Abstract

Introduction

Parkinson’s disease (PD) is the fastest growing neurodegenerative disorder, affecting more than 8.5 million people (Dorsey and Bloem, 2018). The main pathological hallmarks of PD include loss of dopaminergic neurons in the substantia nigra and the presence of intraneural Lewy bodies, with motor and non-motor symptoms (Bloem et al., 2021). The etiology of sporadic PD is complex and influenced by both environmental and genetic factors. Familial monogenic forms defined by rare and pathogenic variants in autosomal dominant (e.g., SNCA, LRRK2, VPS35) or recessive (PRKN, PINK1, PARK7) PD-related genes, account for less than 10% of Mendelian cases (Lesage and Brice, 2009; Karimi-Moghadam et al., 2018). The contribution of genetics in the remaining patients with sporadic forms of PD is not yet well defined. Common variants have also been described as a risk factor for PD (Nalls et al., 2019). The presence of heterozygous variants in the GBA1 gene has emerged as a common risk factor for PD, estimated to occur in about 4–12% of PD patients (Pachchek et al., 2023). The major haplotype (H1) of the microtubule-associated protein Tau (MAPT) gene has been also associated with increased risk of PD (Skipper et al., 2004; Zabetian et al., 2007). Additionally, a specific MAPT H1 sub-haplotype (H1c) has been strongly linked with progressive supranuclear palsy (PSP; Myers et al., 2005). Disease susceptibility may be influenced by a combined effect of more than 90 common low-risk genetic loci defined by large genome-wide association studies (Nalls et al., 2019; Blauwendraat et al., 2020), including those in the SNCA and LRRK2 genes. Although less explored than common single-nucleotide variants (SNVs), copy number variants (CNVs) have been reported, especially in PD-associated genes where pathogenic deletions and duplications have been identified using either a gene candidate approach (PRKN, SNCA, PINK1, PARK7 and ATP13A2) (Toft and Ross, 2010; Pankratz et al., 2011; La Cognata et al., 2017) or genome-wide burden analysis (Liu et al., 2013; Sarihan et al., 2021).

Despite ongoing global scientific efforts in genetic analysis, improvements are still needed in terms of early diagnosis and prognosis, causative treatments, and new therapeutic approaches. As the population ages, the number of PD patients will increase dramatically. It is therefore important to generate reliable evidence on the epidemiology and genetic etiology of PD to enable precision medicine and prevention for neurodegeneration in PD. In particular, three genetic discoveries that have led to new therapeutic approaches (targeting alpha-synuclein, glucocerebrosidase and LRRK2 pathway) are now in clinical development (Sardi et al., 2018).

We had previously performed a comprehensive screening of GBA1 gene variants (Pachchek et al., 2023). Here, we sought to genetically characterize patients with PD or atypical parkinsonism (AP) in the Luxembourg Parkinson’s Study screening for rare SNVs, CNVs, and estimated the effect of common SNVs using polygenic risk scores (PRS).

Materials and methods

Results

Cohort description

After the quality control procedure, the final dataset for the Luxembourg Parkinson’s study comprised 1,490 individuals (667 PD cases, 99 atypical PD cases and 724 healthy controls). Detailed demographic data are summarized in Table 1. The control group had a mean age at assessment of 65.8±11.6years. The PD patients had a mean age of onset (AAO) of 62.3±11.8years. To illustrate the ethnic composition of our cohort, we performed PCA using 1,000 Genome populations as a reference (The 1000 Genomes Project Consortium, 2012) and showed that all our samples clustered strongly with the European ancestry (Supplementary Figure 1).

Table 1

Demographic data of cases and healthy controls from the Luxembourg Parkinson’s study.

	N	Sex (% male)	Age at assessment (mean±SD)	Age at onset (mean±SD)	Family history of PD (n 1st degree relative, %)
Controls	724	54.2	65.8±11.6		132 (18.2%)
PD	667	68.0	73.0±11.0	62.3±11.8	97 (14.5%)
PSP	50	62.0	76.4±6.7	67.6±7.4	5 (10%)
LBD	25	68.0	77.8±9.8	70.5±9.4	3 (12%)
MSA	13	69.2	75.2±7.7	65.8±8.4	1 (7.7%)
CBS	10	30.0	77.1±7.8	69.2±8.6	1 (10%)
FTD-P	1	0	69	58	0 (0%)

Open in a separate window

Counts, means and percentage were calculated after genotyping data quality controls. PD, Parkinson’s disease; PSP, progressive supranuclear palsy; LBD, Lewy body dementia, MSA, Multiple System Atrophy; CBS, corticobasal syndrome; FTD-P, Fronto-temporal dementia with parkinsonism.

