FR2980579A3 - METHOD AND KIT FOR IN VITRO DIAGNOSIS OF BREAST CANCER - Google Patents

Abstract

The present invention relates to a method for in vitro diagnosis of breast cancer in an individual, comprising detecting, in a biological sample derived from said individual, at least one autoantibody directed against a protein from the group comprising GAL3, PAK2 , PHB2, RUVBL1 and RACK1. A kit for carrying out such a method comprises at least one of the group comprising GAL3, PHB2, RUVBL1, RACK1 and PAK2.

Description

The present invention is in the field of diagnosis in oncology. More particularly, it relates to a method for in vitro diagnosis of breast cancer in an individual, in particular breast cancer at an early stage of tumor development, and more particularly in situ carcinomas of the breast. It also relates to a kit for such a diagnosis. Among solid cancers, breast cancers are the most common in women and the leading cause of death among gynecological cancers in developed countries. In general, the observation of a mammary lesion can be done at present in different ways: accidental discovery by self-examination of the breasts, existence of a clinical symptomatology or search for signs of radiographic calls (mammography , MRI, ultrasound) during routine or organized screening for breast cancer. In particular, among breast cancers, the percentage of cancers detected as ductal carcinoma in situ (DCIS) has increased ten-fold over the last 20 years, especially among women over the age of 50; largely explained by the development of mammographic screening. 20 to 25% of new cases of breast cancer diagnosed each year turn out to be CCIS (Jernal et al., 2009). The identification of early forms of breast cancer, and in particular DCIS, is at present a priority objective in oncology, since the early diagnosis of a pre-invasive stage significantly reduces the risk of relapse and to also significantly increase the survival rate of patients. Although the effectiveness of mammography screening for the detection of early forms of breast cancer is undeniable, several studies have shown, however, a significant decrease in the sensitivity of mammography with the increase in the degree of breast density. and in particular with regard to DCIS. The sensitivity of a screening test or diagnosis is defined in the present description conventionally in itself, that is to say as the proportion of positive individuals actually well detected by the test. The specificity of the test corresponds to the proportion of negative individuals actually well detected by the test. Thus, the majority of women for whom mammography examination shows abnormalities do not present cancer, the variability being mainly based on breast density but also on a certain number of parameters such as the evaluation by the radiologist, the age of the woman, poor operating practices, etc. In particular, retrospective studies have shown a substantial level of false negatives, with previous mammograms showing abnormality up to half the cases (Brem et al., 2003). Mammography thus exposes a risk of over-diagnosis corresponding to a false positive: the patient is considered to be a carrier of breast cancer when she is not, thus exposing her to an unjustified treatment with all the side effects and risks that ensue. Considerable efforts have been made to develop an effective screening method for DCIS, with no fully conclusive results to date. The present invention aims to propose a new non-invasive tool for the detection of early forms of breast cancer, in particular DCIS, which is reliable and which has in particular a sensitivity and detection specificity higher than that of the techniques used. by the prior art, including for patients with high breast density. With this in mind, the present inventors have been particularly interested in techniques using the in vitro detection, by techniques derived from molecular biology, of tumor markers present in a biological sample derived from the individual. Such methods have in particular been proposed for tumor markers CA15-3 and carcinoembryonic antigen (CEA), for the detection of advanced breast cancer. These methods have in particular the general advantages of being non-invasive and of a high degree of reproducibility.

However, no biomarker for reliable detection of early forms of breast cancer has yet been identified. It has now been discovered by the present inventors that certain circulating autoantibodies in the blood of cancer patients, alone or in combination, constitute particularly reliable tumor biomarkers for the detection of early forms of breast cancer, and in particular DCIS. . In general, autoantibodies are naturally occurring antibodies directed against an antigen that the individual's immune system has recognized as foreign to the body. In particular, for an individual suffering from cancer, the tumor produces so-called tumor proteins. A tumor protein as defined in the present disclosure may be a protein overexpressed by tumor cells over normal cells, for example due to abnormal transcriptional regulation, as well as an altered form of a wild-type protein expressed by normal cells, this alteration may for example consist of a modification in the primary amino acid sequence, in the secondary structure, tertiary or quaternary, or a posttranslational modification, for example an abnormal glycosylation. The production of such tumor proteins can thus trigger an immune response in the individual, which results in the production of autoantibodies directed against these proteins. These autoantibodies may be present in the form of free circulating antibodies or in the form of circulating immune complexes comprising autoantibodies bound to their target tumor protein. This occurs months or years before the clinical diagnosis of a tumor, and by the nature of the immune response, autoantibodies can be produced in response to the presence of a very small amount of protein circulating tumor, the detection of these autoantibodies proves particularly suitable for the early diagnosis of cancers (Qiu et al., 2008). However, the methods for detecting circulating autoantibodies identified to date have insufficient sensitivity and specificity to constitute reliable tools for diagnosing early pre-invasive forms of breast cancer. The present inventors have now discovered that autoantibodies directed against Galectin-3 (GAL3), p21 activated kinase 2 (PAK2), Prohibitin-2 (PHB2), Receptor for activated C-kinase 1 (RACK1), and RuvB proteins -like 1 (RUVBL1), constituted tumor biomarkers for reliable diagnosis of breast cancer, including in its early forms. Thus, according to a first aspect, the present invention relates to a method for in vitro diagnosis of breast cancer in an individual, in particular early forms of breast cancer, and more particularly to in situ carcinomas of the breast, which comprises the detection, in a biological sample derived from said individual, at least one autoantibody directed against a protein from the group consisting of GAL3, PAK2, PHB2, RUVBL1 and RACK1. According to particular embodiments, the diagnostic method according to the invention comprises the detection, in a biological sample from said individual, of a plurality of autoantibodies, in particular of at least two, preferably at least three, preferably at least four, autoantibodies each respectively directed against one of the proteins comprising GAL3, PAK2, PHB2, RUVBL1 and RACK1. In general, the combined detection of a plurality of antibodies, each directed against one of said different proteins, advantageously increases the reliability of the diagnostic method. In particular embodiments of the invention, the method thus comprises the detection, in said biological sample, of autoantibodies directed against the GAL3 protein, of autoantibodies directed against the protein PAK2, of auto antibodies directed against the protein PHB2, autoantibodies directed against the RUVBL1 protein and / or autoantibodies directed against the RACK1 protein. For example, the method comprises detecting, in the biological sample from the individual, at least two autoantibodies directed respectively against: GAL3 and PAK2; or GAL3 and PHB2; or GAL3 and RUVBL1; or GAL3 and RACK1; or PAK2 and PHB2; or PAK2 and RUVBL1; or PAK2 and RACK1; PHB2 and RUVBL1; or PHB2 and RACK1; or RUVBL1 and RACK1. It may in particular comprise the detection, in the biological sample from the individual, of at least three autoantibodies directed respectively against: GAL3, PAK2 and PHB2; or GAL3, PAK2 and RUVBL1; or GAL3, PAK2 and RACK1; or GAL3, PHB2 and RUVBL1; or GAL3, PHB2 and RACK1; or GAL3, RUVBL1 and RACK1; or PAK2, PHB2 and RUVBL1; or PAK2, PHB2 and RACK1; or PAK2, RUVBL1 and RACK1; or PHB2, RUVBL1 and RACK1. It may also comprise the detection, in the biological sample from the individual, of at least four autoantibodies directed respectively against: GAL3, PAK2, PHB2 and RUVBL1; or GAL3, PAK2, PHB2 and RACK1: or GAL3, PAK2, RUVBL1 and RACK1; or GAL3, PHB2, RUVBL1 and RACK1; or PAK2, PHB2, RUVBL1 and PACK1. In particular embodiments of the invention, the method comprises the detection, in a biological sample from said individual, of a set of autoantibodies directed respectively against the proteins GAL3, PAK2, PHB2, RUVBL1 and RACK1. .

