Two-Stage Whole-Exome Sequencing Improves to Predict a Risk of Adult Moyamoya Disease in 369,570 Individuals

doi:10.21203/rs.3.rs-4952354/v1

Download PDF

Article

Two-Stage Whole-Exome Sequencing Improves to Predict a Risk of Adult Moyamoya Disease in 369,570 Individuals

https://doi.org/10.21203/rs.3.rs-4952354/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Whole-exome sequencings (WES) have an informative in the limelight to identify causative mutations for adult moyamoya disease (MMD), understanding genomic structures of etiology. Here, we conducted inaugural two-stage WES aimed at uncovering coding modifiers implicated in MMD. Our study comprised an initial discovery phase with 105 MMDs and 115 controls, followed by validation phases involving 55 MMDs and 74 controls, alongside 100 disease-free subjects. We extended comparisons of the allele frequencies of 369,121 individuals derived from UK Biobank (UKB) WES data. Mutant allele risk scores (MARS) were created on the basis of WES-driven mutations. Gene-based association and East-Asian pooled analyses were further performed. During the discovery phase, p.G576S (rs1800307-GAA) and p.R4810K (rs112735431-RNF213) reached at a genome-wide significance threshold (P = 2.63×10^-8 and 2.24×10^-16, respectively), with p.R4810K being confirmed in the validation phase (P = 3.08×10^-8). One insertion (p.S2026ins:rs112774151-MUC4) demonstrated the most significance in 160 MMDs and 100 disease-free controls (P = 5.65×10^-16). Fourteen mutations exhibited significant differences in allele frequencies between patients and UKB controlled data (P < 1×10^-8). MARS9 incorporating nine missense mutations resulted in an enhanced predictability for MMD (AUROC = 0.8323). Gene-based associations replicated across all phases for GAA, RNF213, CHMP6, and CARD14 (P < 5×10^-7). For mutations in RNF213, p.V1195M, p.D1331G, p.S2334N, and p.R4810K were validated in East-Asian populations (P < 3×10^-8). Our pioneering study corroborate the significance of p.R4810K and uncover several novel mutations predisposing patients, thereby understanding polygenetic aspect to the etiology of MMD.

Biological sciences/Genetics/Genetic interaction

Biological sciences/Genetics/Genetic markers

Biological sciences/Genetics/Genome

Biological sciences/Genetics/Genomics

Biological sciences/Genetics/Genotype

Biological sciences/Genetics/Medical genetics

Biological sciences/Genetics/Mutation

Biological sciences/Genetics/Neurodevelopmental disorders

Biological sciences/Genetics/Sequencing

Biological sciences/Genetics

Health sciences/Biomarkers

Health sciences/Diseases

Health sciences/Medical research

Health sciences/Neurology

Health sciences/Risk factors

17q25.3

Moyamoya disease

Mutant allele risk score

Whole-exome sequencing

Moyamoya disease (MMD) is a chronic, rare pathology of the cerebral arteries characterized by progressive steno-occlusive changes in the distal internal carotid or proximal middle cerebral arteries without an established cause.¹ The incidence of MMD is markedly greater in Japanese, Korean, and Chinese populations than in European or North American populations.² Mutations in the ring finger protein 213 gene (RNF213) are broadly recognized as significant contributors to MMD development, particularly in East Asian populations.³ Comprehensive meta-analyses have demonstrated that RNF213 rs112735431 and rs148731719 play crucial roles in MMD development, whereas TIMP-2 rs8179090, MMP-2 rs243865, and MMP-3 rs3025058 exhibit an inverse relationship with MMD.⁴ A recent genome-wide association study (GWAS) augmented by fine-mapping confirmed RNF213 in adult MMD patients in the Korean population and identified additional susceptible loci within 17q25.3, such as TBC1D16, CCDC40, GAA, and ENDOV, contributing to MMD susceptibility.⁵ To date, the majority of microarray-based studies on MMD have utilized single-nucleotide variations (SNVs) to explore associations with disease phenotypes across the human genome.

