ITUB20150390A1 - METHOD FOR THE DIAGNOSIS OF ENDOMETRAL CARCINOMA - Google Patents

Description

? METHOD FOR THE DIAGNOSIS OF ENDOMETRAL CARCINOMA?

DESCRIPTION

The present invention relates to a method for the diagnosis of endometrial carcinoma by means of a metabolomic analysis of the blood and the bioinformatic manipulation of metabolic profiles by means of classification models.

Endometrial cancer? the pi? common invasive cancer of the female genital tract and? responsible for 7% of all invasive cancers in women (excluding skin cancers).

Cancer of the endometrium? rare in women of age? less than 40 years old. The peak of incidence? between 55 and 65 years. Clinico-pathological studies and molecular analyzes have supported the classification of endometrial cancer into two broad categories: Type I and Type II.

Type I? the pi? frequent, with a percentage of cases higher than 80%, undermines the proliferative endometrial glands and as such is indicated with the term endometrioid carcinoma. Generally occurs in a picture of endometrial hyperplasia and, like this, is associated with obesity, diabetes, hypertension, infertility? and uncontested estrogenic stimulation. Recent studies have provided further evidence to support the thesis that endometrial hyperplasia? a precursor of endometrial carcinoma (Muller GL et al. Allelotype mapping of unstable microsatellites establishes direct lineage continuity between endometrial precancers and cancers. Cancers Res 56: 4483, 1996). Type II endometrial cancer generally affects women a decade or more. later than type I (65-75 years) and unlike type I it develops mainly on a picture of endometrial atrophy.

Type II accounts for less than 15% of endometrial cancer cases and is poorly differentiated (G3). The subtype pi? common ? the serous one, so called because of the biological and morphological overlap with ovarian carcinoma. The less common histological subtypes also belong to this category: clear cell carcinoma and Malignant mixed malignancy.

Currently, a mass screening on an asymptomatic population in age? premenopausal and postmenopausal for the early diagnosis of endometrial cancer, as is done for cervical cancer with the Pap test, not? feasible.

Studies conducted on exocervical sampling have shown a false negative frequency of about 40-50% as the exfoliated endometrial cells, having undergone the action of the vaginal environment, show alterations for which they lose the characteristics that allow the differentiation of the tumor cell from the normal one. On the other hand, the prognosis? closely linked to earliness? 5-year survival drastically decreases from 78-98% in the case of stage I diagnosis to 3-10% in the case of stage IV diagnosis.

Several thousand human serum metabolites have been identified to date and the application of metabolomics has allowed the development of biomarkers for numerous disorders such as schizophrenia (Kaddurah-Daouk R., Metabolic profiling of patients with schizophrenia, PLOS Med 2006; 8: e363 ), meningitis (Subramanian A. et al., Proton MR / CSF analysis and a new software as predictors for the differentiation of meningitis in children, NMR Biomed 2005; 18: 213-25) and colon cancer (Denkert C. , et al., Metabolite profiling of human colon carcinoma? deregulation of TCA cycle and amino acid turnover, Mol. Cancer 2008; 7: 1-15). However, the use of metabolomics in gynecology has so far been limited to studies concerning ovarian cancer (Fan L. et al. Identification of metabolic biomarkers to diagnose epithelial ovarian cancer using a UPLC / QTOF / MS platform Acta Oncologica, 2012; 51: 473? 479). Studies conducted in gas chromatography coupled with mass spectrometry and chemometric techniques for the diagnosis of endometrial cancer are not reported in the literature to date.

AND? therefore today strongly felt the necessity? of a non-invasive diagnostic system that allows to carry out a screening on the population at risk by age? or for known risk factors, in order to identify this fearful female neoplasm early.

Advantageously, the present invention solves the above problems by means of a non-invasive method for the diagnosis of endometrial carcinoma. There are currently no other non-invasive diagnostic methods that can allow such histological discrimination of this type of tumor.

To follow the object of the invention will come? illustrated in detail.

BRIEF DESCRIPTION OF THE FIGURES

Figure 1 shows the result of the OPLS-DA analysis based on the metabolomic profile data of patients with endometrial cancer and healthy controls.

Scores plots discriminate between the two classes without overlapping. The triangles represent patients with endometrial cancer, while the circles represent healthy patients. The main components PC1 and PC2 shown on the axes describe respectively 16.5% and 14.9% of the global variance.