More specifically, the method comprises detecting, in said biological sample, autoantibodies directed against the GAL3 protein, autoantibodies directed against the PHB2 protein, autoantibodies directed against the RUVBL1 protein, autoantibodies to the RACK1 protein and autoantibodies to the PAK2 protein.

The combined detection of these five types of autoantibody advantageously makes it possible to obtain a particularly high sensitivity, specificity and precision for the diagnosis of breast cancer, in particular as regards DCIS and early forms of breast cancer, all being associated with a great ease of implementation and a reduced implementation cost, in particular because of the number that remains low of autoantibody types detected. Thus, by detecting and / or quantifying the presence of a particular immune response in the tested individual, this immune response being linked to the presence of breast cancer tumor cells, the method according to the invention advantageously constitutes a a reliable and non-invasive means of diagnosing breast cancer, including at early stages. More generally, the present invention relates to the use of at least one protein, preferably a plurality of proteins, among GAL3, PHB2, RUVBL1, RACK1 and PAK2, for the diagnosis or prognosis in vitro of breast cancer, or for the evaluation of the effectiveness of a therapeutic treatment. The proteins GAL3, PHB2, RUVBL1, RACK1 and PAK2 are known per se. GAL3 protein, also named "Mac-2" or "LGALS3" (for lectin, galactoside binding, soluble, 3, or galectin-3) (Genbank Gene ID: 3958, NC_000014.8; NG_017089.1), belongs to the family of soluble lectins linked to β-galactosides. It has several important regulatory roles, particularly in the humoral immune response by T-cell regulatory processes, and in cancer by resistance to apoptosis, regulation of growth processes, adhesion, differentiation and cell angiogenesis, as well as the regulation of cellular invasion and metastasis. GAL3 is secreted into the extracellular matrix, and is strongly expressed in the aggressive breast cancer cell lines MCF10 DCIS.com and MCF1OCA. The overexpression of GAL-3 in non-tumor lines leads to the formation of tumors and metastases, while its suppression leads to the disappearance of the tumorigenic and metastatic properties of the cells.

PHB2 for "prohibitin 2" (Genbank Gene ID: 11331, NC 000012.11, NG 021408.1, GI: 299758502) is a protein that plays an important role in the modulation of estrogen-induced gene transcription by the recruitment of histone deacetylases. . PHB2 levels are higher in the plasma of cancer patients. In addition, the expression of PHB2 is positively correlated with estrogen receptor levels, and inversely with the grade of tumors in human breast cancer biopsies. RACK1, also referred to as GNB2L1 for "guanine nucleotide binding protein (G protein), beta 2-like polypeptide 1" (Genbank Gene ID: 10399, NC_000005.9) is a protein homologous to the heterotrimeric G protein subunit. RACK1 participates in the membrane anchoring of several proteins (notably Protein Kinase C) and in the coordination of growth, adhesion and cell division. RACK1 promotes proliferation, invasiveness and / or metastasis formation in breast cancer in vitro and in vivo. RUVBL1, for RuvB-like 1, also named "Pontine" or "Tip49" (Genbank Gene ID: 8607, NC_000003.11 GI: 224589815) is a protein that is part of several high molecular weight protein complexes strongly involved in functions. Tumors such as regulation of transcription, detection and repair of DNA damage, cell transformation, and cell growth. The level of expression of RUVBL1 is increasing in cancerous tissues, but its level of expression in breast cancer has not been studied at present. PAK2 for "P21-activated kinase 2" (Genbank Gene ID: 5062, NC_000003.11, NG 009227.1 GI: 219521842) is a protein belonging to the family of serine / threonine kinases that link Rho GTPases to the reorganization of the cytoskeleton and to nuclear signaling. PAK2 is unique in the PAK family because it is cleaved and activated by caspase 3 and similar proteases. In addition, PAK2 phosphorylates c-Jun, thereby promoting epidermal growth factor (EGF) stimulated cell transformation. In breast cancer, PAK2 has been recently shown to be highly expressed in human breast cancer cell lines and in human invasive mammary carcinoma tissues, relative to non-tumor human mammary epithelial cell lines and mammary tissues, respectively. adjacent normal. Although these five proteins are known in themselves in connection with breast cancer, the detection of autoantibodies present in a biological sample and directed against one of them has however never been evoked by the art. previous, and less so in connection with the in vitro diagnosis of this cancer. In particular, the present inventors have now discovered that tumor cells of breast cancer, and in particular of DCIS, produced tumor proteins which provoked the production by the body of autoantibodies directed against them, and that, for GAL3, PHB2, RUVBL1, RACK1 and PAK2, these autoantibodies reacted with the protein in wild form, that is to say as expressed by normal cells, in recombinant form or also by chemical synthesis, and that they were particularly sensitive and specific tumor biomarkers, easily detectable and quantifiable, including conventional immunoenzymatic techniques in themselves. In preferred embodiments of the invention, the diagnostic method is performed from a sample of a fluid of the individual. This sample may for example be a sample of blood, urine, lymph, cerebrospinal fluid or a breast fluid of the individual. Preferably, it is a sample of blood serum or a plasma sample. Preferably, the method according to the invention provides for the determination of the autoantibodies directed against a protein according to the invention and present in the biological sample to be tested, and the comparison of the concentration value obtained with a predetermined threshold value. This threshold value is preferably experimentally determined by the same assay method, using the same isoform of the protein, applied to cohorts of populations of the same type (same age) and whose status with respect to cancer. is known. In the last step, the method comprises assigning a healthy or affected status to the individual concerned, depending on the result of this comparison.

