Signature for the prognosis of brain metastasis relapse

Abstract

The present invention relates to a method for the prognosis of brain metastasis relapse in a subject and to a kit useful in said prognosis.

Description

[0001] SIGNATURE FOR THE PROGNOSIS OF BRAIN METASTASIS RELAPSE

[0002] FIELD OF THE INVENTION

[0003] The present invention belongs to the field of medicine, particularly to the field of diagnosis or prognosis, more particularly, the present invention relates to a method for the prognosis of brain metastasis relapse and to a kit useful in said prognosis.

[0004] BACKGROUND OF THE INVENTION

[0005] Brain metastasis is the most common neurological complication of cancer. When metastatic cells reach the brain, prognosis is poor given that local therapies (i.e., surgery and radiation) have limited benefits for patients and the disease inevitably progresses.

[0006] The true prevalence of brain metastasis is not well known but is probably higher than the available epidemiological estimates (8.5-9.6%) and is certainly more frequent than any other primary tumor in the brain. Upon diagnosis, cancer patients with disease disseminated to their brains continue to face a dismal prognosis, which includes higher morbidity, mortality, and a dramatic increase in the cost of their treatment.

[0007] The incidence of brain metastasis continues to increase and yet, current therapies available for patients with disseminated cancer cells in their central nervous system (CNS) have a limited efficacy and fail to improve survival. Consequently, during the past years, there have been recurrent efforts to improve clinical trial design and management specifically concerning this patient population. However, the inclusion of patients with active CNS disease has been limited in the trials of the past, and this represents an unsolved issue.

[0008] The rise in the number of patients with brain metastasis is partially due to the increasing number of systemic therapies that work extra-cranially but are unable to provide the same therapeutic benefit in the brain. Consequently, cancer cells present at this secondary site have additional time to evolve and to grow into clinically detectable lesions.

[0009] Surgical resection has a well-established role in the management of brain metastasis. While much effort has been done to determine markers useful for determining the risk of developing brain metastasis from different primary cancers, not much progress has been made in the obtention of tools for analyzing the risk of relapse of a brain metastasis, and in particular of those removed by surgery.

[0010] Some representative methods for predicting risk of brain metastasis in patients based on gene signatures, are displayed in the examples below.

[0011] US2021371932 provides a lengthy list of markers that might be related to the prognosis of brain metastasis in patients. However, no reference to prognosis of brain metastasis relapse after brain surgery is even suggested.

[0012] WO2012031008 also displays a lengthy list of markers which might be relevant in the diagnosis, prognosis and monitoring of a disease or other medical condition. As in the example of above, no reference as to the relevance of the list of markers for the prognosis of brain metastasis relapse after brain surgery is mentioned.

[0013] WO2011068839 discloses methods for predicting risk of metastasis and / or cancer recurrence based on gene expression profiling. WO2011068839 discloses gene DPYSL3 in Bladder Cancer Module, Breast Cancer Module 23, Pancreatic Cancer Module 2, Prostate Cancer Module 4, Sarcoma Module 2, Liver Cancer Module 2 and Lymphoma Module 7; and gene LMO7 in Brain Cancer Module 2 and Lung Cancer Module 9. WO2011068839 is silent about brain metastasis relapse.

[0014] US2010029748 discloses two sets of genes, one set of 17 genes (table 1) and one set of 18 genes (table 2) that can be used in predicting the risk of cancer metastasis to the brain, and as a screening assay to identify the suitable treatments for brain metastases. Particularly, the genes in table 1 are: overexpression of ANGPTL4, PLOD2, COL13A1 , PTGS2, PELI1 , MMP1 , B4GALT6, HBEGF, CSF3, RGC32, LTBP1 , FSCN1 , and LAMA4 and underexpression of TNFSF10, RARRES3, SCNN1A and SEPP1 are indicative of an increased likelihood of brain metastases (17 genes set). Also, the genes of table 2 are PCDH7, ARHGDIB, ST6GALNAC5, ODZ3, SERPINI1 , GJA1 , GARNL4, PRSS3, SERPINE2, KTELC1 , RGS2, IL1 B, MMP3, TMEM156, FAM59A, IL1A, MOCS1 , TLR4.

[0015] In US2010029748, gene CYP1 B1 is listed in Table 4 as a gene which is differentially expressed by more than 2.5-fold between parental MDA231 and its brain metastatic derivatives, among other 178 genes. T able 3 of US2010029748 discloses another set of 271 genes differentially expressed by more than 2.5-fold between parental CN34 and brain metastatic derivatives. LIS2010029748 does not relate to a signature for predicting the risk of relapse or recurrence of brain metastasis.

[0016] Kamer et al. (Transl Lung Cancer Res 2020;9(3):682-692) discloses a 22 gene signature for predicting the risk of developing brain metastasis, not for brain metastasis relapse. None of the 22 genes are any of the genes of the signature of the present invention.

[0017] Of all the documents of the state of the art, just Soike et al. (Scientific Reports (2019) 9:14385) seems to point to CD138 as a biomarker for recurrence of brain metastasis after surgery.

[0018] There is the need to predict the risk of relapse of brain metastasis after treatment and in particular after surgery. The present invention provides a signature that allows the significant determination of the risk of relapse of brain metastasis after treatment and in particular, after surgery.

[0019] DESCRIPTION OF THE INVENTION

[0020] The present invention is based on a gene signature for the prognosis of brain metastasis relapse in a patient. The core gene signature is composed of genes LMO7 and / or DPYSL3, and / or at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4. The inventors have found that analyzing the expression levels of at least genes LMO7 and / or DPYSL3 and / or at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4 in an isolated sample from a subject who underwent treatment, for example brain surgery, constitutes a reliable method for the prognosis of brain metastasis relapse in the subject.

[0021] More precisely, the inventors have found that when the expression levels of genes LMO7 and / or DPYSL3 and / or at least one of the genes selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4 are compared with those of reference populations with and / or without relapse, an increased expression of at least a gene from LMO7, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4 and / or a decreased expression of at least a gene from DPYSL3, CALM2 and / or MYL6 with respect to a non-relapse reference population, is associated with a higher risk of relapse.

[0022] Measurement of the expression levels of the core signature of genes LMO7 and / or DPYSL3 and / or at least one of the genes selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4 might be combined with measurement of expression levels of additional genes. In particular, with measurement of expression levels of JAGN1.

[0023] The signatures of the present invention are therefore the basis of a reliable methods and kits for the prognosis of brain metastasis relapse after treatment, and more particularly, after brain surgery.

[0024] None of the documents of the state of the art disclose or even suggest the relevance of the signature of genes of the present invention for prognosis of brain metastasis relapse in an individual.

[0025] The first aspect of the present invention relates to a method for the prognosis of brain metastasis relapse in a subject, wherein the method comprises analyzing the expression levels of at least genes LMO7 and / or DPYSL3 in an isolated sample from the subject. The method of the present invention is particularly useful for the prognosis of brain metastasis relapse after treatment in a subject, and in particular, after brain surgery. In a preferred embodiment, the method comprises analyzing the expression level of the genes LMO7 and DPSYL3.