Figure 2 shows, according to the invention, the histological classification (type I carcinoma vs type II carcinoma) obtained with the PLS-DA model. The dots represent the metabolomic profiles of women with type I endometrial cancer, while the triangles represent those of patients with type II endometrial cancer. Only one of these samples is placed by the model in an area not uniquely attributable to the correct area.

DEFINITIONS

By metabolomics we commonly mean the analysis of cellular processes through the study of the metabolic profile of small molecules of an organism. By metabolomic analysis, the inventors intend to refer to the execution of a process aimed at identifying and determining the concentration of the largest possible number of metabolites within a biological sample.

By metabolites we commonly mean small molecules derived from the anabolic or catabolic biological processes of a cell or a set of cells. The inventors intend to refer, with the term metabolites, to all molecules with molecular weight lower than 1000 Daltons potentially identifiable and measurable within a biological sample.

By metabolic profile we mean the specific pattern that the metabolites assume in the patient's blood according to their relative proportions.

The PLS-DA (Partial Least Squares Discriminant Analysis)? a supervised method that uses multivariate regression techniques to extract, through linear combinations of the original variables (X), the information that can predict belonging to a specific class (Y). To evaluate the effectiveness of class discrimination, a permutation test is performed. In each permutation, a PLS-DA model is constructed between the data (X) and the permuted class labels (Y) using the optimal number of components determined by cross validation for the model based on the assignment of the original classes. Two types of statistical tests are performed to measure the capacity? discrimination between classes. The first is based on the prediction accuracy when training the model. The second is based on the separation distance based on the ratio between the sum of the quadratic distances within the classes and between the classes (B / W-ratio).

L? OPLS-DA (Orthogonal Partial Least Squares - Discriminant Analysis)? an important development of the PLS-DA technique which? was proposed to manage orthogonally the variation of classes in the data matrix. OPLS-DA increases the classification performance of PLS-DA models. Classification performance is estimated based on? K-fold cross validation? dividing the data matrix into k random subsets. For each computation cycle, one of the k subsets is kept aside as a test set and the remaining k-1 subsets act as trainers. Each of the k subsets is used once as a test set, generating k precision values. The accuracy of the classification is calculated as the average of the accuracy rates in the k subsets. The model is cross-validated with the? Leave one out cross validation? Method. (LOOCV) in order to validate it. The data matrix before being subdivided into k subsets is scaled to the mean and unit variance. In other words, the mean and standard deviation of the training data are used to indicate the center and scale the test data. Once the model? been trained, it is used to check if the data has generated an? overfitting ?. To do this, a validation set with labels of known class is created and it is checked whether this gives an accuracy rate comparable to that of the training data. Another method? an R2 / Q2 validation plot that helps assess the risk of the current model being spurious, that is, the model only fits well to the set subsets but does not predict Y as well for new observations. The value of R2 represents the percentage variance of the training set that can? be explained by the model. The Q2 value? a cross-validated measure of R2. Does this validation compare the goodness? of adaptation of the original model with the goodness? of adaptation of the different models based on the data in which the order of the Y observations is permuted randomly, while the matrix is kept intact. The criteria for validity? of the model are as follows:

1. All Q2 values on the permuted dataset must be less than the estimated Q2 value on the current dataset. If this is not checked, it means that the model is over-adapted.

2. The regression line (line joining the real Q2 point to the centroid of the permuted cluster of Q2 values) has a negative y-intercept value.

Support Vector Machines (SVMs) are relatively new supervised machine learning techniques used for classification. SVMs were first proposed in 1982 by Vapnik (Vapnik, V. Estimation of Dependences Based on Empirical Data; Springer Verlag: New York, 1982). The basic principle of SVMs, which are essentially binary classifiers,? the following: given a data set with two classes, a linear classifier is constructed in the form of a hyperplane, which has the maximum margin in the simultaneous minimization of the empirical classification error and the maximization of the geometric margin. In the case of data sets that are not linearly separable, the original data is mapped into a pi? high dimensional space and a linear classifier? built in this new space (this is known as the "kernel").

