WO2013098457A1 - Method for classifying non-microcytic lung carcinoma on the basis of identifying an intratumoral immune response - Google Patents

Abstract

The present invention relates to an in vitro method for classifying non-microcytic lung carcinoma on the basis of the differential expression of 50 genes. Said genes identify an intratumoral immune response. The method of the invention is used to differentiate patients with an expression profile of genes associated with an immune response which is associated with a good prognosis and patients without said expression profile, who have worse prognosis. Said classification can be used as a prognostic marker, as a tumour classifier in accordance with the intratumoral immune response (immunoscore) or as a biomarker predicting therapies based on the immune system (immunotherapy). The present invention also relates to a kit which includes a set of probes that recognise the 50 genes of the invention.

Description

 CLASSIFICATION METHOD OF THE NON-MICROCYCTIC CARCINOMA OF LUNG BASED ON THE IDENTIFICATION OF AN IMMUNE INTRATUMORAL RESPONSE The present invention relates to an in vitro method of classification of non-small cell lung carcinoma based on the differential expression of 50 genes. These genes identify an intratumoral immune response. By the method of the invention, patients with an expression profile of genes associated with an immune response that is associated with a good prognosis are distinguished and patients without that expression profile have a worse prognosis. This classification can be used as a prognostic marker, as a tumor classifier as a function of the antitumor immune response ("immunoscore") or as a biomarker predictor of immune system-based therapies (immunotherapy). The present invention also relates to a kit that It comprises a set of probes that recognize the 50 genes of the invention, so the invention could be framed in the medical field.

STATE OF THE TECHNIQUE Lung cancer is the leading cause of cancer death with an annual rate of more than 1.1 million people worldwide, and with a five-year survival rate of only 15%. Approximately 80% of the diagnosed cases are classified as non-small cell lung carcinoma (CNMP) and the remaining 20% correspond to small cell lung carcinoma (CMP). In the CNMP, the most frequent types are squamous or squamous cell carcinoma and adenocarcinoma.

The TNM staging system {7- edition) based on tumor size (T), lymph node involvement (N) and the presence of distant metastases (M) is currently the most used prognostic factor in patients with CNMP. Depending on these parameters, the tumors are classified into: stage I and stage II (in both cases the disease is localized), stage III (disease locally advanced) and stage IV (metastatic disease) (Kligerman S. American Journal of Roentgenology 2010. 194: 562-573).

In early or early stages (stages I and II), surgery with curative intent is the treatment of choice, the benefit of adjuvant chemotherapy to reduce the high recurrence rate after surgical resection ranging between 30- 35% of patients. Specifically, in stage II, adjuvant platinum-based chemotherapy, such as cisplatin, has been shown to improve the survival of certain subgroups but, on the other hand, there is a percentage of patients who despite not relapse after surgery receive adjuvant treatment and which are therefore patients treated in excess. This over-treatment affects problems in these patients associated with the side effects of these treatments. Regarding stages I (which includes subgroups IA and IB), and according to the consensus guide developed by the National Comprehensive Cancer Network (NCCN) in 201 1, in subgroup IA, adjuvant chemotherapy is not indicated, while that in patients of subgroup IB, it is only recommended in those who meet risk factors such as poor degree of differentiation, vascular invasion, wedge resection and minimum margins. Therefore, due to the lack of precision of the current methods to define the prognosis of the early stages of the CNMP, there are currently patients who receive an adjuvant treatment that does not benefit them and also patients who do not receive an adjuvant treatment and who however they have a high probability of tumor recurrence.

Currently, no markers of proven prognostic and predictive value that indicate the patient's progression are known in lung cancer (Karapaniagiotou E, et al. Open Lung Cancer J 2009. 2: 24-30). In CNMP, studies have been developed that use massive analysis platforms to obtain gene expression profiles that can be used as prognostic biomarkers. The results obtained have been disparate in terms of the genes to be included in the biomarker, perhaps due to the use of criteria different in terms of the inclusion of patients in the study, the collection of samples, the choice of tumor stages, the exclusion or not of histological subtypes of great importance in the CNMP, as well as the lack, in some cases, of validation independent (Roepman P. et al. Clin Cancer Res 2009. 15: 284-290; Chen HY et al. New Engl J Med 2007. 356 (1): 1 1 -20); Raponi M et al. Cancer Res 2006 66: 7466-7472; US20090062144; WO2010007093; Raz DJ et al. Clin Cancer Res 2008 14 (17): 5565-5570).

Therefore, there is a need to develop an alternative tool that can be used clinically, which is more effective than standard risk factors in identifying those completely resected patients who can benefit from adjuvant chemotherapy and distinguish them from those classified as low-risk patients. of recurrence and in which chemotherapy would not be necessary. In addition, the antitumor immune response is currently considered a factor related to the prognosis of patients. A robust method that is capable of stratifying patients with CNMP into groups of good and poor prognosis and a method of classification of the intratumoral immune response ("immunoscore") is therefore required. DESCRIPTION OF THE INVENTION

The technical problem that the invention solves is that of providing an alternative in vitro method that determines the existence of an immune response within the non-small cell lung carcinoma (CNMP) to obtain a personalized treatment of the patient.

In the present invention, an in vitro method is described for classifying the CNMP that is characterized by the detection and / or quantification of an expression product of the set of 50 genes, which are shown in Table 1 in the biological sample of a subject. The present invention also relates to the use of the expression products of said 50 genes as CNMP cancer prognostic biomarkers. The method of the invention provides a 50 gene predictor for CNMP. The strategy used to obtain this predictor began with a detection and / or quantification of the global gene expression of CNMP tumors in early stages (I and II). Based on gene expression, a molecular classification and an association with recurrence were performed; the relationship of molecular groups with the most important histological and clinical variables; obtaining a predictor that identifies the molecular groups generated; obtaining a predictor that differentiates a group of patients with a good prognosis compared to a group of patients with a poor prognosis; and validation of the predictors with an external series. Finally, it was found that the method of the invention is useful for CNMP prognosis. The predictor of the invention consists of 50 genes shown in Table 1, hereafter referred to as "50 genes of the invention".

The 50 genes in table 1 are overexpressed in the good prognosis group. The function described for these genes indicates that their overexpression is largely due to the presence of an intratumoral immune response, which is also associated with a better prognosis.

The 50 genes in Table 1 are mostly related to structural elements of immune system cells or immune functions (maturation, recruitment, proliferation and survival), especially in B lymphocytes and intratumoral plasma cells. Many of these genes code for immunoglobulin (Ig) molecules (constant heavy and light chains, the J chain, the variable regions of the heavy and light chains), the B cell receptor (CD79a), the specific lineage marker of B cell (CD19), the specific transcription co-activator in B cells (POU2AF1) or the specific folding factor for IgM assembly (pERpI). There are also genes although they are not expressed exclusively in B cells, they have great influence on homeostasis of this cell type, such as the B cell maturation factor (TNFRSF17), which is a target Putative transcriptional expression of the overexpressed expression factor POU2AF1 (Zao C. et al. 2008 Oncogene 27: 63-75) and the receptor for the B cell activation factor (Trung Chu V. et al. 2007 J. Immunol 179: 5947-5957 ). This is also the case of SLAM7F (CD139), which induces the proliferation and expression of autocrine cytokines on human B lymphocytes (Lee JK, et al. 2007 J Immunol. 179 (7): 4672-8), of CXCL13, a chemoattractant cytokine of B cells (Sáez et al. Blood 201 1, 1 1; 1 18 (6): 1560-9), of IRF4, a member of the family of transcription factors of the interferon regulatory factor, which have shown to have critical functions in several stages of the development of B cells (Havelange V. et al. Blood 201 1, 1 18 (10): 2827-9), or of CD38, which supports the proliferation and survival of B cells ( Malavasi et al. Blood, 201 1, 1 18 (13) 3470-3478). Interestingly, CD38 is strongly expressed in plasma cells as well as CD27, which is a marker of memory cells that is also within the 50 overexpressed genes in the present study. It is interesting to note that Pim-2, a serine / threonine kinase that is also one of the 50 overexpressed genes, has recently been described as an anti-apoptotic mediator in plasma cells (Asano et al. Leukemia 201 1, 25, 1 182 -1 188). Additionally, a gene from table 2 {Homo sapiens ephrin-A4 (EFNA4), transcript variant 3, mRNA), whose reference in the NCBI (National Center for Biotechnology Information, US National Library of Medicine) gene database is NM_182690, is overexpressed in the group of poor prognosis. This gene codes for a protein that prevents extravasation of lymphocytes through the vascular endothelium to reach the tumor, thus effecting an opposite effect to the overexpressed genes in the good prognosis group.