[0026] The second aspect of the present invention relates to a method for the prognosis of brain metastasis relapse in a subject, wherein the method comprises analyzing the expression levels of at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4 in an isolated sample from the subject. The method of the present invention is particularly useful for the prognosis of brain metastasis relapse after treatment in a subject, and in particular, after brain surgery. In a preferred embodiment, the method comprises analyzing the expression level of the genes NDEL1, HMGB2, SH3GLB1 , CALM2, MYL6, RAP1B and KPNA4. In an even more preferred embodiment, the method comprises analyzing the expression level of the genes LMO7, NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1 B and KPNA4; or DPYSL3, NDEL1 , HMGB2, SH3GLB1, CALM2, MYL6, RAPIB and KPNA4; or LMO7, DPYSL3, NDEL1 , HMGB2, SH3GLB1, CALM2, MYL6, RAP1B and KPNA4.

[0027] The third aspect of the present invention relates to a kit comprising means for analyzing the expression levels of genes LMO7 and / or DPYSL3, and / or at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1B and / or KPNA4 in a sample.

[0028] A fourth aspect of the present invention relates to the use of any of the methods or of the kits of the present invention for the prognosis of brain metastasis relapse in a subject. And in particular, after subjecting the subject to treatment of such brain metastasis, and in particular, by brain surgery.

[0029] In a preferred embodiment, the expression levels of genes LMO7 and / or DPYSL3 and / or the expression level of genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4 are compared with those of reference populations with and / or without relapse. An increased expression of any of the genes LMO7, NDEL1 , HMGB2, SH3GLB1 , RAP1 B, KPNA4 and / or a decreased expression of any of the genes DPYSL3, CALM2 and MYL6 when compared with a non-relapse reference population, is associated with a higher risk of relapse.

[0030] In another preferred embodiment, the methods and kits further involve analysis of the expression levels of gene JAGN1. In such methods, a decreased expression of gene JAGN1 , when compared with a non-relapse reference population, is associated with a higher risk of relapse.

[0031] By detection of the expression level of a gene it is meant detection of its mRNA level, but also detection of any corresponding translation product issuing it (e.g. the corresponding protein). The gene expression analysis may be performed in a sample by any possible means. Preferably by detection of mRNA levels by qPCR, preferably by RT-qPCR, by in situ hybridization, preferably RNAscope, and / or by protein detection preferably with antibodies.

[0032] In another embodiment, the methods and kits further comprise analyzing the expression of one or more additional genes, and / or of one or more reference genes in the sample. The one or more additional genes can be selected from CYP1 B1 and / or EDARADD, and from any other set of genes considered to be relevant in relapse of brain metastasis. In the particular case of genes CYP1 B1 and EDARADD, their decreased expression when compared with a non-relapse reference population, is associated with a higher risk of relapse.

[0033] Gene LMO7 (LIM domain only protein 7) has Gene ID: 4008, (updated on 8- May-2022) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene LMO7 in the Ensembl database is ENSG00000136153 and the reference in the Uniprot database is Q8WW11 .

[0034] Gene DPYSL3 (dihydropyrimidinase like 3) has Gene ID: 1809, (updated on 24-Apr-2022) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene DPYSL3 in the Ensembl database is ENSG00000113657 and the reference in the Uniprot database is Q14195. Gene DPYSL3 is also known as TUC4. Gene JAGN1 (jagunal homolog 1) has Gene ID: 84522, (updated on 8-May-2022) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene JAGN1 in the Ensembl database is ENSG00000171135 and the reference in the Uniprot database is Q8N5M9.

[0035] Gene CYP1 B1 (cytochrome P450 family 1 subfamily B member 1) has Gene ID: 1545, (updated on 24-Apr-2022) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene CYP1 B1 in the Ensembl database is ENSG00000138061 and the reference in the Uniprot database is Q16678.

[0036] Gene EDARADD (EDAR associated death domain) has Gene ID: 128178, (updated on 24-Apr- 2022) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene EDARADD in the Ensembl database is ENSG00000186197and the reference in the Uniprot database is Q8WWZ3.

[0037] Gene NDEL1 (Nuclear distribution element-like 1) has Gene ID 81565 (updated on 17-Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene NDEL1 in the Ensembl database is ENSG00000166579 and the reference in the Uniprot database is Q9GZM8.

[0038] Gene HMGB2 (High Mobility Group Box 2) has Gene ID: 3148 (updated on 17-Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene HMGB2 in the Ensembl database is ENSG00000164104 and the reference in the Uniprot database is P26583.

[0039] Gene SH3GLB1 (SH3 Domain Containing GRB2 Like, Endophilin B1) has Gene ID: 51100 (updated on 17-Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene SH3GLB1 in the Ensembl database is ENSG00000097033 and the reference in the Uniprot database is Q9Y371.

[0040] Gene CALM2 (Calmodulin 2) has Gene ID: 805 (updated on 17-Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene CALM2 in the Ensembl database is ENSG00000143933 and the reference in the Uniprot database is P0DP24.

[0041] Gene MYL6 (Myosin Light Chain 6) has Gene ID: 4637 (updated on 17- Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene MYL6 in the Ensembl database is ENSG00000092841 and the reference in the Uniprot database is P60660.

[0042] Gene RAP1 B (RAP1 B, Member Of RAS Oncogene Family) has Gene ID: 5908 (updated on 29- Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene RAP1 B in the Ensembl database is ENSG00000127314 and the reference in the Uniprot database is P61224.

[0043] Gene KPNA4 (Karyopherin Subunit Alpha 4) has Gene ID: 3840 (updated on 17-Jun-2024) in the NCBI (National Center for Biotechnology Information) Entrez Gene Database. The reference for gene KPNA4 in the Ensembl database is ENSG00000186432 and the reference in the Uniprot database is 000629.

[0044] The methods and kits might comprise means for analyzing the expression levels of at least one gene selected from the group consisting of LMO7, DPYSL3, LMO7 and DPYSL3; NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, , the group of genes consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, the group of genes consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4 and DPYSL3, the group of genes consisting of of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4, LMO7 and DPYSL3, the group genes consisting of LMO7, DPYSL3 and JAGN1 , the group genes consisting og LMO7 and JAGN1 , the grpup of genes consisting of DPYSL3 and JAGN1 , the group of genes consisting of DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4 and JAGN1 , the group of genes consisting of LMO7, DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4 and JAGN1 , the group of genes consisting of any of the combination of genes as disclosed above and / or one or both of genes CYP1 B1 and EDARADD, of up to 6 genes, of up to 7 genes, of up to 8 genes, of up to 9 genes, of up to 10 genes, of up to 20 genes, of up to 30 genes, of up to 40 genes, of up to 50 genes, of up to 60 genes, of up to 70 genes, of up to 80 genes, of up to 90 genes, of up to 100 genes.