Considering a set of training data xi n, i = 1,?, M where each xi falls into one of the two categories yi {1,1} the SVM determines the hyperplane whose parameters are given by (w, b) as obtained from the solution of the following convex optimization problem:

subject to conditions

where c? the regularization parameter, which? a compromise between the accuracy of learning and the prediction term, and? a measure of the number of misclassifications. The inclusion of the regularization term reduces the problem of overfitting.

DECISION TREES. Decision trees build classification models based on recursive partitioning of data. Typically, a decision tree algorithm starts with the entire data set, the data is split into two or more? subgroups based on the values of one or more? attributes and then repeatedly divides each subset into subsets pi? small as long as? the size of each subset reaches an appropriate level. The whole modeling process can? be represented in a tree structure, and the generated model can? be summarized as an "if-then" set of rules. Decision trees are easy to interpret, computationally undemanding, and capable of dealing with noisy data. Most decision trees deal with classification problems, such as the object of this invention. In this context, the technique? also referred to as a classification tree. In the tree representation, a node represents a dataset, and the whole dataset? represented as a node at the root.

DETAILED DESCRIPTION OF THE INVENTION

The present refers to a method for the diagnosis of endometrial carcinoma, based on the metabolomic analysis of the blood and on an integration of the results obtained by means of a multivariate analysis using discriminant analysis models chosen from the group consisting of PLS-DA and OPLS-DA, or computer learning models chosen from the group consisting of SVM and decision tree.

Object of the invention? a method for the diagnosis of endometrial cancer based on metabolomic analysis of the blood, said method comprising the following steps:

(I), said training phase, comprising:

- GCMS or GCxGCMS analysis of blood samples taken from patients with endometrial cancer and from healthy controls;

- the integration of the results obtained through a multivariate analysis using at least one discriminant analysis model or a computer learning model to train at least one classification model; (II), said attribution step, comprising the GCMS or GCxGCMS analysis of an unknown blood sample; and its attribution to the class to which it belongs on the basis of the classification model formulated in the training phase.

The multivariate analysis, carried out on the chromatograms collected using - at least one discriminant analysis model selected from the group consisting of: PLS-DA and OPLS-DA, or

- a computer learning model selected by the group consisting of: SVM and decision tree;

has advantageously allowed the satisfactory dichotomous classification (? Healthy patient? vs? Patient with endometrial carcinoma?) of unknown samples. The classification model obtained with PLS-DA multivariate analysis even allowed the histological discrimination of carcinoma (type I carcinoma vs type II carcinoma); there are currently no other non-invasive diagnostic methods that can allow such histological discrimination of this type of tumor. In said training phase (I), samples taken from patients with endometrial carcinoma and from healthy women with similar physical (BMI, age, comorbidities) and social (educational level, socio-economic status) characteristics are analyzed, and for by means of these the classification models are trained. Said training phase? aimed at creating and delimiting the characteristics of the metabolic profile present in the blood of the two groups. Why? you can get a good predictivity? of the classification model? It is necessary to subject to multivariate analysis a number of samples equal to at least 80% of the number of variables identified, called samples belonging to at least 2 different classes.

In said phase (II) of attribution, the unknown samples are subjected to GCMS analysis, and the resulting chromatograms are classified according to the previously trained models, estimating the most? probable class of belonging.

The method for the diagnosis of endometrial carcinoma of the present invention is not based on the measurement of the concentration of the single metabolites, but the entire cluster of metabolites (metabolic profile) is considered as a biomarker, which, in order to be present in different proportions in the 2 groups, they allow the insertion in two different classes of belonging.

Preferably, said training phase (I) further comprises the following sub-phases:

- extraction and derivatization of metabolites from blood samples taken from patients with endometrial carcinoma and from healthy controls;

- GCMS or GCxGCMS analysis of the extracted and derivatized metabolites to obtain a chromatogram for each sample, each chromatogram being a metabolic profile;

- creation of a data matrix of the metabolic profiles of patients with endometrial cancer and healthy controls;

- structuring of at least one classification model following multivariate analysis of the data matrix; said multivariate analysis carried out using at least one discriminant analysis model or a computer learning model to train at least one classification model. Different classification models lend themselves to the purpose of the present invention; preferably said classification models are selected from the group consisting of: PLS-DA, OPLS-DA, SVM and Decision Tree.

Preferably, said attribution phase (II) further comprises the following sub-phases:

- the application of the first three sub-phases of said phase (I) to the unknown sample; And

- the attribution of the metabolic profile to a class based on the classification model trained in phase (I).