In contrast to B / plasma cells, no specific lineage genes were found for T cells, NK cells or macrophages within these 50 genes in table 1 but genes such as CD38, CD27 and granzyme B can be expressed in both lineage T as B although traditionally have associated cytotoxic T lymphocytes and NK cells (Hagn et al. 201 1 Immunology and Cell Biology, (2 August 201 1) | doi: 10.1038 / icb.201 1 .64).

The data presented here indicate that immunovigilance is acting as an important prognostic factor. In fact, the need for an "immunoscore" to make a correct prognosis of cancer is becoming increasingly urgent (Galón, et al. J Transí Med. 2012 Jan 3; 10: 1). Although immunity mechanisms with prognostic value have been mainly related to T cells (Broussad EK et al. J. Clin Oncol 201 1 29 (6) 602-603), the participation of B cells has also been described (Ascierto et al Breast Cancer Res Treat. 2012, 131 (3): 871-80).

The term "predictor" refers herein to a differential gene expression profile or gene expression profile.

"Gene expression profile" means the gene profile obtained after quantification of the expression product of the genes of interest. The term "expression product" means messenger RNA (mRNA), complementary DNA (cDNA), complementary RNA (cRNA) and / or the protein produced by the genes of interest or biomarkers, that is, by the genes of Table 1, in an isolated biological sample.

The expression profile of the genes is preferably performed by determining the level of mRNA derived from its transcription, after extracting the total RNA present in the isolated biological sample, which can be performed by protocols known in the state of the art. The level of mRNA derived from the transcription of the genes in Table 1 can be determined, for example, but not limited to, by amplification by polymerase chain reaction (PCR), back transcription in combination with the chain reaction of the polymerase (RT-PCR), quantitative RT-PCR, back transcription in combination with the ligase chain reaction (RT-LCR), or any other nucleic acid amplification method; analysis in gene expression series (SAGE, SuperSAGE); DNA or RNA microarrays made with oligonucleotides or probes synthesized in situ by photolithography or by any other mechanism; in situ hybridization using specific probes labeled with any method of marking; by electrophoresis gels; by membrane transfer and hybridization with a specific probe; by nuclear magnetic resonance or any other diagnostic imaging technique using paramagnetic nanoparticles or any other type of detectable nanoparticles functionalized with antibodies or by any other means. The gene expression profile could also be obtained by the detection and / or quantification of the proteins resulting from the translation of the mRNA derived from the transcription of the genes in Table 1, for example, but not limited to, immunodetection by immuno blotting, immunohistochemistry. , chromatography or microarray.

The present invention could also refer to an in vitro method for classifying the CNMP that is characterized by the detection of the number of copies in the DNA of the 50 genes shown in Table 1, as well as epigenetic alterations such as hypermethylation of the promoter of the genes or as of the alteration of the mRNA stability due among other factors to transcriptional modifications that affect for example the tail of Poly Adenines. The present invention also relates to the use of these 50 gene alterations as prognostic biomarkers of CNMP cancer, as "immunoscore" or as a biomarker predictive of immunotherapy response.

Finally, the gene expression profile could also be obtained by detecting and / or quantifying the number of copies of the genes present in Table 1, as well as the levels of epigenetic alterations such as the level of promoter methylation or stability levels of the messenger of these same genes. This detection could be carried out, although not limited by microarrays, CGH (Comparative Genomic Hybridization) or FISH (fluorescent in situ hybridization). It could also be made from material included in paraffin.

This invention could also be applied for advanced stages (III and IV).

As described herein, a first aspect of the invention relates to an in vitro method of obtaining useful data for the prognosis of stage I or II CNMP characterized by the detection and / or quantification of the expression product of the genes of the Table 1 in the isolated biological sample of a subject. From now on we will refer to this as the "first method of the invention".

The term "in vitro" refers to the method of the invention being performed outside the subject's body. The term "prognosis" in the present invention refers to the ability to detect patients who have a high or low probability of recurrence after surgery. A high probability of recurrence is associated with a poor prognosis while a low probability of recurrence is associated with a good prognosis. "Recurrence" is understood as the recurrence of the disease, in this case of lung cancer. The terms "probability of non-recurrence" and "probability of ILE (disease-free interval)" are used interchangeably herein.

The term "non-small cell lung cancer", "non-small cell lung carcinoma" (CNMP), "non-small cell lung carcinoma" (NSCLC), or non-small cell lung cancer ("non-small cell lung cancer") , NSCLC) refers to a type of lung cancer or tumor according to histological classification comprising the subtype squamous or squamous cell carcinoma, adenocarcinoma, adenoescamoso, sarcomatoid carcinoma, and large cell carcinoma. "Stage" means the phase or classification of lung cancer based on the TNM classification. The TNM classification refers to tumor size (T), lymph node involvement (N) and involvement of other organs (M). Stage I refers to the IA or IB substations. The IA substation refers to lung tumors classified as T1 N0M0. The IB substation includes lung tumors classified as T2aN0M0. Stage II refers to any of the IIA or IIB substations. Sub-stage IIA refers to lung tumors classified as T1 N1 M0, T2aN1 M0 and T2bN0M0. Sub-stage IIB includes lung tumors classified as T2bN1 M0 and T3N0M0. In the TNM classification, T1 refers to when the tumor <3 cm of maximum dimension, is surrounded by lung tissue or visceral pleura and without invasion proximal to the lobar bronchus in fibrobronchoscopy. T1 a is a tumor <2 cm and T1 b is a tumor> 2cm and <3cm. T2 refers to a tumor> 3 cm in maximum dimension and <7 cm or a tumor with at least one of the following characteristics: infiltrate the main bronchus 2 cm or less from the carina, invade visceral pleura or be associated with atelectasis or pneumonitis obstructive T2a is a tumor> 3 cm and <5 cm and T2b is a tumor> 5 and <7 cm. T3 refers to a tumor> 7 cm or a tumor that affects the costal wall (including tumors of the upper fissure), diaphragm, mediastinal pleura or pericardium; no involvement of the heart, large vessels, trachea, esophagus, vertebral bodies; or a tumor of the main bronchus less than 2 cm from the carina, without infiltration thereof; where atelectasis affects an entire lung and there may be non-malignant pleural effusion. It does NOT refer to the lung tumor without lymph node involvement. N1 refers to the tumor that has involvement of the ipsilateral peribronchial or hilar lymph nodes or both. M0 refers to the lung tumor that does not have distant metastases.

In the present invention the terms "early stages", "initial stages" or "early stages" refer to CNMP stage I or II. The term "immunoscore" refers to a method to classify the intratumoral immune response (Galón 2012). In the present invention, by detecting and / or quantifying the expression of the 50 genes in Table 1, it is possible to identify a group of patients with the presence of an intratumoral immune response associated with a good prognosis against a group in which no intratumoral immune response is identified that is associated with poor prognosis.

The term "immunotherapy" refers to a therapy or treatment against cancer based on or related to the performance of the immune system of the individual in which the tumor occurs, by facilitating an antitumor immune response and recognition or by preventing actions of the immune system that favor tumor growth.

The term "genes from table 1" or "50 genes" refers to the 50 genes described in table 1 shown below.

The terms "Entrez Identifier" or "Entrez ID" refer to the gene reference number in the NCBI (National Center for Biotechnology Information, U.S. National Library of Medicine) gene database.

Below is a brief description of some of the known functions of the genes presented in Table 1: AMPD1: It catalyzes the deamination of adenosine monophosphate (AMP) to inosine monophosphate (IMP) in skeletal muscle and plays an important role in The purines cycle.

TNFRSF17: This receptor is expressed in mature B lymphocytes and is important for the development of B cells and in the autoimmune response. It has as ligand member 13b of the tumor necrosis factor superfamily and activates the nuclear factor of the light chain polypeptide gene enhancer Kappa in B cells (NF-kappaB) and mitogen-activated protein kinase 8 (MAPK8 / JNK). It also binds to other ligands and sends signals of cell survival and proliferation. CD19: A molecule that binds to the B-cell antigen receptor to lower the threshold of lymphocyte stimulation through antigen stimulation.