[0045] In a preferred embodiment, the one or more additional genes to LMO7 and / or DPYSL3 is at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4. Preferably, the one or more additional genes consists of the genes NDEL1 , HMGB2, SH3GLB1, CALM2, MYL6, RAP1 B and KPNA4.

[0046] In another embodiment, the expression levels of genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4 are also compared with those of reference populations with and / or without relapse. An increased expression of genes NDEL1 , HMGB2, SH3GLB1 , RAP1 B or KPNA4 when compared with a non-relapse reference population, is associated with a higher risk of relapse, and / or a decreased expression of any of the genes CALM2 and MYL6 when compared with a non-relapse reference population, is associated with a higher risk of relapse.

[0047] In a preferred embodiment, the reference gene might be selected from actin, beta-2 microglobulin, cytochrome 01 , EMG1 N1 -specific pseudouridine methyltransferase, glyceraldehyde-3- phosphate dehydrogenase, TATA-box binding protein, topoisomerase (DNA) I, tyrosine 3- monooxygenase / tryptophan 5-monooxygenase activation protein zeta and combinations thereof, preferably is actin, beta-2 microglobulin or both.

[0048] As used herein, the expression “reference population” refers to either a population that has suffered brain metastasis and has not had any relapse after treatment and in particular after brain surgery (non-relapse reference population), or to a population that has suffered brain metastasis, which relapsed after treatment and in particular after brain surgery (relapse reference population).

[0049] The sample of the above methods may be a fluid sample, such as a blood sample or a cerebrospinal fluid sample, or a tissue sample, such as a biopsy, obtained when or after the subject is or has been subjected to treatment, and in particular, to brain metastasis surgery. In a preferred embodiment, the sample is a brain sample obtained by brain surgery of a brain metastasis. In a preferred embodiment, the sample is a cDNA sample (complementary DNA), preferably a cDNA obtained from an RNA sample. As used herein, DNA is Deoxyribonucleic acid and RNA is Ribonucleic acid.

[0050] In a preferred embodiment, the subject of the above methods is a human, preferably a human with brain metastasis from a primary cancer, for example, lung cancer, melanoma, colorectal cancer or breast cancer. In an example, the primary cancer is lung cancer, for example, nonsmall cell lung cancer, and in yet another example, lung adenocarcinoma. In another example, the primary cancer is breast cancer HER positive.

[0051] In a preferred embodiment, the gene expression analysis in the above methods is made by quantitative polymerase chain reaction (qPCR), preferably by RT-qPCR, by in situ hybridization, preferably RNAscope, and / or by protein detection preferably with antibodies. In another preferred embodiment, the gene expression analysis is made using RNA probes and / or primers.

[0052] In a more preferred embodiment, the kit comprises means for analyzing the expression levels of LMO7 and / or DPYSL3 and at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4. More preferably, the kit comprises means for analyzing the expression levels of LM07 and / or DPYSL3, and genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4. In another embodiment, the kit comprises means for analyzing the expression levels of at least one gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4. More preferably, the kit comprises means for analyzing the expression levels of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAPI B and KPNA

[0053] In a preferred embodiment, the means for analyzing the gene expression comprises a pair of primers for each one of the genes. For example, the following primers can be used for LMO7 and DPYSL3:

[0054] And for JAGN1 , CYP1 B1 and EDARADD:

[0055] In another preferred embodiment, the means for analyzing the gene expression in the kits described above, further comprises a pair of primers for at least one reference gene. Preferably, the reference gene or genes is or are selected from actin, beta-2 microglobulin, cytochrome C1 , EMG1 N1 -specific pseudouridine methyltransferase, glyceraldehyde-3-phosphate dehydrogenase, TATA-box binding protein, topoisomerase (DNA) I, tyrosine 3- monooxygenase / tryptophan 5-monooxygenase activation protein zeta and combinations thereof, more preferably the reference gene or genes is / are actin, beta-2 microglobulin or both. Primers for analyzing the gene expression of these reference genes are well known by the skilled person and commercially available. In a preferred embodiment, the means for analyzing the gene expression in the kits described above, further comprises a specific probe for each one of the genes. Preferably, the probe is labeled, preferably dual-labeled. As used herein, the term “labeled” means that the probe is marked by means of a fluorescent reporter dye on the 5' base and a quencher located on the 3' base.

[0056] In another preferred embodiment, the means in the kits described above, further comprise at least one of the following: a DNA polymerase, a nucleotide mix, a buffer, DNase-free water, or any combination thereof. In a preferred embodiment of the present invention, a master mix comprising a DNA-binding dye is comprised.

[0057] In another embodiment, the gene expression analysis in the kits disclosed above, is made by the quantification of the proteins encoded by the genes of the signature of the present invention, for example by using specific antibodies.

[0058] A preferred embodiment of the invention consists of an in vitro method for the prognosis of brain metastasis relapse after brain surgery comprising the following steps:

[0059] (a) analyzing the expression levels of at least genes LMO7 and / or DPYSL3 and / or at least one gene selected from NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4 in an isolated sample of a subject who has been subjected to brain metastasis surgery, and

[0060] (b) comparing the expression levels found in step (a) with those of reference populations with and / or without relapse, and

[0061] (c) determining the risk of relapse of the brain metastasis for the subject, wherein the increased expression of any of genes NDEL1 , HMGB2, SH3GLB1 , RAP1 B, KPNA4 or LMO7 and the decreased expression of any of the genes CALM2, MYL6 and / or DPYSL3, when compared with a non-relapse reference population, is associated with a higher risk of relapse.

[0062] In a particular embodiment, the genes analyzed in steps a) and c) are LMO7 and / or DPYSL3, preferably LMO7 and DPYSL3. In another particular embodiment, the genes analyzed in steps a) and c) is at least one, two, three, four, five, six or the seven genes selected from NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, preferably, the genes are NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4.

[0063] The method might additionally include analyzing the expression levels of one or more of genes JAGN1 , CYP1 B1 , EDARADD, wherein the decreased expression of these genes when compared with a non-relapse reference population, is associated with a higher risk of relapse. A preferred embodiment of the present invention corresponds to a kit comprising means for quantifying the expression of substantially only genes LMO7 and / or DPYSL3 and / or at least a gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, and optionally additionally one or more of genes JAGN1 , CYP1 B1 and EDARADD. As used herein, the expression “substantially only” means that if any other gene is quantified, then said other gene or genes is or are reference genes, which are only quantified for serving as a reference for the quantification of the expression, as a reference for the sample load. According to this preferred embodiment, an increased expression of any of genes NDEL1 , HMGB2, SH3GLB1 , RAP1 B, KPNA4 or LMO7 and / or a decreased expression of any of genes DPYSL3, CALM2, MYL6, JAGN1, CYP1 B1 and EDARADD, when compared with a relapse reference population, might be associated with a higher risk of relapse. Contrarily, a decreased expression of any of genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4 or LMO7 and an increased expression of any of genes DPYSL3, CALM2, MYL6, JAGN1 , CYP1 B1 and EDARADD, when compared with a relapse reference population, might be associated with a lower risk of relapse. Preferably the genes analyzed are LMO7 and / or DPYSL3 and at least one gene selected from JAGN1 , CYP1 B1 and / or EDARADD. In another preferred embodiment, the genes analyzed are DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4, and at least one gene selected from JAGN1 , CYP1 B1 and / or EDARADD; or the genes analyzed are LOM7, DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4, and at least one gene selected from JAGN1 , CYP1 B1 and / or EDARADD.

