Identification of predictive biomarkers of response to angiogenesis inhibitors in neuroendocrine tumours

• Autores: Alberto Lens Pardo
• Directores de la Tesis: Rocío García Carbonero (dir. tes.), Beatriz Soldevilla Navarro (dir. tes.), Eva de Lago Femia (tut. tes.)
• Lectura: En la Universidad Complutense de Madrid ( España ) en 2025
• Idioma: inglés
• Títulos paralelos:
  • Identificación de biomarcadores predictivos de respuesta a inhibidores de la angiogénesis en tumores neuroendocrinos
• Tribunal Calificador de la Tesis: Guillermo Velasco Díez (presid.), Pilar García Alfonso (secret.), Oriol Casanovas i Casanovas (voc.), Guillermo de Velasco Oria de Rueda (voc.), Mauro Cives (voc.)
• Programa de doctorado: Programa de Doctorado en Bioquímica, Biología Molecular y Biomedicina por la Universidad Complutense de Madrid
• Menciones:
  • Internacional
• Calificación: Sobresaliente Cum Laude
• Enlaces
  • Tesis en acceso abierto en: Docta Complutense
• Resumen
  • español

    Los tumores neuroendocrinos (TNEs) son un grupo heterogéneo de neoplasias que se origina de las células neuroendocrinas. Estos tumores están muy vascularizados y la angiogénesis juega un papel clave en su desarrollo y progresión. Esto ha propiciado la aprobación de inhibidores de la angiogénesis para su tratamiento. Sin embargo, la eficacia global es limitada y no todos los pacientes se benefician. El objetivo de este proyecto fue identificar biomarcadores predictivos de respuesta a los inhibidores de la angiogénesis en TNEs.Para ello, hemos analizado el transcriptoma de 126 pacientes incluidos en el ensayo clínico fase II/III, doble ciego, aleatorizado AXINET, que evaluó axitinib frente a placebo, ambos en combinación con análogos de somatostatina (ASS), en 256 pacientes con TNEs extrapancreáticos (EP). Además, hemos examinado la asociación entre la expresión génica, y la supervivencia libre de progresión (SLP) y la reducción del volumen tumoral. A continuación, se generó una firma de expresión en R con singscore y se evaluó su capacidad predictiva, y las diferencias moleculares entre pacientes con puntuaciones altas y bajas de la firma mediante un análisis de enriquecimiento de vías de señalización. Los resultados se validaron en una cohorte independiente de 29 pacientes tratados con antiangiogénicos o ASS en el Hospital 12 de Octubre (H12O). La capacidad predictiva de osteopontina (OPN) en plasma, uno de los biomarcadores identificados, se evaluó mediante ELISA tanto en la cohorte de descubrimiento, como en 2 cohortes de validación: en 87 pacientes independientes del ensayo AXINET, y en 99 pacientes tratados con lenvatinib del ensayo TALENT. Finalmente, se modificó genéticamente la expresión de OPN en líneas celulares de TNEs para investigar su papel en los TNEs y en la respuesta a los antiangiogénicos in vitro e in vivo.Se identificaron 11 genes cuya expresión se asocia significativamente con la SLP en pacientes tratados con axitinib y no en aquellos tratados con placebo. A continuación, se seleccionaron los 3 genes (REXO1L2P, ATXN7, SPP1) que mostraron la mejor asociación con la reducción del volumen tumoral en pacientes tratados con axitinib. Los pacientes con puntuaciones altas de la firma de 3 genes mostraron una SLP significativamente mayor y alcanzaron una mayor reducción del volumen tumoral que aquellos con puntuaciones bajas cuando fueron tratados con axitinib, pero no cuando fueron tratados con placebo. Los pacientes con bajos niveles de la firma tenían tumores enriquecidos en vías asociadas a resistencias a inhibidores de la angiogénesis como la hipoxia, mTOR o la transición epitelio-mesénquima. Estos resultados se validaron en 29 pacientes independientes de la cohorte del H12O.También, hemos evaluado y confirmado la capacidad predictiva de OPN, la proteína codificada por SPP1, en el plasma de los pacientes con TNEs. Los pacientes con altos niveles de OPN en plasma tuvieron una SLP significativamente menor que aquellos con niveles bajos cuando fueron tratados con axitinib, pero no cuando fueron tratados con placebo. Este hallazgo se validó en 87 pacientes independientes del ensayo AXINET, y en 99 pacientes con TNE-EP y pancreáticos tratados con lenvatinib en el ensayo TALENT. Finalmente, hemos demostrado que OPN regula la proliferación, migración y clonogenicidad y que promueve la angiogénesis in vitro. Asimismo, OPN induce resistencia a los inhibidores de la angiogénesis en modelos CDX de ratón de TNEs.En conclusión, hemos identificado una firma tumoral compuesta por 3 genes (REXO1L2P, ATXN7, SPP1) y una proteína en plasma, OPN, que predicen la eficacia a los inhibidores de la angiogénesis en TNEs. OPN no solo regula propiedades oncogénicas in vitro, sino que genera resistencia a estos fármacos en modelos CDX de TNE. Los niveles plasmáticos de OPN podrían utilizarse en la práctica clínica para guiar las decisiones terapéuticas relacionadas con el uso de los inhibidores de la angiogénesis.

  • English

    Neuroendocrine tumors (NETs) are rare and heterogeneous malignancies arising from neuroendocrine cells. These tumors are hypervascular and angiogenesis plays a crucial role in their development and progression. Indeed, NETs express high levels of VEGFVEGFR signaling pathway proteins and several omic analysis have highlighted the importance of angiogenesis in these patients. This has led to the approval of antiangiogenic multikinase inhibitors (MKIs) as anticancer therapy for NET patients.

