatory when generating gastric cancer PDXs. Such care would avoid misleading results that could occur if testing of putative gastric cancer therapies is performed in lymphoma PDXs. We propose rituximab treatment of mice to prevent lymphoma development in PDX models, averting the loss of human-derived samples.

Background: Trastuzumab is the only approved targeted therapy in patients with HER2 amplified metastatic gastric cancer. Regrettably, in clinical practice only a fraction of them achieves long term benefit from trastuzumab-based upfront strategy. To advance precision oncology, we investigated the therapeutic efficacy of different HER2-targeted strategies, in HER2 "hyper"-amplified (>8 copies) tumors. Methods:We undertook a prospective evaluation of HER2 targeting with monoclonal antibodies, tyrosine kinase inhibitors and antibody-drug conjugates, in a selected subgroup of HER2 "hyper"amplified gastric patient-derived xenografts (PDXs), through the design of ad hoc preclinical trials.Results: Despite the high level of HER2 amplification, trastuzumab elicited a partial response only in 2 out of 7 PDX models. The dual HER2 blockade with trastuzumab plus either pertuzumab or lapatinib led to complete and durable responses in 5 (71%) out of 7 models, including one tumor bearing a concomitant HER2 mutation. In a resistant PDX harboring KRAS amplification, the novel antibody-drug conjugate trastuzumab deruxtecan (but not trastuzumab emtansine) overcame KRAS-mediated resistance. We also identified a HGF-mediated non-cell-autonomous mechanism of secondary resistance to anti-HER2 drugs, responsive to MET co-targeting.Conclusions: These preclinical randomized trials clearly indicate that in HER2-driven gastric tumors a boosted HER2 therapeutic blockade is required for optimal efficacy, leading to complete and durable responses in most of the cases. Our results suggest that a selected subpopulation of HER2-"hyper"-amplified GC patients could strongly benefit from this strategy. Despite the negative results of clinical trials, the dual blockade should be reconsidered for patients with clearly HER2addicted cancers.

Purpose: Gastric and gastroesophageal adenocarcinomas (GEA) represent the third leading cause of cancer mortality worldwide. Despite significant therapeutic improvement, the outcome of patients with advanced GEA is poor. Randomized clinical trials failed to show a significant survival benefit in molecularly unselected patients with advanced GEA treated with anti-EGFR agents. Experimental Design: We performed analyses on 4 cohorts: IRCC (570 patients), FMI (9397 patients), COG (214 patients) and INT (206 patients). Preclinical trials were conducted in patientderived xenografts (PDXs). Results: The analysis of different GEA patient cohorts suggests that EGFR amplification drives aggressive behaviour and poor prognosis. We also observed that EGFR inhibitors are active in patients with EGFR copy number gain and that co-amplification of other receptor tyrosine kinases or KRAS is associated with worse response. Pre-clinical trials performed on EGFR-amplified GEA PDX models revealed that the combination of an EGFR monoclonal antibody and an EGFR tyrosine kinase inhibitor was more effective than each monotherapy and resulted in a deeper and durable response. In a highly EGFR amplified non-responding PDX, where resistance to EGFR drugs was due to inactivation of the TSC2 tumor suppressor, co-treatment with the mTOR inhibitor everolimus restored sensitivity to EGFR inhibition. Conclusions: This study underscores EGFR as a potential therapeutic target in gastric cancer and identifies the combination of an EGFR TKI and a monoclonal antibody as an effective therapeutic approach. Finally, it recognizes mTOR pathway activation as a novel mechanism of primary resistance that can be overcome by the combination of EGFR and mTOR inhibitors.