[0064] The present invention also refers to the use of signatures of the present invention for deciding or recommending a patient who has suffered from brain metastasis to perform (follow) a treatment to prevent brain metastasis relapse or recommending an alternative medical regime. The signatures of the invention are the following:

[0065] - LMO7;

[0066] - DPYSL3;

[0067] - LMO7 and DPYSL3;

[0068] - At least a gene selected from NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4;

[0069] - NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4;

[0070] - LMO7, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4;

[0071] - DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4; or

[0072] - LMO7, DPYSL3, NDEL1 , HMGB2, SH3GLB1, CALM2, MYL6, RAP1 B and KPNA4. The present invention also refers to a method for deciding or recommending a patient who has suffered from brain metastasis a medical regime to prevent brain metastasis relapse after treatment, and in particular after surgery, the method comprising: a) analyzing the expression levels of at least one gene selected from NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, and / or genes LMO7 and / or DPYSL3, and optionally, additionally one or more of genes JAGN1 , CYP1 B1 and EDARADD in a sample of a subject who has been subjected to brain metastasis treatment and in particular, surgery, b) comparing the expression levels found in step (a) with those of reference populations with and / or without relapse, and c) recommending a preventive medical regime to prevent brain metastasis relapse in the subject when the expression of any of genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4 or LMO7 is increased and the expression of any of genes DPYSL3, CALM2, MYL6, JAGN1 , CYP1 B1 and EDARADD is decreased, in comparison with the non-relapse reference population; or recommending an alternative medical regime if the expression of genes NDEL1 , HMGB2, SH3GLB1 , RAP1 B, KPNA4 or LMO7 is increased and / or the expression of genes DPYSL3, CALM2, MYL6, JAGN1 , CYP1 B1 and EDARADD is decreased, in comparison with the non-relapse reference population.

[0073] In a particular embodiment, the genes analyzed in steps a) and c) are LMO7 and / or DPYSL3, preferably LMO7 and DPYSL3. In another particular embodiment, the genes analyzed in steps a) and c) is at least one, two, three, four, five, six or the seven genes selected from NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, preferably, the genes are NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4. As indicated, genes JAGN1 , CYP1 B1 and EDARADD are optionally analyzed in steps a) and c).

[0074] The present invention also refers to the use of any of the signatures of the present invention for stratifying brain metastasis patients with regard to their likelihood to develop brain metastasis relapse. The signatures of the invention described above.

[0075] In another embodiment, the method for stratifying brain metastasis patients with regard to their likelihood to develop brain metastasis relapse comprises: a) analyzing the expression levels of at least a gene selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4 and LMO7 and DPYSL3, and optionally, additionally one or more of genes JAGN1 , CYP1 B1 and EDARADD in a sample of a subject who has been subjected to brain metastasis treatment and in particular surgery, b) comparing the expression levels found in step (a) with those of reference populations with and / or without relapse, and c) stratifying the likelihood to develop brain metastasis relapse as high if the expression of any of genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B, KPNA4 and / or LMO7 is increased and the expression of genes DPYSL3, CALM2, MYL6, JAGN1 , CYP1 B1 and / or EDARADD is decreased, in comparison with the non-relapse reference population or d) stratifying the likelihood to develop brain metastasis relapse as low if the expression of any of the genes NDEL1 , HMGB2, SH3GLB1 , RAP1 B KPNA4 or LMO7 is decreased and the expression of any of the genes DPYSL3, CALM2, MYL6, JAGN1 , CYP1 B1 and EDARADD is increased, in comparison with the relapse reference population.

[0076] In a particular embodiment, the genes analyzed in steps a), c) and d) are LMO7 and / or DPYSL3, preferably LMO7 and DPYSL3. In another particular embodiment, the genes analyzed in steps a), c) and d) is at least one, two, three, four, five, six or the seven genes selected from NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, preferably, the genes are NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4. As indicated, genes JAGN1 , CYP1 B1 and EDARADD are optionally analyzed in steps a) and c).

[0077] The present invention also refers to a combined use of an expression product of each one of at least genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, and / or LMO7 and / or DPYSL3, and optionally additionally of one or more of genes JAGN1 , CYP1 B1 and / or EDARADD genes as a marker of prediction of the risk of developing brain metastasis relapse in a subject after treatment and in particular after brain surgery; or of prognosis of brain metastasis relapse after treatment and in particular after brain surgery; or as a marker for deciding or recommending a patient who has suffered from brain metastasis to perform a medical regime to prevent brain metastasis relapse; or for stratifying brain metastasis patients with regard to their likelihood to develop brain metastasis relapse.

[0078] All the embodiments of the methods of the present invention are also embodiments of the other aspects of the present invention.

[0079] All individual features (e.g. preferred features) mentioned herein may be taken in isolation or in combination with any other feature (including a preferred feature) mentioned herein (hence, preferred features may be taken in conjunction with other preferred features, or independently of them).

[0080] Throughout the description and the claims, the word "comprises” and the variants thereof are not intended to exclude other technical characteristics, additives, components or steps. Furthermore, the word "comprise" and its variations encompasses the term "consisting of. For persons skilled in the art, other objects, advantages and characteristics of the invention will arise partly from the description and partly from the practice of the invention. The following examples and figures are provided for illustrative purposes and are not intended to limit the scope of this invention.

[0081] BRIEF DESCRIPTION OF FIGURES

[0082] Figure 1. Representative image of patient-derived brain metastasis matched samples (before and after undergoing neurosurgery), stained with DPYSL3-specific antibody or RNA probe.

[0083] Figure 2. Representative image of mouse-derived brain metastasis matched samples, stained with DPYSL3-specific antibody or RNA probe.

[0084] Figure 3. Representative image of patient-derived brain metastasis matched samples, stained with LMO7-specific antibody or RNA probe.

[0085] Figure 4. Representative image of mouse-derived brain metastasis matched samples, stained with LMO7-specific antibody or RNA probe.

[0086] Figure 5. Representative image of patient-derived brain metastasis matched samples, stained with JAGN1 -specific RNA probe.

[0087] Figure 6. Representative image of mouse-derived brain metastasis samples, stained with JAGN1 -specific RNA probe.

[0088] Figure 7. Results corresponding to analysis of LM07 (a), DPYSL3 (TUC4) (b) and JAGN1 (c) in the 8 human samples of Example 3.

[0089] Figure 8. Table with Details of the 8 human samples made use of for the analysis of Example 3.

