Predictive signature of response to antiangiogenics in gastro-enteric-pancreatic and pulmonary neuroendocrine tumours

Abstract

The invention relates to a predictive gene expression signature of response to antiangiogenic drugs, such as axitinib, in gastro-enteric-pancreatic and pulmonary neuroendocrine tumours. The signature is formed by the gene expression of three genes (SPP1, ATXN7 and REXO1L2P), which is converted by means of a bioinformation packet into a unique score that can predict which patients will benefit from treatment with the drug, this benefit being understood as longer survival without progression and larger tumour volume reductions in response to treatment.

Description

[0001]PREDICTIVE SIGNATURE OF RESPONSE TO ANTIANGIOGENETICS IN NEUROENDOCRINE TUMORS OF GASTROENTEROPANCREATIC AND PULMONARY ORIGIN TECHNICAL FIELD OF THE INVENTION The present invention falls within the field of medicine, specifically in the field of oncology. More specifically, the present invention relates to neuroendocrine tumors of gastroenteropancreatic and pulmonary origin. BACKGROUND Neuroendocrine tumors (NETs) are a heterogeneous group of neoplasms that originate from the diffuse neuroendocrine system. Neuroendocrine cells are found in many tissues of the body and that is why these tumors can originate in multiple organs, although the digestive tract, lung and pancreas are their most frequent primary locations (Cuny T et al., 2018. Role of the tumor microenvironment in digestive neuroendocrinetumors. Endocrine-Related Cancer 25(11): R519-44). Historically,NETs have been considered rare diseases due to their low incidence. However, in recent decades, this incidence has increased six-fold and has already exceeded the threshold for rare diseases (5 new cases per 100,000 inhabitants per year). This increase is due to improved diagnostic techniques and case identification (Dasari A et al., 2017. Trends in the Incidence, Prevalence, and Survival Outcomes in Patients With Neuroendocrine Tumors in the United States. JAMA Oncol. 3 (10): 1335-42). Despite this low incidence, NETs are tumors with a very significant prevalence, mainly due to the good prognosis of low-grade NETs. A representative example is that the prevalence of pancreatic NETs is 2 times higher than that of pancreatic adenocarcinoma despite its lower incidence (Modlin IM et al., 2003. A 5-decade analysis of 13,715carcinoid tumors. Cancer.97(4):934-59). NETs are included within the Neuroendocrine Neoplasms (NENs), which are classified according to their histology and proliferative capacity in two main groups: on the one hand, NETs, ​​which are well-differentiated tumors, their histology resembles the endocrine glands from which they originate and they express neuroendocrine differentiation markers. In turn, NETs are divided into Grade 1, 2 and 3 depending on their proliferative capacity. This is measured with the proliferation marker Ki67 and the mitotic index. On the other hand,Neuroendocrine carcinomas (NETs) are characterized by being poorly differentiated and expressing low levels of neuroendocrine markers. These tumors always have a high proliferative capacity (Grade 3) (Rekhtman N 2022. Lung neuroendocrine neoplasms: recent progress and persistent challenges. Mod Pathol.35(Suppl 1):36-50). Angiogenesis plays a crucial role in the development and progression of NETs. These express various receptors on their membrane, such as VEGFR (vascular endothelial growth factor receptor) or PDGFR (platelet-derived growth factor receptor), which, when binding their ligands, activate signaling pathways that promote angiogenesis and other pro-tumorigenic mechanisms such as migration or proliferation (Raymond E et al.,2011. Therapy innovations: tyrosine kinase inhibitors for the treatment of pancreatic neuroendocrine tumors. Cancer Metastasis Rev. 30 Suppl 1:19-26). This has led to the evaluation of the efficacy of various antiangiogenic agents in the context of NETs, ​​including receptor tyrosine kinase inhibitors or monoclonal antibodies against VEGFR such as sunitinib or bevacizumab, respectively. However, only sunitinib demonstrated a significant improvement in progression-free survival (PFS) in advanced pancreatic NETs, ​​achieving an objective response (OR) rate of 9% and a 6-month increase in PFS (Raymond E et al., 2011. Sunitinib malate for the treatment of pancreatic neuroendocrine tumors. N Engl J Med. 364(6):501-13). In the current context, personalized medicine in the field of NETs, ​​and particularly in treatment with antiangiogenic agents, is still far from being implemented. Some biomarkers are known, but none have been validated. Recently,Axitinib, a potent and selective VEGFR-1,2,3 inhibitor, in combination with somatostatin analogues (SSAs) demonstrated a significant improvement in OR (13.2 vs 3.2%, OR 4.58, p = 0.0045) and PFS (16.6 vs 9.9m, HR 0.71, p = 0.017) by centralized radiological review compared to SSAs monotherapy, in 256 advanced extrapancreatic NETs enrolled in the double-blind, randomized, placebo-controlled AXINET trial (Garcia-Carbonero et al. ESMO 2021). Despite this, the magnitude of the benefit was limited, reaching an OR rate of 14% and an increase of about 7 months in PFS. The use of predictive biomarkers allows us to identify and select those patients who will benefit (or not benefit) from a particular drug,avoiding unnecessary toxicities and costs and allowing patients to receive alternative, personalized therapy that is more effective. Several predictive biomarkers have been proposed that could potentially allow selecting which patients could benefit from tyrosine kinase inhibitors or other antiangiogenic agents in the field of NETs, ​​although none of them have been validated or applied clinically. This is the case, for example, of the study in which the plasma of patients with pancreatic NETs and carcinoids treated with sunitinib was analyzed. It was observed that patients with pancreatic NETs with high levels of SDF-1 had worse PFS and OR. In turn, patients with carcinoid tumors with high levels of SDF-1, IL-8, and sVEGFR-3 were also associated with worse PFS and with worse OR in the case of SDF-1 and IL-8 (Zurita AJ et al.,2015. Circulating cytokines and monocyte subpopulations as biomarkers of outcome and biological activity in sunitinib-treated patients with advanced neuroendocrine tumors. Br J Cancer. 112 (7): 1199–205). Furthermore, the mechanisms of acquired resistance to antiangiogenic treatments have been studied. These could be used to monitor treatment response, but not necessarily to predict it (predictive biomarker). In a study using the RIP-Tag2 mouse model, which develops pancreatic neuroendocrine tumors, it was observed that these tumors suffered an increase in the expression levels of VEGFR, FGF1, FGF2, FGF7, FGF8, Ephrin-A1 and Angiopoietin-2 in their tumor cells and an increase in FGF1, FGF2, Ephrin-A1 and Angiopoietin-1 and -2 in their endothelial cells after treatment with monoclonal antibodies against VEGFR-1 and 2 (Casanovas O et al.,2005. Drug resistance by evasion of antiangiogenic targeting of VEGF signaling in late-stage pancreatic islet tumors. Cancer Cell. 8(4):299-309). Other authors have observed these same mechanisms of acquired resistance after treatment with sunitinib, which they managed to reverse through the expression of SEMA3A (Maione F et al., 2012. Semaphorin 3A overcomes cancer hypoxia and metastatic dissemination induced by antiangiogenic treatment in mice. J Clin Invest. 122(5):1832-48). Other resistance mechanisms related to the hydrophobic nature of sunitinib have been described. This causes the tumor cell to sequester the drug in the lysosomes, preventing it from carrying out its function (Wiedmer T et al.,2017. Autophagy Inhibition Improves Sunitinib Efficacy in Pancreatic Neuroendocrine Tumors via a Lysosome-dependent Mechanism. MolCancer Ther. 16(11):2502-15). To date, no combinations of markers have been described that predict the response to the drug axitinib in patients suffering from neuroendocrine tumors. Identifying predictive signatures that allow efficient selection of which patients will respond best to antiangiogenic treatments for NETs is, therefore, of crucial need. The predictive signatures described here allow stratifying patients with neuroendocrine neoplasms of gastroenteropancreatic and pulmonary origin based on the prediction of their response to antiangiogenic agents. Thus, these previously diagnosed patients will be classified into two distinct groups: responders and non-responders. With current tools,It is difficult to predict which patients will respond best and for the longest period of time to antiangiogenic drugs. Some clinical trials have shown that antiangiogenic drugs such as sunitinib or axitinib provide a significant benefit in the OR and PFS of patients with well-differentiated neuroendocrine tumors, compared to placebo treatment. However, the objective response rate of these treatments is only 10% to 14% in NETs. This represents a major problem for patients and for the national health system, making the development of biomarkers such as this patent urgently needed to implement personalized medicine in oncology treatments. The document Lens-Pardo A et al., 2023. 