Characterization of the immune microenvironment and patient-derived tumor organoids to identify biomarkers of response and therapy resistance mechanisms in advanced gastric cancer

Gastric cancer is the fifth leading cause of cancer death worldwide. For this reason, many efforts have been made in recent years to understand the biology of this type of tumor, leading to the approval of new treatments such as immunotherapy. However, until a year ago, it was considered that only those patients who presented high microsatellite instability (MSI-H) or a positive CPS (CPS ≥ 5) really experienced a benefit. In this work we have performed an in-depth study of the characteristics of the immune microenvironment in patients with advanced gastric cancer presenting a CPS ≤ 5, with the aim of finding new biomarkers of response to immunotherapy in this population. First, using a retrospective cohort of 31 patients with advanced gastric cancer, we carried out an integrated transcriptomic and clinicopathological analysis. Transcriptomic analysis was performed with a NanoString expression panel containing 770 genes related to the immune microenvironment. This analysis allowed the generation of a new immune classification based on the composition and functionality of the immune infiltrate, identifying a subgroup of patients characterized by high infiltration with high function, candidates to benefit from immunotherapy. This observation was subsequently validated in a prospective cohort of 23 patients and in an external cohort of patients treated with immune checkpoint inhibitors, demonstrating that our classification, together with CPS, improved the selection of patients who could benefit from this treatment. On the other hand, about a decade ago, a methodology was developed that allows the generation of organoid cultures derived from patient tumor tissue as a disease model. These 3D cultures have revolutionized research in oncology, since they allow the study of cancer in a patient-specific manner, serving as unique functional tools for personalized and precision medicine. In this work, a biobank of 36 tumor organoid lines derived from patients with advanced gastric cancer has been generated, representing all subtypes of this type of tumor, both histologically (diffuse, intestinal, mixed and undifferentiated) and molecularly (HER2-positive, MSI, EBV-positive...). Morphological, genomic and proteomic characterization made it possible to demonstrate that the organoids generated faithfully reproduced the characteristics of the tissues of origin and, therefore, constituted a good model for the patients from which they were derived. Drug sensitivity assays were performed, both with conventional chemotherapy and with treatments directed against the HER2 protein, such as trastuzumab and lapatinib, observing a heterogeneity of response and being able to classify the models as sensitive or resistant to each drug. This allowed us to compare the response of each organoid with the clinical response of the patients from whom they were derived, obtaining a high correlation in practically all cases, highlighting the potential role of organoids as a predictive model of response to therapy in gastric cancer. Likewise, an integrated proteotranscriptomic analysis was also carried out, comparing organoids resistant vs. sensitive to 5FU+Oxaliplatin, which led to the identification of biological processes potentially involved in the resistance to this combined treatment, such as the deregulation of membrane potential or lipid metabolism, among the most striking ones. In addition, the proteomic analysis comparing organoids resistant vs. sensitive to anti-HER2 treatment allowed the identification of resistance mechanisms already described in the literature, as well as new proteins potentially involved in such resistance, such as HSP90AA1 or DEPTOR. Finally, it was also studied how macrophages, the most abundant immune cell type in the microenvironment of gastric tumors, influenced organoid behavior. Specifically, it was observed that the conditioned medium of M1 (M1CM) and M2 (M2CM) macrophages had a heterogeneous influence on viability, proliferation and sensitivity to cisplatin in the different models tested, suggesting these differences could be due to the specific biology of each tumor. However, in general, M2 macrophage conditioned medium promoted survival and proliferation of most of the lines tested, while decreasing sensitivity to cisplatin. Proteomic analysis of the conditioned media allowed the identification of a panel of candidate proteins that could play a role in the observed effects. Among these, FN1 stood out for its central position in the functional interaction network generated from M2CM vs M1CM differential expression data. Gene silencing experiments of FN1 in M2 macrophages, as well as repolarization of M2 macrophages with polarizing agents towards the M1 phenotype, demonstrated the partial reversal of these pro-tumor effects, serving as a proof of concept for possible future therapeutic strategies.