CD27: Member of the tumor necrosis factor receptor superfamily. The receptor has the function of generating and maintaining for a long time the immunity of T cells. The CD70 ligand binds to it and works in the activation of B cells and in the synthesis of immunoglobulins. The adapter proteins called Factor Associated to Receptors of Tumor Necrosis Factors 2 and 5 (TRAF2 and TRAF5) mediate this process. CD27 binding protein (SIVA) is a proapoptotic protein that plays an important role in apoptosis mediated by this receptor.

CD38: It is a multifunctional ectoenzyme that is expressed in a multitude of cells and tissues especially in leukocytes. CD38 also has functions in cell adhesion, signal transduction and calcium signaling.

CD79A and CD79B: code for the lg-alpha and lg-beta proteins that are components of the B lymphocyte antigen receptor. The lg-alpha and lg-beta molecules are necessary for the expression and function of this receptor.

GZMB: Cytolytic T lymphocytes (CTL) and "natural killer" cells (NK) have the ability to recognize, bind and lyse specific target cells. GZMB is crucial for the rapid induction of apoptosis of the target cells through the immune response generated by the cytolytic T lymphocytes or even in the mediated by B lymphocytes. IGHA1 and IGHA2: Antibody with an important presence in the mucous secretions and that represents the first line of defense of the organism. There are two subclasses Immunoglobulin A1 (lgA1) and Immunoglobulin (lgA2). IGHG1: This gene is translocated in chronic B-cell lymphocytic leukemia with the Cyclin D1 gene (CCND1) and in subclasses of MALT lymphomas (Mucosa Associated Lymphoid Tissue) with the genes "LIM homeobox 4" (LHX4) and "Forkhead box P1 "(FOXP1). IGJ: Its function is to join two monomers of either Immunoglobulin M (IgM) or Immunoglobulin A (IgA). It also has the function of binding these immunoglobulins to the secretory component.

@ GL: Each immunoglobulin molecule has two identical heavy chains and two identical light chains. There are two kinds of light chains that are kappa and lambda. This gene encompasses the lambda light chain locus that includes segment V (variable), segment J (junction) and segment C (constant). IGLL1: It is a gene of the immunoglobulin superfamily that codes for the light chain replacing the preB cell receptor. Mutations in this gene can cause B-cell deficiency or agammaglobulinemia.

IRF4: It belongs to the family of interferon regulatory factors. It is lymphocyte specific and negatively regulates Toll-like receptors (or TLR), which is a central molecule in the activation of the innate and adaptive immune response.

KCNN3: Regulates neuronal excitability.

KRT81: It is a member of the keratin family. CXCL9: Its function is not well defined but it seems to be involved in T-cell traffic.

PNOC: It is a neuropeptide that acts as an endogenous ligand of the "Opiate Receptor-Like 1" receptor (ORL1).

POU2AF1: It is a specific co-activator of B cells and its absence seems to be related to defects in the development of B cells and the lack of germ centers.

BFSP2: also called phakinin, it is a structural protein of cytoskeleton filaments. Next to the filensin forms the BF ("beaded filament").

CXCL13: Promotes the migration of B lymphocytes preferably against T lymphocytes and macrophages by calcium stimulation.

PIM2: It is a serin / threonine / protein kinase. Prevents apoptosis and promotes cell survival. It is an anti-apoptotic mediator of plasma cells. SMR3A: It is a functional homolog of the Vcsal gene ("Variable Coding Sequence A1"). It has been associated as a marker of erectile dysfunction associated with both diabetic and non-diabetic etiology.

MZB1: It is associated with the heavy and light chains of immunoglobulin type M (IgM), promoting the assembly of IgM and its secretion.

FKBP11: It belongs to the FKBP family which catalyze the folding of proline-containing polypeptides. Its function is inhibited by FK506 and by rapamycin.

LAX1: A negative regulator of lymphocyte signaling. CPNE5: Calcium-dependent membrane binding protein that appears to be involved in the regulation of molecular phenomena at the interface of the cell membrane and in the cytoplasm. SLAM7: It is involved in the activation of NK cells and in the regulation of B lymphocyte proliferation during the immune response.

DUSP26: It is associated with the inactivation of the protein kinase activated by mitogens 1 and 3 (MAPK1 and MAPK3), as well as with the inhibition of epithelial cell proliferation, which could suggest a role as a tumor suppressor gene.

FCRL2: It is part of the immunoglobulin receptor superfamily. It can be a prognostic marker of chronic lymphocytic leukemia.

FCRL5: It is also part of the immunoglobulin receptor superfamily. It is involved in the development of B cells and lymphomagenesis.

 FCRLA: This receptor mediates the destruction of antigens recognized by Immunoglobulin G (IgG). It is selective B cell protein and may be involved in its development.

DERL3: Protein that is located in the endoplasmic reticulum with the function of degrading misfolded glycoproteins.

MTSS1L: May be involved in actin packaging. It belongs to the MTSS1 family (Type 1 Metastasis Suppressors).

JSRP1: The sarcoplasmic reticulum is a cellular compartment that controls the concentration of intracellular calcium and is involved in the excitation-contraction functions of this cellular compartment. In mice it has been seen This protein interacts with key proteins involved in these excitation-contraction processes.

C5orf20: This gene is expressed in dendritic cells, which are potent antigen presenting cells involved in activating native T cells to initiate the antigen-specific immune response.

MEI1: Defects in its expression are related to stopping in meiosis and is associated with azoospermia phenomena.

GPR114: Protein G associated with receptors with an N-terminus that contains regions rich in serine / threonine. Its expression in cytotoxic lymphocytes has been described. IGHV5-78, FER1L4, IGKV1D-8, KIAA0125, LOC401847, LOC642424, LOC100132941, LOC100133862, LOC100287723, IGHV1-24 and LOC100293440: as of today, the function of these genes is still unknown.

The term "biological sample" includes, but is not limited to, tissues and / or biological fluids of an individual, obtained by any method known to a person skilled in the art that serves this purpose.

The term "subject" refers to an individual, preferably human, who has been diagnosed with CNMP.

A preferred embodiment of the first aspect of the invention relates to a method that further comprises comparing the useful data obtained from the isolated biological sample of a new subject, with the reference expression values for the genes in Table 1 obtained from subjects with CNMP stage I or II in which the prognosis is known (reference sample). The comparison allows the identification of the new subject as a subject of good prognosis or bad prognosis. From now on, we will refer to this method as the "second method of the invention".

The term "reference samples" as understood in the present invention refers, for example, but not limited to samples obtained from individuals having a known molecular profile. This molecular profile can be of good prognosis or of poor prognosis.

A person skilled in the art could classify a new patient in the group of good or in the group of poor prognosis when comparing his expression data for the 50 genes of the invention with the expression data for the 50 genes in the reference samples. These reference samples are a group of samples of which the expression profile of the 50 genes is known and the presence or absence of recurrence. For example, but not limited to, a new subject whose expression profile is similar to the good prognosis reference group can be classified as belonging to the good prognosis group, which has an average 3-year probability of ILE at 85% and / or 5 years of 79%. For example, but not limited to, a new subject whose expression profile is similar to the bad prognosis reference group can be classified as belonging to the bad prognosis group, which has an average 3-year probability of ILE at 62% and / or 5 years of 48%.

The determination of the prognosis of new patients diagnosed with CNMP in stages I or II implies the classification of these patients into one of the two previously defined reference groups: a group with a good prognosis or a group with a poor prognosis. These reference groups consist of the reference samples.

The comparison of the useful data obtained from the biological sample of a new subject, with the reference expression values for the genes of table 1 obtained from subjects with CNMP stage I or II in which the prognosis is known (reference sample ), can be carried out by any statistical prediction method known in the state of the art, such as, but not limited to, in any of the methods described in Simón R. et al. J Clin Oncol 2005; 23: 7332-41. In a preferred embodiment of the second method of the invention, the comparison is made by the nearest compact centroid method. Hereinafter, the "third method of the invention".