When the diagnostic method provides for the assay of a plurality of autoantibodies each directed against a different protein, it preferably advantageously comprises multivariate statistical analysis steps, conventional in themselves, to obtain for each population cohort. given known status, a linear combination formula and a threshold value defining a healthy status and a status achieved. The method according to the invention then comprises a step of applying the linear combination formula to the concentration values measured for each type of autoantibody of the tested individual, the comparison of the final value thus obtained with the threshold value then assigning a healthy or affected status to the individual concerned, depending on the outcome of that comparison. In preferred embodiments of the invention, the detection of each autoantibody comprises contacting the biological sample with an immunological reagent comprising the protein from GAL3, PAK2, PHB2, RUVBL1 and RACK1. against which said autoantibody is directed, and detecting the presence of complexes formed by specific binding of that protein and an autoantibody present in the sample. This step can be carried out for each of the proteins GAL3, PAK2, PHB2, RUVBL1 and / or RACK1.

The process according to the invention can advantageously be carried out by standard enzyme immunoassay techniques per se, such as the so-called enzyme-linked immunosorbent assay (ELISA). This technique advantageously makes it possible to detect and measure the presence of autoantibodies in the biological sample to be tested. It has the advantages of a great ease of implementation, by means of reagents and materials commonly available commercially and in analytical laboratories. The method according to the invention is thus advantageously simple and quick to implement, and it has a high degree of reproducibility. It is particularly particularly suitable for use as a routine examination, and moreover by high throughput screening techniques. In preferred embodiments, the diagnostic method described above comprises indirect ELISA-type assay steps, which include, for each autoantibody to be detected: a preparation step, in which the protein associated with said autoantibody on a support, - a step of contacting the biological sample, to be tested for the presence of the autoantibodies, with said support, - a step of revealing comprising contacting the support with a reagent comprising at least one binding partner of the autoantibody, said binding partner being associated with a detectable compound. The method as described above preferably comprises a series of negative and positive control controls. The binding partner is preferably an antibody, referred to as a "secondary antibody". Preferably, but not exclusively, the detectable compound is a fluorochrome, the detection and quantification of which can be carried out for example by means of a fluorescence microscope. According to preferred embodiments of the invention, when the support is a well plate, a single protein according to the invention can be fixed in each well, allowing the detection of a single type of self-supporting agent. antibodies per well, or a plurality of different proteins can be fixed in the same well, thereby allowing the detection of a plurality of autoantibodies simultaneously in each well. The method according to the invention may also comprise the combined detection, in said biological sample, of an autoantibody directed against a different protein, that is to say not belonging to the list of five protein above, including a protein described as an immunogenic protein associated with breast cancer, or a plurality of autoantibodies directed respectively against a plurality of such different proteins. By way of example, mention may be made of the autoantibodies directed against the PPIA protein, the autoantibodies directed against the PRDX2 protein, the autoantibodies directed against the MUC1 protein, the autoantibodies directed against the HSP60 protein and autoantibodies to the FKBP52 protein. MUC1, for "Mucin" (Genbank Gene ID: 7059, NC_000001.10 NG 029383.1, GI: 339639630), HSP60, for "Heat Shock Protein 60" (Genbank Gene ID: 3329, NC_000002.11, NG_008915.1, GI: 211064491), FKBP52, for "Peptidyl-prolylcis-trans isomerase FKBP52" (Genbank NC 000012.11, GI: 224589803), PPIA, for "Peptidyl-prolylcis-trans isomerase A" (Genbank Gene ID: 8607, NC_000007.13 GI: 224589819 ) and PRDX2 for "Peroxiredoxin-2" (Genbank Gene ID: 5478, NC_000003.11, NG 029697.1, GI: 224589815), have in particular been described as immunogens proteins associated with DCIS. A second aspect of the invention relates to a kit for the in vitro diagnosis of breast cancer in an individual, including early forms of breast cancer, and more particularly DCIS, which comprises at least one protein from the group comprising GAL3. , PHB2, RUVBL1, RACK1 and PAK2. In a preferred embodiment of the invention, said at least one protein is a recombinant protein. However, the invention does not exclude the use in the kit of a protein or proteins extracted from body fluids or from cell cultures of healthy individuals or persons suffering from cancer, or of whole proteins or fragments of proteins containing the same. epitope with which the autoantibody reacts, obtained by chemical synthesis.

