Generación de modelos preclínicos predictivos para la evaluación de nuevas estrategias terapéuticas para pacientes con carcinoma de pulmón no microcítico con mutación en KRAS

Abstract

Lung cancer is the leading cause of death due to cancer, among other reasons due to the lack of effective therapies for some subtypes, such as adenocarcinoma with mutation in the gene homologous to the Kirsten rat sarcoma viral oncogene (KRAS), which accounts for 20-30% of these cases. Despite advances in research and the development of new targeted therapies, there are still no effective targeted therapies approved in the clinic for patients with KRAS-mutated lung adenocarcinoma. Therefore, there is a need to develop new therapeutic strategies for these patients and for this it is necessary to generate new and better preclinical models that are truly predictive of response that allow these new potential therapeutic strategies to be optimally tested. Models derived from patient tumors, such as PDX in vivo and organoids in vitro, more accurately recapitulate the complexity and heterogeneity of human cancer than other models, and could represent better experimental tools not only for testing new therapies but also for the search for biomarkers of response, and, therefore, favor the implementation of precision medicine in NSCLC. In the current study, we have generated and characterized three collections of preclinical NSCLC models: an in vitro model for drug screening, patient-derived xenograft (PDX) derived organoids (PDXDO), and two in vivo models, PDX models and humanized PDX models. We have shown that the 45 established PDX models preserve the histological and molecular characteristics of the tumors of the patients from which they are derived throughout passages. Furthermore, our collection of PDX models faithfully represents the diversity of molecular alterations found in the clinic, and could predict the response to the drugs used in clinical practice. Similarly, PDXDOs maintain the morphology and molecular profile of the tumors from which they derive, including both the presence of the driving genomic alterations of each PDX model and their allelic frequency, which demonstrates the conservation of tumor heterogeneity in in vitro culture. Additionally, we develop humanized PDX models in which we can test combinations with immunotherapy, where we have confirmed the infiltration of human T cells in tumors and a reduction in tumor growth on treatment with antibodies against the programmed death receptor (PD-1). Once the response predictive role of our three collections of preclinical models had been generated, characterized and demonstrated, they were used to test and analyze the response to two therapeutic combinations directed at lung adenocarcinoma with a KRAS mutation, which included the inhibitor of cyclins-dependent kinases 4 and 6 (CDK4/6) plus an inhibitor of extracellular signal-regulated kinase (ERK1/2) or an anti-PD-1 antibody. The combination of the CDK4/6 inhibitor with the ERK1/2 inhibitor proved to be more effective than the treatment of any the drugs alone. In vitro, in all adenocarcinoma PDXDOs, the combination reduced proliferation and activation of the signaling pathways involved. Likewise, in vivo, in the PDXs used, treatment with both drugs produced a significant reduction in tumor growth compared to monotherapy treatments. Furthermore, the combination was also effective in those in vitro and in vivo models that showed resistance to the CDK4/6 inhibitor in monotherapy. The combination treatment increased the inhibition of all the pathways involved in the cell cycle at a transcriptomic level to a greater extent than the treatment in monotherapy, and reversed the activation mediated by the treatment with the CDK4/6 inhibitor in monotherapy on the PI3K and WNT pathways, which could explain the underlying synergy. However, the other combination tested, the CDK4/6 inhibitor plus an anti-PD-1 antibody, did not show greater efficacy compared to the CDK4/6 inhibitor in monotherapy in any of our humanized in vitro and in vivo models nor in an immunocompetent murine model, where the reduction in tumor volume being similar to monotherapy treatment. In summary, we have generated three collections of preclinical models that may serve as valuable tools to test potential new therapeutic strategies for patients with NSCLC. Our collection of PDXDO models allows massive in vitro screening of models that reproduce the heterogeneity of patients, to later confirm the efficacy of the best strategies in our collections of PDX and humanized PDX models in vivo. Additionally, we have demonstrated the efficacy of combining the CDK4/6 inhibitor with the ERK1/2 inhibitor in our preclinical models of KRAS mutated lung adenocarcinoma and have suggested a mechanism that could underlie such a synergy and identified some predictive biomarkers of sensitivity.