Genetic analysis of Mendelian mutations in a large UK population-based Parkinson’s disease study

In a large population-based study, Tan et al. show that pathogenic mutations are present in 1.4% of UK patients with Parkinson’s disease and are more common in those with early-onset or familial disease. Carriers of mutations in PRKN or PINK1 have distinctive clinical features compared to non-carriers.


Research ethics
Ethics approval was provided by West of Scotland Research Ethics Service (reference 11/AL/0163). The study was carried out in accordance with the Declaration of Helsinki and is registered as NCT02881099 at ClinicalTrials.gov.

Clinical data
Family history was obtained through a standardised structured interview based on patient report.
Participants were classified as having a family history if they had one or more first or second degree family members affected by Parkinson's disease.
Patients who had a family history of Parkinson's disease, with only family members affected from the same generation (siblings) were classified to be consistent with autosomal recessive inheritance. Patients with affected family members in other generations (e.g. father, mother, child, grandparent, aunt or uncle), were classified to be consistent with autosomal dominant inheritance.
Participants that did not have a family history of Parkinson's disease were not included in either the recessive or dominant inheritance groups, although recessive disease can become manifest in apparently sporadic cases.
The criteria and calculations to determine levodopa equivalent daily dose (LEDD) dose, motor subtypes, cognitive impairment, depression and anxiety have been previously described (Malek et al., 2015).
Missing AAO data were estimated from the age of diagnosis, using the mean time from disease onset to diagnosis (1.86 years) from patients with non-missing data.

Genetic analysis of mutations
Pathogenicity of a variant was classified based on MDSGene criteria: co-segregation with disease, allele frequency in healthy controls in Genome Aggregation Database (gnomAD; http://gnomad.broadinstitute.org/), the Combined Annotation Dependent Depletion (CADD) score, and functional evidence from in-vivo and in-vitro studies (Lill et al., 2016;Kasten et al., 2018).

Exome sequencing
Exome sequencing was performed by Macrogen (http://www.macrogen.com/) using the Agilent SureSelect capture kit (Santa Clara, CA, USA). Variant calls and individual genotypes that did not meet quality filters were excluded. Samples were aligned to the human genome (build hg19). The Genome Analysis Toolkit (GATK) was used for local realignment, base quality score recalibration, and multi-sample variant calling (Unified Genotyper). GATK Variant Quality Score Recalibration and recommended GATK training sets were used to create a high-quality set of variant calls (Farlow et al., 2016). ANNOVAR was used to annotate variants with information on functional consequence, minor allele frequency (MAF), variant type and previous reporting (Wang et al., 2010).

Identity-by-Descent analysis
IBD analysis was conducted in PLINK v.1.9 using linkage-pruned SNPs with MAF>0.05. SNPs were excluded if they had a squared correlation (r 2 ) of greater than 0.05 within a sliding window of 50 adjacent SNPs scrolling through 5 SNPs at a time. IBD analysis was conducted using default PLINK settings which assesses extended segmental sharing of at least 100 SNPs and total length greater than 1 megabase (Mb) (Purcell et al., 2007).

Mutations of uncertain pathogenicity
We reported variants that were suggested to be potentially pathogenic or protective for Parkinson's disease/parkinsonism from previous studies (excluding those reported in the main text), with references provided in Supplementary Table 3. We screened all the PARK loci (Hernandez et al., 2016) and genes associated with parkinsonism in MDSGene (Lill et al., 2016). Additional genes and variants were identified through ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), Genomics England PanelApp Parkinson's disease panel (https://panelapp.genomicsengland.co.uk/), Online Mendelian Inheritance in Man (https://www.omim.org/) and from literature searches.
APOE ε4 genotypes were extracted from exome data using SNPs rs429358 and rs7412 (Williams-

Literature review
For our discussion of previous literature, we focused on studies relating to the frequency and phenotype-genotype of LRRK2, SNCA, PRKN and/or PINK1 pathogenic mutations. We did not conduct a systematic review of the literature but used MDSGene systematic reviews Trinh et al., 2018) and other recent review articles (from 2013 onwards) to identify relevant studies and present a balanced discussion of previous literature.

Results of haplotype analysis
Haplotype construction showed that 6 of 16 LRRK2 G2019S patients carried Haplotype 2 (European predominant), and 9 carried Haplotype 1 (Jewish/European predominant). Haplotype could not be confirmed for one patient, possibly due to lower imputation quality for one marker rs28903073 (Supplementary Figure 6).
Haplotype construction for the two LRRK2 R1441C carriers showed they shared a common haplotype based on region of 126 kB across 10 SNP markers, indicating that these apparently unrelated individuals shared a common ancestral founder (Supplementary Figure 7). This shared region is consistent with haplotypes previously reported in European patients (Haugarvoll et al., 2008;Criscuolo et al., 2011). However we were unable to distinguish between the two common haplotypes, the Belgian-Western Nebraska haplotype  or the other major haplotype found in Italian, German, Spanish and American patients.
The most common PRKN mutations detected were P113Xfs deletion (5/8 carriers) and the R275W mutation (4/8 carriers). Haplotype reconstruction suggests that the five P113Xfs mutation carriers had a shared segment spanning at least 242 kB, at the markers rs1801474, rs4709583 and rs2075923, indicating a common founder (Supplementary Figure 8). Haplotypes were discordant at the markers rs1801582 and rs3765474.
For PRKN R275W, there were 3 carriers with SNP array data and haplotypes appeared to be shared at all 6 markers genotyped spanning approximately 1Mb (data not shown). One PRKN carrier did not have SNP array data available for haplotype reconstruction. Only two patients carried deletions but these were in different exons so we were unable to observe shared or discordant haplotypes.

Supplementary Tables
Supplementary Table 1 (Gao et al., 2011;Federoff et al., 2012) Mixed evidence as to whether the ε4 allele is associated with increased risk for PD and potentially dementia in PD (Marder et al., 1994;Li et al., 2004;Gao et al., 2011;Federoff et al., 2012)  12:40631791  T  T  T  T  (C/T)  T  T  T  T  T  T  T  T  T  T  T  rs1896252 12