Preferably, the kit comprises at least two proteins from the group comprising GAL3, PHB2, RUVBL1, RACK1 and PAK2, preferably at least three, and preferably at least four such proteins, packaged separately or in the same container.

In particular, the kit may comprise, packaged separately or in the same container: - GAL3 and PAK2; or GAL3 and PHB2; or GAL3 and RUVBL1; or GAL3 and RACK1; or PAK2 and PHB2; or PAK2 and RUVBL1; or PAK2 and RACK1; PHB2 and RUVBL1; or PHB2 and RACK1; or RUVBL1 and RACK1; or - GAL3, PAK2 and PHB2; or GAL3, PAK2 and RUVBL1; or GAL3, PAK2 and RACK1; or GAL3, PHB2 and RUVBL1; or GAL3, PHB2 and RACK1; or GAL3, RUVBL1 and RACK1; or PAK2, PHB2 and RUVBL1; or PAK2, PHB2 and RACK1; or PAK2, RUVBL1 and RACK1; or PHB2, RUVBL1 and RACK1; or - GAL3, PAK2, PHB2 and RUVBL1; or GAL3, PAK2, PHB2 and RACK1: 15 or GAL3, PAK2, RUVBL1 and RACK1; or GAL3, PHB2, RUVBL1 and RACK1; or PAK2, PHB2, RUVBL1 and PACK1. Preferentially, the kit comprises all the proteins GAL3, PHB2, RUVBL1, RACK1 and PAK2, packaged separately or in the same container. The kit may also comprise a different protein, that is to say not belonging to the list of the five proteins above, or a plurality of such different proteins, for example one or more of the proteins from PPIA, PRDX2, MUC1, HSP60 and FKBP52. The kit which is the subject of the invention also preferably comprises reagents and / or materials conventionally used for carrying out ELISA tests, for example well plates, means for detecting complexes formed by binding the autoantibodies and the associated protein, for example comprising secondary anti-human IgG antibodies, etc. Another aspect of the invention is a composition comprising at least one of the group comprising GAL3, PHB2, RUVBL1, RACK1 and PAK2 for use as a breast cancer vaccine. Preferably, the composition further comprises an adjuvant, conventional in itself for entering the constitution of vaccines.

The characteristics and advantages of the method according to the invention will appear more clearly in the light of the following examples of implementation, provided for illustrative purposes only and in no way limitative of the invention, with the support of FIGS. 1 to 3, in FIG. which: - Figure 1 shows, schematically, the strategy applied to identify the autoantibodies according to the invention; FIG. 2A shows the result of a two-dimensional Western Blot analysis of the respectively cytosolic (lane 1), membrane (lane 2) and nuclear (lane 3) fractions obtained by subcellular fractionation of cancer cells derived from the MCF10A line (MCF1ODCIS. com or MCF1OCA1), highlighting the subcellular markers associated with each fraction, HSP90 for the cytosolic fraction, Calnexin for the membrane / organelle fraction, PARP1 for the nuclear fraction (deposition of 1-10 μg / lane); FIG. 2B represents two-dimensional electrophoresis gels of lysates of subcellular fractions, cytosolic (gel 1), membrane (gel 2) and nuclear (gel 3), respectively, of cancer cells derived from the MCF10A line (MCF1ODCIS.com or MCF1OCA1) after screening with a mixture of sera from individuals with DCIS; FIG. 2C shows enlargements of two-dimensional electrophoresis gel regions of lysates of subcellular fractions, respectively cytosolic and membrane, of cancerous cells derived from the MCF10A line (MCF1ODCIS.com or MCF1OCA1) after screening with a mixture of serums of individuals with early breast cancer, including DCIS (1st line), with benign breast lesions (BBL) (2nd line), immunodeficient (AID) (3rd line) and healthy (HC) (4th line) . For each column, the arrows indicate respectively the location of the tasks corresponding to the proteins GAL3 (1st column), RACK1 (2nd column), PAK 2 (3rd column), PHB2 (4th column) and RUVBL1 (3rd column); and FIGS. 3A to 3C show the curves resulting from a multiparametric ROC analysis illustrating the specificity and the sensitivity for the autoantibodies (biomarkers) reacting with the GAL3 protein, the PAK2 protein, the RUVBL1 protein, the PHB2 protein, the RACK1 protein, separately or in combination (panel of five biomarkers), for groups of individuals with DCIS versus healthy individuals (HC) (Figure 3A), individuals with early stage primary breast cancer (PBC) versus healthy individuals (HC) (Figure 3B) and individuals with early breast cancer (DCIS and PBC), versus healthy individuals (HC) (Figure 3C). MATERIAL AND METHODS 1. Patient selection Serum samples from women with breast cancer were obtained from the cancer center of Val d'Aurelle (Montpellier, France) between September 2004 and February 2008. The protocol of the study was approved by the Advisory-Committees-of-Protection-of-the-People of Montpellier (Saint Eloi) and the board of revision of RBM03-63-INSERM. All patients provided informed written consent to participate in the study. Samples were collected, processed and stored in the manner described in the Desmetz et al., 2009 publication. The population used for the research phase, known as the "Discovery Group" (Group 1), consists of 80 individuals. known status: 20 women undergoing surgery and histopathologic diagnosis of early forms of breast cancer (DCIS, n = 10, PBC, n = 10), 20 women of the same age with mild breast lesions (BBL), 20 women of the same age as healthy controls (HC) with negative mammograms, negative breast physical exams for at least the last four years, and no history of malignancy, immunodeficiency, autoimmune disorder, hepatitis, or previous HIV infection, and finally 20 women with autoimmune deficiency (AID) accompanied by inflammation (rheumatoid arthritis, n = 10, systemic lupus erythematosus, n = 10). The serum of an entirely different group is randomly divided into two groups: - a group of learning (Group 2) comprising 91 samples of individuals of the same age (28 DCIS, 29 patients with primary breast cancer at an early stage (PBC), and 34 HC), - a validation group (Group 3) comprising 91 independent 10 age-matched samples (27 DCIS, 30 patients with PBC, and 34 HC). Group 4 of "additional validation" of Example 3 below consists of the addition of Group 2 and Group 3. The DCIS were classified according to the classification of tumors of the World Health Organization. Health and were graded according to Bloom and Richardson. Patients were classified according to their stage of cancer using the American Joint Committee's Cancer Classification Manual, sixth edition. No patients received neoadjuvant therapy, all underwent surgery within four weeks of the initial breast cancer diagnosis. 2. Cell culture The human breast cancer cell lines MCF1ODCIS.com, more precisely comedo-type DCIS, and MCF1OCA1 derived from the MCF10A xenograft model, are cultured by known methods (eg Tait et al., 2007). ). The cells are maintained in modified Dubelcco Eagle medium. DMEM / F12 (1: 1) supplemented with 5% equine serum, 1014 / ml bovine insulin, 25 ng / ml epidermal growth factor, 0.514 / ml hydrocortisone, and 100 ng / ml cholera toxin. The culture medium is supplemented with penicillin and streptomycin at 100 U / ml. The cells are cultured at 37 ° C. under 5% CO2. DMEM / F12, equine serum, penicillin, and streptomycin are from Gibco®, and insulin, epidermal growth factor, hydrocortisone, and cholera toxin are from Sigma-Aldrich. 3. Subcellular fractionation Subcellular fractionation is carried out using the "ProteoExtract® Subcellular Proteome Extraction Kit" (Calbiochem, Merckbiosciences, Germany), following the manufacturer's recommendations. Briefly, different protein extracts, depending on their subcellular location (including cytosolic, membrane / organelle, nuclear fractions), are obtained from the cells using the fraction-specific buffers provided in the kit. The protein fractions are purified by ice-cold acetone and dissolved in two-dimensional electrophoresis gel lysis buffer, as described in the publication of Desmetz et al., 2008. The quality of subcellular proteome fractionation was confirmed using the "ProteoExtract® S-PEK Antibody Control kit" kit (Calbiochem). 4. Two-dimensional electrophoresis (2-DE) and Western Blot analysis All reagents and materials for the 2- DE / western blot experiments are from GE Healthcare. All the experiments were triplicated. The techniques used are as described in Desmetz et al., 2009. Briefly, 150 lig of each subcellular fraction are separated on a 2-DE gel. The separated proteins are transferred to a Hybond-P membrane of polyvinylidene fluoride (PVDF) (Millipore) and incubated with 4 lots of 5 will be "cancer" individuals (DCIS + PBC) and 4 lots of 5 will be "control" individuals. Group I (BBL, HC, AID) at a dilution of 1: 200. The proteins are revealed by a horseradish peroxidase-coupled goat anti-human IgG polyclonal antibody (Jackson ImmunoResearch Laboratories) at a dilution of 1: 3000 and using a West Pico Chemiluminescent Substrate SuperSignal Kit (Pierce). 5. Gel enzymatic digestion and mass spectrometry A silver coloration of the two-dimensional electrophoresis gels makes it possible to visualize the spots which are then excised and digested in the gel with trypsin (Promega), as indicated in Desmetz et al. , 2009. The digests are fully dehydrated in a vacuum centrifuge and re-suspended in 10 μl of 2% formic acid. The fragments are desalted by means of Ziptip® C18 tips (Millipore), eluted with 10 μl of 50% acetonitrile / 0.1% trifluoroacetic acid, and dried before being reconstituted for separation by nano-liquid chromatography (nano). -LC) then analysis by mass spectrometry (MS). The peptides are separated on an UltiMate® 3000 nano-LC system (Dionex) coupled to a micro-fraction collector, with an Acclaim® PepMap® column (C18, 3 i.tm, 100 Å), 75 i.tm/15 cm (Dionex).