Polygenic risk scores

Using significant common SNVs from the largest PD GWAS summary statistics (Nalls et al., 2019), we calculated the PRS in the Luxembourg Parkinson’s study for 724 controls and 667 PD patients. The PRS model was calculated based on 75 clumped SNPs that showed the best prediction at the GWAS value of p threshold of 5e-08 and an observed phenotypic variance R2 of 5.3% (1.9% after adjustment for PD prevalence of 5e-03, empirical value of p=9.9e-05) with an AUC of 62.8%. We found a significant association of PRS with higher PD risk (OR=1.70[1.50–1.93], p=5.9e-17). The distribution of PRS scores in PD cases and healthy controls was significantly different (Wilcoxon test value of p <2.2e-16, Figure 1A). Individuals with the 5 and 10% of highest PRS values had a 9.5-fold [3.9–26.3] (p=1.4e-08) and 5.6-fold [3.3–9.7] (p=1.8e-12) increased risk, respectively, compared to individuals with the lowest 5 and 10% PRS values (Figure 1A). Out of the 3,090 canonical pathways gene-sets representing the most important biological processes and diseases, 17 gene-sets were significantly associated with PD risk (Bonferroni adjusted value of p <0.05, Figure 1B; Supplementary Table 2). Among the enriched pathways, the majority were associated with PD (showing the highest R2 values, Figure 1B) and PD pathogenesis (Alpha synuclein, Parkin and ubiquitination related pathways), Alzheimer disease (AD), signal transduction and metabolism of proteins. No gene-set remained significant after excluding the 90 PD GWAS hits region (1Mb upstream and downstream each locus), indicating the absence of other risk loci acting independently of the known ones.

Open in a separate window

Figure 1

(A) Distribution of the polygenic risk score (PRS) between Parkinson’s disease (PD) patients and controls. (B) Forest plots showing PRS odds ratio (OR) and 95% confidence interval for the significant canonical pathways associated with PD risk (left panel) and the estimation of variance explained by PRS (right panel).

Discussion

The current report is a comprehensive genetic description of participants recruited within the monocentric case–control Luxembourg Parkinson’s study, including patients with PD and atypical parkinsonism, with previously described recruitment design and clinical characteristics (Hipp et al., 2018; Pavelka et al., 2022). Previous long-read sequencing of GBA1 gene in our cohort revealed that 12.1% of PD patients carried GBA1 variants (Pachchek et al., 2023). Analyzing now the complete Neurochip genotyping data, we investigated the potential effect of rare variants, common low-risk variants and CNVs on the PD pathogenesis. Our findings are consistent with those previously reported in European ancestry datasets.

In the LRRK2 gene, two well-established pathogenic SNVs were found in five PD patients and two controls with a frequency similar to previous European ancestry datasets (Correia Guedes et al., 2010; Shu et al., 2019). Pathogenic and probably pathogenic variants in the ATP13A2, PARK7, PRKN and PINK1 genes associated with autosomal recessive PD were found in the heterozygous state, except in one PD patient. The latter carried a homozygous pathogenic PINK1 variant (p.L369P) and had an AAO of 32years (Arena and Valente, 2017) reported a similar finding, where homozygous variants in PINK1 associated with early-onset PD (EOPD) were present in the patient before the age of 45years. In our study, the number of heterozygous SNVs in the recessive PARK7, ATP13A2, PRKN and PINK1 genes was not significantly different between PD cases and controls. Controls carrying these variants were over 50years of age. The majority had no family history of PD and most of PD patients were not of young onset (AAO>50years). Patients with AP carried probably benign heterozygous variants, mainly in LRRK2 and PINK1. Pathogenic LRRK2 variants have been described in patients with primary tauopathies, although at a low frequency (Wen et al., 2021). In particular, LRRK2 has recently emerged as a genetic risk factor associated with PSP progression (Jabbari et al., 2021).