The "closest shrunken centroid method" is the classification method described in Tibshirani R. et al. PNAS 2002, 99: 6567-6572 and applied through the Prediction analysis of microarray tool (PAM). The "PAM" tool was developed by Standford University and is freely accessible.

The determination of the CNMP prognosis of stages I or II can be established, but not limited, by determining a "reference value" for the group of good prognosis (value 1) and another for the group of poor prognosis (value 2 ). The forecast can be made by estimating the distance between the expression values of the new sample and the "reference values" of each of the two groups. If the distance between the new sample and the value 1 is less than the distance between the new sample and the value 2, the favorable forecast can be determined. On the contrary, if the distance between the new sample and the value 1 is greater than the distance between the new sample and value 2, the unfavorable forecast can be determined.

The reference values of each group can be calculated based on the expression values of the 50 genes in the samples of the reference matrix or "development matrix" and will therefore be expressed by a vector of 50 components. The calculation of the reference value of each group (in our case the group of good prognosis and the group of bad prognosis), is obtained by adding to the global average value of all the samples, a second factor defined as the distance (statistic "t") between the average expression value of the 50 genes of said group with respect to the average expression value of the 50 genes of all the samples included in the training matrix. The data of the second factor will be standardized taking into account the variability of expression of each of the 50 genes within the analyzed group and taking into account a convergence value Δ that allows to evaluate the predictive power of each of the genes. It is understood as distance between two samples, groups or subtypes, the quantification of their expression differences.

Although the final reference value or "shrunken centroid" obtained for each group is based on the expression values, its actual value is dimensionless and is not directly proportional to the fluorescence data initially obtained in each sample. Said reference value, in each group, contains 50 components, one for each of the genes analyzed.

Once the reference values for each group have been calculated, the "closest shrunken centroid" method is able to assign new samples (which in our case make up the validation matrix) to each of the defined groups. The distance between the new sample and each of the groups is relative to the difference between the expression values of the 50 genes in the new sample with respect to the components of the compact centroid ("shrunken centroid") that represent each group. The quantification of distances could be measured, but not limited, by Euclidean distance (Tibshirani R. Diagnosis of multiple cancer types by shrunken centroids of gene expression. PNAS 2002; 99 (10): 6567-72). As mentioned earlier, the new sample will be assigned to the group from which it is at a smaller distance. For all that is described herein, a second aspect of the invention relates to an in vitro method for the prognosis of stage I or II CNMP characterized by: to. the detection and quantification of the expression product of the genes of table 1 in a reference sample;

 b. the calculation of a reference value (value 1) for each expression product of the genes in table 1 in the favorable prognostic reference samples (good prognosis group) and the calculation of a reference value (value 2) in unfavorable prognosis reference samples (poor prognosis group) by using the nearest centroid method;

 C. the detection and quantification of the expression product of the genes of table 1 in the biological sample of a new subject in which the prognosis is unknown (study sample);

 d. the comparison by using the closest compact centroid classification method of the values obtained in the detection and quantification of the expression product of the genes of table 1 in the study sample with the reference values obtained in the groups of good and bad prognosis. and. the association of the study sample to the group of good prognosis or the group of poor prognosis as established in the method of the nearest compact centroid.

Hereinafter this method will be called "fourth method of the invention".

A preferred embodiment of the fourth method of the invention relates to the method where the nearest compact centroid method is carried out through the application of Microarray Analysis Prediction (PAM).

A preferred embodiment of the first and second aspects of the invention refers to the method where the reference sample and the study samples have been previously normalized before comparison.

"Normalization" means the use of a control sample that serves to eliminate experimental variations between the different samples. Another preferred embodiment of the first and second aspects of the invention refers to the method that also comprises the detection and / or quantification of at least one expression product of the genes described in Table 2.

Another preferred embodiment of the first and second aspects of the invention relates to the method where the expression product is messenger RNA. An even more preferred embodiment refers to the method where the detection and / or quantification of messenger RNA is performed by microarrays. A more preferred embodiment also refers to the method where the detection and / or quantification of messenger RNA is performed by RT-PCR.

Another preferred embodiment of the first and second aspects of the invention relates to the method where the expression product is a protein. An even more preferred embodiment refers to the method where the detection and / or quantification of the protein is performed by immuno blotting, immunohistochemistry, chromatography or microarrays.

The detection and quantification of the expression product (mRNA, complementary RNA obtained from cDNA, complementary DNA or protein) can be performed using methods known to those skilled in the art. For example, by determining the level of mRNA derived from its transcription, after extracting the total RNA present in the isolated biological sample, which can be done by protocols known in the state of the art. For this, the isolated biological sample can be physically or mechanically treated to break up the cellular tissue or structures and release the intracellular components to an aqueous or organic solution to prepare the nucleic acids for further analysis. Nucleic acids are extracted from the sample by procedures known to those skilled in the art and commercially available. The level of mRNA derived from the transcription of the genes in Table 1 can be determined, for example, but not limited to, by amplification by chain reaction of the polymerase (PCR), retrotranscription in combination with the polymerase chain reaction (RT-PCR), quantitative RT-PCR, retrotranscription in combination with the ligase chain reaction (RT-LCR), or any other amplification method of nucleic acids; serial analysis of gene expression (SAGE, SuperSAGE); microarrays, micromatrices or DNA chips made with oligonucleotides deposited by any mechanism or made with oligonucleotides synthesized in situ by photolithography or by any other mechanism; in situ hybridization using specific probes labeled with any method of marking; by electrophoresis gels; by membrane transfer and hybridization with a specific probe; by nuclear magnetic resonance or any other diagnostic imaging technique using paramagnetic nanoparticles or any other type of detectable nanoparticles functionalized with antibodies or by any other means.

The present invention demonstrates that the detection and quantification of the total mRNA of a biological sample of a subject with stage I or II CNMP is useful for the prognosis of said disease. Therefore, in a preferred embodiment of this aspect of the invention, the expression product detected and quantified is mRNA.

Therefore, another preferred embodiment of the first aspect of the invention relates to a method where the expression product is mRNA. "Microarray" (expression microarray, chip or microarray) is understood as the set of probes (oligonucleotides or cDNAs) arranged in an orderly manner on a solid surface, which allows simultaneous analysis of the expression of the entire genome of an organism. Each of the probes specifically represents a gene determined by having a sequence complementary to the mRNA transcribed by said gene, thus enabling the measurement of the expression levels of all the genes that make up the genome at the same time and in a single experiment. For the use of microarrays and obtaining data from them, the experimental phase of the microarray can consist of the steps described below. First, the total RNA is retrotranscribed using as a primer a messenger specific primer (PolidT) and a retrotranscriptase enzyme. Using as a template the double stranded cDNA obtained above, the cRNA was synthesized, while the process of amplification and marking of the sample was carried out. The labeled cRNA obtained was purified by columns. The cRNA is fragmented into smaller sequences and hybridized to the microarray. Said hybridization process is carried out in a hybridization oven for a long period of time. In this process the labeled cRNA binds specifically to the oligonucleotides synthesized in the microarray. Subsequently, the microarray is washed to remove all excess cRNA not bound to the oligonucleotides. In accordance with the present invention, the expression product, preferably mRNA or cDNA or complementary RNA (cRNA) obtained from cDNA, can be labeled or labeled by techniques well known in the state of the art. Detectable labels include, for example, radioactive isotopes, fluorescent labels, chemiluminescent labels, bioluminescent labels or enzymatic labels. The fluorescent labels may be different in the case of the mapping of the expression product of the biological sample and the expression product of the control sample. On the other hand, the detection and quantification can also be carried out by RT-PCR, whereby another preferred embodiment of the first aspect of the invention refers to a method according to claim 7 wherein the detection and / or quantification of the mRNA is performed by RT-PCR or preferably by real-time RT-PCR. The RT-PCR process can be carried out in two phases:

- Retrotranscription: the union between a primer and the mRNA is produced by a joint incubation process of both products. Then the retrotranscription itself takes place using reverse transcription enzymes.