Preferably, the method of the present invention provides a classification model trained to a dichotomous classification? Healthy patient? or? Patient with endometrial cancer ?. Even more? preferably, said classification model? also trained for a histological classification of carcinoma? type I? or? type II ?.

Preferably, said metabolite extraction? carried out after adding a known aliquot of a reference compound to the sample; preferably, said reference compound? I re-titled it.

Preferably, said extraction? carried out using an extraction mixture consisting of an aqueous mixture of an alcohol and an aprotic polar solvent, preferably CH3OH / H2O / CHCl3, even more? preferably in volume ratio 2-3 / 0.5-0.5 / 0.5-1.

In a preferred embodiment, said extraction and derivatization sub-phase comprises:

i) agitation of the sample obtained from the treatment with the extraction mixture;

ii) centrifugation of said sample obtained from i);

iii) derivatization of the supernatant obtained from ii) by treatment with methoxylamine hydrochloride in pyridine;

iv) silanization of the supernatant obtained from iii) with a silanizing agent selected from the group consisting of: N, O-bis (trimethylsilyl) trifluoroacetamide (BSTFA), N-methyl-N- (trimethylsilyl) trifluoroacetamide (MSTFA), hexamethyl disilazane (HMDS ), 1- (trimethylsilyl) imidazole (TMSI), N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA), 1- (tertbuthyldimethylsilyl) imidazole (TBDMSIM) in optional presence of trimethylchlorosilane (TMCS).

To obtain a separation between the metabolites useful for the purposes of this invention? it is possible to operate both in one-dimensional gas chromatography and in two-dimensional gas chromatography; two-dimensional gas chromatography? preferable as the best resolving power of the technique offers better classification accuracy. However, as shown in the EXAMPLES section? can also operate with the pi? common one-dimensional gas chromatography.

The gas chromatograms obtained, preferably in the mode? SCAN, are integrated in order to identify all the peaks that have an area greater than 10 times the background noise of the gas chromatographic trace.

Using the peak of the reference compound (preferably ribitol) as a reference both for quantitative analysis and for centering retention times, each peak is identified on the basis of an m / z quantization signal and at least 2 m / z of qualification. Following the integration, quantification is carried out using the method of normalized percentage areas. The results obtained from this quantization (normalized percentage areas) are transferred to a matrix in which each sample represents a line and the columns are represented by the various metabolites uniquely identified by means of their gas chromatographic retention time, compared to the retention time of the reference compound. The first column of the matrix is used to define the membership class of the sample. In the scenario pi? simple can there be only two classes? Healthy patient? and? Patient with endometrial carcinoma ?, plus? below there are evidences of the functioning of the invention on the basis of this dichotomous classification.

Multivariate statistical data analysis (PLS-DA and OPLS-DA) and machine learning (SVM and decision tree) are performed on normalized and corrected chromatograms (on the ribitol peak area) using SIMPCA-P 13.0 (Umetrics ), RapidMiner 5.3 (Rapid-I) and R (Foundation for Statistical Computing, Vienna). The values are centered on the mean and the variance is normalized.

For the metabolic profile, the OPLS-DA model has shown capacity? modeling and predictivity? satisfactory using one predictive component and three orthogonal components (R2Ycum = 0.995, Q2

cum = 0.985). Figure 1 shows the separation between classes obtained with the OPLS-DA model.

AND? Furthermore, a classification based on the histology of the carcinoma was constructed using a PLS-DA model. As shown in Figure 2, only one sample is placed in a dubious area of the class definition space.

The present invention may? be better understood in the light of the following non-limiting embodiments.

EXAMPLES

The diagnostic methodology object of the present invention? was developed starting from the metabolomic analysis, carried out on blood samples collected from patients with a certain diagnosis of endometrial carcinoma, before the hysterectomy and by a group of control women, with similar physical and socio-economic characteristics but with a healthy uterus . The information about the isotype and the degree of the neoplasm were collected after the hysterectomy on the basis of the pathological evidence obtained from the analysis of the explanted organ.