28MO Predictive biomarkers of response to axitinib in patients with advanced EP-NETs enrolled in the AXINET trial (GETNE 1107): Underlying molecular mechanisms. ESMO Sarcoma & Rare Cancers) describes a predictive signature that includes, among others,the genes RPS10-NUDT3, MX2, C18orf25, SSP1, CLDN14, REXO1L2P, KLHDC3, ATXN7 and CUBN. The signature described in this document has a weak predictive capacity (HR: 0.32) despite using a combination of 9 genes. Furthermore, a signature composed of 9 genes would be very complicated to implement at the hospital level, since it would require a larger analysis, increasing costs and analysis time. The problem that the present invention seeks to solve is to identify a predictive signature to determine the response to antiangiogenic drugs in patients with neuroendocrine tumors of gastroenteropancreatic and pulmonary origin, with maximum predictive capacity by minimizing the components of the signature. DESCRIPTION OF THE INVENTION BRIEF DESCRIPTION OF THE INVENTION In a first aspect, a predictive signature is described consisting of the genes REXO1L2P, ATXN7 and SPP1 and / or the proteins ataxin-7 or osteopontin, or a combination thereof,for use in predicting the therapeutic efficacy of an antiangiogenic drug (monoclonal antibodies or tyrosine kinase inhibitors) in a subject having a gastroenteropancreatic or pulmonary neuroendocrine neoplasm. In a second aspect, a method is described for predicting the therapeutic efficacy response of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin to an antiangiogenic drug (monoclonal antibodies or tyrosine kinase inhibitors) comprising the following steps: a) selecting a previously obtained sample of plasma or tumor from a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin; b) determining the expression profile of the REXO1L2P, ATXN7 and SPP1 genes and / or the expression of the ataxin-7 or osteopontin proteins,or a combination of them; c) classify the patient as a responder or non-responder to antiangiogenic treatment based on the levels detected in stage b). In a third aspect, an antiangiogenic drug is described for use in the treatment of a neuroendocrine neoplasia of gastroenteropancreatic or pulmonary origin of a subject, characterized in that the subject has a REXO1L2P, ATXN7 and SPP1 gene signature value between 0.415 and 0.705, and / or an osteopontin concentration between 7.1 ng / mL and 43.1 ng / mL. In a fourth aspect, a device is described for use in the selection of a treatment with an antiangiogenic drug for a subject suffering from a neuroendocrine neoplasia of gastroenteropancreatic or pulmonary origin, comprising: (a) one or more devices to determine the level of expression of the REXO1L2P, ATXN7 and SPP1 genes and / or of at least the ataxin-7 or osteopontin proteins, or a combination thereof,in a sample previously obtained from a subject; (b) a processor; and (c) a storage medium comprising a computer application that, when executed, is configured to: (i) calculate the expression level of the biomarkers REXO1L2P, ATXN7 and SPP1 and / or the expression level of the proteins ataxin-7 and / or osteopontin in the sample; (ii) calculate, from the expression level of the biomarkers from step (i), whether the subject's sample has high or low values ​​for the biomarker signature, with high values ​​starting at 0.415 and low values ​​below 0.410; and (iii) output from the processor the selected treatment; wherein: if the sample has high values ​​for the biomarker signature, it is indicative that the subject responds to treatment with an antiangiogenic drug, and if the sample has low values ​​for the biomarker signature,is indicative that the subject is not responding to treatment with an antiangiogenic drug. In a fifth aspect, a product is described which is a computer program comprising a non-transitory computer-readable storage device having computer-readable program instructions incorporated therein, which causes the computer to: (i) access and / or calculate the expression of the biomarkers REXO1L2P, ATXN7 and SPP1 and / or the expression level of the proteins ataxin-7 and / or osteopontin in a sample on one or more devices; (ii) calculate from the expression value of the three genes or the two proteins from step (i) whether the sample from a subject is positive or negative for the biomarker signature and; (iii) provide a result,wherein the result is the selection of a treatment with an antiangiogenic drug or the prediction of the capacity to respond to an antiangiogenic drug; wherein the sample is positive if the value of the gene signature is greater than 0.415 or the protein concentration is less than 44.5 ng / mL; wherein the sample is negative if the value of the gene signature is less than 0.410 or the protein concentration is equal to or greater than 45.0 ng / mL. In a sixth aspect, a kit is described comprising the reagents necessary to detect and measure the gene expression of the REXO1L2P genes,ATXN7 and SPP1; and / or the expression levels of the proteins ataxin-7 and osteopontin (secreted phosphoprotein 1) in a tumor sample from a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin from a subject; and / or to measure the expression levels of ataxin-7 and osteopontin (secreted phosphoprotein 1) in the plasma of a subject with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin. DESCRIPTION OF THE FIGURESFigure 1. Association of 20 genes with PFS in patients with NETs treated with axitinib. A. Cox regression for the 5000 genes using continuous gene expression and PFS,in the two treatment arms separately. B. Volcano plot representation of LOG2 (Hazard Ratio) on the x-axis versus –LOG10 (P-value). The names of the 20 identified candidates are indicated. Figure 2. Association of the signature with PFS by Cox regression and by Kaplan-Meier survival curves and the Mantel-Cox statistical test. A. Patients with high levels of the signature (median PFS of 38.5 months) treated with the combination of axitinib with somatostatin analogues have a significantly better PFS (p<0.0002) than patients with low levels (median PFS of 11.6 months). These differences were not observed in patients treated with placebo. B. Analysis of adding axitinib to analogues in patients with high (responders, left graph) and low levels of the signature (non-responders,Figure 3. Association of the signature with response evaluating the differences in the percentage of tumor reduction using a Mann-Whitney test. A. Patients treated with axitinib (left graph) with high levels of the signature have a significantly greater tumor reduction % than those with low levels of the signature (-11.3% vs 1.0%; p = 0.0026). These differences were not observed in patients treated with placebo. B. Flowchart of patients treated with axitinib (left graphs) and with placebo separately (right graphs). Each bar represents a patient and represents the % tumor reduction, the color being the response according to RECIST criteria. Dark green represents the complete response,Partial response in lighter green; stable disease in yellow; and progressive disease in red. Only patients with a high signature treated with axitinib are enriched in complete and partial responses. Figure 4. Differential expression analysis between patients with high and low levels of the signature. Gene group enrichment analysis using Hallmarks and KEGG signaling pathways. Green: enriched in patients with high expression of the signature. Red: enriched in patients with low expression. Figure 5. Forest plot showing that the predictive value of the signature is independent (p=0.003) of other clinicopathological variables.Figure 6. The signature values ​​of patients with high and low levels of the signature are similar in different patient cohorts. The signature