Mass spectrometry analyzes to identify proteins are performed using an Applied Biosystems MALDI TOF / TOF® 4800 Proteomics Analyzer Mass Spectrometer. The spectra of m / z 700-400 are acquired in positive mode using 1500 laser pulses. The precursor ions of the most abundant peptides with a signal-to-noise ratio greater than or equal to 50 are selected for MS / MS analysis using 3500 laser pulses of m / z 300-1500. The MS / MS spectra are compared to the Uniprot human protein database (uniprot_sprot300108) of the European Institute of Bioinformatics, using the ProteinPilot® 2.0 software and the Paragon method (Ab Sciex, Software revision 50861). Proteins corresponding to a single peptide with a high degree of confidence (> 95%) are considered as positively identified. 6. Biomarker Analyzes The autoantibodies to the GAL3, PAK2, PHB2, RACK1 and RUVBL1 antigens were detected using MaxiSorp® high protein binding plates (Nunc, Roskilde, Denmark) and assayed as described. below. More specifically, the immuno-enzymatic method used for the analyzes of these biomarkers is an indirect ELISA type method, conventional in itself. Negative and positive controls were also performed: a-positive controls: a range of dilutions of the antibody to be assayed b-negative controls: absence of antigen, absence of antibodies to be assayed, absence of detection antibodies, absence of detectable substrate, substrate alone, buffer alone The ELISA method used comprises the following steps, for each antigen: - Antigen fixation (Ag): 100 μl of a solution of Ag (50 ng / well) diluted in phosphate buffered saline (PBS) are dispensed into each of the wells used in the MaxiSorp® plate (with the exception of the wells for negative controls without antigen). The plate is then incubated at 4 ° C overnight, the wells are washed (0.1% PBS, Tween®20) to remove excess antigen, 2 times, the non-specific sites of the plates are blocked for 2 hours at room temperature (RT) in 50 mM HEPES, 200 mM NaCl, 0.08% triton X-100, 25% glycerol, 1 mM DTT, 40 mM NaOH and 1% BSA at pH 7.5, then the plates are washed twice (other blocking buffers, such as PBS, 0.1% Tween®20, 2% BSA (Desmetz et al., 2009) or PBS, 0.1% milk, may be used with similar results), - fixation of the autoantibody to be assayed: 100 μl of the patient's sera diluted 1/500 in PBS are deposited in the wells (with the exception of the controls), the plates are incubated 2 h at RT, the plates are washed 4 times, revelation of the fixed autoantibodies: the plates are incubated with the anti-human IgG secondary polyclonal antibody for 1 h at RT with shaking. After 4 washes, the plates are incubated with the substrate (tetramethylbenzidine, TMB) for 15 min and the absorbance at 450 nm determined after adding H2SO4 to stop the reaction. For each serum the test is triplicated. The antigens PAK2, RACK1, PHB2 and RUVBL1 used are recombinant proteins from Abnova, and GAL3 is a recombinant protein from Calbiochem. 7. Methods of statistical analysis The data quantified by ELISA are expressed by the medians. The differences between the groups are analyzed by a Wilcoxon test. The differences are considered statistically significant when p <0.05.