We called CNVs from the genotyping data of individuals in the Luxembourg Parkinson’s study and after a stringent quality control and filtering steps, we screened for CNVs overlapping PD causal genes. We identified 12 PD patients who carried CNVs exclusively in the PRKN gene, of which three CNVs were validated by MLPA and were reported in patients having a disease AAO≤50years. Especially, we described a heterozygous exon1-4 duplication in a patient with EOPD (AAO of 18years) who did not present any rare variant in the PD-related genes studied here. Moreover, we validated by MLPA a homozygous duplication of PRKN exon1 in another PD patient with a late disease-onset (69years). Duplications of PRKN exons were previously reported as ‘likely pathogenic’ (Schüle et al., 2015). Indeed, both homozygous and compound heterozygous PRKN deletions and duplications have previously been associated with early-onset and familial forms of PD (Elfferich et al., 2011; Kim et al., 2012; Huttenlocher et al., 2015; Ahmad et al., 2023). This was recently reproduced in a large CNV study of 4,800 clinical exome sequencing reports (Pennings et al., 2023). In a Latin American PD cohort, CNVs in PRKN were significantly associated with disease progression, with a prevalence of 5.6% in EOPD cases (Sarihan et al., 2021).

We found that six PD cases and one PSP case carried digenic variants in two different PD-related genes (LRRK2-PRKN, LRRK2-PINK1, PINK1-ATP13A2) with AAO greater than 50years. Hitherto, only a few studies have identified digenic variants of PD-related genes [LRRK2-PRKN (Dächsel et al., 2006), PINK1-PARK7 (Tang et al., 2006) or PRKN-PINK1 (Hayashida et al., 2021)]. Previous studies reported that carriers of digenic variants in PRKN and PINK1 develop the disease at a younger age and exhibit distinctive symptoms such as schizophrenia, facial dyskinesia, grimacing and severe dysarthria (Funayama et al., 2008) and also epilepsy and essential tremor (Hayashida et al., 2021). However, the digenic variants reported in this study differ from those previously described, and carriers of these variants have an older age at onset. Nonetheless, the ambiguity surrounding digenic variants persists due to the limited number of reported cases. A detailed familial and clinical study could be carried for every individual, to confirm that the combination of these heterozygous variants, in the context of a digenic inheritance, may point out the phenotype observed in PD and PSP cases.

In our study, we did not find a significant overrepresentation of rare heterozygous SNVs and CNVs in PRKN. In particular, heterozygous pathogenic PRKN variants were not significantly more frequent in controls than in PD cases. Homozygous or compound heterozygous variants in this gene were the most common cause of EOPD (Kilarski et al., 2012), while heterozygous loss of PRKN function may be a potential risk factor for developing PD (Klein et al., 2007; Huttenlocher et al., 2015; Castelo Rueda et al., 2021; Lubbe et al., 2021) and therefore identifying individuals at increased risk might be useful in the prodromal phase. However, this role of heterozygous PRKN is still under debate, as previous reports suggested a lack of association with PD (Kay et al., 2010). Recently, in a larger association study, Yu and colleagues fully sequenced PRKN in a PD cases/controls cohort from European ancestry, including 1965 late-onset and 553 early-onset, and concluded that heterozygous SNVs or CNVs in PRKN are not associated with EOPD (Yu et al., 2021). They reported that 1.52% of PD and 1.8% of controls were carriers. Here, using a SNP array based on CNVs and SNPs screening, we showed similar percentages (1.64% of PD and 0.82% of controls) with non-significant differences between controls and mainly late-onset PD cases.

Potential neuroprotective PD therapies and clinical trials are now targeting specific PD subtypes based on genetic markers causing or increasing the disease risk, such as therapies targeting LRRK2, GBA1 and alpha-synuclein (Sardi et al., 2018). Parkin-proved disease is characterized by a slow motor progression, preserved cognition and a limited increase in dopaminergic medication over time (Menon et al., 2023). Moreover, severe loss of dopaminergic neurons was observed in homozygous PRKN carriers without Lewy bodies formation, which is one of the major markers of idiopathic PD (Mata, 2004). Confirming the potential role of heterozygous PRKN variants in the pathogenesis of PD will be crucial, despite the lack of data describing PD conversion of individuals carrying these genetic risk factors.