Collection of samples

The samples were taken from 88 women with endometrial cancer and 80 healthy women, who voluntarily donated blood samples. I study ? been approved by the ethics committee of the university? of Magna Grecia in Catanzaro and healthy patients and volunteers signed an informed consent about the purposes of the study. Blood samples were taken immediately prior to the hysterectomy using BD Vacutainer? Tubes, serum? was frozen at -80 ° C until the time of analysis. The diagnostic suspicion of endometrial cancer following hysteroscopic examination with biopsy of the endometrial lesion? was confirmed by the anatomopathological examination of the uterus following the hysterectomy. AND? A control group was also set up by taking blood samples from women without signs of endometrial cancer and with similar physical and socio-economic characteristics (weight, height, body mass index, age, marital status, educational level, etc.).

The demographic and clinical characteristics of the cases and controls are shown in table 1 while in table 2 the anatomopathological characteristics of the investigated tumors are listed.

Table 1: Characteristics of the study population

Table 2: anatomopathological characteristics of the investigated tumors

Extraction and derivatization of metabolites

Fifty microliters of serum were transferred into 2 mL Eppendorf tubes and added 20 L of a 1 g / L solution of ribitol and 200 L of a mixture consisting of 2.5 parts of Methanol, 1 part of Water and 1 part of Chloroform ( CH3OH: H2O: CHCl3, 2.5: 1: 1).

The solution is vortexed for 30 seconds.

The samples were then centrifuged at 16000 rpm for 10 minutes at 4 ° C. A? 200 L aliquot of the supernatant? was collected and transferred into new 2 mL Epperndorf tubes and added with 200 L of H2O and vortexed for 30 seconds and centrifuged again at 16000 rpm for 5 minutes at 4 ° C.

A? 350 L aliquot of the supernatant? was collected again and transferred into 1.5 mL glass vials, and lyophilized.

The lyophilized sample? was treated with 50 L of methoxylamine hydrochloride 20 mg / mL in pyridine. The reaction ? was carried out at 37 ° C under stirring (350 rpm) for 90 minutes. Upon completion, 50 L of N, O-bis (trimethylsilyl) trifluoroacetamide (BSTFA) with 1% trimethylchlorosilane (TMCS) was added to each vial and the silanization reaction? was carried out at 37 ° C for 60 minutes under stirring (350 rpm).

MDGCMS analysis

For two-dimensional gas chromatography? was used a primary column (placed in the first oven), of the type SLB-5ms 30,0 m x 0,25 mm ID with 1 m of film thickness [silphenylene polymer, practically equivalent in polarity? to poly (5% diphenyl / 95% methylsiloxane)] (J&W Agilent) what? been connected to position 1 of the 7-port interface (SGE). A BPX-505.0m x 0.50mm ID with 0.25m film thickness? been connected to position 7 of the interface. A BPX-501.5m x 0.25mm ID, 0.25m? been fixed at position 6 and connected to a flame ionization detector (FID) set at 320 ° C, while the analytical column of 5.0 m (chemically identical to the one connected to the FID)? been connected to the qMS system.

The column connected to the FID? was used to reduce flows in the second dimension and to verify that an unrepresentative compound was not the result of a random fluctuation in chromatography.

AND? A 40 L external capillary (20 cm x 0.71 mm OD x 0.51 mm ID stainless steel) was used to connect ports 3 and 4 of the SGE interface.

The same heat program for the two ovens was: 80 ° C for 1 minute then heating up to 320 ° C at 3 ° C / minute and maintained for 4 minutes.

The initial helium pressure (constant linear velocity)? was set at 129.6 kPa. The initial auxiliary helium pressure of the APC (advanced pressure control), also operating in conditions of speed? linear constant? was set at 90.4 kPa.

The injection volume at 1 L with a split ratio: 1: 5. The modulation period? was set at 4.1 s (accumulation period 4.0 seconds, injection period 0.1 seconds). The conditions of the quadrupole mass spectrometer were: modality? ionization rate: electron impact (70 eV), mass range: 40-600 m / z, scan rate: 10,000 amu / second. GCMS analysis

For one-dimensional gas chromatography? A column of the type CP-Sil 8 CB GC Column, 30 m, 0.25 mm, 1.00 m, (Agilent J&W) was used. The GC thermal program had an initial temperature of 100 ° C for 1 minute then heating up to 320 ° C at 4 ° C / minute and 4 minutes of hold time for a total run time of 60 minutes.