- Subsequent PCR: amplification of the cDNA obtained in the previous phase is produced by the polymerase chain reaction (PCR) technique. For each sample and for each transcript of the genes analyzed, the PCR reaction will be carried out individually. This process involves the cyclic repetition of 3 phases: cDNA denaturation phase, oligonucleotide specific binding phase of the gene under study to the denatured cDNA strand and elongation phase from the bound oligonucleotide by which a new strand of synthesized will be synthesized. CDNA Being a process that is measured in real time, it is necessary to use a fluorescent molecule to monitor what happens throughout the process. Quantification on the other hand can also be performed by determining the level of protein derived from the translation of transcribed mRNAs from the 50 genes of the invention. This protein quantification can be performed by any method known to a person skilled in the art that serves this purpose, such as, but not limited to, immunodetection methods (such as western blot, ELISA, immunohistochemistry, immunocytochemistry, immunofluorescence), methods based on isobaric (such as iTRAQ - isobaric Tag for Relative and Absolute Quantitation-, or ICAT -Isotope-Coded Affinity Tag-) or in isotopic (such as SILAC -Stable Isotopes Labeling by Amino Acids in Cell Culture-) or based on fluorescent ( such as 2D-DIGE -Difference in Gel Electrophoresis-), as well as methods based on mass spectrometry (MRM, -Multiple Reaction Monitoring-). Therefore, in another preferred embodiment of this aspect of the invention is the method where the expression product is a protein. Another preferred embodiment of the first and second aspect of the present invention relates to the method where the detection and / or quantification of the Protein is made by immuno blotting, immunohistochemistry, chromatography or protein expression arrays.

The terms "amino acid sequence" or "protein" are used interchangeably herein, and refer to a polymeric form of amino acids of any length, which may or may not be chemically or biochemically modified. The term "residue" corresponds to an amino acid.

Another preferred embodiment of the first and second aspects of the present invention relates to the method where the biological sample is selected from the list comprising: tissue, blood, plasma, serum, lymph, bronchoalveolar lavage or ascites fluid.

Another also preferred embodiment of the first and second aspects of the present invention relates to the method where the biological sample is fresh, frozen, fixed or fixed and embedded in paraffin.

Another preferred embodiment of the first and second aspects of the invention relates to a method where the subject is a human.

A third aspect of the invention relates to the in vitro use of the expression products of the genes in Table 1 as a prognostic marker of stage I or II CNMP. A fourth aspect of the invention relates to the in vitro use of the expression products of Table 1 to classify the intratumoral immune response in stage I or II CNMP.

The authors of the present invention have found that the genes in Table 1 are overexpressed in those patients who have a better prognosis, and that these genes are mostly related to the immune response. In this way, overexpression of the genes in table 1 It is associated with the presence of an intratumoral immune response, which correlates with a better clinical prognosis. On the other hand, a lower expression of these 50 genes is associated with the absence of an intratumoral immune response, which correlates with a worse clinical prognosis.

A fifth aspect of the invention relates to the in vitro use of the expression products of Table 1 as a biomarker predictive of therapeutic response to immunotherapy in the CNMP of stage I or II. A sixth aspect of the invention relates to a kit comprising the probes that recognize the messenger RNA, product of the expression of the genes of table 1, or the cRNA or cDNA to said mRNA, or antibodies that recognize a protein product of expression of the genes in table 1. The amount of probes used for each gene may vary in number. Preferably the kit comprises probes, which consist of probes that recognize the messenger RNA resulting from the expression of the genes in Table 1. More preferably the probes are the sequences described as SEQ ID NO: 1 to SEO ID NO: 66 and that specifically recognize the 50 genes in Table 1. Hereinafter we will refer to this kit as the "first kit of the invention".

A preferred embodiment of the sixth aspect of the invention relates to the kit which further comprises at least one probe or an antibody that recognizes an expression product of the genes in Table 2. Hereinafter we will refer to this kit as the "second kit of the invention".

Another preferred embodiment of the sixth aspect of the invention relates to the kit being able to comprise at least one retrotransciptase, or an RNA polymerase or a fluorophore. Thus, a preferred embodiment of the third aspect of the invention refers to a kit that also comprises at least one of the reagents selected from the list comprising: retrotranscriptase, an RNA polymerase or a fluorophore. In addition, the kit may comprise a mixture of tri-phosphate deoxynucleotides (dNTPs), a mixture of tri-phosphate nucleotides (NTPs), deoxyribonuclease (DNase), ribonuclease inhibitors (RNase), Dithiothreitol (DTT), inorganic pyrophosphatase (PPi) and the necessary buffers for the enzymes provided in the kit. In addition, the present invention also relates to the kit where the probes or antibodies are preferably located on a solid support, for example, but not limited to, glass, plastic, tubes, multiwell plates, membranes, or any other known support. Thus, a preferred embodiment of the sixth aspect of the invention refers to a kit where the probes or antibodies are preferably located on a solid support.

A seventh aspect of the invention relates to the use of the kit of the sixth aspect of the invention to obtain useful data for the prognosis of CNMP stages I or II. In addition, data collection may be useful for the administration of adjuvant treatment, for example chemotherapy. As regards the use of the first kit of the invention for the evaluation of the need to provide such treatment.

An eighth aspect of the invention relates to the use of the kit of the sixth aspect of the invention for obtaining useful data for the classification of the intratumoral immune response of the CNMP of stages I or II. The kit of the present invention can be used to know if there is an intratumoral immune response in the patient. A ninth aspect of the invention relates to the use of the kit of the sixth aspect of the invention to obtain useful data to predict the response to immunotherapy of the stage I or II CNMP.

Throughout the description and the claims the word "comprises" and its variants are not intended to exclude other technical characteristics, additives, components or steps. For those skilled in the art, other objects, advantages and features of the invention will be partly derived from the description and in part of the practice of the invention. The following examples and figures are provided by way of illustration, and are not intended to be limiting of the present invention. DESCRIPTION OF THE FIGURES

Fig. 1. Shows the probability of ILE in the two main histological subtypes of CNMP. Kaplan-Meier curve showing the probability of ILE of the two main histological subtypes of the CNMP, adenocarcinoma and squamous carcinoma in the development matrix. ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the subgroups analyzed are due to chance.

Fig. 2. Shows the probability of disease-free interval in stages I and II. . Kaplan-Meier curve showing the probability of ILE for stages I and II of CNMP in the development matrix. ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the subgroups analyzed are due to chance. Fig. 3. It shows the hierarchical grouping of the samples of the development matrix analyzed according to their molecular profile with 3,232 genes.

The hierarchical clustering of 84 samples with 3,232 genes is shown (see filtering 3 of example 1) according to the method described in Quackenbush J. Nat Rev Genet. 2001; 2 (6): 418-27. The samples are differentiated according to the histological subtype: continuous line, adenocarcinoma subtype; striped line, squamous subtype; continuous line terminated in ^* , other CNMP subtypes. "Molecular profile" is defined: as the set of genomic data (in our cases mRNA expression levels) capable of characterizing and identifying a subject or sample. The molecular subtypes found show a clear association with the histological subtypes of the tumors. Fig. 4. It shows the probability of ILE as a function of the molecular groups obtained in the development matrix from 3,232 genes.

Kaplan-Meier curve showing the probability of ILE of the two main molecular subtypes of the CNMP found in the development matrix. ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the subgroups analyzed are due to chance.

Fig. 5. It shows the hierarchical grouping of the samples of the development matrix analyzed according to their molecular profile with 2,160 genes.

Analysis of the overall gene expression pattern of tumors of the developmental matrix to obtain molecular groups using the list of 2,160 genes (see filtering 4 of example 1). A, the hierarchical molecular clustering of 84 samples with 2,160 genes is shown. B, grouping perfected by the "k-means" method described in Quackenbush J. Nat Rev Genet. 2001; 2 (6): 418-27. In both cases it results in three molecular groups (Group 1, 2 and 3) or "clusters".

Fig. 6. It shows the probability of ILE in the three molecular groups obtained based on their molecular profile with 2,160 genes in the development matrix. Kaplan-Meier curve showing the probability of ILE of the three molecular groups obtained using the list of 2,160 genes and the "k-means" technique. ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the analyzed subgroups are due to chance, (x), indicates the number of samples in each of the analyzed groups.

Fig. 7. Shows the probability of ILE in the validation matrix samples according to the classification of 3 molecular groups. Kaplan-Meier curve showing the probability of ILE for the validation matrix samples (external series, Roepman et al.) Grouped according to molecular profiles (Group 1, Group 2 and Group 3) previously observed in The development matrix and defined through a predictor of 1,000 genes generated with the application "PAM". ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the analyzed subgroups are due to chance, (x), indicates the number of samples that there are in each of the analyzed groups.