Following GWAS, whole-exome sequencing (WES) offers another perspective on MMD pathogenesis by pinpointing functional mutations or allied mechanisms.^6,7 Wiedmann et al. (2023) identified a novel potential AGX12 locus associated with MMD in Northern European populations that is implicated in the regulation of nitric oxide metabolism.⁷ Nevertheless, studies employing WES have generally included a more limited patient cohort than GWASs do, without separate phases for discovery and validation or considering age among MMD patients. To address these gaps, we undertook the inaugural two-stage WES study to identify and confirm coding modifiers in adult MMD patients through a hospital-based and multicenter patient-control investigation. Furthermore, we assessed WES data from MMD patients against another independent data from Korean populations and contrasted mutant allele frequencies between Asian- and European-ancestry populations using ~ 450K WES entries from the UK Biobank (UKB).⁸

Study subjects and UK Biobank resources

This research included a cohort of adult patients diagnosed with MMD or undergoing revascularization procedures, alongside a control group devoid of MMD evidence on radiological assessments across multi-centered institutions.⁵ The discovery phase included 105 MMD patients and 115 controls from December 2021 to April 2022, and the subsequent validation phase included 55 MMD patients and 74 controls from November 2022 to May 2023. The definition of adult MMD patients hinged upon chronological age at diagnosis (18 years or older) and singular or bilateral angiographic manifestations of MMD vasculopathy, excluding cases of intracranial irradiation, fibromuscular dysplasia, Marfan syndrome, tuberous sclerosis, and polycystic kidney disease.^9,10 Control individuals were identified among patients attending hospital visits for general health screenings or headache assessments and confirmed to be devoid of cerebrovascular disorders by "The First Korean Stroke Genetics Association Research (The FirstKSGAR)" consortium.¹¹ Subsequent analyses leveraged data from the Korean Association Resources (KARE) cohort, including WES data from 100 healthy, disease-free participants integrated by the Korean Genome and Epidemiology Study (KoGES) consortium since 2001.¹² These data were produced on the HiSeq 2500 system, achieving a read depth of 200× (Illumina, San Diego, CA, USA). For the UKB WES dataset, analyses were conducted using three identified MMD patients and 369,118 individuals without cerebrovascular disease.⁸

Clinical and radiological data from patients were meticulously reviewed by three medical doctors, focusing on sex, age at diagnosis, hypertension, type 2 diabetes mellitus, hyperlipidemia, smoking history, initial symptoms, surgical revascularization interventions, clinical outcomes, and angiographic progression. All the methodologies conformed to the applicable guidelines and regulations. All methods were carried out in accordance with relevant guidelines and regulations. The study including all protocols and patients informed consent has been approved by three Institutional Review Boards (IRBs) and Ethics Committees, such as the Hallym University (HIRB-2022-042-CR), the Hallym University Chuncheon Sacred Heart Hospital (2019-06-006), and the Seoul National University Hospital (1806-105-951). Access to all UKB data was acquired following project approval (application number 80385).

Whole-exome sequencing

Genomic DNA (gDNA) at a concentration of 100 ng/µl was extracted from the peripheral blood of the study participants. The PicoGreen assay and agarose gel electrophoresis were employed to assess the quantity and quality of gDNA, respectively. Subsequently, the DNA underwent a washing and amplification process.⁵ The final, purified product was quantified using quantitative PCR in accordance with the qPCR Quantification Protocol Guide (KAPA Library Quantification Kits for Illumina Sequencing platforms), and its quality was ascertained using a TapeStation DNA Screentape D1000 (Agilent Technologies Inc, CA, USA). For sequencing, the NovaSeq 6000 platform was utilized, employing paired-end reads and achieving a coverage depth of 300× (Illumina, San Diego, CA, USA). Comprehensive details are provided on the library construction and sequencing procedures (Supplementary Note S1).

