Specific metabolic markers of neurodegenerative diseases such as Parkinson's disease and their use in diagnosis

Abstract

The invention relates to the identification of new combinations of metabolites specific to neurodegenerative diseases, in particular Parkinson's disease (PD), for use as biomarkers in the early diagnosis of neurodegenerative diseases, in particular Parkinson's disease in the prodromal stage.

Description

Title of the invention: SPECIFIC METABOLIC MARKERS OF NEURODEGENERATIVE DISEASES SUCH AS PARKINSON'S DISEASE AND THEIR USE IN DIAGNOSIS

[0001] The invention relates to the identification of new combinations of specific metabolites of neurodegenerative diseases, in particular Parkinson's disease (PD), for use as biomarkers in the early diagnosis of neurodegenerative diseases, in particular Parkinson's disease in the prodromal stage.

[0002] In the description below, references in brackets ([ ]) refer to the list of references at the end of the text. State of the art

[0003] Parkinson's disease (PD) is one of the most widespread neurodegenerative diseases in the world, affecting approximately 1% of adults over 60 years of age [1]. From its onset, the disease progresses gradually and continuously through three stages: preclinical, prodromal, and clinical. These stages correspond respectively to the onset of neurodegeneration without symptoms (preclinical phase), then the progression of the neurodegenerative process leading to the appearance of non-motor symptoms (prodromal phase), and finally to the appearance of motor symptoms (clinical phase) [2]. It is these symptoms, present only in the last stage, that are sufficiently characteristic of PD to allow for its diagnosis [3]. However, they appear when 70-80% of the dopaminergic neurons of the nigrostriatal pathway have already degenerated [1], therefore too late to consider neuroprotective intervention [4].Furthermore, post-mortem histopathological analyses of brains show that currently the accuracy of the diagnosis of Parkinson's disease in patients with motor symptoms for less than five years [5] is only 53%. Thus, the late diagnosis and its imprecision [3, 6, 7] contribute significantly to limited and purely symptomatic patient management and therapies, which, moreover, become ineffective after several years.

[0004] Moreover, although many molecules have already been tested and have shown promising theoretical or preclinical evidence of efficacy, there is still no neuroprotective agent or therapy that can effectively slow, stop or reverse neurodegeneration in patients with PD, partly because these therapeutic strategies are applied too late [8-10].

[0005] Thus, being able to detect PD with a high degree of specificity, during the early phases, represents a crucial challenge to enable the success of curative or preventive strategies for PD, and the development of new therapies capable of delaying or slowing its progression [11, 12].

[0006] The "prodromal" phase is characterized by a set of non-motor symptoms including neuropsychiatric disorders (e.g., apathy, depression, anxiety). Although they are expressed in 40 to 50%

[13] of Parkinson's patients, they are not exclusive to the disease and can be found in other pathologies.

[0007] Thus, finding reliable and specific biomarkers of the prodromal phase, suitable and highly predictive for clinical use, is a significant challenge, which motivated the inventors' study. Their initial hypothesis is that the metabolome—i.e., the overall pool of metabolites—which results from gene expression but also reflects the impact of numerous environmental factors

[15] , could contain potential biomarkers of the disease [16-19], now considered a multifactorial disease resulting from complex interactions between genetic and environmental factors

[14] . Moreover, metabolic alterations have already been described in patients exhibiting neuropsychiatric symptoms similar to those observed in the prodromal phase of PD

[20] .Furthermore, it has been shown that analyzing metabolic dysregulations observed in the serum of newly diagnosed PD patients has allowed for their precise differentiation from healthy subjects

[21] . Finally, alterations in the plasma metabolome have been correlated with disease progression in Parkinson's patients

[22] . However, the study of the metabolome as a tool for predicting the prodromal phase of PD has never been carried out. Accessing this prodromal phase requires the use of animal models. Nevertheless, although numerous animal models of PD have been developed, none recapitulates all the phenotypic and etiological characteristics of the disease

[23] .

[0008] Thus, although early detection of PD represents a key step for the development of curative therapies or therapies capable of delaying or slowing its progression, there is currently no specific tool to enable the diagnosis of patients with PD before the onset of motor symptoms, particularly during the prodromal phase. The discovery of easily measurable and highly predictive biomarkers of the prodromal phase of the disease appears to be a milestone for accelerating the development of curative therapies and improving patient care. Description of the invention

[0009] Also, the Inventors have developed a novel strategy enabling them to offer New biomarkers capable of specifically and early diagnosis of neurodegenerative diseases, such as Parkinson's disease (PD), particularly prodromal Parkinson's disease, have been developed. This strategy, based on the use of different animal models and patient cohorts, overcame inherent limitations and increased the incomplete predictive value of each individual model, revealing a characteristic set of deregulated metabolites usable as potential biomarkers. Finally, the study was extended to human serum samples from two biobanks to validate the clinical relevance of this new PD biomarker.

[0010] Thus, through behavioral and histological assessment and using metabolomic approaches based on nuclear magnetic resonance (NMR) spectroscopy followed by advanced multivariate statistics, the Inventors investigated metabolic changes (i) in the serum and brain tissue of three different animal models of PD, two rodent models and one non-human primate model, mimicking the different phases of the disease [24-26], including the prodromal phase, and (ii) in the serum of two cohorts of patients with de novo PD (n=95) (USA, Italy). Dysregulations common to both serum and tissue samples were observed in the animals, indicating that blood could reflect central (brain) dysfunction

[27] , at least partially.

[0011] The 6-hydroxydopamine (6-OHDA) rat model was initially used because it stably mimics each stage of PD, without changes between stages, and it has already been validated for the study of prodromal phases

[24] . Furthermore, it represents a phenotypically accurate PD model, with good isomorphism and good predictive value

[28] , although it is etiologically limited. The α-synuclein rat model

[29] was subsequently used because it has the advantage of being progressive in establishing the phenotype, which makes longitudinal studies possible. Moreover, it is closer to human pathology, particularly with regard to the anatomopathological features and pathophysiological mechanisms of the disease

[30] . Finally, non-human MPTP (l-methyl-4-phenyl-l,2,3,6-tetrahydropyridine) primates were used because they share greater genome sequence identity with humans and significant neuro-anatomical similarities.

[0012] Metabolic changes were analyzed at the blood (serum) level to find metabolites that are easy to transpose to clinical practice, and in specific brain regions differently affected by the neurodegenerative process (the dorsal striatum (DS) and the nucleus accumbens (NAcc)) in order to potentially associate changes in these metabolites with the pathophysiological processes involved in PD.

[0013] From a translational perspective, the metabolomics study was extended to sera from newly diagnosed Parkinson's patients from two bio different banks (United States, Italy). The translational strategy revealed common metabolic dysregulations in the serum of different animal models and patients with PD. Some of these were also found in tissue samples (animal models), perhaps reflecting pathophysiological mechanisms associated with the development of PD.

[0014] Thus, this interspecies study has strengthened the relevance of circulating metabolites for the diagnosis of Parkinson's disease (PD). Furthermore, it proposes a composite biomarker constructed with a combination of six metabolites that can distinguish PD-mimicking animals from control (sham) animals, regardless of the stage of the disease, and particularly from the prodromal phase. Importantly, the same combination of metabolites has been used to distinguish PD patients from healthy subjects.

[0015] It is interesting to note that in animals, certain metabolic dysregulations appeared before motor symptoms and could thus represent early biomarkers of the disease.

[0016] Thus, the Inventors have demonstrated that the relative concentrations of 6 metabolites: acetoacetate, betaine, β-hydroxybutyrate (BHB), creatine, pyruvate, and valine, combined using a logistic regression method, can constitute a precise composite biomarker of Parkinson's disease in animals and humans. The use of this biomarker made it possible to distinguish American Parkinson's patients from healthy controls with high sensitivity (0.95), and from Italian Parkinson's patients with slightly lower sensitivity (0.83). The divergent pre-analytical procedure applied for serum collection in the two cohorts (i.e., the tubes used, sample processing, etc.) could constitute a major source of experimental variability that could explain this difference [31, 32].Furthermore, it should be noted that the human samples were collected according to a very strict and standardized experimental protocol, but one that was not designed for metabolomics. This reflects the power of metabolomics and the robustness of the results, and supports the view that serum biomarkers, which are easier to use and less invasive than imaging methods, could have similar efficacy

[33] .

[0017] It is interesting to note that among the six metabolites used to construct the composite biomarker of PD in animal models or in humans, the dysregulation of betaine, BHB, and pyruvate was corrected by chronic treatment with a dopaminergic agonist (i.e., pramipexole or Pra), which also corrects motivational deficits in 6-OHDA rats mimicking the prodromal phase. This pharmacological validation underscores their potential involvement in the cellular dysregulations implicated in neuropsychiatric symptoms. associated with this prodromal stage. The metabolic dysregulations reported in this study could therefore also provide some clues to the mechanisms involved in the progression of PD.

[0018] In summary, the present study reported, for the first time, common serum metabolic alterations in three animal models mimicking Parkinson's disease (PD), although from different cohorts and species. Consequently, this common signature is likely to reflect specific alterations related to the pathophysiology of PD. In particular, specific dysregulations characteristic of the prodromal stage were identified. In addition to providing clues for further investigation of metabolic alterations in animal models, metabolic dysregulations were also found in the sera of Parkinson's and newly diagnosed (i.e., clinically diagnosed) patients from two biobanks not specifically designed for metabolomics.Logistic regression, using the same metabolite levels as those used for animals, allowed very adequate discrimination between healthy subjects and patients with MP in two cohorts of different origins (at least 0.83).