Fig. 8. Shows the probability of ILE in the validation matrix samples according to the classification established by the predictor of 50 genes. Kaplan-Meier curve showing the probability of ILE of the two molecular groups obtained in the validation matrix using the 50 gene predictor. ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the subgroups analyzed are due to chance, (x) indicates the number of samples in each of the branches of the Kaplan-Meier curve.

Fig. 9. Probability of ILE in the validation matrix samples according to the classification established by the predictor of 50 genes independently for stages I and II. Kaplan-Meier curve showing the probability of ILE of the two molecular groups obtained in the validation matrix with the 50 gene predictor generated with the "PAM" application for: A, stage I, and B, stage II. ILE, disease free interval; p, is the probability associated that the differences found in the ILE between the subgroups analyzed are due to chance, (x) indicates the number of samples in each of the branches of the Kaplan-Meier curve.

EXAMPLES

The following specific examples provided in this patent document serve to illustrate the nature of the present invention. These examples are included for illustrative purposes only and should not be construed as limitations on the invention claimed herein. So, The examples described below illustrate the invention without limiting its scope.

Example 1: Obtaining the 50 gene predictor

Materials and methods

1 .1 .1 Patient selection

 This study included 84 patients (12 women and 72 men with a mean age of 66.5 -range 36-82 years) diagnosed in early stages (60 stage I patients and 24 stage II patients) of CNMP during the years 2001 to 2008 at the San Carlos Clinic Hospital (HCSC) in Madrid. All patients met the following inclusion criteria: patients with completely resected tumors, no mediastinal lymph node involvement, no chemotherapeutic treatment and of which there was frozen tumor material in the HCSC biobank belonging to the RETICS subprogram of the Carlos III Health Institute (number of file RD090076 / 0102). The data collected for the study are divided into clinical data of the patient (age of diagnosis, sex and smoking habit) and histological data of the tumor (histological subtype, tumor size, tumor stage -7- TNM Classification (Kligerman S. American Journal of Roentgenology 2010. 194: 562-573) -, degree of differentiation, keratinization, presence of polymorphonuclear lymphocytes - PMN-, lymph node involvement, k-ras mutations, necrosis, tumor stroma, chronic inflammation, presence of intratumoral lymphocytes -TIL-, location by pulmonary lobes and type of recurrence - regional or distant madness).

1 .1 .2. Tumor samples RNA extraction and purification.

Following the freezing protocol of the samples included in the HCSC biobank, CNMP tumors were collected immediately after surgery and frozen and stored at -80 ^Q C. Histopathological review of the frozen tumors was carried out with the so that all patients included in the study had a tumor representation as a minimum of 70% in the sample used. At the same time, samples of non-tumor pulmonary parenchyma were also collected from these same patients, which were also frozen following the same protocol. RNA from these latter samples was used to create the control sample (a pool of normal tissue RNA). In all cases, the total ribonucleic acid (RNA or RNA) was extracted directly from the frozen samples using Trizol® and a tissue homogenizer. He was subsequently treated with DNAse and quantified in the NanoDrop ND-1000® spectrophotometer. The quality of the extracted RNA was measured in Bioanalyzer 2100® using the RIN (or RNA Integrity Number) and only samples with a good RNA quality (RIN> 7.5), were included for the study.

1 .1 .3. Expression profile by microarray.

The expression profile of the 84 tumors was determined using Agilent® full genome oligonucleotide microarrays (G41 12F) following the protocol provided by the manufacturer. Briefly, double marking was used, with cyanine-5 (Cy5) for each of the 84 tumors included in the study and with cyanine-3 (Cy3) for the control sample, consisting of a "pool" of 42 parenchyma samples not lung tumor. This control sample was introduced in each of the experiments (the same in all of them) in order to identify and correct the technical variations introduced during the experimental phase of the analysis. After this correction (called normalization) the data generated is the ratio between the fluorescence of the tumor and the control sample. "Spikelns", which are 10 control transcripts synthesized in vitro that derive from the adenovirus E1 A transcriptome, that do not interact with human mRNA and whose initial concentration is known, were included during the stages of screening and hybridization. The "a priori" knowledge of the initial concentration of each of the "Spikelns", allows us to predict at what level of fluorescence these transcripts should emit once hybridized in the microarray and therefore can be used as quality control of the experimental phase . The microarrays were scanned and quantified using the Agilent® scanner and the Feature Extraction® program (10.7.3) respectively. For the normalization of the extracted data, the Lowess or "Locally Weighted Scatterplot Smoothing" technique was used. {Cleveland WS: Journal of the American statistical Association 1979, 74: 829-836; Cleveland WS, et al. Journal of the American Statistical Association 1988, 83: 596-610.)

1 .1 .4. Analysis of data

 To obtain the method of the invention, an initial list of 41,000 probes present in the Agilent® full genome oligonucleotide microarray was started. From a filtering process, a molecular classification was reached that ultimately resulted in the creation of the predictor of the invention composed of only 50 genes. The method was developed following the following filtering steps:

1 .- Filtered by "flags": exclusion of probes with low fluorescence or problems during the hybridization process in more than 10% of the samples. The new listing included 24,617 probes.

2.- Average of the probes with the same identifier in order to work with unique expression values for each gene. The new listing included 17,881 genes.

3.- Filtering by expression: gene selection with a variation of expression at least 3 times with respect to the median of that gene in at least 10% of the samples. The new listing included a total of 3,232 genes (Fig. 3). Once the molecular groups were generated from this list of 3,232 genes, the molecular classification obtained was evaluated to determine whether or not there was an association with the disease-free interval (ILE) (Fig. 4). 4.- Histological filtering: the genes that characterize the histological differences between the main histological subtypes of the CNMP (adenocarcinoma and squamous carcinoma) were eliminated. For this, the differentially expressed genes (p-value <0.01 and expression difference> 1, 5) were selected and the generated list (1,072 genes) was excluded from the initial list (3,232 genes). Therefore, a list of 2,160 genes (genes shown in Table 2) that are used for the final molecular classification of the 84 tumors is generated. The strategy used for the discovery of the molecular groups consisted in applying first a method of unsupervised analysis, clustering or hierarchical clustering (Fig. 5A), and then an improvement of the molecular groups obtained by means of a second method, method of k-Means (Fig. 5B), which allows to reduce intra-group heterogeneity and increase inter-group variability. The list of 2,160 genes is used to initially construct the molecular classification (which has 3 groups). Once these molecular groups were generated, the molecular classification obtained was evaluated to determine whether or not there was an association with the disease-free interval (ILE) (Fig. 6). The ILE is defined as the time that elapses from the date of surgery until the patient's recurrence is confirmed.

In the statistical analysis, Kaplan-Meier curves and the log-rank test have been used to assess the probability of each molecular subtype with respect to recurrence (Clark TG. British Journal of Cancer 2003. 89: 232-238). In addition, the "hazard ratio" for molecular groups is calculated using the Cox proportional regression method.

Likewise, an analysis of the molecular pathways that are significantly altered between the molecular groups obtained was performed. It was carried out using the GSEA tool ("Gene Set Enrichment Analysis" or gene set enrichment analysis) (Subramanian A et al. PNAS 2005 102 (43) 15545-15550 and Mootha VK et al. Nat Gen 2003). Only molecular pathways with a minimum representation of more than 15 genes were evaluated and 100,000 permutations were used to ensure the results. In order to obtain the GSEA results, the original list of 17,881 genes was used since when analyzing molecular pathways, all available genes that meet quality controls (17,881 genes) should be included, since non-significant differences in expression in a group of genes they can nevertheless be keys, to define which signaling paths ("pathways") are altered between the groups.

1 .1 .5. Validation of the 3 molecular groups in an external series.

 For the validation of the molecular classification obtained, the data matrix published by the Roepman et al. Group (Roepman P et al. Clin Cancer Res 2009. 15: 284-290) was used. The validation matrix includes the expression data of 162 patients diagnosed with the same histological subtypes as those of the invention.