Read alignment, base recalibration, and variant calling

A comprehensive overview of the methodologies was described in elsewhere (Supplementary Fig. 1). With the exception of the UKB WES data, which were provided in the PLINK binary format, paired-end sequences in Fastq format were subjected to a quality assessment, filtering out those with a quality score and read length less than 20 and 70, respectively, using the NGSQCToolkit v2.3 (https://github.com/mjain-lab/NGSQCToolkit).¹³ Following this initial quality control, the sequences were mapped and aligned utilizing the BWA-MEM program against the hg19 reference genome (build 37).¹⁴ SAM files generated from this process were subsequently transformed into binary alignment map (BAM) files and organized by chromosome using SAMtools v1.9,¹⁵ facilitating the evaluation of the alignment’s read conversion rate. To prepare the aligned BAM files for variant calling, PCR duplicates were excised, and the files underwent local realignment and base recalibration, processes facilitated by the Picard MarkDuplicates v2.22.4 (http://picard.sourceforge.net) and Genome Analysis Toolkit (GATK) v4.1 (http://www.broadinstitute.org/gatk) programs, respectively.

For each sample, a single-sample genomic variant call file (gVCF) was produced using HaplotypeCaller in “-ERC GVCF” mode.¹⁶ Subsequent to gVCF creation, a joint genotyping analysis was conducted, in which all gVCF files were merged via the GATK GenotypeGVCFs tool. To ensure variant quality, SNVs and short insertions/deletions were subjected by the variant quality score recalibration to filtration based on tranche scores of 99.5 and 95.0, respectively. All the aforementioned procedures adhered to the guidelines recommended in the GATK Best Practice (https://gatk.broadinstitute.org/hc/en-us/articles/360035535932-Germline-short-variant-discovery-SNPs-Indels-).

Variant annotation and quality controls

A comprehensive total of 244,623 sites were meticulously annotated within exome regions using the ANNOVAR program (http://www.openbioinformatics.org/ANNOVAR/).¹⁷ Of these, 70,333 and 61,513 sites were subsequently refined for the discovery and validation sets, respectively, along with an additional cohort comprising 160 MMD patients and 100 healthy controls. This refinement process entailed implementing stringent quality control measures, including the enforcement of a genotyping call rate of ≥ 95%, the exclusion of monomorphic sites, and the maintenance of Hardy-Weinberg equilibrium with P value ≥ 1×10^− 6 within the control group. All participants underwent rigorous checks to ascertain their relatedness to one another utilizing the PLINK 1.9 program, employing the “--genome” flag (https://www.cog-genomics.org/plink/1.9/)¹⁸ to calculate the statistics of identical by descent (IBD) paired with a genetic relatedness metric (PI_HAT). The derived genetic relatedness values ranged from 0 to 0.0489 among the adult MMD patients and from 0 to 0.177 within the control participants, confirming the absence of relatedness among all individuals involved.

Statistical analyses

The data are presented as subject numbers (percentages) for discrete variables and means ± standard deviations for continuous variables. All summary statistics were computed using STATA software v17.0 (Stata Corp., TX, USA). A two-stage, whole exome-wide association study (WEWAS) was performed, assuming an additive inheritance model after adjusting for age, sex, and four principal component analysis (PCA) values, utilizing the PLINK v1.9.¹⁸ PCA values were transformed into four multidimensional scales and calculated at each stage of WEWAS and within the 1000 Genome Reference Panel Phase 3 (ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release), employing the shared variants. The inflation factor was derived by executing PLINK v1.9 with the options “—assoc” and “--adjust”.¹⁸ Manhattan and regional association plots were generated using “qqman” in the R package v3.6.2 (https://cran.r-project.org/web/packages/qqman) and LocusZoom v1.3 (http://locuszoom.org/),¹⁹ our modified scripts of which was written by Python v.2.7.5 and R v.3.6.2 languages. Subsequently, we compared the differential allele frequencies with those of the UKB WES data.