Beyond the effects of rare variants, we have demonstrated that individuals carrying the MAPT H1 haplotype are at higher risk to develop PD and PSP. These findings are consistent with previous studies that have assessed the H1 haplotype as a PD (Zabetian et al., 2007) and PSP (Baker et al., 1999) risk factor. We did not detect the association of PD or any forms of atypical Parkinsonism with H1c sub-haplotype which was strongly associated with risk for PSP and CBD (Myers et al., 2005; Kouri et al., 2015) but not PD (Zabetian et al., 2007). Next, we estimated the total cumulative contribution of common low-risk SNVs by calculating the PRS. Our PRS model of disease risk showed an expected trend similar to previous reports showing that PRS discriminates PD cases from unaffected individuals (Dehestani et al., 2021). Several polygenic analyses have become standard tools for assessing the risk for complex disorders and an accurate method for predicting disease status and identifying high-risk individuals (Lewis and Vassos, 2020). Next, we looked up at how thousands of biological pathways might contribute to the risk of developing PD. In addition to pathways already associated with PD and AD, molecular processes underlying proteins metabolism, signal transduction and post-translational protein modification were among the most important contributors to PD risk. The metabolic dysfunction, energy failure and redox imbalance observed in PD were considered obvious features to qualify PD as a complex metabolic disorder (Anandhan et al., 2017). In addition, disruption of any stage in the protein life cycle could engender PD pathology (Langston and Cookson, 2020). Comparing our results with a previous large-scale gene set-specific PRS studies that reported the involvement of multiple processes in the etiology of PD (Bandres-Ciga et al., 2020), similar molecular processes were found here. However, other processes such as immune response, synaptic transmission and endosomal-lysosomal dysfunction were not highlighted which may be due to the smaller sample size in our dataset. Pathway PRSs are expected to provide important insights into the complex heterogeneity of PD and how patients respond to treatment, by generating biologically traceable therapeutic targets from polygenic signals (Choi et al., 2023). We are aware that our study has several limitations: (1) the sample size was not large enough to have sufficient statistical power to perform further analysis, such as GWAS for PD risk or AAO, genome-wide CNV burden or human leukocyte antigen (HLA) association; (2) not all known variants associated with PD can be accurately assessed by the NeuroChip and we might have missed some mutated alleles, even though we confirmed all the identified variants by Sanger sequencing; (3) we used best practices to call CNVs from genotyping data (Sarihan et al., 2021), and thus we will always miss small CNVs that are systematically filtered out. Moreover, we could validate only few of the called CNVs with MLPA and (4) our analysis revealed a higher incidence of first-degree family history among controls. Therefore, caution must be exercised when searching for recessive disease forms. Although proxy cases have proven their effectiveness in large scale study investigating common variants (Nalls et al., 2019) and have also highlighted their usefulness in detecting rare variants (Makarious et al., 2023).

Conclusion

In conclusion, our study has successfully performed a comprehensive genetic baseline characterization of the Luxembourgish PD case–control cohort, investigating rare variants, CNVs and PRSs. Our findings do not support an association between PD risk and rare heterozygous PRKN variants. We also described a possible role of LRRK2 in AP and new possible digenic inheritance patterns in PD. Together with other studies in different European populations, our findings will advance the understanding of PD pathogenesis and genetics and could redefine the development of future therapeutic targets and therapies.

Data availability statement

Genetic and clinical data for this manuscript are not publicly available as they are linked to the internal regulations of the Luxembourg Parkinson’s Study. Requests for accessing the datasets can be directed to request ul.inu@dp-recn. The code behind the analyses is available under Apache-2.0 license. More information on how to request access to the data and the code is available at https://doi.org/10.17881/ea65-t027.

Ethics statement

The studies involving humans were approved by the National Research Ethics Committee (CNER Ref: 201407/13). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