The initial helium pressure (constant linear velocity of 39 cm / s)? was set at 83.7 kPa. The injection volume at 2 L with a split ratio: 1: 5. The conditions of the quadrupole mass spectrometer were: modality? of ionization: electron impact (70 eV), mass range: 35-600 m / z, scanning speed: 3,333 amu / second with a solvent cut time of 4.5 minutes.

Creating a data matrix

In a TIC chromatogram, more than 100 percentages are normally detected. of 250 signals, some of these peaks have not been further investigated why? no matches were found in other samples, why? in too low concentration or why? of poor quality? spectral in order to be confirmed as metabolites.

A total of 198 endogenous metabolites such as amino acids, organic acids, carbohydrates, fatty acids and steroids were detected. For the identification of the peak, the linear retention index (LRI) was used, setting as tolerance a difference between the tabular Kovats index and the maximum experimental one of 10, while the minimum of compatibility for research in the libraries and was placed at? 85%. Two libraries were used: NIST11 and a specially developed library derivatizing over 500 metabolites under the same conditions as the analyzed samples. The peak areas were normalized and corrected to the ribitol signal. The results were summarized in a comma separated matrix file (CSV) and loaded into appropriate software for statistical processing.

The gas chromatograms obtained in the modality? SCAN have been integrated to identify all peaks that have an area greater than 10 times the background noise of the gas chromatographic trace. Any peak? been identified on the basis of a quantization m / z signal and at least 2 qualification m / z signals. Following the integration yes? proceeded to quantify with the method of normalized percentage areas, the peak of the ribitol? was used as a reference for both quantitative analysis and to center retention times. The results obtained from this quantization (normalized percentage areas) were transferred to a matrix in which each sample represents a line and the columns are represented by the various metabolites uniquely identified by means of their gas chromatographic retention time. The first column of the matrix is used to define the membership class of the sample. In the scenario pi? simple can there be only two classes? Healthy patient? and? Patient with endometrial carcinoma ?, plus? below there are evidences of the functioning of the invention on the basis of this dichotomous classification. Further evidence has been obtained regarding the possibility? of the different classification models tested to also predict the histotype of the neoplasm and the grading.

Statistic analysis

Multivariate statistical data analysis (PLS-DA and OPLS-DA) and machine learning (SVM and decision tree) were performed on normalized and corrected chromatograms (on the ribitol peak area) using SIMPCA-P 13.0 ( Umetrics), RapidMiner 5.3 (Rapid-I) and R (Foundation for Statistial Computing, Vienna).

Were the values centered on the mean and variance? been normalized. Results

For the metabolic profile, the OPLS-DA model has shown capacity? modeling and predictivity? satisfactory using one predictive component and three orthogonal components (R2Ycum = 0.995, Q2

cum = 0.985). The other classification models have shown good (even if lower than the OPLS-DA) capacity? classification. Different approaches are possible for the definitive attribution of the class to which the unknown sample belongs. Can you? use the response of a single model or integrate the responses of the various models into a pi? complex decision algorithm.

Table 3 reports some diagnostic performance estimation indices used to evaluate the investigated models. The sensitivity? ? been calculated as TP / (TP + FN), where TP represents the number of true positives, that is? samples correctly diagnosed as having endometrial cancer by the proposed model, and FN the number of false negatives, that is? samples incorrectly identified as negative. The specificity? ? been estimated as TN / (TN + FP), where TN represents the number of true negatives, that is? samples correctly diagnosed as healthy and FP represents false positives, that is? the number of subjects misdiagnosed as healthy. The positive likelihood ratio (PLR)? was calculated as Sensitivity / (1-Specificity), while the negative (NLR) was calculated as (1-Sensitivity) / Specificity. The negative predictive value (NPV)? was estimated as TN / (TN + FN), while the positive one (VPP) as TP / (TP + FP). Accuracy represents the percentage of all correct assignments and? was estimated as (TP + TN) / (TP + FP + TN + FN) while the reproducibility? as the number of successful reassignments in 10 replicates of one sample run.