[0019] Overall, the present interspecies study demonstrated the utility and reproducibility of untargeted metabolomics as a non-invasive approach for the search for PD biomarkers. This approach appears particularly promising for PD patients, as serum is readily available and the biomarker can be relevant even in the early stages of the disease when other methods fail.

[0020] The present invention therefore relates to an in vitro method for the early diagnosis of a neurodegenerative disease from a biological sample of a patient comprising: a) the in vitro measurement in said biological sample of the expression levels (E) of at least 6 of the following metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine; b) comparison of the expression levels of metabolites measured in step a) with those of a reference biological sample or reference values; where a difference (e.g. joint dysregulation) in the expression levels of metabolites measured in step a) compared to those of the reference biological sample or reference values ​​is an indication that said patient has the neurodegenerative disease.

[0021] According to a particular embodiment of the process of the present invention, step a) further comprises the in vitro measurement of the expression level of at least one metabolite selected from the group consisting of alanine, lactate, DMSO2, glycine, serine, threonine, myo-inositol, and leucine.

[0022] According to a particular embodiment of the process of the present invention, step a) comprises or consists of measuring the expression levels of acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine.

[0023] For the purposes of this invention, "early diagnosis" means a diagnosis made at a stage of the neurodegenerative disease before the distinguishing symptoms have appeared. For example, a diagnosis would be considered early if made at the prodromal stage of PD.

[0024] For the purposes of this invention, "neurodegenerative disease" refers to a progressive pathology affecting the brain or, more broadly, the central nervous system, leading to the death of nerve cells. The most well-known and common examples are Alzheimer's disease and Parkinson's disease, but others exist (multiple sclerosis, Huntington's disease, Creutzfeldt-Jakob disease, amyotrophic lateral sclerosis (also known as Lou Gehrig's disease), etc.). Parkinson's disease is a particular example.

[0025] For the purposes of this invention, "biological sample" means a biological sample (e.g. of body fluid (serum, blood, urine, saliva, etc.), or of tissue (brain, bone marrow, liver, skin, etc.)) taken after a 12-hour fast from a patient.

[0026] For the purposes of this invention, "reference biological sample" means a biological sample (e.g., of body fluid (serum, blood, urine, saliva, etc.), or of tissue (brain, bone marrow, liver, skin, etc.)) taken after a 12-hour fast from a healthy individual (control / sham), i.e., not exhibiting any neurological disease, in particular any symptoms of PD or that could be a precursor to PD, no family history of movement disorder and / or PD, and no specific medical condition.

[0027] For the purposes of this invention, "reference value" means the value of the expression level of a metabolite in a healthy individual, that is to say, an individual with no neurological disease, in particular no symptoms of PD or that may be a precursor to PD, no family history of movement disorder and / or PD, and no specific medical condition.

[0028] The term "difference in expression levels" in the context of the present invention refers to the variation in the expression levels of metabolites measured in a patient compared to the expression levels of the same metabolites in a healthy individual (reference values), i.e., an individual without any neurological disease, in particular without any symptoms of Parkinson's disease (PD) or any symptoms that could be precursors to PD, without a family history of movement disorders and / or PD, and without any specific medical condition. For example, in the case of Parkinson's disease.

[0029] The present invention relates to an in vitro method for the early diagnosis of Parkinson's disease from a biological sample of a patient comprising: a) the in vitro measurement in said biological sample of the expression levels (E) of the following metabolites: acetoacetate, labetaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine; b) Calculating the logistic regression Logit(P) according to the following formula: -45.60 + 0.28 Ebhb + 4.54 EPyruvate — 11.23 EVaiin + 7.05 ECacetoacetate + 2.85 ECreatine — 8.27 E Betaine or Ebhb, Epyri|VUle, Eyaiine, EAcetoacetate? Ecreatine and EBetaine correspond respectively to the measured expression levels of B-hydroxybutyrate (BHB), pyruvate, valine, acetoacetate, creatine and betaine, in the biological sample; and

[0030] where a calculated value of Logit(P) greater than 0.316 ua is the indication that said patient has Parkinson's disease.

[0031] The term "calculated Logit(P) value" in the context of the present invention refers to a standard value obtained from the application of a logistic regression using the measurement of expression levels of the metabolites mentioned above. In the case of Parkinson's disease, in the logistic regression using the following six metabolites: acetoacetate, betaine, β-hydroxybutyrate (BHB), creatine, pyruvate, and valine, any individual obtaining a value greater than 0.316 (arbitrary unit, AU) will be considered to have PD.

[0032] According to a particular mode of the process of the present invention, the patient's biological sample comes from a patient in the prodromal stage of MP.

[0033] The term "prodromal stage of PD" in the context of the present invention refers to a stage where a definitive and / or certain diagnosis of PD could not be established based on the symptoms presented. For example, this includes any patient presenting with non-motor symptoms frequently observed in patients who eventually develop the disease and which could represent early warning signs (neuropsychiatric symptoms (apathy, depression, anxiety), loss of smell, constipation, sleep disturbances, digestive disorders, cognitive deficits) and / or any patient with a family history of the disease.

[0034] According to a particular mode of the process of the present invention, the biological sample is, for example, a serum sample.

[0035] According to a particular embodiment of the present invention, the measurement step of expression levels is carried out for example by NMR spectroscopy measurements.

[0036] The present invention also relates to a biomarker for the early diagnosis of a neurodegenerative disease comprising the combination of at least six of the following metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine.

[0037] According to a particular embodiment of the present invention, the biomarker further comprises the combination of at least one metabolite selected from the group consisting of alanine, lactate, DMSO2, glycine, serine, threonine, myo-inositol, and leucine.

[0038] According to a particular embodiment of the present invention, the biomarker comprises or consists of the combination of the following six metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine.

[0039] The present invention also relates to the in vitro use of a combination of metabolites according to the present invention, as a biomarker of a neurodegenerative disease, in particular Parkinson's disease. Brief description of the figures

[0040] [Fig. 1] Fig. 1 shows a schematic of the experimental procedures applied to different animal models of PD. All serum samples were collected after fasting and according to the same protocol. They were analyzed by 950 MHz ¹H NMR. Brain samples from 6-OHDA rats were used for histological and metabolomic analyses performed by 500 MHz HRMAS ¹H NMR.

[0041] [Fig. 2] [Fig. 2] depicts striatal dopaminergic denervation induced by bilateral lesions of the substantia nigra pars compacta (SNc) with 6-OHDA, resulting in apathetic-like behavior and fine motor dysfunction. (A) PD Progression Score (PMP) used to classify 6-OHDA animals according to disease progression. The PMP score is based on the sum of histological and behavioral components (self-administration and gait readjustment), taking values ​​from 0 to 4 for each. In this way, a score can be assigned to each animal from 0 (control / sham animal) to 3-7 (prodromal-type animal), or 8-12 (clinical-type animal). (B) Examples of coronal sections of control (sham) and 6-OHDA rat brains, labeled for tyrosine hydroxylase (TH) on three representative striatal levels localized relative to bregma, and corresponding schemes selected from the atlas of Paxinos and Watson

[78] .The areas used for quantifying dopaminergic denervation in the different subregions analyzed are illustrated. Scale bars represent 2 mm. (C) Quantification of TH labeling loss at striatal levels indicated in (B), expressed as a percentage of the mean value obtained for control (sham) animals (n=22). A large decrease in TH-positive neurons in the DS and a smaller decrease in the Nacc was observed in prodromal (n=14) and clinical (n=15) type rats. (D) SNc 6-OHDA lesion induced a severe instrumental deficit in an operant self-administration procedure. Sucrose. Mean number of sucrose doses obtained per session. (E) SNc 6-OHDA lesion reduced the number of readjustments in a gait readjustment test, only in clinical-type animals. Mean number of forelimb readjustments for two trials before (left bar) and after (right bar) injection of 6-OHDA (prodromal and clinical-type animals) or saline (control / sham). Data are presented as means ± SEM (standard error of the mean) and tested by one-way ANOVA or repeated-measures ANOVA (RM-ANOVA), followed by a Tukey or Sidak post-hoc test. ***: p < 0.001 compared to control (sham); ##: p < 0.01 clinical-type compared to prodromal-type; $$$: p < 0.0001 before surgery compared to after surgery.

[0042] [Fig. 3] Figure 3 represents ambulatory motor activity measured in an open area. Dopaminergic lesions did not affect horizontal ambulatory activity measured over a period of 15 minutes. Data are presented as mean value for each minute ± SEM.