levels are represented on the Y axis in patients with high and low levels of the signature in the two study cohorts: AXINET and H12O treated with axitinib and sunitinib, respectively. A Kruskall–Wallis test was performed and pairwise comparisons were made using Dunn's test. P-values ​​<0.05 were considered statistically significant. (****P <0.0001; ns, nonsignificant). Figure 7. Analysis of the association of the signature with PFS using Cox regression,Kaplan-Meier survival curves and Mantel-Cox statistical test in patients from the validation cohort treated with sunitinib or somatostatin analogues. Figure 8. Differential expression analysis between patients with high and low levels of the signature. Gene cluster enrichment analysis using Hallmarks and KEGG signaling pathways. Signaling pathways with an FDR <0.01 were considered significant.05. Green: enriched in patients with high expression of the signature. Red: enriched in patients with low expression.Figure 9. Plasma OPN levels are associated with PFS exclusively in patients with NETs treated with axitinib. Association of plasma OPN levels with PFS using Cox regression and using Kaplan-Meier survival curves and the Mantel-Cox statistical test. A. Validation cohort in which the predictive signature had not been analyzed. Patients with low OPN levels (by median; median PFS not reached) treated with the combination of axitinib with somatostatin analogues have a significantly better PFS (p=0.015) than patients with high levels (median PFS of 14.49 months) (HR: 0.36 (0.15 – 0.84), p=0.019). No significant differences in response were observed in patients treated with the combination of placebo and somatostatin analogues (PFS OPN High: 35.6 months; PFS OPN Low: 13.6; HR: 1.78 (0.81 – 3.94), p=0.152). B. Discovery cohort in which a tumor sample had been previously analyzed for signature discovery. Patients with low OPN levels (median; median PFS 38.5 months) treated with the combination of axitinib with somatostatin analogues have a significantly better PFS (p=0.049) than patients with high levels (median PFS 19.26 months). No significant differences in response were observed in patients treated with the combination of placebo and somatostatin analogues (PFS OPN High: 11.0 months; PFS OPN Low: 13.1; HR: 0.98 (0.53 – 1.84), p=0.958). DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a predictive tool for the response to treatment with antiangiogenic agents (monoclonal antibodies or tyrosine kinase inhibitors) based on the levels of 3 independent genes, REXO1L2P, ATXN7 and SPP1, and / or their proteins. encoding, ataxin-7 or osteopontin, or a combination of them,in the tumor and plasma of patients with neuroendocrine neoplasms of gastroenteropancreatic or pulmonary origin. In a first aspect, the present invention consists of a predictive signature for use in predicting the therapeutic efficacy of an antiangiogenic drug (monoclonal antibodies or tyrosine kinase inhibitors) in a subject having a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin. For the purposes of the present invention, therapeutic efficacy is understood as progression-free survival (PFS) and tumor volume reduction. Preferably, the predictive signature comprises the REXO1L2P, ATXN7, and SPP1 genes for use in predicting the therapeutic efficacy of antiangiogenic drugs (monoclonal antibodies or tyrosine kinase inhibitors) in a patient having a gastroenteropancreatic or pulmonary neuroendocrine neoplasm. In a preferred embodiment, the predictive signature consists of the REXO1L2P,ATXN7 and SPP1. Alternatively, the predictive signature comprises the following proteins ataxin-7 (ATXN7) and osteopontin (secreted phosphoprotein 1, SPP1) for use in predicting the therapeutic efficacy of antiangiogenic drugs (monoclonal antibodies or tyrosine kinase inhibitors) in a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin. The REXO1L2P gene, which stands for “Exonuclease 1 Homolog-Like 2, Pseudogene”, is not known to encode any protein, and as its name indicates, it is a pseudogene of the REXO1 gene (ENST00000425429.3, GENCODE V44). This gene encodes an exonuclease located in the nucleus that participates in the hydrolysis of phosphodiester bonds in nucleic acids. A recent study indicates that Circ-CCDC66 could increase REXO1 levels, promoting the progression of cervical cancer (Zhang Y et al.,2021. Circ-CCDC66 upregulates REXO1 expression to aggravate cervical cancer progression via restraining miR-452-5p. Cancer Cell International. 21 (1): 20). The ATXN7 gene encodes the ataxin-7 protein (ATXN7), which is involved in spinocerebellar ataxia 7 and lattice degeneration (ENST00000674280.1, GENCODE V44). These diseases are caused by an expansion of the “CAG” repeats in coding regions of the protein, giving rise to polyglutamine tails that alter its correct functioning (Palhan VB et al., 2005. Polyglutamine-expanded ataxin-7 inhibits STAGA histone acetyltransferase activity to produce retinal degeneration. Proc Natl Acad Sci US A. 102 (24): 8472-7). ATXN7 has also been described as potentially playing a relevant role in cancer. A first study showed that circulating ATXN7 contributes to the progression and resistance to doxorubicin in breast cancer (Wang H et al.,2022. CircATXN7 contributes to the progression and doxorubicin resistance of breast cancer via modulating miR-149-5p / HOXA11pathway. Anticancer Drugs. 33(1):e700-10) and accelerates the malignancy of large cell lung cancer (Li D et al., 2022. Downregulation of circATXN7 represses non-small cell lungcancer growth by releasing miR-7-5p. Thorac Cancer.13(116-159):15). In turn, fusions of ATXN7 have been described in colorectal cancer (Kalvala A et al., 2016. Rad51C-ATXN7 fusion geneexpression in colorectal tumors. Mol Cancer. 15(1):47) and mutations that cooperate with RAS thyroid cancer (Montero Conde et al., 2017. Transposon. mutagenesis identifies chromatin modifiers cooperating with Ras in thyroid tumorigenesis and detects ATXN7 as a cancer gene.Proc Natl Acad Sci US A. 114(25):E4951-60). Moreover, low levels ofATXN7 have been described to be associated with worse survival in hepatocarcinoma (Han C et al.,2016. ATXN7 Gene Variants and Expression Predict Post-Operative Clinical Outcomes in Hepatitis B Virus-Related Hepatocellular Carcinoma. Cell Physiol Biochem.39(6):2427-38). Ataxin-7 (O15265 ATX7_HUMAN, UniProtKB) is one of the components of the STAGA co-activator complex, which is involved in the regulation of transcription and chromatin modification (Martinez E et al., 2001. Human STAGA Complex Is a Chromatin-Acetylating Transcription Coactivator That Interacts with Pre-mRNA Splicing and DNA Damage-Binding Factors In Vivo. Mol Cell Biol. 21(20):6782-95). Specifically, this protein mediates the interaction between the STAGA complex and CRX, a transcription factor involved in the differentiation of photoreceptor cells (Palhan, 2005). The SPP1 gene (ENST00000395080.8, GENCODE V44) encodes the osteopontin protein (P10451OSTP_HUMAN, UniProtKB). Osteopontin is a matricellular phosphoprotein and is synthesized by osteoblasts, osteocytes,and other hematopoietic cells. It has also been described that cells of the immune system such as neutrophils, dendritic cells, NK cells and T and B lymphocytes are capable of secreting it (Zhao H et al., 2018. The role of osteopontin in the progression of solid organ tumor. Cell Death Dis. 9 (3): 1-15). This protein is involved in inflammation, immune response (controls cytokine production and cell trafficking), bone calcification, allowing osteoclasts to anchor to bone mineral, and in cancer (Bellahcène A et al., 2008. Small integrin-binding ligand N-linked glycoproteins (SIBLINGs): multifunctional proteins in cancer. Nat Rev Cancer. 