[0090] Figure 9. Table displaying protein quantification results corresponding to LM07 and DPYSL3, obtained with the 8 human samples of Example 3 (before and after undergoing relapse). Figure 10. Prognostic ability of relapse signatures in the RENACER cohort (N=144 patients, divided in tertiles of 48 patients according to expression level of signature). A, Prognostic ability of the 7-gene signature comprising NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1B, KPNA4. B, Prognostic ability of the 7-gene signature in the lung primary origin patients (N=71 patients, divided in two groups). C, Prognostic ability of the 7-gene signature in the lung primary origin patients (N=73 patients, divided in two groups). D, Prognostic ability of the 8-gene signature including all genes in A plus LMO7. E, Prognostic ability of the 8-gene signature including all genes in A plus DPYSL3. F, Prognostic ability of the 9-gene signature including all genes in A plus LMO7 and DPYSL3. G. Prognostic ability of DPYSL3 alone. H. Prognostic ability of the combination of LMO7 and DPYSL3.

[0091] Figure 11. Mapping of two selected marker genes of the persistent cluster (highlighted with a dashed line) in the integrated LIMAP, showing specific expression of both genes in the cluster of interest.

[0092] EXAMPLES

[0093] Material and methods

[0094] Established H2030-BrM (Luciferase+ / GFP+ (Green Fluorescent Protein)) metastases were obtained from 6 individual mice (nude) previously implanted intracranially with the human cell line. The initial brain metastasis was obtained using neurosurgery and mice were followed over time until relapse was confirmed using bioluminescence imaging (Living Image software, version 4.5). Mice were sacrificed at endpoint (approximately 6 weeks post-surgery) and the relapsed tumor was microdissected. Both the initial brain metastasis and the relapsed one from the same mice (matched samples) were profiled with RNAseq.

[0095] Animal studies

[0096] All animal experiments were performed in accordance with a protocol approved by the CNIO (IACUC.030-2015), Institute de Salud Carlos III (CBA35_2015-v2), and Comunidad de Madrid Institutional Animal Care and Use Committee (PROEX250 / 15 and PROEX135 / 19). Females athymic nu / nu (Harlan) 4-10 weeks of age were used. Housing and husbandry conditions were in accredited by Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). Mice were SPF (specific pathogen-free) with microbiological and environmental parameters constantly monitored. Brain colonization assays were performed by injecting 40,000 cancer cells intracranially (the right frontal cortex, approximately 1.5 mm lateral and 1 mm caudal flow bregma, and to a depth of 1 mm) by using a gas-tight Hamilton syringe and a stereotactic apparatus. Brain colonization was analyzed in vivo and ex vivo by BLI. Anesthetized mice (isoflurane) were injected retro-orbitally with Dluciferin (150 mg / kg; Syd Labs) and imaged with an IVIS machine (Perkin Elmer). Bioluminescence analysis was performed using Living Image software, version 4.5. Brain tumor resection was performed by adapting previously described procedures (Morrissy et al., 2016. Nature 529: 351 - 357). In brief, after exposing the skull, a craniotomy was performed surrounding the brain metastasis area, which was visualized by GFP, using an excitation light source at 460-495 nm (FS / ULS-02 B2, BLS Ltd) and goggles carrying emission filters (FHS / EF-3GY1 , BLS Ltd). The skull and the dura were lifted with micro-dissecting forceps, and the bulk of the tumor was then removed using a microcurette guided by GFP. When hemostasis was obtained, the surgical wound was sutured using interrupted stitching with absorbable sutures. Animals received meloxicam at 5 mg / kg once per day during 72 h and dexamethasone at 13 mg / kg once per day during 48 h to contain brain edema.

[0097] Cell culture

[0098] Human BrM cell line has been previously described and obtained from the same batches that were previously published (Nguyen et al., 2009. Cell 138: 51 - 62). The cell line was validated by morphological analysis and its behavior in vivo. The cell line was tested mycoplasma-free. H2030- BrM3 (abbreviated as H2030-BrM) (Origin: Massague lab, Memorial Sloan Kettering Cancer Center) was cultured in RPMI1640 media supplemented with 10% FBS (fetal bovine serum), 2 mM L-glutamine, 100 lU / ml penicillin / streptomycin, and 1 pg / ml amphotericin B.

[0099] Transcriptomics of relapsed tumors and statistical analysis

[0100] 500 ng of total RNA samples were used. Sample RNA Integrity numbers were 8.6 on average (range 5.9-9.5) when assayed on an Agilent 2100 Bioanalyzer. Sequencing libraries were prepared with the QuantSeq 3‘mRNA-Seq Library Prep Kit (FWD) for Illumina (Lexogen, Cat. No. 015) by following manufacturer’s instructions. Library generation was initiated by reverse transcription with oligodT priming, and a second strand synthesis was performed from random primers by a DNA polymerase. Primers from both steps contain Illumina-compatible sequences. Libraries were completed by PCR (This kit generated directional libraries stranded in the sense orientation: the readl , the only read in single read format, had the sense orientation (-library-type fr-secondstrand in TopHat, - stranded = yes in HTSeq)}. cDNA libraries were purified, applied to an Illumina flow cell for cluster generation and sequenced on an Illumina NextSeq 550 (with v2.5 reagent kits) by following manufacturer’s protocols. Eighty-five-base-pair single-end sequenced reads followed adapter and polyA tail removal as indicated by Lexogen. The resulting reads were fed to Xenome to separate the xenograft-derived human and mouse reads. Human reads were analyzed with the nextpresso pipeline as follows: sequencing quality was checked with FastQC v0.11.0. Reads were aligned to the human genome (GRCh38 (Genome Reference Consortium)) with TopHat-2.0.10 using Bowtie 1.0.0 and Samtools 0.1.19, allowing 3 mismatches and 20 multihits. The Gencode v29 gene annotation for GRCh38 was used. Read counts were obtained with HTSeq. Differential expression and normalization were performed with DESeq2, filtering out those genes where the normalized count value was lower than 2 in more than 50% of the samples. From the remaining genes, those that had an adjusted p value below 0.05 FDR were selected. GSEAPreranked was used to perform gene set enrichment analysis for several gene signatures on a pre-ranked gene list, setting 1 ,000 gene set permutations. Only those gene sets with significant enrichment levels (FDR q-value < 0.25) were considered.

[0101] RNAscope

[0102] Detection of the genes of interest in human and mouse tissue by RNA in situ hybridization. Paraffin-embedded tissue from mouse brain metastases was sectioned and RNAscope 2.5 VS probes for Hs-DPYSL3 (540289, ACDbio), Hs-UWO7 (1169419-C1 , ACDbio), Hs- JAGN1 (1169429-C1 , ACDbio) were assayed on the Roche Ventana Discovery XT using the standardized automated protocols at the CNIO Histopathology Core Facility.

[0103] Antibodies made use of were:

[0104] DPYSL3: AB5454, Sigma Aldrich. LMO7: HPA020923, Sigma Aldrich.

[0105] Results

[0106] Example 1 .

[0107] H2030-BrM metastasis samples obtained after neurosurgery, as well as samples resulting from relapse (reappearance after surgery), were subjected to RNAseq analysis.

