Periodic Reporting for period 1 - FRANKIE (Impact of exposure to chemicals to the onset and progression of Lewy Body Dementia)
Période du rapport: 2023-07-01 au 2025-06-30
Neurodegenerative diseases, such as Lewy Body Dementia (LBD) and Parkinson’s disease (PD) represent a major and growing public health challenge in Europe. These diseases severely affect quality of life and place significant economic and emotional burdens on individuals, carers, and healthcare systems.
LBD and PD are related to misfolding of alpha-synuclein in the brain. While genetic factors have been extensively characterised, the contribution of environmental chemical exposure remains underexplored.
The FRANKIE project addresses this gap by linking exposure to chemicals with the molecular markers of onset and progression of synucleinopathies. It aims to identify molecular mechanisms and blood-based biomarkers for earlier diagnosis and intervention. Using a unique longitudinal cohort of 256 patients, including 140 individuals with prodromal idiopathic rapid-eye movement sleep behaviour disorder (iRBD) sampled in 3 time points and 114 patients diagnosed with PD, FRANKIE integrates non-targeted high-resolution mass spectrometry with advanced statistical modelling and network-based metabolomics to:
1. Characterize the chemical exposure using targeted and untargeted mass spectrometry analysis.
2. Identify the chemicals and/or their mixture associated with the increased degree of cognitive impairment.
3. Associate individual chemicals or their mixture with effect biomarkers.
4. Identify prodromal and clinical biomarkers.
This interdisciplinary approach, combining analytical chemistry, exposomics, metabolomics, clinical neurology, and biostatistics, advances beyond the state-of-the-art by identifying candidate biomarkers for PD risk and earlier detection, and by linking real-world chemical exposure profiles to molecular signatures of neurodegeneration. It applies advanced LC-HRMS technology, machine learning and network-based metabolomics.
The results of the FRANKIE project could yield clinically useful blood-based biomarkers to enable early detection of PD, facilitating earlier intervention and medical care. Furthermore, linkage to chemical exposure could provide a foundation for exposure reduction strategies, aiming to reduce long-term disease incidence.
LBD and PD are related to misfolding of alpha-synuclein in the brain. While genetic factors have been extensively characterised, the contribution of environmental chemical exposure remains underexplored.
The FRANKIE project addresses this gap by linking exposure to chemicals with the molecular markers of onset and progression of synucleinopathies. It aims to identify molecular mechanisms and blood-based biomarkers for earlier diagnosis and intervention. Using a unique longitudinal cohort of 256 patients, including 140 individuals with prodromal idiopathic rapid-eye movement sleep behaviour disorder (iRBD) sampled in 3 time points and 114 patients diagnosed with PD, FRANKIE integrates non-targeted high-resolution mass spectrometry with advanced statistical modelling and network-based metabolomics to:
1. Characterize the chemical exposure using targeted and untargeted mass spectrometry analysis.
2. Identify the chemicals and/or their mixture associated with the increased degree of cognitive impairment.
3. Associate individual chemicals or their mixture with effect biomarkers.
4. Identify prodromal and clinical biomarkers.
This interdisciplinary approach, combining analytical chemistry, exposomics, metabolomics, clinical neurology, and biostatistics, advances beyond the state-of-the-art by identifying candidate biomarkers for PD risk and earlier detection, and by linking real-world chemical exposure profiles to molecular signatures of neurodegeneration. It applies advanced LC-HRMS technology, machine learning and network-based metabolomics.
The results of the FRANKIE project could yield clinically useful blood-based biomarkers to enable early detection of PD, facilitating earlier intervention and medical care. Furthermore, linkage to chemical exposure could provide a foundation for exposure reduction strategies, aiming to reduce long-term disease incidence.
To derive knowledge on PD onset and progression and linking chemical exposure to perturbed molecular pathways, the FRANKIE project included 7 work packages and multiple interdisciplinary activities. Work performed so far and main results per work package:
- WP 1 – Planning and management
WP1 covered finances, budgeting, and task scheduling. A career development plan was prepared in collaboration with supervisors, addressing training needs, transferable skills, teaching, publication planning, and conference participation.
- WP 2 – Training
Technical training included operation of LC-HRMS instruments, hands-on QA/QC procedures, data analysis, health and safety training, ethics and data protection training and science-to-policy training. Complementary and transferrable skills training included grant writing training, project management training and communication & dissemination training.
- WP 3 – Data acquisition and processing
Analytical method used for data acquisition was firstly thoroughly characterized by analysing 602 chemically diverse standards consisting of native metabolites and chemical exposure agents. Analytical separation was run using two alternative column chemistries, using silica hydride in aqueous normal phase and pentabromobenzyl column in reverse phase. Based on retention properties of standards on both columns, retention mechanisms were estimated and retention prediction models built using support vector regression. Mass spectral data was used to construct spectral libraries, which are used in metabolite annotation.
Plasma samples were analysed using this method and strict quality control protocols. Acquired data was processing using open-source software MzMine 4.
- WP 4 – Data analysis
Data was checked for compliance for strict quality control parameters. Statistical models and machine learning models were constructed and robustly cross validated and candidates for biomarkers proposed. Exposures were annotated using suspect screening approach and associated with neurocognitive outcome.