8 (3): 212-26). In cancer it plays a specific role, promoting proliferation through metalloproteinases and the HIF-1 signaling pathway. In turn,It inhibits apoptosis by activating its receptor CD44 and PI3K. Another of its functions in the tumor context is to mediate the epithelial-mesenchymal transition (EMT) and "wound healing" through an increase in TWIST, SNAIL or SLUG also mediated by the PI3K-Akt pathway. Osteopontin is also essential for the survival of endothelial cells through its integrin receptor αvβ3 and promotes chemoresistance through the MEK / ERK1-2 pathway (Zhao, 2018). Furthermore, OPN and VEGF have been shown to co-express and can induce each other, affecting macrophages and endothelial cells. This axis is important for proliferation, motility and formation of the endothelial tube (Ramchandani D and Weber GF, 2015. Interactions between osteopontin and vascular endothelial growth factor: Implications for cancer. Biochim Biophys Acta.1855(2):202-22.). In a preferred embodiment, the antiangiogenic drugs are selected from the group consisting of Axitinib, Sunitinib,Cabozantinib, Lenvatinib, Surufatinib, Nintedanib, Famitinib, Pazopanib, Regorafenib and Bevacizumab. In another preferred embodiment, the gastroenteropancreatic or pulmonary neuroendocrine neoplasms are any neuroendocrine neoplasm (G1 / 2 / 3) with primary origin in the lung, pancreas, small intestine, rectum, colon or thyroid or other organs of the digestive system. In a preferred embodiment, the invention relates to both sets, one of genes including REXO1L2P, ATXN7 and SPP1, and another of ataxin-7 or osteopontin (secreted phosphoprotein 1) proteins, or a combination thereof, for use in predicting the response to treatment with antiangiogenic agents (monoclonal antibodies or tyrosine kinase inhibitors) of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin. Some studies suggest that patients with Grade 1 tumors, with functional tumors or with fewer previous lines of treatment may respond better to these therapies,However, this has not been validated (García-Carbonero. ASCO 2021). Functionality is a common characteristic of NETs, ​​whereby these tumors hypersecrete some hormones or bioactive substances into the bloodstream, producing a series of specific symptoms related to the hypersecreted hormone. The signature genes and / or their encoding proteins have been linked to various pathological processes, including cancer or drug resistance. However, to date, no document has been described demonstrating the predictive antiangiogenic effect of these genes and their respective encoding proteins in patients with any type of solid organ or hematological tumor. The identification, development, and implementation of these biomarkers allow for the optimization of the benefit-risk and benefit-cost balances of cancer therapies.avoiding unnecessary toxicities for patients and reducing expenses on ineffective treatments without clinical benefit in certain patients. Of course, this will also allow the patient to receive other therapies that could be effective and therefore improve their prognosis and quality of life. In a second aspect, the invention relates to a method for predicting the response of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin to an antiangiogenic agent (monoclonal antibodies or tyrosine kinase inhibitors), which comprises the following steps: a) selecting a sample, previously obtained,of plasma or tumor from a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin; b) determine the expression profile of the signature; and c) classify the patient as a responder or non-responder to antiangiogenic treatment based on the levels detected in step b). The sample in step a) may consist of a sample of paraffin-embedded, fresh or frozen tissue or a biological fluid, such as blood, plasma, saliva or urine, among others, previously extracted from a subject, preferably human. In step b), the expression profile of the signature is performed by PCR or any technique known in the state of the art that is used to quantify the expression of RNAs or to quantify the expression of proteins. In a more preferred embodiment, it is performed by one of the following methods: PCR of the REXO1L2P, ATXN7 and SPP1 genes,Immunohistochemistry of the proteins ataxin-7 and osteopontin (secreted phosphoprotein 1) in the tumor sample from the patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin; or ELISA (enzyme-linked immunosorbent assay) of the proteins ataxin-7 and osteopontin in the plasma of the patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin. In a more preferred embodiment, gene expression is determined by RT-qPCR, more preferably with Taqman probes. The result of the gene or protein expression profile obtained by the method of the invention can be included with the rest of the parameters of interest (clinical characteristics).which will allow the patient to be classified based on their capacity to respond to antiangiogenic drugs. In the procedure described here, a patient's response is evaluated as progression-free survival (PFS) from the start of treatment until disease progression or death. The median PFS for patients with high levels of the gene signature (divided by median) is 38.5 months, while those with low levels of the gene signature have a median PFS of 11.6 months. The expression of the 3 genes is mathematically transformed into a "score" or numerical score of the signature. In a study conducted on 2 independent cohorts, patients with "high levels of the signature" or a "high level of the signature" were defined as patients whose signature values ​​ranged between 0.415 and 0.705, and patients with "low levels of the signature" or "a low level of the signature" were defined as patients whose signature values ​​ranged between -0.03 and 0.410. In a preferred embodiment, “high signature levels” or a “high signature level” refers to patients whose signature values ​​range between 0.415 and 0.705. The cutoff is set by median. In a particular embodiment, in the validation cohort with sunitinib in pancreatic tumors, patients with signature values ​​ranging between 0.359 and 0.640 are defined as having “high levels of the signature” or a “high level of the signature”, and patients with “low levels of the signature” or “a low level of the signature” are defined as patients with signature values ​​ranging between 0.142 and 0.358. In the sunitinib cohort, since there are fewer patients, there is a tendency towards patients with lower signatures, so the population median is lower. However, it is observed that there is no overlap between cases with “high” or “low” levels of both cohorts, being significantly different. As can be verified in the examples,Patients with high levels of the signature have an objective response rate (OR) of 15.6% compared to 11.1% observed in patients with low levels of the signature. Objective response criteria in NETs do not work as well as in other solid tumors, since these tumors are less proliferative, have relatively small sizes and, for this reason, small variations in the measurement of radiological images can introduce error. Similarly, antiangiogenic drugs do not always show a large tumor reduction; in many cases, the tumor remains the same size, but in a necrotic state that prevents observing the benefit because a decrease in size is not observed. In this sense, although drastic changes in the OR are not observed, a notable increase in PFS is observed.The time it takes to progress to antiangiogenic agents is 3 times longer in patients with high levels of the signature compared to those with low levels. Similarly, plasma osteopontin protein levels also allow stratification of patients with NETs. SPP1 or osteopontin is higher in patients with low levels of the signature. It is observed that patients with low levels of osteopontin (by median) have a PFS of 38.5 months, while those with high levels of the protein have only 19.26 months. This same thing was observed in a validation cohort where the PFS of patients with low levels of osteopontin did not reach the median PFS, while those who expressed high levels had a median PFS of 15.5 months. In another particular embodiment, in a plasma sample from the discovery cohort,Non-responders (low signature levels) are defined as patients with plasma osteopontin (OPN) levels between 46.9 ng / mL and 538.5 ng / mL, and responders (high signature levels) are those with plasma OPN levels between 7.1 ng / mL and 43.1 ng / mL. Note that responders have low plasma OPN levels and would correspond to patients with high signature levels who have low levels of the SPP1 gene that encodes OPN. In the validation cohort, the plasma OPN ranges of non-responder patients (low signature levels) range between 45.0 ng / mL and 193.4 ng / mL, and those of responder patients (high signature levels) range between 6.0 ng / mL and 44.5 ng / mL. With the 3-gene or 2-protein signature as described in the present invention, better patient monitoring and evaluation is achieved,allowing for more personalized decision-making. The predictive capacity of these sets is independent of other clinical variables of great relevance to the survival and drug response of neuroendocrine tumors, such as primary site, grade, number of prior lines of treatment, functionality, or