[0043] [Fig. 4] [Fig. 4] represents the evolution of the serum metabolic profile of modified 6-OHDA rats as a function of the different phases of MP. (A) Example of a 950 MHz µH NMR spectrum (spectral areas: 0.5-4.7 ppm and 6-8.5 ppm) acquired with the CPMG (Carr-Purcell-Meiboom-Gill) pulse sequence. Assignment: 1-isoleucine; 2-leucine; 3-valine; 4-BHB; 5-lactate; 6-alanine; 7-arginine; 8-lysine; 9-acetate; 10-glutamine; 11-methionine; 12-acetone; 13-acetoacetate; 14-glutamate; 15-pyruvate; 16-citrate; 17-Asparagine; 18-Creatine; 19-Phosphocreatine; 20-DMSO2; 21-Choline; 22-Phosphocholine; 23-Glucose; 24-Betaine; 25-Myo-inositol; 26-Glycine; 27-Glycerol; 28-Threonine; 29-Glycerophosphocholine; 30-Serine; 31-Ascorbate; 32-Glycerate; 33-Proline; 34-Deoxycytidine Triphosphate; 35-Tyrosine; 36-Histidine; 37-Phenylalanine; 38-Formalate. Macromolecules are not specified (see Table 2).(BC) Orthogonal partial least squares (OPLS) statistical model constructed with NMR data and PMP scores from the animals. R² = 0.926; Q² = 0.604; 1 predictive component and 3 orthogonal components; CV-ANOVA (analysis of variance of cross-validated residuals) p-value = 3.47. (B) Graph of OPLS scores against the first predictive component and the first orthogonal component. A clear gradation of color is observed from left to right, showing that metabolic profiles change with progression of MP. (C) Graph of OPLS weights, plotted in one dimension to mimic an NMR spectrum, with the NMR variables on the x-axis. The weights are colored according to their correlation with the PMP score, from light gray (weak correlation) to dark gray (strong correlation). Metabolites with a correlation > 0.5 are considered the most important in discrimination. The positive peaks indicate metabolites that... are upregulated with increasing score, while negative peaks are those that are downregulated with score evolution. (D) Relative amplitude of the metabolites most involved in metabolic gradation observed with OPLS, i.e. alanine, betaine, BHB, dimethyl sulfone (DMSO2), glycine, lactate, pyruvate, serine, threonine, valine, in control (sham) animals (n=22), prodromal (n=14) and clinical (n=15) 6-OHDA animals.

[0044] Mean ± SEM, one-way ANOVA, followed by a Tukey post-hoc test and a correction for multiple comparisons. *: p < 0.05; **: p < 0.01; ***: p < 0.001. ua: arbitrary units.

[0045] [Fig. 5] Fig. 5 represents the gradation of the brain metabolic profile associated to the PMP score of the animals. (A) Example of HRMAS 'H' NMR spectrum at 500 MHz. Assignment: 1-lactate; 2-alanine; 3-GABA; 4-acetate; 5-glutamate; 6-glutamine; 7-N-acetylaspartate; 8-creatine; 9-phosphocreatine (PCR); 10-phosphoethanolamine; 11-glycerophosphocholine; 12-phosphocholine (PC); 13-choline; 14-scyllo-inositol; 15-taurine; 16-myo-inositol; 17-ascorbate; 18-glycine; 19-glutathione. (B) Score curve of the OPLS model, constructed with the HRMAS 'H' NMR spectra of DS (i, top) and Nacc (ii, bottom). There is a clear gradation of colors from left to right, showing that metabolic profiles evolve with the progression of MP, particularly for the DS samples (i). DS: R2Y= 0.883; Q= 0.703, 1+3 components, CV ANOVA p-value= 0.0003; (ii) Nacc: R2Y= 0.690; Q2= 0.523, 1+1 components, CV ANOVA p-value= 0.0005.(C) Relative amplitude of metabolites in DS (i) and Nacc (ii) in control (sham), prodromal and clinical-type 6-OHDA animals for the 7 key metabolites involved in the metabolic gradation observed in OPLS, i.e. alanine, choline, glutamate, lactate, PC, PCR and taurine. Mean + SEM, one-way ANOVA, followed by a Tukey post-hoc test and correction for multiple comparisons*: p < 0.05.

[0046] [Fig. 6] Figure 6 represents the impact of chronic treatment with a do agonist paminergic (Pra) on behavioral and metabolic dysregulations in 6-OHDA rats. (A) Self-administration performance. Mean number of sucrose doses obtained per session after lesioning or after chronic administration of saline (Veh) or Pra (last 3 days). Prodromal-type animals treated with Pra show a complete reversal of the effect of lesioning on the number of sucrose doses obtained compared to prodromal-type animals treated with Veh. Clinical-type animals treated with Pra show a moderate increase in the number of sucrose doses obtained compared to clinical-type animals treated with Veh, without reaching the performance of control (sham) animals. (B) Serum metabolic dysregulations. Percentage change in serum metabolites between control (sham) and control (sham) animals. prodromal type (solid bar) and between prodromal-type animals treated with Veh or Pra (hatched bar). Alanine, betaine, BHB, and pyruvate vary in the opposite direction to the effect of the lesion. (C) Brain metabolic dysregulation. Percentage change in brain tissue metabolites normalized to control (sham) animals treated with Veh in the DS and in the Nacc for prodromal (i) and clinical (ii)-type animals. Animals treated with pramipexole (hatched bar) showed an inverse change in metabolite levels compared to animals treated with Veh (solid bar). Mean ± SEM, one-way ANOVA, followed by Tukey's post-hoc test. **: p < 0.01; ***: p < 0.001; #: p < 0.05

[0047] [Fig. 7] Figure 7 shows the absence of metabolic dysregulation in serum in animals that received a GFP (Green Fluorescent Protein) control virus infusion. The histogram represents the levels of metabolites corresponding to the different samples taken from the GFP animals for acetoacetate, betaine, BHB, creatine, glycine, myo-inositol, pyruvate, and serine, which are the main metabolites involved in discriminating the 3 groups in the OPLS-DA (discriminant analysis) of α-synuclein animals. White bars correspond to samples at week 0, light gray bars to 3 weeks after virus infusion, and dark gray bars to 10 weeks after. Data are presented as mean ± SEM and are tested by one-way ANOVA followed by a Tuckey post-hoc test.

[0048] [Fig. 8] [Fig. 8] represents the metabolomics study by NMR on serum samples from other PD models in different species. (A) Rat α-synuclein animal model. OPLS-DA scores constructed from serum α-H NMR spectra, compared to the first two orthogonal predictive components. The three groups, mimicking the three different phases of PD, are clearly discriminated. R²Y = 0.929; Q² = 0.518; 1 predictive component and 3 orthogonal components; CV ANOVA p-value = 0.008. (B) Relative amplitude of the metabolites most involved in discriminating between prodromal α-synuclein-type animals (n=11), clinical α-synuclein-type animals (n=11), and control (sham) animals (n=12), i.e., acetoacetate, betaine, BHB, creatine, glycine, myo-inositol (myo), pyruvate, and serine. (C) Non-human primate model MPTP. OPLS-DA scores constructed from serum α-H NMR spectra, compared to the first two predictive and orthogonal components.Control (sham) and clinical-type animals are clearly differentiated. R2Y = 0.998; Q2 = 0.963; 1 predictive component and 1 orthogonal component; CV ANOVA p-value = 0.0005. (D) Histogram showing the relative quantitative variations of metabolites between MPTP (n=8) and control (sham) (n=6) animals for acetoacetate, alanine, betaine, BHB, creatine, lactate, and . Pyruvate and valine, representing the main metabolites involved in the discrimination of the 2 groups in the OPLS-DA. Mean ± SEM, mixed model followed by a Tukey post-hoc test or the Mann-Whitney t-test *: p < 0.05; **: p < 0.01; ***: p < 0.001.

[0049] [Fig.9] Figure [Fig.9] shows the effect of pramipexole on BHB and betaine levels in patients with PD in the US cohort. Histogram showing the relative quantitative values ​​of metabolites in the control group (white bar), patients with PD (dark bar), and patients with PD treated with Pra (hatched bar). Data are presented as mean ± SEM.

[0050] [Fig. 10] Figure 10 represents the logistic regression used to construct a composite biomarker from six serum metabolites: BHB, acetoacetate, valine, creatine, betaine, and pyruvate. (A) Receiver operating curve (ROC) with serum sample data from control (sham) and prodromal (6-OHDA and α-synuclein rats) animal models (AUC = 0.936, sensitivity: 0.853, specificity: 0.88). (B) ROC with patient and corresponding control data from the US cohort (AUC = 0.88, sensitivity: 0.957, specificity: 0.714).

[0051] [Fig. 11] [Fig. 11] represents the logistic regression curve for the serum metabolite panel including: BHB, acetoacetate, valine, creatine, betaine, and pyruvate. (A) ROC curve from serum samples of all MP-type animal models. AUC = 0.954, sensitivity: 0.85, specificity: 0.915. (B) Logistic regression algorithm for all MP-type animals versus the control (sham) group. EXAMPLES EXAMPLE 1: MATERIALS AND METHODS Plan

[0052] Fig. 1 illustrates a diagram of the entire protocol applied to the animal models. The rats of the first cohort (Fig. 1(i)) were trained for two weeks on a sucrose self-administration task until they achieved stable performance and were subjected to a step readjustment test before receiving a bilateral intracerebral injection of 6-OHDA (n=29) or physiological saline (NaCl, n=22) into the SNc. After stabilization of the lesion (approximately 3 weeks), self-administration was resumed for one week, and the step readjustment test was repeated to monitor the evolution of the animals' performance, before subjecting each group (6-OHDA and NaCl) to 2 weeks of daily injections of Pra 0.2 mg / kg (6-OHDA Pra n=15; NaCl Pra n=1) or physiological saline (NaCl 0.09%; 6-OHDA NaCl n=14; NaCl NaCl n=1). Self-administration was continued throughout the Pra treatment, and a final step readjustment test was performed at the end of treatment. Serum samples were collected after intracerebral injection of 6-OHDA or NaCl, following stabilization of self-administration performance, and at the end of Pra treatment. Brains were collected at the end of the behavioral procedure, frozen in liquid nitrogen, and maintained at -80°C before being processed for histology and HRMAS*H NMR experiments.

[0053] The second cohort of rats ([Fig. 1] (ii)) received an intracerebral injection of AAV-hA53Ta-syn (n=12) or AAV-GFP (n=12) into the SNc. Serum samples were collected longitudinally during the study, before, 3, and 10 weeks after AAV injection.