[0108] When the original tumors and the relapsed tumors were compared, only 5 genes of the whole transcriptome were significantly deregulated (1 upregulated: LMO7; 4 down regulated: CYP1 B1 , DPYSL3, EDARADD, JAGN1 , in comparison with the initial brain metastasis).

[0109] Table 1 . Statistical relevance of the five genes of the signature of the invention. Gene Iog2 Fold Change pvalue padj

[0110] CYP1 B1 -3,647148 1.40E-18 2.16E-14

[0111] LMO7 2,0089955 6.06E-09 4.67E-05

[0112] DPYSL3 -1 ,351161 5.76E-08 0,000296

[0113] EDARADD -1 ,146406 5.10E-06 0,0196618

[0114] JAGN1 -1 ,918629 9.53E-06 0,0294155 is the result of dividing the FPKM (Fragments per kilo base per million mapped reads) of the second value (relapsed metastasis) on the FPKM of the first value (initial brain metastasis) to get the Fold Change (FC). Then, the Iog2 fold change value from FPKM value is obtained.

[0115] Pvalue and padj: differential expression and normalization was performed with DESeq2 (Love et al. 2014. Genome Biol. 2014;15(12):550), filtering out those genes where the normalized count value was lower than 2 in more than 50% of the samples. From the remaining genes, those that had an adjusted p-value below 0.05 FDR (False Discovery Rate) were selected.

[0116] The main conclusion from the experimental model was that when the expression level of the gene LMO7 is increased and the expression levels of genes CYP1 B1 , DPYSL3, EDARADD, JAGN1 are decreased comparing to a reference population, then the subject presents a higher risk of relapse.

[0117] Confirmation of the relevance of expression levels of genes LMO7 and / or DPYSL3 was achieved both in a lung cancer model and in human brain metastasis samples pre- and post-relapse. Corresponding results are displayed in Examples 2 and 3.

[0118] Example 2.

[0119] A validation of candidate genes in situ in experimental models of relapse was performed in the H2030-BrM lung cancer model by evaluation by RNAscope for LMO7, DPYSL3 JAGN1 , and also with antibodies for LMO7 and DPYSL3. Corresponding results are displayed in Figures 1-6.

[0120] Example 3. Validation of biomarkers of relapse in human samples.

[0121] Analysis of rare matched human brain metastasis obtained from neurosurgery after an initial diagnosis (first surgery), and the one generated after some time after relapse post-surgery (relapse post-Sx, when the local relapse appears), was performed in samples from 8 patients. The markers previously validated in mouse models (i.e. LMO7, DPYSL3, JAGN1) were tested in matched human brain metastasis pre- and post-relapse.

[0122] Results obtained are displayed in Figure 7. Basically, it was confirmed that LMO7 (a), DPYSL3 (also known as TLIC4) (b) and JAGN1 (c) followed and reproduced the data on experimental models. I.e. LMO7 is upregulated in relapse, while DPYSL3 and JAGN1 are downregulated. While JAGN1 is only validated at RNA level (as corresponding antibodies were not functional), LMO7 and DPYSL3 have been validated at the protein level (as well as RNA).

[0123] Upon analysis of this samples, it was concluded that LMO7 is upregulated, as predicted from experimental models, in relapse samples in 50% of the cases. While DPYSL3 and JAGN1 are downregulated, as predicted from experimental models, in relapse samples in 75% of the cases.

[0124] A table displaying the details of the 8 patient samples made use of is displayed in Figure 8. Results corresponding to LMO7 and DPYSL3 are displayed in Tables of Figures 9a and 9b, respectively.

[0125] As it can be observed, 4 out of the 8 patients whose samples have been analyzed, display alteration expression levels of both DPYSL3 and LMO7.

[0126] Example 4. Additional markers for the prognostic cluster

[0127] Material and methods

[0128] Generation of BrM relapse mouse samples

[0129] To generate the materials for the scRNAseq profiling of naive and relapsed tumors, it was used the BrM relapse after neurosurgery published in Zhu et al. (EMBO Mol Med. 2022. A clinically compatible drug-screening platform based on organotypic cultures identifies vulnerabilities to prevent and treat brain metastasis. EMBO Mol Med. Mar 7;14(3):e14552). This model consists in the intracranial injection of 40,000 cells from the human H2030-BrM cell line (Luciferase+ / GFP+ (Green Fluorescent Protein)) and once the tumor is established (3-4 weeks), the brain metastasis is resected using neurosurgery and mice are followed over time during relapse using bioluminescence imaging (Living Image software, version 4.5).

[0130] To generate the samples used for this material, mice were sacrificed at several timepoints for analysis, i.e. naive / non-resected control (n=1), 24h after surgery (n=7), 2 weeks after surgery (n=3) and 4 weeks after surgery (n=2), cancer cells were isolated by sorting for GFP and profiled with scRNAseq. Genes LMO7 and DPYSL3 were obtained by RNAseq bulk from naive tumors (obtained with the surgery) vs relapsed tumors in the same area at end point, whereas the cluster of 7 genes were obtained with scRNAseq from naive tumors before the surgery and relapsed tumors at different points during the relapse. scRNAseq profiling of BrM relapse cancer cells

[0131] Single cell RNA-sequencing data generation

[0132] Cells were counted and their viability was checked using the Countess III cell counter (Thermofisher). Each cell suspension was loaded into one port of a Chromium Next GEM Chip G (10x Genomics) with a target output of 10,000 cells for the naive, relapse 24h and relapse 2 weeks samples, and of 5,000 cells in the relapse 4 weeks samples. Single cells were encapsulated into emulsion droplets using the Chromium Controller (10x Genomics). ScRNA-seq libraries were prepared using the Chromium Next GEM Single Cell 3' Kit v3.1 (10x Genomics) following the manufacturer instructions and each library was amplified using a SureCycler 8800 thermal cycler (Agilent Technologies). The average size of each library was then calculated using a High sensitivity DNA chip on a 2100 Bioanalyzer (Agilent Technologies) and the concentration was determined using the Qubit fluorometer (Thermofisher). Individual libraries were diluted to 10 nM and pooled for sequencing. Library pool was loaded at 700 pM onto a P3 flow cell (100 cycles) of the NextSeq 2000 (Illumina) in paired-end configuration (28bp Readl , 10bp Indexl , 10pb Index 2 and 90bp Read2). FastQ files for each sample were obtained using cellranger mkfastq pipeline (10x Genomics).