time since diagnosis, allowing for unbiased stratification. Finally, these signatures are easy to clinically implement given the preferred techniques for their determination, since they are technologies already available in many hospitals and are affordable. In a third aspect, the invention relates to a kit comprising the reagents necessary to detect and measure gene expression of the REXO1L2P genes,ATXN7 and SPP1 by PCR in the tumor of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin; and / or the expression levels of ataxin-7 and osteopontin (secreted phosphoprotein 1) by immunohistochemistry in the tumor of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin; and / or to measure the expression levels of ataxin-7 and osteopontin (secreted phosphoprotein 1) by ELISA (enzyme-linked immunosorbent assay) of the ataxin-7 and osteopontin proteins in the plasma of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin. EXAMPLES Example 1: RNA extraction from paraffin-embedded tissues and preparation of plasma samples. The samples were obtained from the different hospitals that participated in the AXINET clinical trial.with the approval of the ethics committee of the different institutions and with informed consent from the patients for the use of their samples. RNA from the paraffin-embedded samples was extracted using the RNeasy FFPE kit (QIAGEN, Valencia, CA, USA) according to the manufacturer's instructions. RNA concentration and integrity were then assessed using a Qubit, ®Fluorimeter (Life Technologies, Carlsbad, CA, USA) or a Nanodrop 1000 (NanoDrop, Wilmington, DE, USA), respectively. Sample purity was assessed using readings at 280 nm, and the absorbance ratio at 260 / 280 nm was calculated. An optimal value was between 1.6 and 2. Plasma samples were obtained by centrifuging the blood at 3000 rpm for 10 minutes at 4°C and stored at -80°C. For evaluation, plasmas were diluted with 1% PBS / BSA. Example 2: Identification of the predictive signature for antiangiogenic agents. The Clariom S Human arrays gene expression platform (Affimetrix, SantaClara, CA, USA), which contains probes for more than 23,000 unique transcripts, was selected to analyze a cohort of 126 patients with advanced extrapancreatic NETs who had a paraffin-embedded tumor sample and who had consented to its use for translational studies.These patients were part of the AXINET clinical trial, which evaluated the efficacy of an antiangiogenic agent, axitinib (a VEGFR 1-2-3 inhibitor), in combination with octreotide (an SSA analogue), compared with placebo in combination with octreotide. Thirty-eight patients had a NET of pulmonary origin, 80 had a NET of gastrointestinal origin, and 10 had a NET of other locations (Table 1). Table 1: Sample characteristics. To identify potential predictive biomarkers for axitinib, the association between the expression of the 5000 genes with the highest MAD (median absolute deviation) and PFS was studied using Cox regression. Since only genes that specifically predicted response to axitinib were sought, the association with PFS was analyzed separately in patients treated with axitinib and in patients treated with placebo as a control. Both arms received the somatostatin analogue. In the axitinib arm, genes with a p <0.01 were considered significant, while in the placebo arm the p <0.05 was used to eliminate the largest number of genes associated with PFS in general and not with PFS to axitinib in particular. 20 genes were identified that were significantly associated with PFS in the axitinib arm (p <0.01) and 256 genes in the placebo arm (p <0.05). (Figure 1A).Of the 20 potential predictive biomarkers, 10 were associated with a higher risk of progression (HR >1) and 10 with a lower risk of progression (HR <1). Furthermore, none of the genes in the placebo arm were concordant with those identified in the axitinib arm. A volcano plot was also performed, plotting the LOG2 (Hazard Ratio) versus the LOG10 (P-value) on the x-axis. The names of the 20 identified candidates are indicated (Figure 1B). The goal was then to reduce the number of genes in the signature for two reasons: first, to select those most relevant for predicting response to axitinib; and second, to facilitate clinical implementation, as such a complex signature would be difficult to implement in a clinical setting. This process of selecting the most relevant genes is one of the differences with other previously published signatures (Lens-Pardo et al. 2023).28MOPredictive biomarkers of response to axitinib in patients with advanced EP-NETs enrolled in the AXINET trial (GETNE 1107): Underlying molecular mechanisms. ESMO Sarcoma & Rare Cancers). To do this, we first performed an interaction test between the expression of the 20 candidate genes and the treatment arm. This test identified genes with significantly different associations with PFS between the axitinib and placebo treatment arms. Those with a significant interaction (p <0.05) were selected, as they were significantly associated with PFS in the axitinib arm but not in the placebo arm. Nine genes with a p-value less than 0.05 were identified. In addition, the SPP1 and MX2 genes were included as they had practically significant p-values ​​of 0.052 and 0.07, respectively (Table 2). Table 2: Genes with significant interaction. Next, in order to select the most relevant genes for the response to treatment with axitinib, we sought to identify the genes that were most associated with a reduction in tumor volume in response to treatment with axitinib. To do this, a regularized linear regression model was constructed. This method allowed us to build a model using the expression of the 11 genes in Table 2 as independent variables and the percentage of tumor volume reduction as the dependent variable and to select those independent variables that best explain the response variable. In this case, using an alpha value of 0.5, a lambda of 8.3233 and the cross-validation method with n-1 iterations (59), the model selected 3 genes that best explained the variable of tumor volume reduction, minimizing the mean square error (Table 3).Therefore, after this selection process we have reached a signature composed of 3 genes, one associated with a better response ATXN7 and 2 associated with a worse response, REXO1L2P and SPP1. Table 3. Genes selected by the model. This method not only allowed us to reduce the number of genes, but also reduced model overfitting by selecting genes based on an independent criterion (% tumor reduction instead of PFS), which is also highly relevant when evaluating treatment response. This gene selection process differs from previously published studies, such as that shown by Lens-Pardo et al. 2023, in that genes were selected using an interaction test, which was not done in the state-of-the-art document. Furthermore, the selection based on percentage tumor reduction in this first publication, of which we are the inventors, was performed using simple linear regression using the coefficients (slopes) of the regression model. This may be overfitted, since no cross-validation method was performed as in the present regularized linear regression model.Furthermore, this first model consisted of 9 genes, which provided redundant information to the model, something that has been eliminated with the regularized regression of the present invention. Finally, having 9 genes instead of 3 greatly hinders its clinical application. Example 3. Predictive value of response to axitinib of the signature Once the genes of interest were identified, we wanted to assign a single numerical value to the set of the signature of the three genes. To do this, we used an R package called singscore. This uses ranking statistics to convert the individual expression of the three genes of the signature into a single value. In this way, a signature value was obtained for each of the patients analyzed. This package is used to calculate the signature score from a set of genes and evaluate gene expression in biological samples.Signscore generates a ranking based on the expression of the genes selected for each patient, where a high position indicates high expression and vice versa, normalizing it to values ​​between -1 and 1. The final score depends on the average of the gene positions in said ranking, and a score is assigned per patient that varies between 0 and 1. Therefore, in the ranking statistics, the package converts the individual expression of the three genes into a single value, allowing a signature