[0054] Monkeys (n = 8) received intramuscular injections of MPTP (0.3 to 0.5 mg / kg) every 4 to 5 days for 3 weeks

[77] . Repeated administration of low doses of MPTP was used to mimic a moderate stage, with slow disease development and the triggering of moderate dopaminergic lesions. Serum samples were collected at the beginning of the experiment and after the monkeys had reached a stable parkinsonian state ([Fig. 1] (iii)).

[0055] All serum samples were analyzed by 1H NMR at 950 MHz and brain samples were subjected to 1H HRMAS NMR at 500 MHz. Animals

[0056] Rats: The experiments were carried out on adult male Sprague-Dawley rats (January, Le Genest-Saint-Isle, France), weighing approximately 300 g (7 weeks old) at the start of the experiment. They were housed under standard laboratory and ethical conditions with a reversed light-dark cycle (12 h / light-dark cycle, with the lights turned on at 7 p.m.) and with food and water available ad libitum.

[0057] Monkeys: The experiments were carried out on adult male Macacafascicularis, weighing between 5 and 8 kg, aged 3 to 5 years and housed in standard conditions (12:12 hour light-dark cycle; 23°C; 50% humidity).

[0058] The protocols used complied with the European Union Animal Protection Act 2010 and French Directive 2010 / 63, and were approved by the French National Ethics Committee (2013 / 113) No. 004 and by the local ethics committee C2EA84 and CELYNE C2EA. Surgery

[0059] Bilateral injection of 6-OHDA: As previously described [18, 23], rats received a subcutaneous injection of desipramine (15 mg / kg) 30 minutes before the 6-OHDA injection to protect noradrenergic neurons. All animals had The animals were then anesthetized by intraperitoneal injection of ketamine-xylazine (100-7 mg / kg) and placed in a stereotaxic frame (Kopf Instruments). The stereotaxic coordinates of the injection sites were as follows, according to the stereotaxic atlas of Paxinos and Watson

[78] and relative to the bregma: incisor bar placed at -3.2 mm: anteroposterior (AP) = -5.4 mm / lateral (L) = + / -1.8 mm / dorsoventral (V) = -8.1 mm. The animals received a bilateral injection of 2.3 µl of 6-OHDA (3 mg / ml) or physiological saline (0.9% NaCl; control / sham), at a rate of 0.5 µl / min. For animals injected with 6-OHDA that experienced transient starvation 2-3 days post-operation, supplementation with a high-calorie liquid diet and palatable feed was implemented for 1-2 weeks and discontinued 10 days prior to serum sampling. Animals that did not recover were excluded.

[0060] Bilateral α-synuclein injection: As previously described

[19] , rats were anesthetized with 2.5% isoflurane and placed in a stereotaxic frame (Kopf Instruments) to receive two bilateral injections of AAV-hA53Ta-syn (100-7.0 x 0.12vg / ml) or AAV-GFP (7.0 x 0.12vg / ml, control / sham animals) into the SNc (AP=-5.1 and -5.6mm / L = + / - 2.2mm / V = ​​8mm from the bregma). After awakening, the animals were returned to the animal facility. Behavioral assessment

[0061] 6-OHDA rats: The rats were subjected to 2 behavioral tests, 3 weeks after the operation, when the 6-OHDA lesion had stabilized

[79] .

[0062] Operant self-administration of sucrose (motivational component): Rats were trained to self-administer a 2.5% sucrose solution in operating chambers (Med Associates, St. Albans, VT, USA) containing an active, reinforced lever, pressing which resulted in the delivery of 0.2 ml of a sucrose solution associated with a low-intensity light stimulus, and an inactive, unreinforced lever, pressing which resulted in neither the delivery of sucrose nor the light stimulus. This task was performed with a fixed ratio of 1, with a single press of the reinforced lever resulting in the delivery of a reward. Each session ended when 100 rewards were obtained, or at the end of the allotted time (1 hour). The number of rewards obtained was counted for each session using the MED-PC IV software.

[0063] Gait readjustment test (motor component): The animals, restrained by the hind third of their bodies, were moved a length of 90 cm in a straight and regular left-to-right motion, and vice versa, along a smooth-surfaced table. The number of forelimb adjustments made during the movement was counted

[80] . The test was performed three times, by two different experimenters, blinded (unaware of the nature of the experimental groups).

[0064] MPTP Monkeys: The severity of the parkinsonian condition was assessed using a scale rating taking into account classic motor symptoms (bradykinesia, rigidity, tremors, blocking, posture and arm position), spontaneous activities (arm movements, spontaneous eye movements and activity in the housing cage) and other activities (vocalization, triggered eye movements and feeding)

[81] .

[0065] Histology: Immunostaining of TH and quantification of denervation in 6-OHDA rats.

[0066] Brain harvesting: Rats were sacrificed by decapitation 1 day after the last sucrose self-administration session and 1 hour after the last Pra or Veh injection (i.e., 6 weeks after the 6-OHDA or saline injection). The brains were immediately frozen in liquid nitrogen and stored at -80°C. They were then sliced ​​using a cryostat thermostated at -20°C. Sections of the regions of interest were taken for both TH immunostaining and metabolomics analysis. For HHRMAS 1-cell NMR, thick sections of DS and Nacc tissue were grouped in inserts and maintained at -80°C until NMR analysis.

[0067] For TH immunostaining, 14 µm coronal sections of the striatal levels of interest

[18] were taken and stored at -20°C. Immunostaining was then performed as previously described [18,23]. Briefly, after fixation with 4% paraformaldehyde, the striatal sections were incubated with an anti-TH antibody (mouse monoclonal MAB5280, Millipore, France, 1:2500) overnight at 4°C. Subsequently, the sections were incubated with a biotinylated goat anti-mouse IgG antibody (BA-9200, Vector Laboratories, Burlingame, CA, USA; 1:500) and immunoreactivity revealed by the Avidine-peroxidase conjugate (Vectastain ABC Elite, Vector Laboratories, Burlingame, CA, USA).

[0068] The extent of striatal dopaminergic denervation was quantified using the ICS Framework computerized image analysis system (Calopix, version 2.9.2, TRIBVN, Châtillon, France) coupled to an optical microscope (Nikon, Eclipse 80i). After drawing masks of the three striatal levels of interest, the optical densities (OD) were measured for each striatal subregion (DS and Nacc). The ODs were expressed as percentages relative to the mean optical density obtained from homologous regions of the operated control animals (intracerebral injection of saline solution).

[0069] Progression of Parkinson's Disease (PD) Score in the 6-OHDA Model

[0070] In order to assess the progression of PD (PMP) in 6-OHDA rats, a scale called the "PMP score," combining 3 criteria, was developed: 1) criterion 1: assessment of neuropsychiatric symptoms (i.e., apathetic behavior), 2) criterion 2: assessment of motor symptoms (i.e., motor deficit) fine), these two criteria are also used clinically in scales for the diagnosis of PD, and 3) criterion 3: the level of dopaminergic denervation in DS, available only after postmortem histological analysis in the patient. Each criterion was quantified for each animal, with values ​​ranging from 0 to 4, as illustrated in Figure 2A. The values ​​assigned to each criterion were summed to produce the PMP score, ranging from 0 to 12.

[0071] For the evaluation of behavioral deficits in the self-administration and gait readjustment tests (criteria 1 and 2, respectively), the performance of each animal was compared before and after the injection of 6-OHDA (or saline (sham)). A value of 0 for each of these two criteria corresponded to an absence of deficit, or a deficit of less than 30%, i.e., the normal daily variability of the tests used. The following subcategories were defined as follows: 1 = mild deficit (30%–50%); 2 = moderate symptoms (50%–70%); 3 = severe symptoms (70%–90%); 4 = significant or total deficit (>90%).

[0072] Given that motor symptoms appear after 70% loss of striatal dopaminergic neurons in patients with PD, the value of dopaminergic denervation in DS (criterion 3) was used to define the boundary between the prodromal and clinical phases.

[0073] Regarding the PMP score obtained by adding the values ​​of the 3 criteria, the animal mimicking PD was classified as: 1) an asymptomatic rat with minor lesions but no behavioral symptoms (PMP score = 1 to 2), 2) a prodromal-type animal, i.e., presenting only neuropsychiatric disorders and a small DS lesion (PMP score = 3 to 7), and 3) a clinical-type animal, i.e., presenting neuropsychiatric disorders, motor symptoms, and an extensive DS lesion (PMP score = 8 to 12). The control (sham) animals were assigned a score of 0.

[0074] Given the small number of asymptomatic animals and the fact that they have no clinical relevance, they were excluded from the study. Human cohorts

[0075] Blood samples from patients with Parkinson's disease and corresponding control subjects were obtained via 1) the Parkinson's Disease Biomarker Program Consortium (PDBP), supported by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health (NIH-USA), and 2) the Santa Lucia Foundation (Rome, Italy). All subjects (patients and controls) provided their written informed consent.

[0076] The inclusion criteria for patients in the United States included a recent diagnosis (<1 year) of PD according to the new Movement Disorders Society (MDS) criteria. For the Italian cohort, the diagnostic criteria were less restrictive, and patients diagnosed longer ago than those in the United States cohort were included. Patients were included (0 to 3 years after diagnosis). In both cohorts, patients had not received any antiparkinsonian treatment at the time of inclusion, did not present with dementia or psychiatric disorders, nor any other medical condition that could compromise the study. The inclusion criteria for the controls were: absence of neurological diseases, absence of a family history of movement disorders, and absence of specific medical conditions.