 The term "training matrix" or "development matrix" refers to the HCSC biobank samples (n = 84). The term "validation matrix" refers to the set of samples published by Roepman et al used for the validation of molecular classification. "Matrix" means the set of expression data obtained in a series of patients using microarrays.

For validation a common data matrix has been generated that includes 246 samples (84 of the development matrix + 162 of the validation matrix) each with 17,881 genes. With the development matrix, a predictor was obtained through the PAM application (Microarray Prediction Analysis) (Tibshirani R. et al. PNAS 2002; 99 (10): 6567-72) that was evaluated in the validation matrix by studying its association, through the Kaplan-Meier curve, with the ILE. The Cox proportional regression model was used to confirm the prognostic power of our predictor.

1 .1 .6. Obtaining and validating the Predictor of 50 genes. The 3 molecular groups generated by histological filtering were grouped into 2 groups, a good prognosis group or group 3 and a poor prognosis group or a 1 + 2 group, due to the prognostic similarity of both molecular groups. Thus, with two groups and starting from 2,160 genes, 50 genes (the genes of the invention shown in Table 1) capable of classifying new samples based on these two prognostic groups in CNMP are selected by PAM application. stages I or II. Based on this predictor of 50 genes, the validation matrix samples were classified in the group with a good prognosis or a group with a poor prognosis. The Kaplan-Meier curves and the Cox proportional regression model were used to validate the prognostic power of our predictor (Figs. 8, 9A and 9B).

1 .1 .7. Explanation of the analysis with PAM (Tibshirani R. et al. PNAS. 2002, 99: 6567-6572).

To exemplify this description, we will use as an example the creation of the 50 gene predictor for two molecular groups (good and bad prognosis) mentioned in the previous section. Thus, using the PAM application as a classification tool, the forecast classification process requires as a starting point the calculation of a "reference value" for each of the two groups. These "reference values" are obtained from the samples of the patients that make up the so-called "training matrix" or "development matrix" and of which "a priori" their classification is known (as they were with them with the that defined what the group of good and bad prognosis was). From the patients of the group of good prognosis we will obtain the "reference value 1" and from the patients of the group of poor prognosis we will obtain the "reference value 2". Each of the reference values will be expressed as a vector of 50 components (one for each of the genes of the invention) and will be calculated as the sum of two subvectors each also expressed with 50 components. The first subvector is common for the two reference values while the second is specific for each of the two reference values to be calculated. The first subvector consists of 50 components, each of which corresponds to the average expression value of one of the 50 genes throughout all the samples that make up the training or development matrix regardless of the group in which they are classified (i.e. the 84 tumors of our matrix). The second subvector will also be defined by 50 components (each of which represents a gene) that will be defined by a "t" statistic that compares for that gene the differences between the first subvector and the average expression value of that gene in the Samples included in the group for which the reference value is to be calculated (either the good prognosis group (29 samples) or the poor prognosis group (55 samples)). The data of the second subvector will be standardized taking into account the variability of expression of each of the 50 genes within the analyzed group and taking into account a convergence value Δ that allows to evaluate the predictive power of each of the genes. The aforementioned transformations will mean that although the "reference value" or "shrunken centroid" obtained for each group is based on expression values, its actual value is dimensionless and is not a reflection of the initial fluorescence data of each sample. Once the "reference value" or "shrunken centroid" is calculated for each group, the PAM is able to assign the new samples, which in this example formed the validation matrix (162 samples), to each of the previously defined groups . The application of this invention to know the prognosis of the new patients is done by calculating the distance between the expression values of the 50 genes of the new sample with respect to the 50 components of the "reference value" or "shrunken centroid" of each group. If the distance between the new sample and the "reference value 1" is less than the distance between the new sample and the "reference value 2", the favorable prognosis for the new patient can be determined. On the contrary, if the distance between the new sample and the "reference value 1" is greater than the distance between the new sample and "reference value 2", the unfavorable prognosis for the new patient can be determined. During these last calculations, factors that correct the result are also introduced, taking into account the variability of expression within the groups and the probability of belong to a certain group taking into account their sample size with respect to the population analyzed. The quantification of distances is measured using the Euclidean distance. 1.2.- RESULTS

1 .2.1. Association analysis of the ILE with the clinical and histopathological variables.

 A first statistical analysis was carried out to verify if there was an association between the most important histopathological variables in the routine management of the CNMP (the histological classification of the tumor, the stage, etc.), with the ILE. The Kaplan-Meier curves obtained did not show a statistically significant association of the ILE with the histopathological type (Fig. 1), the stage (Fig. 2) or with any other variable analyzed (data not shown). Only the presence of mutations in the K-Ras gene showed a tendency towards association with a worse prognosis (p = 0.07).

1 .2.2. Molecular groups from 3,232 genes.

 By means of the hierarchical clustering method (Pearson and Average linkage centering (Quackenbush J. Nat Rev Genet. 2001; 2 (6): 418-27), two main molecular subtypes are identified that show a clear association with the most represented histological subtypes in our series, molecularly separating, tumors of the adenocarcinoma subtype of squamous subtype tumors (Fig. 3). These 2 molecular subtypes do not show statistically significant differences with the ILE (p = 0.350) (Fig. 4).

In view of these results, we conclude that the molecular groups obtained using the list of 3,232 genes (which are the genes that vary their expression at least 3 times with respect to the median of that gene in at least 10% of the samples; step 3 of the filtrate explained above) are conditioned by the histology of the tumors. Importantly, there are no statistically significant differences over time. of recurrence when both molecular groups are compared and remember that they did not exist when both groups classified according to histological criteria were compared. Taking into account that the oncological criterion for the management of CNMP patients indicates that the histology of the tumors is only important in metastatic disease (stage IV) and only in relation to the indicated treatment, we exclude from the initial list of 3,232 genes those that characterize the histological differences of the 84 tumors by means of a filtrate that included: T-Test ap <0.01 with correction for multiple comparisons of Benjamini and Hochberg (B&H) (Benjamini Y and Hochberg Y. Journal of the Royal Statistical Society. 1995) and an expression difference of more than 1, 5 times. The genes that met these filtering criteria were excluded, resulting in a list of 2,160 genes that were used to obtain the molecular classification and which are the genes included in Table 2.

1 .2.3. Molecular groups with 2,160 genes. Association with ILE.

After the grouping of the 84 patients according to the gene expression profile using the list of 2,160 genes and the hierarchical clustering method (Fig. 5A) subsequently perfected by the k-means method, 3 molecular groups were obtained that were named as Group 1, Group 2 and Group 3 (Fig. 5B).

These three groups were statistically significantly associated with the ILE (log-rank p = 0.004), showing in the Kaplan-Meier curve, 2 molecular groups with a poor prognosis regarding recurrence (Group 1 and Group 2) and one group. molecular prognosis (Group 3) (Fig. 6). The "Hazard ratio" (HR, that is, the risk or probability of relapse that one group has with respect to another) of the groups of poor prognosis versus the group of good prognosis is 6.4 for Group 1 (IC 95 %: 1, 8-22.3; p = 0.004) and 4.9 for Group 2 (95% CI: 1, 4-17.8; p = 0.014). There is no statistically significant difference in risk between Group 1 and Group 2 (p = 0.526). a) - Multivariate analysis

 In this analysis, mutations for k-ras were included due to a tendency (p = 0.07) for association with the ILE and the classification by Stage since it is the main prognostic factor for the CNMP.

After adjusting for Stage and K-ras status, Cox's multivariate proportional hazards model confirmed the molecular classification as an independent prognostic factor to assess the risk of recurrence (HR Group 1 vs. 3 = 1 1 .170; 95 % Cl: 2.9 to 43.4; p = 4.9E-04; HR Group 2 vs. 3 = 7.521; 95% Cl: 2.0 to 28.8; p = 0.003); HR Group 1 vs. 2 = not significant). b) -. Study of molecular pathways.

It was observed that the molecular classification in 3 groups was related to the involvement of molecular pathways related to the immune system such as the T-cell, B-cell, Inflammation and Th1 response pathway that differentiate Group 3 from Group 2 and especially Group 3 from the Group one . On the other hand, the alteration of genes involved in cell cycle pathways and DNA repair mechanisms confers the main biological differences between Group 2 and Group 1. c) - Statistical analysis of the clinical and histological variables included in the study.