The performance of the tests based on the individual values of the autoantibodies and on the combined values (linear combination) are based on the analysis of the ROC (Receiver Operating Characteristic) curves. The generalized ROC criterion determines the best linear combination (virtual marker) of tumor markers, which is such that the area under the ROC curve (AUC) is maximized (Piura et al., 2011). The accuracy, sensitivity, and specificity of the autoantibody-based or combination-based assays were evaluated using the optimal cutoff value calculated to maximize the Youden index. This index is defined as the sum of the sensitivity and the specificity minus 1.

Statistical analyzes were performed with InStat (v3.06, Graph Pad), STATA 11.0 (2009 StataCorp), mROC (Kramar et al., 2001). EXAMPLE 1 FIG. 1 represents, schematically, the steps of the approach used by the inventors to arrive at the invention, which are described in detail hereinafter. 1 / Step 1 of research: identification of autoantibodies specifically associated with early forms of breast cancer, in particular DCIS The cytosolic, nuclear and membrane fractions respectively obtained by subcellular fractionation of cells of the MCF1ODCIS.com line or the MCF1OCA1 cell line, were subjected to two-dimensional Western Blot analysis.

The purity of these fractions was confirmed by the presence of the respectively associated subcellular markers: HSP90 for the cytosolic fraction (fraction 1), Calnexin for the membrane / organelle fraction, and PARP1 for the nuclear fraction (Figure 2A). The molecular weight marker (PM) is shown on the right in this figure.

Proteins from each of these fractions were then screened with serum samples from the discovery group (Group 1), for each category CCIS / PBC, BBL, AID and HC. The solutions obtained were subjected to two-dimensional electrophoresis and the electrophoresis gels submitted either to a Western blot with the sera, or to silver staining, to highlight the spots corresponding to the proteins having a reactivity with respect to these sera. . A total of 85 proteins showed remarkably high reactivity with serum samples from patients with early forms of breast cancer (DCIS / PBC), while not reactive with other serum categories. The gels obtained for each of the subcellular fractions, after screening with the sera of the CCIS / PBC patients, are shown in FIG. 2B, each spot corresponding to a protein reactive with respect to the serum CCIS / PBC. These spots were excised and digested in the gel with trypsin, and the protein material obtained was analyzed by MALDI-TOF MS / MS mass spectrometry. 37, 22, and 8 proteins with different biological functions were identified respectively in the cytosolic, membrane and nuclear fractions. GAL3, RACK1 and PAK2, localized in the cytosolic fraction, and PHB2 and RUVBL1, localized in the membrane or organelle, were selected for subsequent analyzes. The selection of these five proteins was based on published information associating them with the development of epithelial cancer and its progression, particularly breast cancer.

Experiments carried out on cells of different cell lines (MCF1ODCIS.com and MCF1OCA1) made it possible to identify the same five proteins in the same respective fractions. FIG. 2C illustrates, for each of these five proteins, their presence in the gels after screening with the DCIS / PBC sera, and their absence after screening with the other control sera (BBL, AID, HC) (black arrows in the figure) . This first step demonstrates that autoantibodies capable of reacting respectively with the proteins GAL3, PHB2, RUVBL1, RACK1 and PAK2, are present in the serum of patients suffering from DCIS / PBC, and absent in the serum of healthy, immunogenic individuals. deficient or with benign breast lesions (BBL). 2 / Step 2 of tests and validation: statistical analysis of the panel of the five biomarkers The levels of the five biomarkers in the sera (autoantibodies directed respectively against GAL3, PAK2, PHB2, RACK1 and RUVBL1) were determined by setting The sera of the learning group (Group 2) are used for specific tests of the ELISA type, using the respective antigens: GAL3, PAK2, PHB2, RACK1 and RUVBL1. First, the reproducibility of the experiment was verified. All samples were blinded and duplicated in at least two independent experiments. Reproducibility was tested by performing the experiment on duplicate representative samples. In addition, samples from the same patients were tested on different plates, on different days. For all five biomarkers, the replicate samples showed a strong correlation, with R2 values> 0.99. In addition, a variability of less than 10% between the results indicates that the method used is reliable and reproducible. The median values for each individual biomarker were measured.