[0077] The US cohort comprised 21 patients with PD not receiving antiparkinsonian treatment, 9 patients with PD receiving Pra treatment, and 30 age- and sex-matched controls. The Italian cohort comprised 21 patients with PD and 23 age- and sex-matched controls. All clinical and demographic information collected from both human cohorts is summarized in Table 1 below (data are represented as mean ± SEM, M: Male; F: Female; *2 patients of unknown sex).

[0078] [Table 1] NIH Italy Control MP Against MP Age 61.4 ±1.8 62.0 ±2.1 57.26 ± 1.89 60.33 ± 2.00 Sex 16 M / 14 F § M / 12 F 9 M / 14 F 14 M / 7 F*

[0079] NMR experiments

[0080] 1H NMR 1As LH of rat brain samples 6-OHDA

[0081] Just before the HRMAS analysis, the D2O 1Oql were added to the inserts containing the brain biopsies, which were then sealed and placed in a 4 mm zirconium rotor.

[0082] Data acquisition: All HRMAS NMR spectra were acquired, as previously described

[82] , on a Bruker Advance III spectrometer (Billerica, USA) (IRMaGe, CEA Grenoble, France) at 500 MHz. The samples were spun at 4000 Hz at the magic angle (54.7°) and the temperature was maintained at 4°C for all experiments. The one-dimensional spectra were acquired using a CPMG pulse sequence (TE=30 ms, 256 repetitions, 17 min). The residual water signal was pre-saturated during a 1.7 s relaxation period.

[0083] Data processing: Quantification was performed using jMRUI software with a quantification based on a quantum estimation procedure (QUEST)

[83] . This procedure requires the use of a metabolite database and a Complete spectral assignment. Nineteen metabolites were assigned and quantified: acetate, alanine, ascorbate, choline, creatine, gamma-aminobutyrate (GABA), glutamate, glutamine, glycine, glycerophosphocholine, glutathione, lactate, myo-inositol, α-acetylaspartate, phosphocholine, phosphocreatine, phosphoethanolamine, scyllo-inositol, and taurine. The metabolite amplitude calculated by QUEST was normalized to the total spectral signal. CRLBs were calculated for each metabolite as an estimate of the standard deviation of fit. 1H serum NMR

[0084] Serum collection: In rats, blood was collected from the caudal vein after 2 hours of fasting and under gas anesthesia with isoflurane (2%) and was stored on ice before being rapidly centrifuged at 1600 g for 15 min at 4°C. The supernatant serum was collected and stored at -80°C until the day of MRI. The time before freezing never exceeded 30 min

[84] .

[0085] For monkeys, blood was collected from the saphenous vein after a 12-hour fast, under anesthesia (atropine 0.05 mg / kg im followed by zoletil 15 mg / kg im) and subjected to the same protocol as rat blood.

[0086] For humans, blood was collected after an overnight fast, kept at room temperature for 15 to 60 min before being centrifuged at 1200 to 1500g at 4°C for 10 to 15 min. The supernatant serum was then collected and stored at -80°C.

[0087] On the day of the NMR experiments, the samples were slowly thawed on ice, then rapidly centrifuged to remove any cryoprecipitates. The NMR tubes were filled with 60 µl of serum sample and 120 µl of 0.1M phosphate saline buffer (PBS) in D2O (50% D2O, pH = 7.4), and stored at 4°C until NMR acquisition.

[0088] Data acquisition: All animal and human serum samples were subjected to the same NMR protocol. The 1H NMR experiments were performed on a Bruker Advance III 950 MHz NMR spectrometer (IBS, Grenoble, France) using a 3 mm cryoprobe. Spectra were recorded using a CPMG pulse sequence. The residual water signal was pre-saturated during a 2-second relaxation period.

[0089] Peak assignment was performed using 2-dimensional homo-mononuclear 'H-'H (TOCSY) and heteronuclear 'H-13C (HSQC) experiments, and databases. Where necessary, the addition of selected metabolites was used to clarify ambiguous identifications.

[0090] Data processing: The time signals were transformed using the Fourier method and manually phased with Bruker's Topspin software, version 3.6.2. Subsequently, further preprocessing steps (baseline correction, alignment, segmentation) were performed using the online NMRProcFlow v 1.2 software. (http: / / nmrprocflow.org). Segments of 0.001 ppm were performed over the spectral range 0-10 ppm, excluding the residual water peak, macromolecule / lipid signals, and other regions containing pollutants (δ 5.0-4.7; δ 5.25-5.38; δ 2.2-2.0; δ 1.4-1.25; δ 1.15-1; δ 0.9-0.5). Each segment was normalized to the sum total of the segments. Statistical analysis

[0091] Multivariate analysis: Data from liquid NMR or HRMAS were imported into SIMCA v.14 (Malmö, Sweden) for multivariate statistics. An unsupervised principal component analysis (PCA) was first used for the overall visualization of the distribution of all samples, followed by an OPLS to find the discriminating metabolites associated with a specific disease stage. For the latter, either a continuous variable (i.e., the PMP score) or a discrete variable (i.e., group membership) was used to label each sample, leading to OPLS-DA in the latter case. The total number of components was determined using the cross-validation procedure, which produces the R2Y and Q2 factors that indicate the goodness of fit and predictability of the model, respectively. A model is considered robust and predictive when both factors are >0.5.Scores were plotted in 2D against the first two components of the OPLS models, while weights were plotted in 1D to mimic an NMR spectrum but with positive / negative peaks indicating upregulated / downregulated metabolites, respectively. In this "statistical spectrum," the correlation of each NMR variable with the continuous ranking variable (score), ranging from 0 to 1, was color-coded from dark gray (high correlation) to light gray (low correlation). Metabolites with a correlation >0.5, considered the most discriminating, underwent univariate analysis. Cross-validated analysis of variance of predictive residuals (CV-ANOVA) was used to assess the model's significance.

[0092] Metabolites that varied significantly in at least three of the four models studied (6-OHDA, α-synuclein, MPTP, and human) were subjected to multiple logistic regression to evaluate their relevance as a potential predictive biomarker of PD using R software (version 3.6.1 - R core team). Regressions, including cross-validation, were performed for prodromal-type animals (6-OHDA and α-synuclein prodromal-type rats). Clinical-type animals were used for external validation (6-OHDA and α-synuclein clinical-type rats, and all primates). The same analyses were performed for the US patient cohort. The Italian patient cohort was used for external validation. All combinations of the selected metabolites were tested, and ROC curves were then generated for each of them to evaluate the quality of fit using the area under the curve (AUC) and the optimal threshold that maximizes sensitivity and specificity.

[0093] Univariate analysis:

[0094] All univariate analyses were performed using Graphpad Prism 8 software (San Diego, USA). All results are expressed as mean values ​​± SEM with a significance level set at 0.05.

[0095] 6-OHDA models: For self-administration tests and step readjustment tests, a univariate analysis of the effect of Pra was performed by applying a repeated measures (RM) factor ANOVA followed by a Sidak post hoc test.

[0096] For the histological and metabolomic data, a one-way ANOVA, followed by a Tukey post hoc test with correction for multiple comparisons, was performed.

[0097] α-Synuclein Models: For the longitudinal study conducted in α-synuclein animals, some values ​​were missing due to artifacts during NMR measurement. The data were therefore analyzed by fitting a proposed mixed model using a compound symmetry covariance matrix and were fitted using the restricted maximum likelihood method (REML). This model was followed by a Tukey post-hoc test.

[0098] MPTP Monkeys: Given the small number of animals, the non-parametric Mann-Whitney test was used.

[0099] Humans: The data underwent a logarithmic transformation given the non-Gaussian distribution of metabolite levels. First, t-tests were performed individually in the two different cohorts. Then, a two-way ANOVA was applied with i) cohort origin and ii) group (control or MP) as factors, followed by Sidak's post-hoc test. EXAMPLE 2: RESULTS

[0100] The 6-OHDA rat model allows the study of the different stages of Parkinson's disease

[0101] Although in humans the diagnosis of PD relies on a scale that integrates assessments of neuropsychiatric and motor symptoms

[22] , its definitive confirmation is based on post-mortem histological assessments, primarily of neuronal loss in the nigrostriatal pathway leading to denervation of the substantia nigra. In order to characterize 6-OHDA rats with respect to the clinical progression of the disease, a score, called the Parkinson's Disease Progression Score (PMP), based on the same type of criteria as those used in humans, i.e., the neuropsychiatric component—assessed by self-administration performance (motivation)—, fine motor skills—assessed by performance on the step readjustment test—, and the extent of substantia nigra lesion—assessed by post-mortem histological analysis— was assigned to them. Based on this score, the animals with MP were divided into two categories, namely prodromal type animals and clinical type animals.

[0102] Tyrosine hydroxylase immunoreactivity (TH-IR) revealed that bilateral injection of 6-OHDA into the SNc led to partial nigrostriatal dopaminergic lesion resulting in dopaminergic denervation in the DS and, to a lesser extent, in the Nacc (Figures 2B and 2C). Indeed, TH-IR quantification showed a large loss (62.2%) of dopaminergic projections in the DS of prodromal-type animals compared to controls (sham), with an even greater loss in clinical-type animals (71.96%). A slight loss of TH-IR was also observed in the Nacc of 6-OHDA-treated animals (Figure 2C). This denervation pattern preserves learning abilities and overall ambulatory activity of animals

[18] , allowing the study of motivational processes, without the potential bias related to locomotor alterations often present in animal models ([Fig.3]) [18, 23].Motivation was measured by the sucrose self-administration procedure. Before the injection of 6-OHDA (saline solution for shams), the rats were trained in the self-administration task, which allowed them to reach their maximum performance level (approximately 80 rewards per one-hour session). During the post-injury phase (i.e., after the 6-OHDA / saline solution injection), the performance of the control rats (sham, saline solution injection) remained stable, while that of the 6-OHDA animals decreased considerably [30 to 66%] ([Fig.2] D).