[0133] Single cell RNA-sequencing data processing and analysis

[0134] Sequencing quality was checked with FastQC

[0135] (http: / / www.bioinformatics.babraham.ac.uk / projects / fastqc / ). Reads were aligned to the Human reference genome GRCh38, v46 gene annotation from GENCODE (Frankish et al. GENCODE 2021. Nucleic Acids Research, 49(D1), D916-D923.) using STARsolo (STAR 2.7.8a) (Dobin et al., 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21). Each time point, specifically "Naive," "24h_P-Sx," "2 weeks," and "4 weeks," was subsequently analyzed with the Seurat pipeline (Hao et al., 2021. Integrated analysis of multimodal single-cell data. Cell, 784(13), 3573-3587. e29) separately. Initially, the data was normalized to correct for technical variations, followed by the identification of the top 3000 variable features using the "vst" method. The data was then scaled to standardize gene expression levels across cells. Principal Component Analysis (PCA) was performed on these variable features, and an elbow plot was generated to determine the optimal number of principal components, selecting 30 principal components (PCs) for further analysis. Uniform Manifold Approximation and Projection (UMAP) was applied to visualize the cells in reduced dimensions. Neighbors were identified based on the UMAP dimensions, and clustering was conducted at a resolution of 0.1 to group cells with similar expression profiles into clusters. To find marker genes for each cluster, the FindAIIMarkers function was used with thresholds set to a minimum percentage of 0.25 and a log fold change threshold of 0.25. This allowed us to obtain the gene markers for each cluster, including the remaining cluster 3 in the naive time point.

[0136] To ensure a more accurate comparison and alignment of cellular states and types across the dataset, it was crucial to integrate data from different time points and correct for batch effects. Therefore, Harmony (Korsunsky et al. 2019. Fast, sensitive and accurate integration of single-cell data with Harmony. Nature Methods, 76(12), 1289-129) was run on the non-integrated Seurat object using sample information to correct for these effects. Following integration, clustering was re-identified on the harmonized data using a resolution of 0.15 to ensure that the clusters reflect the integrated dataset accurately.

[0137] Signature mapping was performed using the run_wmean function from the decoupleR (Badia-I- Mompel et al., 2022. decoupleR: ensemble of computational methods to infer biological activities from omics data. Bioinformatics Advances, 2(1)) package to map gene expression signatures to single-cell data. This approach involves mapping predefined gene signatures such as cluster 3 in the naive time point onto the single-cell data to identify the activity of these signatures within each cell cluster.

[0138] Assessment of prognostic ability of scRNAseq signature in RENACER samples

[0139] Human samples bulk RNA-sequencing data processing

[0140] Raw RNA-sequencing (RNA-seq) reads were processed as follows: it was used FASTQC vO.12.1 to check the quality of the sequencing reads and BBDuk (BBMap v38.93) to remove the adapter sequences and read ends with Phred quality scores lower than 10, discard reads shorter than 20 bp. Trimmed reads were aligned to the human genome (GRCh38) with STAR v0.74.0 1 and Samtools v1.14 2. Finally, mapped reads were counted and aggregated to a matrix of gene-level counts with HTSeq vO.13.5 3. Genes with less than 10 counts across samples were filtered out. RNA-seq based counts were normalized using a variance stabilizing transformation with DESeq2 v1.34.0 4 and then corrected for batch variation while preserving variation associated with the primary tumour with limma v3.50.1 5. Only 144 patients from the RENACER cohort were considered, who had available RNA-seq data and information of overall survival.

[0141] Evaluation of the prognostic effect of the genes involved in the cluster-based signature derived from single cell RNA-sequencing

[0142] Univariate Cox’s proportional hazard regression models were fitted including overall survival as a dependent variable, each individual gene of the signature as fixed factors.

[0143] Patients were split into low and high at the median. It was considered features as significant only if the 95% confidence intervals around the hazard ratio (HR) do not include the one.

[0144] The univariate Cox regression analysis was performed using the survival R package v3.7.0.

[0145] Definition of a 20-gene signature score associated with the relapse to surgery

[0146] It was quantified a signature score using the weighted mean method from decoupleR 6 v2.8.0. It was assigned a positive unit weight to genes associated with relapse and a negative unit weight to genes associated with survival. Survival curves were estimated using the Kaplan-Meier estimator and plotted using the survminer R package vO.4.9.

[0147] Robustness analysis of the 20-gene signature

[0148] It was defined the gene ranking using the following drop out inference approach:

[0149] Considering the signature genes (20 genes) used to build the relapse post-surgery score using the data from the RENACER cohort.

[0150] It was computed the scores 1000 times, by dropping a variable percentage of the signature genes each time. This percentage ranges between 5% and 95%.

[0151] Checking all the occurrences where a gene was present, we calculated the average Mean Square Error (MSE) to compare the actual predictions with the predictions from the original signature (full score with all genes).

[0152] It was built our ranking on the basis that a gene removal has a bigger effect on the predictions (higher MSE is associated with a gene being most useful / important). To assess the stability of this ranking, we applied bootstrap on the average MSE of each gene creating 100 samples of size equal to the minimum number of occurrences of all genes. The different gene rankings derived from the bootstrap samples showed a nice overlap of the gene ranks.

[0153] This analysis revealed that original signature could be reduced to 7 genes (i.e. dropping out 35% of the original signature genes), as it showed a reasonable compromise between the loss in performance (MSE = 5.77) and the ability to predict poor prognosis (HR = 4.67 [High group - Low group]; HR=1.92 [Mid group vs Low group]). The 7 top-ranked genes were included in the signature.

[0154] Conservation of the prognostic effect of the signature in non-lung primary tumours as well as in only lung primary tumours

[0155] As the cluster-based signature derived from scRNA-seq was obtained using a lung cancer BrM mouse model and the RENACER cohort comprises a great number of brain metastatic patients diagnosed with lung primary tumours, we investigated whether the 7-gene signature was conserved across non-lung tumours. We therefore separated patients in two groups: a non-lung patient group (n=73) and only-lung patients group (n=71), and evaluated the prognostic power of the signature based on the HR estimate. We observed similar results in both groups which confirmed the validity of the signature regardless of the primary tumor type.

[0156] Results

[0157] Generation of a signature of the potential “seeds of BrM

[0158] After performing integration of the different timepoints in the preclinical scRNAseq data, it was identified a cluster persistent in all timepoints which also comprises the majority of cells present 24h after surgery. This subpopulation was named “remaining cluster” and hypothesize it may represent the potential cells surviving surgery and populating the relapsed BrM. To extract the signature of the “remaining cluster” subpopulation, it was selected the cells belonging to this cluster from the naive sample, due to its decreased noise compared to the database integrated with all timepoints as well as the more similar situation with the RENACER human sample database which is comprised mostly of naive BrMs.

[0159] Curation of the “seeds of BrM relapse” signature to obtain a prognostic signature

[0160] It was hypothesized that the signature of the potential seeds of relapse may represent a tool to identify patients with high risk of relapse (and consequently worse prognosis). To generate a prognostic signature of BrM relapse with clinical value, we interrogated the prognostic potential of each individual gene from the remaining cluster signature in the RENACER cohort of human BrM samples. We then selected the top 20 genes according to their individual concordance index (11 good prognosis genes and 9 bad prognosis genes) and combined them to generate a preliminary relapse signature that had prognostic ability in the RENACER cohort. We then applied stability analysis to reduce the signature to the minimum number of genes that showed a reasonable compromise between the loss in performance and the ability to predict poor prognosis. The final prognostic signature was composed of 7 genes (NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1B, KPNA4) (Figure 10A) and maintained its prognostic ability after stratifying RENACER patients into lung primary origin BrM and non-lung primary origin BrM (Figure 10B, C), thus confirming its validity regardless of primary tumor type.