From these values, a univariate analysis was performed. For each biomarker, the ROC curve representing the individual values obtained was established for CCIS versus HC, PBC versus HC, (DCIS + PBC) versus HC. In particular, it appears that the autoantibodies directed respectively against GAL3, PHB2 and RACK1 make it possible to discriminate between subjects with PBC and healthy individuals. This difference is also significant for the specific autoantibodies of GAL3 and RACK1, for the cumulative group (PBC + DCIS) versus HC. Multivariate analysis models mROC were also generated to evaluate the reliability of a diagnostic test based on the combined detection, in the serum from the individuals, of specific autoantibodies of GAL3, PHB2, RACK1, RUVBL1 and PAK2. Surprisingly, it emerges that it is possible to discriminate between individuals newly diagnosed with DCIS and HC healthy subjects, with an area under the AUC (area under the curve) curve. 0.85 (with a 95% Cl confidence interval between 0.76 and 0.95) (Figure 3A). These results clearly show the improvement of the DCIS diagnostic test against each individual biomarker, using the panel of five biomarkers. In patients with PBC, an AUC AUC of 0.81 (95% Cl [0.72-0.88]) is obtained. The discrimination between the group containing individuals with early stages of breast cancer (DCIS + PBC) and the group of healthy individuals (HC) is high, with a AUC of 0.81 (95% CI [0.74- 0.86]). The sensitivity, specificity and overall accuracy of the diagnosis of DCIS, PBC and the group (DCIS + PBC) were thus high for the panel of the five autoantibodies according to the invention. The test performance of these five autoantibodies as biomarkers was validated on an independent validation group (Group 3), which contains 91 additional patients, again by ELISA tests. The diagnostic performance of DCIS, early PBC and the cancer group (DCIS + PBC) was maintained, with a AUC of 0.81 (95% Cl [0.70-0.93]), 0.78 ( 95% Cl [0.66-0.90]), and 0.80 (95% Cl [0.70-0.89]), respectively. EXAMPLE 2 Using the same ELISA assays as previously described, the combination of the five autoantibodies was evaluated for the detection of breast cancers in sera of 68 healthy women (HC), 15 patients with a diagnosis of mastopathies, and breast cancer patients, 59 with primary invasive breast cancer (PBCI) and 56 with 25 DCIS. From the levels of autoantibodies measured in the sera, a statistical analysis was performed. The effectiveness of the combination for the diagnosis of cancer was evaluated by the area under the ROC curve. For the Cancer group (PBCI + DCIS) versus HC group, an area under the curve of 0.81 was obtained, with a 95% CI confidence interval between 0.74 and 0.87. The following linear combination formula was established: 3.04 x [GAL3] - 5.62 x [PAK2] + 2.41 x [PHB2] + 5.51 x [RACK1] - 5.31 x [RUVBL1] where [] expresses the concentration of the autoantibody in the serum. For the DCIS group versus HC group, an area under the curve of 0.85 was obtained, with a 95% CI confidence interval between 0.78 and 0.92. The following linear combination formula was established: 2.05 x [GAL3] - 5.89 x [PAK2] + 1.93 x [PHB2] + 7.68 x [RACK1] - 5.93 x [RUVBL1] where [] expresses the concentration of the autoantibody in the serum. These results clearly demonstrate that the method according to the invention, in particular using the detection in the serum of the individual of a set of autoantibodies directed respectively against one of the five proteins among GAL3, PHB2, RACK1, RUVBL1 and PAK2, is particularly reliable for performing in vitro diagnosis of early forms of breast cancer, and in particular DCIS. EXAMPLE 3 An additional experiment was carried out according to the same protocol as in Example 1 above, for step 2 of tests and validation, with the following differences. The population used for this example, called the "Additional Validation Group" (Group 4), comprises 182 independent samples of individuals of known status and age: 68 HC, 55 DCIS and 59 breast cancer patients at one time. early stage (PBC). The performance of the tests based on the individual values of the autoantibodies and on the combined values (linear combination) are based on the analysis of the curves ROC (for the English Receiver Operating Characteristic), the analysis SVM (for English Support Vector Machine) and Linear Discriminant Analysis (LDA). The SVM analysis identifies maximum margin hyperplanes for separating two classes of samples in n-dimensional space. The separation boundary is chosen as the one that maximizes the margin. The LDA analysis makes it possible to define a linear combination of markers characterizing or separating two classes of samples. To determine the robustness of the classification algorithms (ROC, SVM and LDA) and to evaluate the misclassification rate, cross-validation methods and statistical interference based on succession resampling (for bootstrapping) are used on the same additional validation group (Group 4). Cross-validation is based on sampling techniques. In practice, the sample (Group 4) is divided k times, then one of the k samples is selected as the validation sample and the other k-1 samples constitute the training set. We start again k times to estimate the prediction error. The performance of each classification algorithm is evaluated with cross-validation where k = 10 ("10-fold cross-validation").