[0103] With regard to fine motor skills, all animals exhibited a similar number of leg readjustments prior to the injection of 6-OHDA / physiological saline, which was not altered after the injection of 6-OHDA, except in the clinical type group where it was considerably reduced ([Fig.2] E).

[0104] Serum metabolic signatures co-evolve with the progression of MP in rats 6-OHDA

[0105] Figure 4A illustrates a typical ¹H NMR spectrum of rat 6-OHDA serum, acquired at a very high magnetic field (23 T). This allowed the identification of approximately 50 metabolites in the serum samples. For each compound, the ¹H and ¹³C assignment of their chemical groups, as well as the multiplicity of the peaks, are presented as follows: singlet (s), doublet (d), doublet doublet (dd), multiplet (m) (Table 2).

[0106] [Table 2A] Metahotite Groupe ’H ppni :SC ppm Multiplicité 3-hydroxy bttfy rate ï-CHs 1.19 24.4 d haif g-cjjg hatî a'çHs ’ _2.30_ 2.39 49.2 49,2 dd dd S-CH 4.15 m Acetoacetate 3.44 s 2.27 s Acétate CHs 1.91 259 s Acetçne CH2CG 2.22 32,9 s Avorte CHs 1.4? 18.8 d a-CH 3.77 53.5 0 Aibumtn iysyl 6-ÔHï 2.88 42.0 t e-CHi 2.95 41.9 t e-GH2 3.01 42.0 t Arginine vü:-L 1.64 V-CHi 1,69 3-CH, 1.89 6-Chü 3.23 43.2 T Aspartate ha» p-CH2 2.66 q haït 0-CHs 2.80 dd Bêlai ne Choiesteroi CH2 3.9 s CH- C18 Hn HDL} 3.26 0.86 s m CI8 (In VLDL1 0.68 026 and C27 0.83 25.2 m CM i ne N (CHah 3^21 56.7 s 3.51 4.06 Citrate hait CHa ha!f CH- 2.83 2.66 d Créait ne Créatinine CHs 3.03 s ___CHs CH- __3^92___ 3.Ô4 a s Chfe 4.05 s .Dimethylamine CHs 2.71 s Ethanof CHs 1,17 21.8 t CHsCOH 3.SS d Fatty actàs (mairiy LDL} CH^CH^n 0.34 16.5 m (Ci-Mn 1.27 32.2 m Faiiy ac ids (mainty VLDL) CHsCHïCHs O 0.86 m ÇH2CH2CO CHÆHÆHÏ CD 1.57 1.29 27,4 m m Acides pr as CH3CH2 0.93 21.05 m DHÿOH^CHs) n 1.24 34.4 m CHaCH^CHg) n 1.26 25.2 m CHs 1.26 19.2 m CH-CHiCHiCwC 1.30 1.68 29.2 m . Métabolite Groupe "'gh-c^c rH ppm 2.00 :5G ppm Multiplicité Acide Gras 29.7 m CH2C0 2.22 36.3 m Ç^CCH^ CH=CHCH2C H^CH ___2.72__ 5.26 ,,,,^.1,,,,, 130.6 ,,,,,,^ m CH«CHCH2C H-OH 5-29 132.2 m Formate CH 3 45 s Fructose 3,99 m 4.01 dd Fucose1 p-Galactose 4.54 d Gtucose H2 3.24 76.3 t H4 3.40 72.4 t H4 3.41 72.4 t HS 3.46 78.6 m H3 3.48 78.5 t H2 æ.63 74.3 q 83 3.71 75.6 t hait CH^C6 hatf CH2-C6 3.72 3.76 B3.S 63.4 q m 85 3.82 74.2 ddd bail CH^Ctô 3.84 63.4 m hait CH.-CS 3.89 63.5 dd H1 4.54 98.7 d 81 5.23 94.9 d G-Mâmate haït 3-C8;. 2.04 m 2.12 m hait y-CBï 2.34 33.9 m hatf vCBa 2.36 m 3.74 m Glutamiae 2.08 2.09 haït p-CHù 2.11 29.7 m half v-CHs 2,44 33.8 m 2.46 57.4 m 3.74 Gtycerç: ha« CHa 3.56 65.6 q ha» CHa 3.65 65.6 q .Cà-H 3.87 74.6 m Cdyceropbosphüch dine 3.22 s NCHa 3.56 68.7 m OCHa 4.29 62.2 m Glyceroi backbone 4.06 [Tableau 2B] MetaMite Group 'H ppm !3C ppm Multiplicité PGLYsand TAGa CHOCOR 4.22 ! 5.20 ! ! Glycine CHs 3.55 44.3 S Histidine dd 398 1 | dd H4 7 04 j î :S H2 7.7S j S Isofeucine 5-CH; 3-CH3 0.93 j 13,9 t 1.00 17,15 d bail \-CHs 1.24 haif y-CHs 1.46 i i 1.96 3.65 j j Lactate CHs 1.32 22.7 d CH 4.11 71.2 q Lactose 3.55 i j 3.66 3.97 i 4.45 j Leucine B-CHs 0.95 i j d &-CH; 0.96 d 1.66 m 1.70 42.7 m 1.73 m a-CH 3.71 j Lysine Y-CH2 1.43 j m y-ch2 1.49 m S-CHa 1.72 m p-CHs 1.88 j m g-CHs 1.91 ! j m 3.02 ! 3.74 J j t Mannose 4.69 ! d 5.18 d Methanol CHsOH 3.35 S Méthionine 2,64 j 31,4 t Méthionine 2.15 ! j s Myo-inositof 4 05 j j 3.27 i N-acetyl-glycopratem 1 NHCOCH 2.04 24.7 s N-acetyl-glycopnoteia 2 NHCOCH 3 2.07 25 S Métabolite Group ’H ppm :3C ppm MuitiplicHâ Phetlylaianine half p-Cî-fe 3.26 a-CH 3.97 H2, HS 7.31 d H4 7.35 m H3, HS 7.40 m Proline y-CHs 1.98 m 2.01 m half p-CHs 2.05 m half p-CHs 2.34 m half ô-CH2 3.33 m a-CH 4.12 m Succinate 2.39 s Threonirte y-ch3 1.31 d a-CH 3.55 d p-CH 4.23 m Trehaioss 3.40 74.3 Tyrosine 6.38 d H2, HS 7.18 d Valine CHS 0.98 19.4 d CHS 1.03 20.6 d p-CH 2.26 m a-CH 3.60 63.4 d Urea NHsC--0NH £ 5.77 Xylose 3.41 78.6 The OPLS score graph, created using NMR data and PMP scores for each animal (i.e., the "6-OHDA serum OPLS model"), shows a clear gradation in dot colors, from "cold" (white) on the left to "warm" (dark gray) on the right. This indicates that the metabolic profile changes in tandem with disease progression (Figure 4B). Among the 14 most relevant metabolites (i.e., correlation with PMP score >0.5), ten were significantly altered in at least one MP-type group compared to control (sham) animals (Figure 4C): alanine, betaine, β-hydroxybutyrate (BHB), dimethyl sulfone (DMSO2), glycine, lactate, pyruvate, threonine, serine, and valine. These metabolites did not change identically with disease progression.First, 6 metabolites were significantly altered from the "prodromal" phase: alanine, BHB, glycine, lactate, and serine increased, while betaine decreased. Their levels... These levels then remained stable in the "clinical" phase. Next, one metabolite, pyruvate, increased progressively according to the different phases of Parkinson's disease. Finally, three metabolites were modified only in the clinical stage: DMSO2 and valine decreased, while threonine increased (Figure 4D).

[0107] Metabolic profiles of brain tissue in the 6-OHDA rat model strengthen blood results

[0108] High-resolution magic angle spinning (HRMAS) NMR spectroscopy was used to study the metabolic profiles of intact brain biopsies (DS and Nacc) from 6-OHDA animals, sampled at the same level as the sections used for histology in order to quantify dopaminergic denervation. A representative spectrum of a DS biopsy from a 6-OHDA rat is shown in Figure 5A, with the labeling of the assigned and quantified metabolites. For the DS samples, the OPLS analysis score plot (i.e., the "OPLS 6-OHDA DS model") revealed a co-evolution of brain metabolic profiles with the animals' PMP score, as in serum (Figure 5B(i)). Significant deregulations of the main metabolites are illustrated in Figure 5C(i).Compared to control animals (sham), one metabolite, taurine, was significantly altered in the prodromal type group (increase), while four metabolites were significantly altered in the clinical type group: alanine, lactate and phosphocreatine increased and glutamate decreased.

[0109] OPLS analysis constructed with Nacc data (i.e., the "OPLS 6-OHDA Nacc model") showed the same trend, although less clear (Figure 5B(ii)). Univariate statistics revealed a significant increase in alanine and lactate in the prodromal type group, whereas the clinical type group was characterized by a significant increase in phosphocholine (PC) and a decrease in choline compared to the control (sham) animals (Figure 5C(ii)).