    While these drugs are active in pancreatic and extrapancreatic NETs, the global magnitude of the effect is limited, not all patients benefit from this therapeutic strategy, and even those that initially respond eventually develop resistance and experience disease progression over time. In this context, the main goal of this project was to identify predictive biomarkers of response to angiogenesis inhibitors in NETs and unravel the molecular mechanisms involved. To this aim, we analyzed biological samples of patients enrolled in the randomized, double-blind, placebo-controlled, international phase II/III AXINET clinical trial assessing axitinib versus placebo, both in combination with somatostatin analogues (SSA), in 256 patients with advanced extrapancreatic NETs (EudraCT Number 2011-001550-29) First, we analyzed the transcriptomic profile of 126 NET patients (discovery cohort) enrolled in the AXINET trial using Clariom Human S arrays and studied the association of gene expression with progression-free survival (PFS) and tumor shrinkage. To do so, we have performed Cox regression, interaction tests, and regularized linear regression separately in axitinib/SSA- and placebo/SSA-treated patients. A gene expression signature was established using the R package singscore and its predictive capacity was evaluated with Kaplan-Meir curves and Cox regression models. Clinicopathological and molecular differences between high- and low-score signature patients were evaluated using Chi-square and differential expression followed by gene set enrichment analysis, respectively. These results were validated using the same methodology in an independent cohort of 29 NET patients treated with MKIs or SSA at Hospital 12 de Octubre. The predictive capacity of osteopontin (OPN) plasma levels, one of the identified predictive biomarkers, was evaluated as previously described using ELISA in the discovery cohort and in two validation cohorts: an independent cohort of 87 patients from the AXINET trial and a second cohort of 99 NET patients treated with lenvatinib, another MKI, in the TALENT clinical trial. Finally, we overexpressed and knockdown OPN in NET cell lines to study its role in NETs proliferation, migration and clonogenicity. Similarly, we also evaluated its role in modulating response to MKIs in vitro, using tube formation and cell viability assays, and in vivo using NET cell-line derived xenografts (CDX) in the chorioallantoic membrane (CAM) of the chicken embryo and mouse models.

    In the first stage of the project, we identified 11 genes that were significantly and exclusively associated with PFS in axitinib-treated and not in SSA-treated patients. Next, linear regularized regression identified the 3 genes (REXO1L2P, ATXN7 and SPP1) that better explain tumor shrinkage in response to axitinib. Patients with high scores of this 3gene signature had significantly better PFS than patients with low signature scores (PFS:

    38.5 vs. 11.6 months; HR: 0.31 (0.16 – 0.59), p < 0.001) when treated with axitinib and notably, the signature score was not associated with PFS in patients treated with placebo.

    Similarly, high score patients treated with axitinib had significantly increased tumor shrinkage compared to those with low score (p = 0.004) while this difference was not observed by signature score in placebo-treated patients. These results were validated in an independent cohort of 29 NET patients treated with MKIs or SSA at Hospital 12 de Octubre. Interestingly, low-score patients (non-responders) were enriched in molecular pathways previously associated with resistance to anti-angiogenic therapy (AAT) such as hypoxia, mTORC1 signaling or EMT.

    To facilitate the clinical implementation of the signature, we evaluated the predictive capacity of the protein product of one gene of the signature, OPN (SPP1), assessed in patient’s plasma. In the discovery cohort, patients with high OPN plasma levels had significantly worse PFS than patients with low levels when treated with axitinib (PFS:

    13.2 vs. 38.5 months, HR: 2.04 (0.99 – 4.22), p = 0.053) but not when treated with placebo. We confirmed these findings in an independent cohort of 87 AXINET patients (validation cohort), being PFS significantly lower in patients with high plasma OPN levels as compared to those with low levels (median PFS: 14.0 vs. 71.1; HR: 2.81 (1.18 – 6.65), p = 0.019) when treated with axitinib and not when treated with placebo. The predictive capacity of OPN was further validated in patients enrolled in the TALENT single arm trial that assessed lenvatinib in pancreatic and extrapancreatic NETs. Among these patients, those with high OPN plasma levels had significantly worse PFS than patients with low levels (PFS: 13.1 vs. 22.3 months; HR: 1.81 (1.09 – 3.00), p = 0.022).

    Finally, in vitro studies demonstrated that OPN upregulation and knockdown modulates proliferation, migration and clonogenicity in NET cell lines. Moreover, OPN upregulation also promoted angiogenesis in vitro and resistance to MKIs in NET CDXs in vivo, confirming its pivotal role in modulating response to MKIs in NETs.

    In conclusion, this study identified a 3-gene signature (REXO1L2P, ATXN7 and SPP1) that predicts the efficacy of anti-angiogenic MKIs in NETs. OPN (SPP1) plasma levels alone, as a more readily accessible surrogate marker, can also predict efficacy of these drugs, enabling an easy implementation of this predictive biomarker in the clinical setting.

    Moreover, OPN promotes the acquisition of key oncogenic functions such as proliferation, migration and clonogenicity in NET cell lines and drives resistance to these drugs in NET CDX mouse models. OPN plasma levels could be used in clinical practice to guide therapeutic decisions regarding angiogenesis inhibitors, increasing efficacy and reducing cost and unnecessary toxicity of these drugs, thereby leading to an overall improvement in their benefit-risk balance in NET patients.