A special case of this method (called "leave-one-out cross-validation") is also used, in which a sample is removed from Group 4, a combination is made with the rest of the samples, then the model is validated on the removed sample and repeat until exhaustion of samples. The performance of each classification algorithm is also estimated by a .632+ bootstrap resampling technique. Statistical analyzes are performed with InStat (v3.06, GraphPad), STATA 11.0 (2009 StataCorp), mROC (Kramar et al., 2001), TANAGRA (v1.4.41) and the MultiExperiment Viewer software (Mev, version 4 ). The levels of the five biomarkers in the sera (autoantibodies directed respectively against GAL3, PAK2, PHB2, RACK1 and RUVBL1) were determined by using the sera of the additional validation group (Group 4) specific type tests. ELISA, using the respective antigens: GAL3, PAK2, PHB2, RACK1 and RUVBL1. First, the reproducibility of the experiment was verified, as shown in Example 1 above. Again, for all five biomarkers, the replicate samples showed a strong correlation, with R2 values> 0.99, and a variability of less than 10% between the results. The median values for each individual biomarker were measured. From these values, a univariate analysis was performed. For each biomarker, the ROC curve representing the individual values obtained was established for CCIS versus HC, PBC versus HC, (DCIS + PBC) versus HC. Figures 3A to 3C respectively illustrate the curves obtained. The results obtained demonstrated that the autoantibodies directed respectively against GAL3, PHB2, and RACK1, used individually, make it particularly possible to discriminate between subjects with PBC and healthy individuals with areas under the AUC curve ( for the English "area under the curve") respective of 0.67 (95% Cl [0.57-0.77]), 0.62 (95% Cl [0.52-0.72]) and 0 61 (95% Cl [0.51-0.71]). This difference is also particularly significant for the autoantibodies specific for GAL3 and RACK1, for the cumulative group (PBC + DCIS) versus HC with AUCs respectively of 0.61 (95% Cl [0.53-0.69]) and 0.59 (95% Cl [0.51-0.67]). For each of the individual autoantibodies, the overall diagnostic accuracies of DCIS, PBC and the group (DCIS + PBC) were further determined to be 60 to 65%, 64 to 69% and 50 to 55%, respectively. %, demonstrating that each of these autoantibodies taken individually allows the in vitro diagnosis of breast cancer, including its early forms. Multivariate analysis models were also generated to evaluate the reliability of a diagnostic test based on the combined detection, in the serum from the individuals, of specific autoantibodies of GAL3, PHB2, RACK1, RUVBL1 and PAK2 respectively. . The mROC curves obtained are illustrated in FIGS. 3A to 3C, respectively for DCIS versus HC, PBC versus HC and (DCIS + PBC) versus HC. The AUC values are substantially the same as those obtained for Example 1. For patients with DCIS and HC healthy subjects, the area under the AUC curve is 0.85 (95% CI [ 0.76-0.95]); for patients with PBC, it is 0.81 (95% Cl [0.72-0.88]); and for patients with early-stage breast cancer (DCIS + PBC), it is 0.81 (95% CI [0.74-0.86]). The overall diagnostic accuracies of DCIS, PBC and the group (DCIS + PBC) for the panel of five autoantibodies are 77%, 76% and 74%, respectively. SVM analysis The SVM classification method gives results very similar to the mROC method in terms of sensitivity, specificity and overall diagnostic accuracy of the DCIS, PBC and the group (DCIS + PBC). In particular, the overall diagnostic accuracies of DCIS, PBC and the group (DCIS + PBC) for the panel of five autoantibodies are 80%, 74% and 70%, respectively. LDA analysis The LDA classification method gives very similar results to the mROC and SVM methods in terms of sensitivity, specificity and overall diagnostic accuracy of the DCIS, PBC and the group (DCIS + PBC). In particular, the overall diagnostic accuracies of DCIS, PBC and the group (DCIS + PBC) for the panel of five autoantibodies are 77%, 76% and 74%, respectively. The test performance of these five autoantibodies as biomarkers was validated for each classification algorithm (mROC, SVM and LDA) using standard cross-validation methods ("10-fold 30 cross-validation"). "Leave-one-out cross-validation" and "bootstrapping"). The diagnostic performance of the DCIS, early PBC and the cancer group (DCIS + PBC) has been maintained regardless of the classification algorithm used, with overall diagnostic accuracies of between 67% and 79%, between 70% and 75%, and between 67 and 74%, respectively.

The results above clearly demonstrate that the detection in the serum from an individual of autoantibodies specific for GAL3, PAK2, PHB2, RUVBL1 and / or RACK1 makes it possible to establish a diagnosis of early breast cancer which is precise, specific and sensitive, and which gives particularly particularly advantageous results in terms of compromise between high diagnostic reliability and low cost and difficulty of implementation of the test. The method according to the invention advantageously makes it possible to effectively detect in vitro non-invasive early forms of breast cancer, such as early DCIS or PBC, which can potentially evolve into invasive forms of breast cancer. Thus, at-risk patients can be screened and treated more quickly, which increases the chances of recovery, avoids cumbersome treatments for patients and reduces expenses. The very good reliability of this test makes it possible to avoid false-positive or false-negative type of diagnostic errors, compared with conventional methods of diagnosing breast cancer. The method according to the invention is also easy and quick to implement, which makes it particularly suitable for application as a routine examination.

BIBLIOGRAPHIC REFERENCES Brem R.F. et al., 2003, Improvement in sensitivity of screening mammography with computer-aided detection: a multiinstitutional trial. AJR Am J Roentgenol 181: 687-93 Desmetz C. et al., 2008, Proteomics-Based Identification of HSP60 as a Tumor-Associated Antigen in Early Stage Breast Cancer and Ductal Carcinoma in situ. Journal of Proteome Research 7: 3830-7 Desmetz C. et al., 2009, Identification of a new panel of serum autoantibodies associated with the presence of in situ carcinoma of the breast in younger women. Clin Cancer Res 15: 4733-41 Jernal A. et al., 2009, Cancer statistics. CA Cancer J. Clin. 59: 225-49 Kramar A. et al, 2001, mROC: a computer program pour combinateur tumour markers dans predicting disease states. Comput Methods Programs Biomed, 66: 199-207. Piura E. et al., 2011, Autoantibodies to tailor-made panels of tumor-associated antigens in breast carcinoma. J Oncology, 2011: 982425 Qiu J. et al., 2008, Occurrence of autoantibodies to annexin I, 14-3-3 theta and LAMR1 in prediagnostic lung cancer will be. J Clin Oncol 26: 5060-6 Tait L.R. et al., 2007, Dynamic stromal-epithelial interactions during progression of MCF1ODCIS.com xenografts. Int J Cancer 120: 2127-34

Claims (4)

CLAIMS1 - A method for in vitro diagnosis of breast cancer in an individual, characterized in that it comprises the detection, in a biological sample from said individual, of at least two autoantibodies each directed respectively against one of the proteins set comprising GAL3, PAK2, PHB2, RUVBL1 and RACK1. 2 - Process according to claim 1, characterized in that said biological sample is a sample of a fluid of the individual, including a blood sample of the individual. 10 3 - Process according to any one of claims 1 to 2, characterized in that the detection of each autoantibody comprises contacting said biological sample with an immunological analysis reagent comprising the protein from GAL3, PAK2, PHB2, RUVBL1 and RACK1 against which said autoantibody is directed, and detecting the presence of complexes formed by specific binding of said protein and an autoantibody present in said sample. 4 - Use of at least two proteins among GAL3, PHB2, RUVBL1, RACK1 and PAK2 for the in vitro diagnosis of breast cancer in an individual. Kit for the in vitro diagnosis of breast cancer in an individual, characterized in that it comprises at least two proteins among the group comprising GAL3, PHB2, RUVBL1, RACK1 and PAK2. Kit according to claim 5, characterized in that said at least two proteins are recombinant proteins.