Table 3? Diagnostic performance of the investigated models

To identify the metabolites that contributed most to the separation of classes? The score of the important variables in the projection (VIP) for each component was calculated. The VIP scores represent the weighted sum of the squares of the pls loading, taking into account the quantity? of y-variance explained in each dimension. Two peaks showed a VIP score greater than 2 in both PLS-DA and OPLS-DA models (both in the endometrial vs control classification and in the type i vs type ii classification). These have also been identified as important nodes in the decision tree, these observations suggest a great importance of these variables in the classification processes (data not shown). The first metabolite (VIP-score = 2.3; spectrometric similarity = 91%; LRI = 11)? turned out to be a signal attributable to the amino acid glutamine, while the second (VIP-score = 2.1; spectrometric similarity = 89% LRI = 16)? found to be a signal attributable to -glucono lactone.

Claims (10)

CLAIMS 1. A method for the diagnosis of endometrial cancer based on metabolomic analysis of blood, said method comprising the following steps: (I) training phase comprising: - GCMS or GCxGCMS analysis of blood samples taken from patients with endometrial cancer and from healthy controls; - the integration of the results obtained through a multivariate analysis using at least one discriminant analysis model or a computer learning model to train at least one classification model; (II) attribution phase comprising the GCMS or GCxGCMS analysis of an unknown blood sample and its attribution to the class on the basis of the classification model formulated in the (I) training phase. 2. Method according to claim 1 wherein - said discriminant analysis model? selected from the group consisting of: PLS-DA and OPLS-DA, or - said computer learning model? selected from the group consisting of: SVM and decision tree. 3. Method according to one or more? of claims 1-2 wherein step (I) comprises the following sub-steps: a) extraction and derivatization of metabolites from blood samples taken from patients with endometrial carcinoma and from healthy controls; b) GCMS or GCxGCMS analysis of the extracted and derivatized metabolites to obtain a chromatogram for each sample; c) creation of a data matrix of the metabolic profiles of patients with endometrial cancer and healthy controls; d) structuring of at least one classification model following multivariate analysis of the data matrix; where dictates multivariate analysis? performed using at least one discriminant analysis model or a computer learning model to train at least one classification model. 4. Method according to one or more? of the preceding claims in which said phase (II) further comprises the following sub-phases: a) extraction and derivatization of metabolites from at least one unknown blood sample; b) GCMS or GCxGCMS analysis of the extracted and derivatized metabolites to obtain at least one chromatogram of the unknown sample; c) creation of a metabolic profile from said chromatogram of the unknown sample; d) attribution of the metabolic profile to a class based on the classification model trained in phase (I). 5. Second method or more? of the previous claims where the number of blood samples taken from patients with endometrial cancer and from healthy controls? equal to at least 80% of the number of metabolic profile variables. 6. Method according to one or more? of the preceding claims in which said classification model? trained in a dichotomous classification? healthy patient? or? Patient with endometrial cancer ?. 7. Method according to one or more? of the previous claims n which classification model? also trained for a histological classification of carcinoma? type I? or? type II ?. 8. Method according to one or more? of the preceding claims wherein said extraction and derivatization comprises i) agitation of the sample obtained by adding an extraction mixture; ii) centrifugation of said sample obtained from i); iii) derivatization of the supernatant obtained from ii) by treatment with methoxylamine hydrochloride in pyridine; iv) silanization of the supernatant obtained from iii) with a silanizing agent selected from the group consisting of: N, O-bis (trimethylsilyl) trifluoroacetamide (BSTFA), N-methyl-N- (trimethylsilyl) trifluoroacetamide (MSTFA), hexamethyl disilazane (HMDS ), 1- (trimethylsilyl) imidazole (TMSI), N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA), 1- (tertbuthyldimethylsilyl) imidazole (TBDMSIM); where said extraction mixture? consisting of an aqueous mixture of an alcohol and an aprotic polar solvent. 9. Method according to one or more? of the preceding claims wherein said metabolite extraction? carried out by adding an aliquot of reference compound, preferably ribitol. 10. Method according to one or more? of the preceding claims further comprising the following steps: -Integration of the chromatograms obtained according to one or more? of the preceding claims, where said integration provides for the identification of all peaks which have an area greater than 10 times the background noise of the chromatographic trace; using the peak of the reference compound as a reference both for quantitative analysis and for centering retention times, where each peak? identified on the basis of: - a quantization signal m / z; and of - at least 2 m / z qualification signals; - quantification with the method of normalized percentage areas; - transfer of the data obtained from said quantification to a matrix in which each sample represents a line and the columns are represented by the various metabolites uniquely identified by means of their chromatographic retention time.