- WP 5 – Biomarker identification and ssPA
Candidates for biomarkers were annotated against in-house and publicly available mass spectral libraries and experimental or predicted retention times. Structural and correlation networks were constructed and used for annotation propagation. Pathway analysis on differential features was completed using Mummichog algorithm.
- WP 6 – Data interpretation
Biomarker candidates were associated with clinical variables. Features sharing same biological pathway were put in biological context and biological relevance crosschecked with other published scientific results.
-WP 7 – Dissemination, communication and exploitation
Dissemination of the project progression, insights and achievements to the scientific public achieved by publishing scientific paper, preparation of the second paper and two oral presentations on conferences. Communication to general public achieved by presenting on seminars, social media and websites.
- WP 1 – Planning and management
WP1 covered finances, budgeting, and task scheduling. A career development plan was prepared in collaboration with supervisors, addressing training needs, transferable skills, teaching, publication planning, and conference participation.
- WP 2 – Training
Technical training included operation of LC-HRMS instruments, hands-on QA/QC procedures, data analysis, health and safety training, ethics and data protection training and science-to-policy training. Complementary and transferrable skills training included grant writing training, project management training and communication & dissemination training.
- WP 3 – Data acquisition and processing
Analytical method used for data acquisition was firstly thoroughly characterized by analysing 602 chemically diverse standards consisting of native metabolites and chemical exposure agents. Analytical separation was run using two alternative column chemistries, using silica hydride in aqueous normal phase and pentabromobenzyl column in reverse phase. Based on retention properties of standards on both columns, retention mechanisms were estimated and retention prediction models built using support vector regression. Mass spectral data was used to construct spectral libraries, which are used in metabolite annotation.
Plasma samples were analysed using this method and strict quality control protocols. Acquired data was processing using open-source software MzMine 4.
- WP 4 – Data analysis
Data was checked for compliance for strict quality control parameters. Statistical models and machine learning models were constructed and robustly cross validated and candidates for biomarkers proposed. Exposures were annotated using suspect screening approach and associated with neurocognitive outcome.
- WP 5 – Biomarker identification and ssPA
Candidates for biomarkers were annotated against in-house and publicly available mass spectral libraries and experimental or predicted retention times. Structural and correlation networks were constructed and used for annotation propagation. Pathway analysis on differential features was completed using Mummichog algorithm.
- WP 6 – Data interpretation
Biomarker candidates were associated with clinical variables. Features sharing same biological pathway were put in biological context and biological relevance crosschecked with other published scientific results.
-WP 7 – Dissemination, communication and exploitation
Dissemination of the project progression, insights and achievements to the scientific public achieved by publishing scientific paper, preparation of the second paper and two oral presentations on conferences. Communication to general public achieved by presenting on seminars, social media and websites.
Results beyond the state of the art of FRANKIE project are:
- A robust, reproducible, and well characterized analytical method for large scale metabolomic and chemical exposomic profiling of human plasma. All methodological resources are available to enable replication, reuse and harmonization of small molecule profiling with other laboratories. This advances the state of the art, as current metabolomic and exposomic workflows are often poorly characterized, lack detailed quality control, and are rarely published in full reproducible form. The developed workflow therefore sets a new benchmark for transparency, cross-study comparability, and open-science alignment, addressing key reproducibility gaps in molecular phenotyping.
- The project identified metabolic candidates associated with PD risk and prodromal stages, with the most promising biomarkers belonging to sphingolipid and bilirubin metabolic pathways. While previous studies have implicated isolated metabolites within these classes, this work extends the evidence by identifying larger number of sphingoid and bilirubin analogues, revealing a broader network of dysregulated pathways. These findings advance the current understanding of metabolic signatures of PD and open new directions for mechanistic and longitudinal validation, as well as indicate potential new drug targets.
To ensure further uptake and impact, the following actions are foreseen:
- replication and validation of the results by an independent cohort
- integration with proteomic, genomic, and exposomic dataset to build a systems-level model of PD pathophysiology
- engagement with biomarker consortia to crosscheck and align candidate metabolites
- dissemination of protocols and resources to openly-available data repositories to facilitate interoperability and standardisation
- A robust, reproducible, and well characterized analytical method for large scale metabolomic and chemical exposomic profiling of human plasma. All methodological resources are available to enable replication, reuse and harmonization of small molecule profiling with other laboratories. This advances the state of the art, as current metabolomic and exposomic workflows are often poorly characterized, lack detailed quality control, and are rarely published in full reproducible form. The developed workflow therefore sets a new benchmark for transparency, cross-study comparability, and open-science alignment, addressing key reproducibility gaps in molecular phenotyping.
- The project identified metabolic candidates associated with PD risk and prodromal stages, with the most promising biomarkers belonging to sphingolipid and bilirubin metabolic pathways. While previous studies have implicated isolated metabolites within these classes, this work extends the evidence by identifying larger number of sphingoid and bilirubin analogues, revealing a broader network of dysregulated pathways. These findings advance the current understanding of metabolic signatures of PD and open new directions for mechanistic and longitudinal validation, as well as indicate potential new drug targets.
To ensure further uptake and impact, the following actions are foreseen:
- replication and validation of the results by an independent cohort
- integration with proteomic, genomic, and exposomic dataset to build a systems-level model of PD pathophysiology
- engagement with biomarker consortia to crosscheck and align candidate metabolites
- dissemination of protocols and resources to openly-available data repositories to facilitate interoperability and standardisation