Parkinson's disease (PD) is the second most common neurodegenerative disease and its prevalence is expected to double over the next 30 years, making it a leading cause of neurological disability [GBD 2016 Neurology Collaborators, 2019; Dorsey et al, 2018]. PD is characterized by motor symptoms, such as muscle stiffness, tremor, slowness of movement (bradykinesia) and postural instability, and non-motor symptoms, such as sphincter disorders, postural hypotension, cognitive disorders, depression, hyposmia, constipation and REM sleep behavioral disturbance. Unfortunately, the mechanisms leading to neuronal dysfunction and death in PD remain poorly known and there are currently no therapies capable of modifying their course [Bloem et al, 2021].

In this study we aim at defining a new set of biomarkers based on the combination between PET, blood metabolomics and natural language extracted from the keywords of electronic health records.

Much evidence suggests the existence of a preclinical stage of disease that begins many years before PD is diagnosed, when an individual appears normal (there are no symptoms or signs), but is already developing typical neuropathological changes. The prodromal phase follows, in which symptoms and signs are present, but are still insufficient to define the disease. In the clinical phase, cardinal motor symptoms become sufficiently evident to diagnose PD [Schaeffer et al, 2020; Toulouse et al, 2021]. For this reason, it is important to have reliable biomarkers that can help in early diagnosis, especially considering that disease-modifying therapies have a greater chance of success if they are started early, before a considerable number of dopaminergic neurons have undergone. death. Furthermore, there is also a need to better define PD subtypes that not only have different clinical presentation and prognosis, but also differ in the underlying pathogenetic mechanisms, requiring personalized therapeutic approaches [Tolosa et al, 2021]. Such biomarkers should be sensitive, specific, non-invasive, inexpensive, easily detectable and measurable [Du et al, 2021].

In recent years, numerous biomarkers of risk and / or prodromal phase have been identified and combined in search criteria for prodromal PD, with the aim of calculating the probability with which a patient is in the prodromal phase of PD. Apart from specific genetic risk markers, including above all GBA and LRRK2 mutations, REM sleep behavioral disturbance and PET / SPECT abnormalities are currently considered the most important prodromal biomarkers, capable of predicting PD with a high probability. [Berg et al, 2015; Heinzel et al, 2019]. However, new biomarkers are needed for a better understanding of the prodromal phase and its potential clinical-pathological subtypes and for a more precise calculation of the probability of MP [Bloem et al, 2021; Schaeffer et al, 2020].

Goals: The main objective of the present study is to identify new, more reliable biomarkers of PD and to develop a new, more accurate predictive model of disease.

The design is that of a longitudinal observational study. Participants will be divided into 6 groups (each with at least 20 subjects) based on clinical characteristics: 1) patients with clinically defined PD; 2) patients with clinically probable PD; 3) patients with neurodegenerative parkinsonism, such as multiple system atrophy (MSA), progressive supranuclear palsy (PSP), Lewy body disease (DLB) or cortico-basal degeneration (CBD); 4) patients with secondary parkinsonism (vascular, iatrogenic, psychogenic, etc.); 5) patients with "probable" or "possible" prodromal PD; 6) healthy subjects with risk or prodromal factors for PD; 7) healthy subjects of the same age and sex without any risk or prodromal factor for PD.

All participants will undergo a careful medical history, general and neurological physical examination, neuropsychological tests, structural brain imaging (MRI or CT) and PET with F-DOPA (or SPECT with DATSCAN), venous blood sampling for routine blood chemistry ( including blood count, erythrocyte sedimentation rate, urea and electrolytes, thyroid function, vitamin B12 and folic acid), metabolomic analysis, including lipidomics, genetic analysis, mononuclear cell separation, and the search for known biomarkers of PD. The severity and progression of the disease will be assessed through the use of specific scores, including the Unified Parkinson's Disease Rating Scale (UPDRS) and the Hoehn and Yahr scale.

Metabolomic analysis, including lipidomics, will be performed through liquid chromatography coupled with tandem mass spectrometry and genetic analysis using next generation sequencing (NGS). Structural imaging and PET with F-DOPA (or SPECT with DATSCAN) of each subject will be subjected to texture analysis, using dedicated software, in search of new, more reliable neuroradiological markers. In the texture analysis, 103 structural parameters will be extracted from the 89 regions of the Hammers brain, for a total of 8779 parameters. For each brain region the parameters will be reduced by Principal Component Analysis (PCA), selecting only the principal components that express 99.5% of the total variance. The ultimate goal will be to quantify, in each individual subject, the risk of developing PD using a convolutional neural network (CNN) with inputs consisting of the various clinical parameters evaluated and the main components selected from metabolomics and PET data.

All participants will be clinically re-examined on a three-monthly basis.

The analysis will be carried out together with the partner "NIM Competence Center for Digital Healthcare GmbH" (NIM), holder of the "GATEKEEPER 1st open call" research grant. The data, appropriately anonymized, will be included in the European Gatekeeper digital medicine platform by the NIM partner.

Patients with clinically defined or probable PD will need to meet the criteria for the diagnosis of PD (clinically defined or probable) prepared by the Movement Disorder Society (MDS) in 2015 [Postuma et al, 2015]. Patients with neurodegenerative parkinsonism will have to meet the published diagnostic criteria for MSA [Gilman et al, 2008], PSP [Höglinger et al, 2017], DLB [McKeith et al, 2017] and CBD [Armstrong et al, 2013]. Patients with prodromal PD will have to meet the diagnostic criteria proposed by the MDS for the diagnosis of "probable" prodromal PD if the risk percentage of developing PD> 80%, or "possible" if the risk percentage of developing PD> 50 % [Berg et al, 2015; Heinzel et al, 2019]. The risk or prodromal factors proposed by the MDS to calculate the probability of prodromal PD include: male sex, exposure to pesticides, exposure to solvents, caffeine intake, smoking habit, familiarity (first degree relative with PD), hyperechogenicity of the substantia nigra on transcranial ultrasound, possible subthreshold parkinsonism (as assessed by UPDRS-III), REM sleep behavioral disorder (RBD), olfactory impairment, constipation, excessive daytime sleepiness, orthostatic hypotension, urinary dysfunction, severe erectile dysfunction, depression / anxiety, activity of the dopamine transporter, RBD confirmed by polysomnography, genetic tests, type II diabetes mellitus, physical activity, serum urate levels, cognitive functions assessed with the Mini Mental State Examination [Berg et al, 2015; Heinzel et al, 2019]. Healthy subjects with risk or prodromal factors for PD should have a probability of developing PD <50% according to the criteria of the MDS [Berg et al, 2015; Heinzel et al, 2019].

In addition to the selected patients, a study will also be conducted on patient data in the reference database of the proposing institution.

- Exclusion: Patients who do not consent to study participation

Participants will be divided into 6 groups (each with at least 10 subjects) based on clinical characteristics: 1) patients with clinically defined PD; 2) patients with clinically probable PD; 3) patients with neurodegenerative parkinsonism, such as multiple system atrophy (MSA), progressive supranuclear palsy (PSP), Lewy body disease (DLB) or cortico-basal degeneration (CBD); 4) patients with secondary parkinsonism (vascular, iatrogenic, psychogenic, etc.); 5) patients with "probable" or "possible" prodromal PD; 6) healthy subjects with risk or prodromal factors for PD; 7) healthy subjects of equal age and sex without any risk or prodromal factor for PD.