[0110] Although some metabolites such as pyruvate and citrate are not detected in tissue samples, due to their immediate transformation upon death (post-mortem effect), these experiments show dysregulation of alanine and lactate levels common to serum and tissue samples. [YES] Pramipexole modifies serum and tissue metabolic dysregulations induced by 6-OHDA

[0112] It was investigated whether the serum metabolic profile was altered by a dopaminergic agonist commonly used in clinical practice, Pramipexole (Pra), known to improve only neuropsychiatric symptoms when used at low doses in animal models [23,24]. As expected, after 15 days of subchronic Pra administration, deficits in the self-administration task measured in 6-OHDA rats were fully or partially improved in animals of the type prodromal and clinical type, respectively. In contrast, the performance of rats treated with Veh was not improved (Figure 6A).

[0113] Regarding serum metabolic profiles, to simultaneously visualize the effect of 6-OHDA injury and treatment, metabolite levels in prodromal animals were normalized to control (sham) animals (injury effect, solid bar), and metabolite levels in Pra-treated prodromal animals were normalized to prodromal animals receiving Veh (Pra effect, hatched bar). In prodromal animals, alanine, BHB, and pyruvate levels increased after injury (Figure 4D and 6B, solid bar) and were significantly reduced by Pra (Figure 6B, hatched bar), while betaine decreased after injury but increased in Pra animals.In the clinical-type group, only the alanine level, which was elevated after injury, was slightly reduced by Pra, consistent with the modest behavioral effect observed with self-administration and the lack of effect on motor tasks, given the low dose of Pra

[24] (data not shown). The other metabolites were dysregulated similarly in both groups, regardless of the treatment received (Veh or Pra).

[0114] Unlike serum, for which several longitudinal samples could be collected from the same animal, brain samples could only be collected at the end of the experiment, when all animals had received chronic treatment with Pra or Veh. Therefore, control (sham) animals treated with Veh were used as a reference to normalize metabolite levels for prodromal-type animals treated with Veh (Lesion Effect, solid bar). Metabolite levels in Pra-treated prodromal-type animals were normalized relative to prodromal-type animals that received Veh (Pra Effect, dashed bar). As in serum, Pra administration caused a decrease in some metabolites previously elevated by 6-OHDA lesion. This was true for alanine in DS and Nacc, and for lactate in Nacc in prodromal-type animals (Fig. 6C i).In the clinical-type group, the alanine level after Pra was also reversed, just as in the prodromal-type animals, and lactate was significantly altered in the DS. In addition, a reversal of phosphocreatine and phosphocholine levels was observed in the DS and Nacc, respectively (Fig 6C ii).

[0115] Overall, significant alterations in serum and tissue metabolites were measured in the 6-OHDA rat model and associated with the different phases of the disease and, interestingly, were detected as early as the prodromal stage. Moreover, some of these disturbances were partially reversed by chronic administration of Pra, which also corrected the neuro- ropsychiatric.

[0116] In order to test the specificity of the biomarkers found in the 6-OHDA model with regard to the progression and pathophysiological mechanisms of the disease, the serum metabolomic study was extended to two other animal models of MP.

[0117] 6-OHDA, α-synuclein rats and MPTP monkeys share perturbations serum metabolics

[0118] First, using the same rat strain (Sprague Dawley) as in the 6-OHDA study, an α-synuclein rat model was selected, in which overexpression of human α-synuclein A53T was induced in the SNc using adeno-associated viral (AAV) vectors. This targeted expression leads to progressive neurodegeneration in the targeted region, allowing the different stages of PD to be tracked in the same animal

[19] , namely the prodromal-like phase around 3 weeks and the clinical-like phase around 10 weeks post-injection.

[0119] An OPLS-DA model was used to assess whether a specific metabolic signature was associated with these different phases (i.e., the "serum α-synuclein OPLS-DA model"). Sham animals (controls that received an injection of a GFP-encoding control virus) showed no difference between the three different time points 0, 3, and 10 weeks post-injection ([Fig. 7]), whereas for α-synuclein animals, the three groups corresponding to the three different samples characteristic of the different phases are clearly separated in the OPLS-DA model (Figure 8A), indicating that each phase was characterized by a different metabolic signature. In particular, a significant increase in BHB, glycine, pyruvate, and serine, and a decrease in betaine, were observed over time, as in 6-OHDA rats.Furthermore, a significant decrease in myo-inositol and an increase in acetoacetate and creatine were observed only in the α-synuclein model (Figure 8B).

[0120] These results revealed similar metabolic disturbances between the two rat models of PD, which are associated with disease progression. To investigate further, a metabolic analysis was performed in a non-human primate model of PD, e.g., the MPTP model, which mimics the clinical phase of PD and exhibits greater homology with humans

[25] .

[0121] The OPLS-DA model shown in Figure 8C (i.e., the "MPTP serum OPLS-DA model") demonstrated a clear discrimination between the control (sham) and MPTP groups. On the one hand, non-human MPTP primates exhibited a significant increase in lactate and a decrease in valine, as in 6-OHDA rats. On the other hand, they exhibited a significant increase in acetoacetate and creatine, as in the α-synuclein rodent model. Finally, a significant increase in BHB and pyruvate was observed, as well as a significant decrease in... ficative of betaine, as in the two rodent models (figure 8D).

[0122] Taken together, these results reveal common metabolic alterations associated with PD progression in three different animal models of PD, generated in two different species. Thus, to validate the clinical relevance of these markers, these preclinical results were compared with those obtained in samples of patients with PD.

[0123] Serum metabolic signatures of newly diagnosed parkinsonian patients Diagnosed and animal models of PD show similarities

[0124] Each cohort was first analyzed individually, i.e. newly diagnosed parkinsonian patients (n = 21 in each cohort) against age- and sex-matched controls (n = 30 / 23 NIH / Italy).

[0125] With the exception of certain metabolites, e.g., alanine (data not shown), the observed variations were consistent between the two cohorts, as shown in Table 3. Table 3 presents the main individual and collective changes in metabolites between patients with PD and controls in the Italian and US (NIH) cohorts, analyzed individually by t-test and collectively by two-way ANOVA, followed by a Sidak post-hoc test. *: p < 0.05; **: p < 0.01; ***: p < 0.001; ns: not significant.

[0126] [Tables3] iiaiie NiH Anova with 2 faders p va: lie Chanaemsnt p Origin of cofiortss iniéraction Aeàtoaûèiate 17.26 0.05 28.12 0.12 ns Béfasne ■9.74 0.35 -28.27 0.01 HS ** ns BHB 28.28 0.24 103.23 0.09 w ■+ ns S. 62 0.37 0.14 •"* 0.08 ns 11.04 0.13 + 4.61 0.04 » ns

[0127]

[0128] For example, a marginal increase in BHB and acetoacetate was observed in Italian patients with PD (+28.3%, p = 0.24 and +17.2%, p = 0.05 respectively), and was even higher in US patients (+103.2%, p = 0.09 and +28.1%, p = 0.12).

[0129] To increase statistical power, the two cohorts were pooled (MP n = 42, control n = 53). However, since the variance of the data was mainly affected by the origin of the sample, i.e., Italy or the United States, regardless of the case (MP or control), a two-way ANOVA was performed with i) the origin of the cohort and ii) the case (control or MP) as factors. Extract information related to PD only. Table 3 illustrates the metabolites significant for the case factor, without interaction between cohort origin and case. The increase in BHB and acetoacetate, observed in each cohort, appeared significant when pooled. A significant decrease in betaine in the serum of PD patients was also detected, as in all animal models. In addition, a significant decrease in valine was observed, as in the 6-OHDA rat model. Finally, a decrease in leucine was noted. Furthermore, analysis of a small number of serum samples from other newly diagnosed Parkinson's patients treated with Pra 0.5–100 mg / day (n = 9) revealed a trend toward normalization of BHB and betaine levels compared to untreated PD patients, as observed in 6-OHDA rats ([Fig. 9]).

[0130] Serum metabolites as potential markers for the early diagnosis of Parkinson's disease

[0131] To evaluate the potential of the identified metabolites to distinguish prodromal-type animals from control (sham) animals, multiple logistic regression models were developed, based on all possible combinations of acetoacetate, betaine, creatine, BHB, pyruvate and valine levels, grouping prodromal-type 6-OHDA and α-synuclein rats.

[0132] The best regression model, evaluated by ROC curves (see Materials and Methods), retained all six metabolites, using a regression based on the following predictive algorithm (p: 1.107x10 15):

[0133] Logit(P) = log(P / (lP)) = -54.72 + 0.49BHB + 0.39Pyruvate - 0.42Valine + 0.06Acetoacetate + 0.52Creatine - 0.10Betaine

[0134] The ROC curve had an area under the curve (AUC) of 0.936. Applying the optimal threshold of 0.618 (sensitivity 0.85, specificity 0.88) (Figure 10A), an accuracy of 0.925 was found, correctly predicting 37 of the 40 animals included in the validation cohort (clinical-type rats and all primates).

[0135] Regarding the human samples, the US cohort was used to build the model because i) it is the largest cohort and ii) it includes only recently diagnosed patients, unlike the Italian cohort, which includes patients with a longer duration of Parkinson's disease. The latter was used as an external validation cohort. Again, the best model retained the six combined metabolites using the following regression algorithm (p: 1.95 x 10⁴):

[0136] Logit(P) = log(P / (lP)) = -45.60 + 0.28BHB + 4.54Pyruvate - 1 l.23Valine + 7.05Acetoacetate + 2.85Creatine - 8.27Betaine.

[0137] The following ROC curve had an AUC of 0.88. The optimal threshold was 0.316, and the The corresponding sensitivity and specificity were 0.957 and 0.714, respectively (Figure 10B). The Italian cohort classification prediction detected 82.6% of patients with PD, with 17.4% false negatives.