NCER-PD consortium member

Geeta Acharya, Gloria Aguayo, Myriam Alexandre, Wim Ammerlann, Katy Beaumont, Camille Bellora, Jessica Calmes, Lorieza Castillo, Gessica Contesotto, Daniela Esteves, Guy Fagherazzi, Jean-Yves Ferrand, Manon Gantenbein, Marijus Giraitis, Jérôme Graas, Gaël Hammot, Anne-Marie Hanff, Estelle Henry, Michael Heymann, Alexander Hundt, Sonja Jónsdóttir, Jochen Klucken, Rejko Krüger, Pauline Lambert, Victoria Lorentz, Paula Cristina Lupu, Guilherme Marques, Deborah Mcintyre, Chouaib Mediouni, Myriam Menster, Maura Minelli, Ulf Nehrbass, Fozia Noor, Magali Perquin, Rosalina Ramos Lima, Eduardo Rosales, Estelle Sandt, Margaux Schmitt, Amir Sharify, Kate Sokolowska, Hermann Thien, Johanna Trouet, Olena Tsurkalenko, Michel Vaillant, Mesele Valenti, Luxembourg Institute of Health, Strassen, Luxembourg; Muhammad Ali, Giuseppe Arena, Rudi Balling, Michele Bassis, Regina Becker, Ibrahim Boussaad, Kathrin Brockmann, Piotr Gawron, Soumyabrata Ghosh, Enrico Glaab, Elisa Gómez De Lope, Valentin Groues, Anne Grünewald, Wei Gu, Linda Hansen, Michael Heneka, Sascha Herzinger, Jacek Jaroslaw Lebioda, Yohan Jaroz, Quentin Klopfenstein, Jochen Klucken, Rejko Krüger, Zied Landoulsi, Tainá M. Marques, Patricia Martins Conde, Patrick May, Francoise Meisch, Michel Mittelbronn, Michel Mittelbronn, Sarah Nickels, Marek Ostaszewski, Clarissa P. C. Gomes, Sinthuja Pachchek, Claire Pauly, Laure Pauly, Lukas Pavelka, Armin Rauschenberger, Rajesh Rawal, Dheeraj Reddy Bobbili, Kirsten Roomp, Isabel Rosety, Stefano Sapienza, Venkata Satagopam, Sabine Schmitz, Reinhard Schneider, Jens Schwamborn, Ekaterina Soboleva, Rebecca Ting Jiin Loo, Christophe Trefois, Carlos Vega, Maharshi Vyas, Paul Wilmes, Evi Wollscheid-Lengeling, Luxembourg Centre for Systems Biomedicine, University of Luxembourg, Esch-sur-Alzette, Luxembourg; Guy Berchem, Nico Diederich, Marijus Giraitis, Linda Hansen, Jochen Klucken, Rejko Krüger, Laura Longhino, Romain Nati, Beatrice Nicolai, Claire Pauly, Lukas Pavelka, Elodie Thiry, Liliana Vilas Boas, Gelani Zelimkhanov, Centre Hospitalier de Luxembourg, Strassen, Luxembourg; Michel Mittelbronn, Friedrich Mühlschlegel, Laboratoire National de Santé, Dudelange, Luxembourg; Alexandre Bisdorff, Rene Dondelinger, Centre Hospitalier Emile Mayrisch, Esch-sur-Alzette, Luxembourg; Sylvia Herbrink, Centre Hospitalier du Nord, Ettelbrück, Luxembourg; Roseline Lentz, Parkinson Luxembourg Association, Leudelange, Luxembourg; Michele Hu, Oxford Parkinson’s Disease Centre, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom; Richard Wade-Martins, Oxford Parkinson’s Disease Centre, Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, United Kingdom; Clare Mackay, Oxford Centre for Human Brain Activity, Wellcome Centre for Integrative Neuroimaging, Department of Psychiatry, University of Oxford, Oxford, United Kingdom; Daniela Berg, Kathrin Brockmann, Thomas Gasser, Inga Liepelt, Center of Neurology and Hertie Institute for Clinical Brain Research, Department of Neurodegenerative Diseases, University Hospital Tübingen, Tübingen, Germany; Brit Mollenhauer, Paracelsus-Elena-Klinik, Kassel, Germany; Katrin Marcus, Ruhr-University of Bochum, Bochum, Germany; Robert Liszka, Westpfalz-Klinikum GmbH, Kaiserslautern, Germany; Walter Maetzler, Department of Neurology, University Medical Center Schleswig-Holstein, Kiel, Germany; Mariella Graziano, Association of Physiotherapists in Parkinson’s Disease Europe, Esch-sur-Alzette, Luxembourg; Nadine Jacoby, Private Practice, Ettelbruck, Luxembourg; Jean-Paul Nicolay, Private Practice, Luxembourg, Luxembourg.

Author contributions

ZL: Conceptualization, Data curation, Investigation, Methodology, Writing – original draft. SP: Conceptualization, Data curation, Investigation, Methodology, Writing – review & editing. DB: Data curation, Writing – review & editing. LP: Data curation, Writing – review & editing. PM: Conceptualization, Supervision, Writing – review & editing, Investigation, Methodology. RK: Conceptualization, Supervision, Writing – review & editing, Investigation, Methodology.

Acknowledgments

Funding Statement

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. The National Centre of Excellence in Research on Parkinson’s Disease (NCER-PD) is funded by the Luxembourg National Research Fund (FNR/NCER13/BM/11264123), the PEARL program (FNR/P13/6682797 to RK), MotaSYN (12719684 to RK), MAMaSyn (to RK), MiRisk-PD (C17/BM/11676395 to RK and PM), the FNR/DFG Core INTER (ProtectMove, FNR11250962 to PM and ZL), and the PARK-QC DTU (PRIDE17/12244779/PARK-QC to RK and SP).

Footnotes

References