score to be obtained for each patient analyzed. The predictive value of the signature was then evaluated.For this purpose, patients with NETs were divided into patients with “high” and “low” levels of the signature based on the population median of the signature, and the association of the signature with PFS was analyzed in each treatment arm using Cox regression with the signature values ​​as a continuous and dichotomous variable and by representing the Kaplan-Meier survival curves and the Mantel-Cox statistical test. First, the association between PFS and the signature was studied as a continuous variable. It was observed that the signature is positively associated with PFS in the axitinib and SSA arm (HR: 0.0007 (3.21∙10-5 - 0.0172) p = 0.00000696) but not in patients treated with placebo and SSAs (HR: 9.369 (0.863 – 2.626); p = 0.398).In turn, when the association between PFS and the signature was analyzed as a categorical variable (High vs Low by median), patients with high levels of the signature had significantly better PFS than those with low levels (HR: 0.31 (0.16-0.59); p = 0.0003), but exclusively in patients treated with the combination of axitinib and SSAs and not in those treated with placebo and SSAs (HR: 1.51 (0.86 – 2.63), p = 0.21). The same was observed in the Kaplan-Meier curves and in the Mantel-Cox statistical test: patients with high levels of the signature had a better PFS of 38.5 months compared to 11.6 months in patients with low levels (p = 0.0002) when treated with axitinib. No significant differences (p = 0.15; High PFS: 8.8 months; Low PFS: 12.0;) were observed in response in patients treated with the combination of placebo and somatostatin analogues (Figure 2A).Similarly, the response to axitinib and placebo was compared in patients with high levels of the signature (responders). Patients with high levels of the signature who received axitinib had a significantly superior PFS (p = 0.0002) than those patients who received the analogs as monotherapy (PFS Axi + SSAs: 38.5 months; PFS Pla + SSAs: 8.8; HR: 0.32 (0.17 – 0.60), p = 0.0003). These differences were not observed (p = 0.17) in those patients who had low levels of the signature (resistant), in PFS when adding axitinib to SSA analogs (PFS Axi + SSAs: 11.6 months; PFS Pla + SSAs: 12.0; HR: 1.49 (0.84 – 2.65), p = 0.17). (Figure 2B). These data demonstrate that the three-gene predictive signature is associated with PFS exclusively in patients treated with axitinib. Notably, the three-gene predictive signature, in addition to being easier to implement and measure given its smaller gene count, has better predictive capacity than the nine-gene signature.The HR of the 3-gene signature was 0.31 compared with 0.32 for the 9-gene signature. Furthermore, the association of the three-gene signature with treatment response, specifically with tumor volume reduction in response to axitinib, was also studied (Figure 5). A significantly greater percentage tumor reduction was observed in patients with high levels of the signature (median) compared to those with low levels of the signature when treated with axitinib (-11.3% vs. 1.0%; p = 0.0026). This was not observed in patients treated with placebo (0% vs. -3.5%; p = 0.0103, right graph), where patients with high levels of the signature (responders) who only received placebo had a smaller reduction in tumor volume (Figure 3A). These differences were also reflected in the response according to RECIST criteria.The percentage of responders (PR + CR) to axitinib was higher in those with high levels of the signature compared to those with low levels (15.6% vs. 11%), and stable disease showed greater tumor reduction. These differences were not seen in patients treated with placebo (0% vs. 8.8%). Quite the opposite, “high” patients treated with placebo fared worse than those with low levels, suggesting that treatment with axitinib may be necessary (Figure 3B). A p-value <0.05 was considered significant. Differentially regulated signaling pathways between patients with high and low levels of the signature were also studied to understand the mechanisms involved (Figure 4). Signaling pathways with an FDR <0.05 were considered significant. Patients with high levels of the signature showed an enrichment of genes related to the Notch pathway (NES: 1.81; FDR=0.011) while patients with low levels showed enrichment in pathways such as epithelial-mesenchymal transition (NES: - 2.83; FDR=4.7x10. -19 ), angiogenesis (NES: -2.20; FDR=2.6x10 -4 ), hypoxia (NES: -1.64; FDR=0.002), the mTOR pathway (NES: -1.75; FDR=2.6x10 -4 ) or glycolysis (NES: -1.69; FDR=7.7x10 4). In turn, a multivariate analysis of the predictive signature was also performed to ensure that its predictive value was independent of other clinicopathological characteristics. Figure 5 presents the data from the Forest plot shown below in Table 4. It is observed that the predictive value of the signature is independent (p = 0.003) of other variables such as the number of previous lines of treatment, the location of the primary tumor, the grade, the time since diagnosis or the presence of carcinoid syndrome. This multivariate analysis was also carried out using the 3-gene signature as a continuous variable, obtaining the same result (HR: 0.0004 (6.7∙10-6- 0.024); p = 0.000183). Table 4. Relationship of the gene signature with other clinicopathological characteristics V ariable N P-valor Example 4. Determining high and low values ​​for the signature. Although the cutoffs determined for the categorical analysis of the variables, it was observed that the association with PFS was more significant when its association with the score was studied as a continuous variable. This implies that the higher the score values, the better the prognosis. It was observed that the differences in response are diluted when they approach the cutoff value. Regarding the ranges observed in the validation cohort with sunitinib in pancreatic tumors, being a small number of patients, the ranges largely coincide with those observed in the discovery cohort. Patients with high levels were defined as those ranging between 0.359 and 0.640; and patients with low levels those between 0.142 and 0.358. The score values ​​by cohort are observed in Figure 6.No differences in signature levels were observed between patients considered high in both cohorts. Nor were any differences observed in signature levels between patients with low expression in both cohorts. In contrast, we observed significant differences between patients with high expression of the signature in one cohort and those with low expression in the other, and vice versa (High Axi vs. Low H12O, P < 0.0001; and Low Axi vs. High H12O, P < 0.0001). This confirms that the cutoffs established to differentiate patients with high and low expression of the signature are adequate and appear to be useful in multiple cohorts.Finally, plasma OPN levels were determined to define responder patients, who correspond to those with high levels of the signature and low levels of OPN protein in plasma, and non-responders, who correspond to those with low levels of the signature and high levels of OPN protein in plasma, in the ELISA analysis. In the discovery cohort, responder patients (low OPN levels) were defined as those with plasma OPN values ​​ranging from 7.1 ng / mL to 43.1 ng / mL, and non-responders (high OPN levels) were defined as those with between 46.9 ng / mL and 538.5 ng / mL. In the validation cohort, plasma OPN levels in responder patients (low OPN levels) ranged from 6.0 ng / mL to 44.5 ng / mL, and those in non-responders (high OPN levels) ranged from 45.0 ng / mL to 193.4 ng / mL It is important to note that the plasma OPN ranges were not derived from the groups with high and low levels of the signature. All patients were recategorized based on their OPN levels relative to the median plasma OPN. This implies that patients with high and low levels of the signature are not the same as those with high or low levels of OPN. Example 5: Validation of the signature To validate the predictive value of the signature, studies were conducted using patients completely independent of those previously used, and other techniques were employed to facilitate its clinical implementation. First, 29 patients who received standard treatment at Hospital 12 de Octubre (Madrid, Spain) were analyzed, either with sunitinib, a tyrosine kinase inhibitor similar to axitinib that inhibits VEGFR, PDGFR, KIT, RET, CSF1R, and FLT3; or somatostatin