[0138] Finally, a multiple logistic regression was performed using serum data from all MP animals (prodromal and clinical type) from all models using the same six metabolites and was evaluated with the following ROC curve showing an AUC of 0.954. ([Fig. 11]). List of References

[0139] 1. Samii A, Nutt JG, Ransom BR. Parkinson's disease. Lancet. 2004;363:1783-1793. doi:10.1016 / S0140-6736(04)16305-8 2. Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson’s disease. Mov Disord. 2015. doi:10.1002 / mds.26424 3. Hughes AJ, Ben-Shlomo Y, Daniel SE, Lees AJ. What features improve the accuracy of clinical diagnosis in Parkinson’s disease: A clinicopathologic study. 4. Stocchi F. Therapy for Parkinson ’ s Disease : What is in the Pipeline ? Neuro-therapeutics. 2013;11:24-33. doi:10.1007 / sl3311-013-0242-l 5. Adler CH, Beach TG, Hentz JG, et al. Low clinical diagnostic accuracy of early vs advanced Parkinson disease: Clinicopathologic study. Neurology. 2014. doi: 10.1212 / WNL.0000000000000641 6. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: A clinico-pathological study of 100 cases. J Neurol Neurosurg Psychiatry. 1992. doi:10.1136 / jnnp.55.3.181 7. Hughes AJ, Ben-Shlomo Y, Daniel SE, Lees AJ. What features improve the accuracy of clinical diagnosis in parkinson’s disease: A clinicopathologic study. Neurology. 2001. doi:10.1212 / wnl.42.6.1142 8. Salamon A, Zâdori D, Szpisjak L, Klivényi P, Vécsei L. Neuroprotection in Parkinson’s disease: facts and hopes. J Neural Transm. 2020. doi: 10.1007 / s00702-019-02115-8 9. Olanow CW, Kieburtz K, Schapira AH V. Why hâve we failed to achieve neuroprotection in Parkinson’s disease? Ann Neurol. 2008;64(SUPPL. 2). doi:10.1002 / ana.21461 10. Beal MF, Rascol, Marek, et al. Bioenergetic approaches for neuroprotection in Parkinson’s disease. Ann Neurol. 2003. doi: 10.1002 / ana. 10479 11. Jankovic J. Parkinson’s disease: Clinical features and diagnosis. J Neurol Neurosurg Psychiatry. 2008. doi:10.1136 / jnnp.2007.131045 12. Michell AW, Lewis SJG, Foltynie T, Barker RA. Biomarkers and Parkinson’s disease. Brain. 2004. doi:10.1093 / brain / awhl98 13. Chaudhuri KR, Healy DG, Schapira a H. Non-motor symptoms of Parkinson’s disease: diagnosis and management. Lancet Neurol. 2006;5(3):235-245. doi: 10.1016 / s 1474-4422(06)70373-8 14. Kaur R, Mehan S, Singh S. Understanding multifactorial architecture of Parkinson’s disease: pathophysiology to management. Neurol Sci. 2019. doi:10.1007 / sl0072-018-3585-x 15. Johnson CH, Gonzalez FJ. Challenges and opportunities of metabolomics. J Cell Physiol. 2012. doi:10.1002 / jcp.24002 16. Ala-Korpela M. Potential rôle of body fluid 1H NMR metabonomics as a prognostic and diagnostic tool. Expert Rev Mol Diagn. 2007;7(6):761-773. doi: 10.1586 / 14737159.7.6.761 17. Hatano T, Saiki S, Okuzumi A, Mohney RP, Hattori N. Identification of novel biomarkers for Parkinson’s disease by metabolomic technologies. J Neurol Neurosurg Psychiatry. 2015:1-7. doi:10.1136 / jnnp-2014-309676 18. Bogdanov M, Matson WR, Wang L, et al. Metabolomic profiling to develop blood biomarkers for Parkinson’s disease. Brain. 2008; 131(2):389-396. doi: 10.1093 / brain / awm304 19. Lei S, Powers R. NMR Metabolomics Analysis of Parkinson’s Disease. Curr Metabolomics. 2013;l(3):191-209. doi:10.2174 / 2213235X113019990004 20. Quinones MP, Kaddurah-Daouk R. Metabolomics tools for identifying biomarkers for neuropsychiatrie diseases. Neurobiol Dis. 2009;35(2): 165-176. doi: 10.1016 / j.nbd.2009.02.019 21. Nagesh Babu G, Gupta M, Paliwal VK, Singh S, Chatterji T, Roy R. Sérum metabolomics study in a group of Parkinson’s disease patients from northem India. Clin ChimActa. 2018;480(January):214-219. doi: 10.1016 / j.cca.2018.02.022 22. Lewitt PA, Li J, Lu M, Guo L, Auinger P. Metabolomic biomarkers as strong correlates of Parkinson disease progression. Neurology. 2017;88(9):862-869. doi: 10.1212 / WNL.0000000000003663 23. Vingill S, Connor-Robson N, Wade-Martins R. Are rodent models of Parkinson’s disease behaving as they should? Behav Brain Res. 2018. doi: 10.1016 / j.bbr.2017.10.021 24. Drui G, Carnicella S, Carcenac C, et al. Loss of dopaminergic nigrostriatal neurons accounts for the motivational and affective déficits in Parkinson’s disease. Mol Psychiatry. 2014;19(3):358-367. doi:10.1038 / mp.2013.3 25. Bourdenx M, Dovero S, Engeln M, et al. Lack of additive rôle of ageing in nigrostriatal neurodegeneration triggered by a-synuclein overexpression. Acta Neuropathol Commun. 2015. doi:10.1186 / s40478-015-0222-2 26. Seo J, Lee Y, Kim BS, et al. A non-human primate model for stable chronic Parkinson’s disease induced by MPTP administration based on individual behavioral quantification. J Neurosci Methods. 2019. doi: 10.1016 / j.jneumeth.2018.10.037 27. Wong ML, Licinio J. Peripheral is Central to the question. Mol Psychiatry. 2005. doi: 10.1038 / sj.mp.4001674 28. Magnard R, Vachez Y, Carcenac C, et al. What can rodent models tell us about apathy and associated neuropsychiatrie symptoms in Parkinson’s disease? Transi Psychiatry. 2016;6(3):e753. doi:10.1038 / tp.2016.17 29. Decressac M, Mattsson B, Lundblad M, Weikop P, Bjôrklund A. Progressive neurodegenerative and behavioural changes induced by AAV-mediated ove-rexpression of a-synuclein in midbrain dopamine neurons. Neurobiol Dis. 2012. doi: 10.1016 / j.nbd.2011.12.013 30. Dehay B, Femagut PO. Alpha-synuclein-based models of Parkinson’s disease. Rev Neurol (Paris). 2016. doi: 10.1016 / j.neurol.2016.04.003 31. Teahan O, Gamble S, Holmes E, et al. Impact of analytical bias in me-tabonomic studies of human blood sérum and plasma. Anal Chem. 2006. doi:10.1021 / ac051972y 32. Lôpez-Bascôn MA, Priego-Capote F, Peralbo-Molina A, Calderôn-Santiago M, Luque De Castro MD. Influence of the collection tube on metabolomic changes in serum and plasma. Talanta. 2016. doi: 10.1016 / j.talanta.2015.12.079 33. Pagano G, Niccolini F, Politis M. Imaging in Parkinson's disease. Clin Med JR Coll Physicians London. 2016. doi:10.7861 / clinmedicine.l6-4-371

Claims

Claims

1. An in vitro method for diagnosing Parkinson's disease at the de novo or prodromal stage from a biological sample of a patient comprising: a) measuring in vitro in said biological sample the expression levels of the following at least 6 metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine; b) comparing the expression levels of the metabolites measured in step a) with those of a reference biological sample or with reference values; wherein a difference in the expression levels of the metabolites measured in step a) with those of the reference biological sample or with reference values ​​is an indication that said patient has Parkinson's disease.

2. The method of claim 1, wherein step a) further comprises measuring in vitro the expression level of at least one metabolite selected from the group consisting of alanine, lactate, DMS02, glycine, serine, threonine, myo-inositol, and leucine.

3. The method of claim 1, wherein step a) comprises measuring the expression levels of the following metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine.

4. An in vitro method for the diagnosis of Parkinson's disease at the de novo or prodromal stage from a biological sample of a patient comprising: a) measuring in vitro in said biological sample the expression levels of the following metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine; b) the calculation of the logistic regression Logit(P) according to the following formula: -45.60 + 0.28 Ebhb + 4.54 EPyruvate — 11.23 EVaiine + 7.05 EAcetoacetate + 2.85 ECreatine ~ 8.27 EBetaine or Ebhb, EPyruvate, Ey^iæ, EAcetoacetate, Ecreatine and EB^taine correspond respectively to the measured expression levels of B-hydroxybutyrate (BHB), pyruvate, valine, acetoacetate, creatine and betaine, in the biological sample; and where a calculated value of Logit(P) greater than 0.316 au is the indication that the said patient has Parkinson's disease.

5. A method according to any one of claims 1 to 4, wherein the biological sample is a serum sample.

6. In vitro use of measuring the expression level of at least six of the following metabolites: acetoacetate, betaine, B-hydroxybutyrate (BHB), creatine, pyruvate, and valine, as a diagnostic biomarker for Parkinson's disease in the de novo or prodromal stage.

7. Use according to claim 6, further comprising measuring the expression level of at least one metabolite selected from the group consisting of alanine, lactate, DMS02, glycine, serine, threonine, myo-inositol, and leucine.

Images