[0161] Fine-tuning of the prognostic signature

[0162] It was also explored the prognostic potential in the RENACER cohort of the genes LMO7 and DPYSL3 individually, together, and as an addition to the aforementioned 7-gene signature. DPYSL3 alone shows prognostic potential, although not statistically significant, as a bad prognosis gene (Figure 10G). This prognostic ability improves when adding LMO7, with an almost significant result (Figure 10H). Finally, adding LMO7 and DPYSL3, both alone and together, to the 7-gene signature improved its prognostic ability between the low-expression and mediumexpression groups, while it did not show a benefit when comparing the low-expression and the high-expression patient groups (Figure 10D, E and F). Thus, the utilization of a 9-gene signature allows a better distinction between the low-expression good prognosis patients and the mediumexpression bad prognosis patients, allowing to establish a clearer patient threshold for the introduction of a potential future therapy based on the elimination of the seeds of relapse.

[0163] Altogether, these results support the inclusion in the present patent of the following relapse signatures:

[0164] - 7-gene signature (NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1B, KPNA4)

[0165] - 8-gene signature (NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1B, KPNA4, LMO7)

[0166] - 8-gene signature (NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1B, KPNA4, DPYSL3)

[0167] - 9-gene signature (NDEL1, HMGB2, SH3GLB1, CALM2, MYL6, RAP1B, KPNA4, LMO7, DPYSL3)

[0168] 2-gene signature (LMO7, DPYSL3). The seven-gene signature allows a good separation of high and low-risk of relapse in a sample. This result may be combined with the nine-gene signature, which has better discrimination between high and medium risk. Aside from the aforementioned genes, two marker genes of the remaining cluster (BTG1 and CEBPB), although not with prognostic ability per se, show value as potential indicators of higher relapse risk due to their specific expression in the remaining cluster (Figure 11).

Claims

CLAIMS1. A method for the prognosis of brain metastasis relapse in a subject, wherein the method comprises analyzing the expression levels of at least genes DPYSL3 and / or LMO7 in an isolated sample from the subject.

2. The method of claim 1 , wherein the method comprises analyzing the expression levels of DPYSL3 and LMO7 in an isolated sample from the subject.

3. The method according to claim 1 or 2, wherein the expression levels of genes DPYSL3 and / or LMO7 are compared with those of reference populations with and / or without relapse, and wherein an increased expression of gene LMO7 and a decreased expression of gene DPYSL3 when compared with a non-relapse reference population, is associated with a higher risk of relapse.

4. The method according to any of claims 1 to 3, further comprising analyzing the expression of one or more additional genes, and / or of one or more reference genes in the sample.

5. The method according to claim 4, wherein the one or more additional genes is selected from the group consisting of NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4.

6. The method according to claim 5, wherein the one or more additional genes consists of the genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4.

7. The method according to claim 5, wherein the genes are DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAPI B and KPNA4.

8. The method according to claim 5, wherein the genes are LMO7, DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4.

9. The method according to any of claims 5 to 8, wherein the expression levels of genes LMO7, DPYSL3, NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and / or KPNA4 is compared with those of reference populations with and / or without relapse, and wherein an increased expression any of genes LMO7, NDEL1 , HMGB2, SH3GLB1 , RAP1 B and / or KPNA4, and / or a decreased expression of any of genes DPYSL3, CALM2 and / or MYL6 when compared with a non-relapse reference population, is associated with a higher risk of relapse.

10. The method according to claim 4, further comprising analyzing the expression levels of gene JAGN1 , and comparison of the corresponding expression levels with those of reference populations with and / or without relapse, wherein the decreased expression of gene JAGN1 , when compared with a non-relapse reference population, is associated with a higher risk of relapse.

11. The method according to claims 4 to 10, wherein the reference gene is selected from actin, beta-2 microglobulin, cytochrome 01 , EMG1 N1 -specific pseudouridine methyltransferase, glyceraldehyde-3-phosphate dehydrogenase, TATA-box binding protein, topoisomerase (DNA) I, tyrosine 3- monooxygenase / tryptophan 5-monooxygenase activation protein zeta and combinations thereof, preferably is actin, beta-2 microglobulin or both.

12. The method according to any one of claims 1 to 11 , wherein the subject has been subjected to treatment of the brain metastasis, preferably by brain surgery.

13. The method according to any one of claims 1 to 12, wherein the gene expression analysis is made by qPCR, preferably by detection of mRNA levels by qPCR, preferably by RT-qPCR, by in situ hybridization, preferably RNAscope, and / or by protein detection preferably with antibodies.

14. The method according to any one of claims 1 to 13, wherein the sample is a brain sample obtained by brain surgery of a brain metastasis.

15. The method according to any one of claims 1 to 14, wherein the sample is a fluid sample, such as a blood sample or a cerebrospinal fluid sample, or a tissue sample, such as a biopsy.

16. The method according to any one of claims 1 to 15, wherein the sample is a cDNA sample, preferably a cDNA obtained from an RNA sample.

17. The method according to any one of claims 1 to 16, wherein the subject is human, preferably a human with brain metastasis from a primary cancer.

18. The method of claim 17, wherein the primary cancer is selected from lung cancer, melanoma, colorectal cancer or breast cancer.

19. The method of claim 18, wherein the primary is lung cancer.

20. The method of claim 19, wherein the lung cancer is selected from non-small cell lung cancer or lung adenocarcinoma.

21. A kit comprising means for analyzing the expression levels of genes DPYSL3 and / or LMO7 in a sample.

22. The kit of claim 21 , further comprising analyzing the expression levels of genes NDEL1 , HMGB2, SH3GLB1, CALM2, MYL6, RAP1 B and KPNA4.

23. The kit of claim 21 , further comprising means for analyzing the expression levels of gene JAGN1 , and / or one or more additional genes, and / or of one or more reference genes in the sample.

24. The kit according to any one of claims 21 to 23, wherein the means for analyzing expression level of a gene involves inclusion of at least a pair of primers for any such gene and / or of at least one specific probe for any such gene.

25. The kit according to any one of claims 21 to 24, further comprising at least one of the following: a DNA polymerase, a nucleotide mix, a buffer, DNase-free water, or any combination thereof.

26. Use of the method according to any one of claims 1 to 20, or of the kit according to any one of claims 21 to 25 for the prognosis of brain metastasis relapse and / or for deciding or recommending a patient who has suffered from brain metastasis to follow a treatment to prevent brain metastasis relapse or recommending an alternative medical regime and / or for stratifying brain metastasis patients with regard to their likelihood to develop brain metastasis relapse.

27. Use of an expression product of each one of at least genes NDEL1 , HMGB2, SH3GLB1 , CALM2, MYL6, RAP1 B and KPNA4, and / or LMO7 and / or DPYSL3, and optionally additionally of one or more of genes JAGN1 , CYP1 B1 and / or EDARADD genes as a marker of prediction of the risk of developing brain metastasis relapse in a subject after treatment and in particular after brain surgery.

Images