analogues.The group of patients treated with sunitinib was enriched in primary pancreatic tumors since this inhibitor is only approved for pancreatic NETs (Table 5). In contrast, those treated with analogues were enriched in less pretreated patients, many of them in first line.Table 5. Patient group in the TKI trial (N = 11) N (%) Gender Time since DxFemales 4 (36.4) < 12 months 2 (18.2)Male 7 (63.6) > 12 months 8 (72.79)Grade No. Previous treatmentsG1 1 (9.1) 0 5 (45.5)G2 9 (81.8) 1 0 (0.0)G3 1 (9.1) +2 6 (54.5). Location Best response RECISTPancreas 9 (81.8) PD 5 (45.5) Using the R package singscore, the individual signature value was calculated in these patients and they were divided into patients with high and low levels of the signature based on the population median of the signature and the association of the signature with PFS was analyzed in each cohort using Cox regression and the signature values ​​as a continuous variable and as a dichotomous variable (High vs Low by median) and by representing the Kaplan-Meier survival curves and the Mantel-Cox statistical test (Figure 7A). Those patients with high levels of the signature had better PFS than those with low levels (HR: 0.32 (0.16-0.59); p = 0.09), but exclusively in patients treated with sunitinib and not in those treated with SSAs (HR: 0.54 (0.17 – 1.70), p = 0.292).The same was observed in the Kaplan-Meier curves and the Mantel-Cox statistical test: patients with high levels of the signature had a better PFS compared to patients with low levels (p = 0.084) when treated with sunitinib (left graph). These differences were not observed in patients treated with placebo (p = 0.28) (right graph). The association between PFS and the signature was then studied as a continuous variable. It was observed that the signature tended to be positively associated with PFS in the sunitinib arm (HR: 0.00014 (1.98x10-8 - 1.06); p = 0.05) but not in patients treated with SSAs (HR: 0.22 (0.0019 – 25.11); p = 0.53). The individual association of the 3 genes of the signature with PFS was also studied in these cohorts. As expected, none of the genes were significantly associated with PFS in patients treated with analogues (SPP1, HR 1.78 (0.54-5.86), p = 0.341; REXO1L2P, HR2.2 (0.77-6.54), p = 0.14; ATXN, HR 1.28 (0.36-4.58), p = 0.69). However, in patients treated with sunitinib, the expression levels of SPP1 (HR 1.76 (0.91-3.41), p = 0.09) and REXO1L2P (HR 10.12 (0.91-112), p = 0.059) tended to be associated with PFS. ATXN7 levels showed no association with PFS in patients treated with sunitinib (HR: 0.63 (0.0002-734), p = 0.693). Differential expression analysis was performed between patients with high and low levels of the signature. Next, gene cluster enrichment analysis was performed using Hallmarks and KEGG signaling pathways. Signaling pathways with an FDR <0.05 were considered significant. Figure 8 shows that patients with high levels of the signature had an enrichment of genes related to the β-catenin pathway (NES: 1.61; FDR = 0.054) and cytokines, while (NES: 1.49; FDR = 0.017) that patients with low levels showed an enrichment in pathways such as hypoxia (NES: -1.43; FDR=0.021), the mTOR pathway (NES: -1.56; FDR=6.4∙10. -3 ) or glycolysis (NES: -1.99; FDR=4.0∙10 -6). Example 6: Validation of the signature in plasma from patients with NETs In order to continue validating the predictive signature, but also to improve its implementation in the clinical context, osteopontin (OPN) levels were evaluated in the plasma of patients with NETs. OPN is the protein encoded by the SPP1 gene, which is highly expressed in patients with low levels of the signature. For this purpose, the DuoSet Elisa KIT (DY1433; R&DSystems) was used. Two cohorts of patients with NETs from the AXINET clinical trial were analyzed. The first, consisting of 106 patients, consisted of patients whose tumor sample had already been used for the discovery of the predictive signature; the second consisted of 87 patients from the AXINET clinical trial, which were completely independent and represent an external validation cohort. The clinical characteristics of these two cohorts are well balanced and representative of the entire trial cohort (Table 6). Table 6.DISCOVERY trial patient group (N= 106) N (%). Plasma OPN levels were measured in patients with NETs according to the manufacturer's instructions. Patients were then divided into high and low OPN patients based on the population median plasma OPN levels, and the association of OPN levels with PFS was analyzed in each cohort using Cox regression with OPN levels as a continuous and dichotomous variable, and by plotting Kaplan-Meier survival curves and the Mantel-Cox statistical test (Figure 9). The association between PFS and OPN levels was studied as a continuous variable. In the discovery cohort, we observed that OPN levels were negatively associated with PFS in the axitinib and SSA arm (HR: 1.02 (1.01-1.04); p = 0.00794) but not in patients treated with placebo and SSAs (HR: 0.996 (0.99 – 1.01; p = 0.451).The same occurred in the validation cohort, OPN levels were negatively associated with PFS in the axitinib / SSA arm (HR: 1.01 (1.00-1.02); p = 0.03) but not in patients treated with placebo and SSAs (HR: 0.99 (0.98 – 1.01; p = 0.552). In turn, 106 patients in whom a tumor sample had been previously analyzed for signature discovery were analyzed. Patients with low OPN levels (median; median PFS 38.5 months) treated with the combination of axitinib and somatostatin analogs have a significantly better PFS (p = 0.049) than patients with high levels (median PFS 19.26 months). Cox regression showed a nearly significant HR of 2.04 (0.99 – 4.22; p=0.053). No significant differences in response were observed in patients treated with the combination of placebo and somatostatin analogues (PFS OPN High: 11 months; PFS OPN Low: 13.1; HR: 1.02 (0.54 - 1.90), p=0.958, Figure 9A).Similarly, in the validation cohort, 87 completely independent patients were analyzed in whom the predictive signature had not been analyzed. Patients with low OPN levels (by median; median PFS not reached) treated with the combination of axitinib with somatostatin analogues, have a significantly better PFS (p = 0.015) than patients with high levels (median PFS of 14.49 months). Cox regression showed a significant HR of 2.81 (1.18 – 6.65; p = 0.019). No significant differences in response were observed in patients treated with the combination of placebo and somatostatin analogues (PFS High OPN: 35.6 months; PFS Low OPN: 13.6; HR: 0.56 (0.26 – 1.24), p = 0.152, Figure 9B). Example 7.Comparison with other prognostic signatures In order to demonstrate the predictive value of the three specific genes selected in the signature of the invention from a set of genes already described in the state of the art, three other genes were randomly selected from the 9 listed in the document Lens-Pardo A et al., 2023. (28MO Predictive biomarkers of response to axitinib in patients with advanced EP-NETs enrolledin te AXINET trial (GETNE 1107): Underlying molecular mechanisms. ESMO Sarcoma & Rare Cancers). This document describes a predictive signature that includes the genes RPS10- NUDT3, MX2, C18orf25, SSP1, CLDN14, REXO1L2P, KLHDC3, ATXN7 and CUBN. The signature described in this document has a predictive capacity (HR = Hazard ratio) of 0.32 despite using a combination of 9 genes, 6 more genes than the signature of the invention. The predictive capacity of a signature of the genes MX2, CUBN, CLDN14 was calculated.This signature is also significantly associated with PFS in the axitinib arm but less significantly, and with a lower HR than the signature proposed in this patent (HR: 0.33 (0-16-0.77); p = 0.0014). Similarly, when one of the genes of the signature of the invention (ATXN7) was combined with two other random genes (CUBN and CLDN14), a model was obtained that, although also significantly associated with PFS, did so with a lower predictive capacity (HR: 0.51) compared to the HR: 0.31 of the signature of this application. Furthermore, these last two signatures analyzed have not been obtained by methods that allow avoiding overfitting of the model and that are only associated with PFS and not with the objective response as the signature of the invention does.Therefore, it would not be obvious to select the three genes of the claimed signature from a larger list, since each combination gives a different prediction and, as has been shown, a lower one in cases where other genes have been selected.