Regarding the clinical variables of the patients included in the study, smoking was associated statistically significantly with the molecular classification obtained (p = 0.002). In the case of histological variables of the tumor, lymph node involvement (p = 0.041), despite having only 3 patients diagnosed with N1, and chronic inflammation (p = 0.001) are also statistically significantly associated with molecular subtypes. . d) - External serial validation and obtaining a predictor for 3 molecular groups.

 Using the development matrix (84 tumors), through the use of PAM, a first predictor of 1,000 genes was identified that identified the patients in the 3 molecular groups, two of poor prognosis (group 1 and group 2) and one of Good prognosis (group 3). For the evaluation of the prognostic power of said predictor, the data of the 162 tumors of the validation matrix were used. These samples were classified in the 3 molecular groups using said predictor (1, 000 genes). The Kaplan-Meier curve for the validation matrix samples revealed a statistically significant association of these three molecular groups with the ILE (log-rank p = 0.022) (Fig. 7). The "Hazard Ratio" (HR) of the groups of poor prognosis versus that of good prognosis is 2.4 times for Group 1 (p = 0.012) and 2.5 times for Group 2 (p = 0.019).

1 .2.4. Obtaining the predictor of 50 genes.

As previously observed in the results obtained in the development matrix, of the three groups obtained through gene expression analysis, the behavior of Group 1 and Group 2 is similar with respect to recurrence, there is no significant statistical difference for risk. between these two groups (p = 0.526). Therefore, both groups were included in only one and a second predictor of 50 genes was generated, using PAM, to differentiate patients with poor prognosis (Group 1 and 2) and patients with good prognosis (Group 3). Table 3 includes the value of the compact centroid ("shrunken centroid") for the groups of good and bad prognosis obtained with the samples of the development matrix. This second predictor encompasses the so-called "50 genes of the invention" (see table 1) and the evaluation of the prognostic power of the same was carried out again in the validation matrix, obtaining the Kaplan-Meier curves that show an association statistically significant of the two groups obtained with the ILE (log-rank p = 0.001) (Fig. 8). The HR for the Group with a poor prognosis is 3.4 compared to the one with a good prognosis (95% CI: 1, 6-7.3; p = 0.001). 1 .2.5. Utility of the predictor of 50 genes in CNMP separated by stage.

 One of the main criticisms of the state of the art (Subramanian J. et al. J Nati Cancer Inst 2010; 102: 1 -1 1) regarding the usefulness of the predictors generated for the CNMP is that it is necessary to demonstrate its usefulness for predict the prognosis of patients independently of the stage in which they were classified. To do this, we separated the 162 patients from the validation matrix in patients classified in stage I (1 10 patients) and classified as stage II (52 patients). The 50 gene predictor was used to obtain the high and low risk molecular groups (i.e., poor and good prognosis, respectively) and their association with the ILE was studied using the Kaplan-Meier curves. Both in stages I separately (Fig. 9 A) and in stages II (Fig. 9 B) a statistically significant association of the groups with the ILE was observed (p = 0.013 and p = 0.029 respectively) and the HR of the bad group prognosis with respect to the good prognosis were in stage I of 3.2 (95% CI: 1, 2-8.3; p = 0.018) and in stage II of 3.5 (95% CI: 1, 1 - 12; p = 0.041).

1 .2.6. Sensitivity and specificity of the 50 gene predictor.

 The sensitivity and specificity values of the predictor for the classification of the samples in the molecular groups identified are shown in Table 4.

Therefore, and based on the results shown, the present invention demonstrates the usefulness of the method of the invention, as well as the use of the invention. genes described in table 1 as prognostic markers of stage I or II CNMP.

Below is the table 2 referred to previously. When the "ID Entrez" is not indicated or is genes that have no information in the NCBI database and in which the name of the genome oligonucleotide microarray probe has been indicated in the gene symbol complete used (Agilent®, G41 12F).

Claims

one . In vitro method of obtaining useful data for the prognosis of stage I or II non-small cell lung cancer characterized by the detection and / or quantification of the expression product of the genes of table 1 in the isolated biological sample of a subject.

2. A method according to claim 1 further comprising comparing the useful data with reference expression values for the expression product of the genes of table 1 in stage I or II non-small cell lung cancer obtained from subjects in the that the prognosis is known (reference sample) for identification of the subject as a subject of good prognosis or poor prognosis.

3. Method according to claims 1 or 2 wherein the comparison is made by the method of the nearest compact centroid.

4. In vitro method for the prognosis of stage I or II non-small cell lung cancer characterized by:

 to. the detection and quantification of the expression product of the genes of table 1 in a reference sample;

 b. the calculation of a reference value (value 1) for each expression product of the genes in table 1 in the favorable prognostic reference samples (good prognosis group) and the calculation of a reference value (value 2) in unfavorable prognosis reference samples (poor prognosis group) by using the nearest centroid method;

 C. the detection and quantification of the expression product of the genes of table 1 in the biological sample of a new subject in which the prognosis is unknown (study sample);

d. comparison by using the closest compact centroid classification method of the values obtained in the detection and quantification of the expression product of the genes of table 1 in the study sample with the reference values obtained in the groups of good and bad prognosis, e. the association of the study sample to the group of good prognosis or the group of poor prognosis as established in the method of the nearest compact centroid.

5. Method according to claim 4 wherein the nearest centroid method is carried out through the application of Microarray Analysis Prediction (PAM).

6. Method according to any one of claims 1 to 5 wherein the reference sample and the study samples have been previously normalized before comparison.

7. Method according to any one of claims 1 to 6 which further comprises the detection and / or quantification of at least one expression product of the genes described in Table 2.

8. Method according to any one of claims 1 to 7 wherein the expression product is messenger RNA.

9. Method according to claim 8 wherein the detection and / or quantification of messenger RNA is performed by microarrays.

10. Method according to claim 8 wherein the detection and / or quantification of messenger RNA is performed by RT-PCR.

eleven . Method according to any one of claims 1 to 7 wherein the expression product is a protein.

12. Method according to claim 1 wherein the detection and / or quantification of the protein is carried out by immuno blotting, immunohistochemistry, chromatography or microarray.

13. A method according to any of claims 1 to 12 wherein the biological sample is selected from the list comprising: tissue, blood, plasma, serum, lymph, bronchoalveolar lavage or ascites fluid.

14. Method according to any of claims 1 to 13 wherein the biological sample is fresh, frozen, fixed or fixed and embedded in paraffin.

15. Method according to any of claims 1 to 14 wherein the subject is a human.

16. In vitro use of the expression products of the genes in Table 1 as a prognostic marker for stage I or II non-small cell lung cancer.

17. In vitro use of the expression products in Table 1 to classify the intratumoral immune response in stage I or II non-small cell lung cancer.

18. In vitro use of the expression products of Table 1 as a biomarker predictor of therapeutic response to immunotherapy in stage I or II non-small cell lung cancer.

19. Kit comprising probes consisting of probes that recognize the messenger RNA, product of the expression of the genes in table 1, or the complementary DNA or RNA complementary to said messenger RNA, or antibodies that recognize an expression product protein of the genes in table 1.

20. Kit according to claim 19 comprising probes, which consist of probes that recognize the messenger RNA resulting from the expression of the genes in Table 1.

twenty-one . Kit according to claim 19 wherein the probes are the sequences SEQ ID NO: 1 to SEQ IDNO: 66.

22. Kit according to any of claims 19 to 21 which further comprises at least one probe or an antibody that recognizes an expression product of the genes in Table 2.

23. Kit according to claim 22 comprising at least one probe that recognizes an expression product of the genes in Table 2.

24. Kit according to any of claims 19 to 23 which further comprises at least one of the reagents selected from the list comprising: a retrotranscriptase, an RNA polymerase or a fluorophore.

25. Kit according to any of claims 19 to 24 wherein the probes are located on a solid support.

26. Use of the kit according to claims 19 to 25 for obtaining useful data for the prognosis of stage I or II non-small cell lung carcinoma.

27. Use of the kit according to claims 19 to 25 for obtaining useful data for the classification of the intratumoral immune response of stage I or II non-small cell lung carcinoma.

28. Use of the kit according to claims 19 to 25 to obtain useful data to predict the immunotherapy response of stage I or II non-small cell lung carcinoma.