Claims

CLAIMS1. A predictive signature consisting of the REXO1L2P, ATXN7 and SPP1 genes, and / or the ataxin-7 or osteopontin proteins, or a combination thereof, for use in predicting the therapeutic efficacy of an antiangiogenic drug (monoclonal antibodies or tyrosine kinase inhibitors) in a subject having a gastroenteropancreatic or pulmonary neuroendocrine neoplasia.

2. The predictive signature for use according to claim 1, characterized in that the antiangiogenic drug is selected from the group consisting of Axitinib, Sunitinib, Cabozantinib, Lenvatinib, Surufatinib, Nintedanib, Famitinib, Pazopanib, Regorafenib and Bevacizumab.

3. The predictive signature for use according to claims 1 or 2, characterized in that the gastroenteropancreatic or pulmonary neuroendocrine neoplasia is any neuroendocrine neoplasia (G1 / 2 / 3) with primary origin in the lung, pancreas, small intestine, rectum, colon or thyroid or other organs of the digestive system.4.A method for predicting the therapeutic efficacy of the response of a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin to an antiangiogenic drug (monoclonal antibodies or tyrosine kinase inhibitors), comprising the following steps: a) selecting a previously obtained sample of plasma or tumor from a patient with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin; b) determining the expression profile of the REXO1L2P, ATXN7, and SPP1 genes and / or the expression of the ataxin-7 or osteopontin proteins, or a combination thereof; and c) classifying the patient as a responder or non-responder to the antiangiogenic treatment based on the levels detected in step b).

5. The method according to claim 4, characterized in that the sample is selected from the group consisting of: paraffin-embedded tissue, fresh tissue, frozen tissue, or a biological fluid.The method according to claim 5 characterized in that the biological fluid is plasma.

7. The method of claims 4 to 6, characterized in that the expression profile is carried out by PCR of the REXO1L2P, ATXN7 and SPP1 genes; or immunohistochemistry of the ataxin-7 and / or osteopontin proteins in the sample; or ELISA of the ataxin-7 and / or osteopontin proteins in a plasma sample.

8. The method of any of claims 4 to 8, characterized in that the classification of the subject in step c) is determined based on the expression levels of the genes or the concentration of the signature proteins.

9. The method of claim 8, characterized in that a responding patient corresponds to an expression of the signature between 0.415 and 0.705 and a low level of the signature corresponds to the expression of the three genes between -0.03 and 0.410.

11. The method according to claim 8, characterized in that a responding patient or one with low OPN expression corresponds to a plasma osteopontin (OPN) concentration between 7.1 ng / mL and 43.1 ng / mL.

12. The method according to any one of claims 9 to 11, characterized in that the median progression-free survival (PFS) of a subject with a high level of the gene signature or a low level of protein expression is approximately 38.5 months.An antiangiogenic drug for use in the treatment of a neuroendocrine neoplasm of gastropancreatic or pulmonary origin of a subject, characterized in that the subject has a gene signature value (REXO1L2P, ATXN7 and SPP1, of between 0.415 and 0.705, and / or an osteopontin concentration between 7.1 ng / mL and 43.1 ng / mL.

14. The antiangiogenic drug for use according to claim 13, characterized in that the drug is selected from the group consisting of Axitinib, Sunitinib, Cabozantinib, Lenvatinib, Surufatinib, Nintedanib, Famitinib, Pazopanib, Regorafenib and Bevacizumab.

15. A device for use in the selection of a treatment with an antiangiogenic drug for a subject suffering from a neuroendocrine neoplasm of gastropancreatic or pulmonary origin, comprising:. (a) one or more devices for determining the expression level of the REXO1L2P, ATXN7 and SPP1 genes and / or of at least the ataxin-7 or osteopontin proteins, or a combination thereof, in a sample previously obtained from a subject; (b) a processor; and (c) a storage medium comprising a computer application that, when executed, is configured to: (i) calculate the expression level of the REXO1L2P, ATXN7 and SPP1 biomarkers and / or the expression level of the ataxin-7 and / or osteopontin proteins in the sample; (ii) calculate, from the expression level of the biomarkers from step (i), whether the subject's sample has high or low values ​​for the biomarker signature, with high values ​​starting at 0.415 and low values ​​below 0.410; and (iii) output from the processor the selected treatment; where: if the sample has high values ​​for the biomarker signature above 0.415,is indicative that the subject is responding to treatment with an antiangiogenic drug, and if the sample has low values ​​for the biomarker signature below 0.410, it is indicative that the subject is not responding to treatment with an antiangiogenic drug.

16. The device according to claim 15 characterized in that the antiangiogenic drug is selected from the group consisting of Axitinib, Sunitinib, Cabozantinib, Lenvatinib, Surufatinib, Nintedanib, Famitinib, Pazopanib, Regorafenib and Bevacizumab.

17. A product which is a computer program comprising a computer-readable non-transitory storage device having computer-readable program instructions incorporated therein, which causes the computer to: (i) access and / or calculate the expression of the REXO1L2P biomarkers,ATXN7 and SPP1 and / or the expression level of the proteins ataxin-7 and / or osteopontin in a sample on one or more devices; (ii) calculate from the expression value of the three genes or the two proteins from step (i) whether the sample from a subject is positive or negative for the biomarker signature and; (iii) provide a result, wherein the result is the selection of an antiangiogenic drug treatment or the prediction of the responsiveness to an antiangiogenic drug; wherein the sample is positive if the gene expression value is greater than 0.415 or the protein concentration is less than 44.5 ng / mL; where the sample is negative if the gene expression value is less than 0.410 or the protein concentration is equal to or greater than 45.0 ng / mL.

18. A kit comprising the reagents necessary to detect and measure the gene expression of the REXO1L2P, ATXN7 and SPP1 genes; and / or the expression levels of the proteins ataxin-7 and osteopontin (secreted phosphoprotein 1) in a tumor sample from a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin from a subject; and / or to measure the expression levels of ataxin-7 and osteopontin (secreted phosphoprotein 1) in the plasma of a subject with a neuroendocrine neoplasm of gastroenteropancreatic or pulmonary origin.

Images