We additionally assessed mutant allele risk scores (MARS) by summing the risk alleles from significant mutations (i.e., 0, 1, or 2) to gauge the diagnostic probability for MMD. The predictive accuracy for diagnosing MMD was determined by the area under the receiver operating characteristic curve (AUROC). A MARS-based examination correlated with MMD incidence was performed using multivariate logistic regression models.

A gene-based association assessment was carried out utilizing the Sequence Kernel Association Test v2.2.5 in R package through the “SKAT_CommonRare” function, which was designed to appraise the collective impact of rare and standard variants (http://cran.r-project.org/web/packages/SKAT).²⁰ Variants with a minor allele frequency (MAF) < 0.01 were classified as rare, whereas those with a MAF ≥ 0.01 were considered common. P values for each gene were derived after resorting to the bootstrap method for resampling 10,000 times, thus computing Q statistics that reflect cumulative chi-square distributions.

A pooled analysis of RNF213 mutations, obtained from exome sequencing, was performed employing the inverse variance method (http://www.well.ox.ac.uk/gwama/),²¹ integrating our results and those of Moteki et al.²² Cumulative estimations were generated under fixed- and random-effect models. The Cochran Q test was used to determine I² statistics, and the q value was used to assess genetic heterogeneity.

Clinical characteristics of study populations

This investigation included a two-phased WEWAS based on participants from hospital settings and expanded upon its comparisons by employing multicenter biobank WES data from the KoGES and UKB. Detailed descriptions of the study methodologies and baseline characteristics for each group of participants are illustrated in Fig. 1 and documented in Table 1. The study’s scope included a hospital-based and multicenter patient-control approach featuring a discovery stage (105 MMD patients and 115 controls, from December 2021 to April 2022) and a validation stage (55 MMD patients and 74 controls, from November 2022 to May 2023). The proportion of female participants among MMD patients was comparable across the two stages. Similarly, the mean age at diagnosis did not significantly differ between MMD patients and controls across stages. With the exception of hypertension incidence, no other significant differences were observed in the analyzed variables (Supplementary Table S1). Routine angiography was conducted for 138 patients (86.3%). The mean follow-up periods were 64.8 ± 42.9 and 55.0 ± 35.9 months for the two stages, respectively. Within this cohort, 10 patients exhibited angiographic progression. For further comparative analysis, the KARE cohort (encompassing 100 disease-free healthy controls) and UKB WES data (including 3 MMD patients) were utilized. The UKB WES included two Asian-ancestry patients and one European-ancestry patient, alongside 9,641 Asian individuals and 359,477 European individuals who had cerebrovascular disease-free records.

Table 1

Clinical characteristics of study populations
Variables^a	Discovery stage		Validation stage		KARE^b	UKB-ASN^c		UKB-EUR^c
	105 MMDs	115 controls	55 MMDs	74 controls	100 DF controls	2 MMDs	9,641 controls	1 MMD	359,477 controls
Female, N (%)	71 (72.4)	154 (61.7)	40 (72.7)	36 (48.7)	60 (60)	1 (50)	4,739 (49.2)	0 (0)	196,584 (54.7)
Age, years ± S.D.	40.3 ± 12.3	43.5 ± 12.2	35.8 ± 11.4	38.3 ± 9.9	49.3 ± 7.9	60.5 ± 0.7	52.9 ± 8.3	68	56.6 ± 8.0
Cigarette smoking, N (%)	12 (11.4)	18 (15.6)	19 (34.6)	11 (14.9)	35 (35)	0 (0)	3,192 (33.1)	0 (0)	215,453 (59.9)
Hypertension, N (%)	26 (24.5)	28 (24.4)	13 (23.6)	9 (12.2)	0 (0)	1 (50)	2,523 (26.2)	1 (100)	89,813 (25)
Diabetes mellitus, N (%)	6 (5.7)	11 (9.6)	0 (0)	5 (6.8)	0 (0)	0 (0)	1,391 (14.3)	0 (0)	15,929 (4.4)
Hyperlipidemia, N (%)	20 (19.1)	16 (13.9)	8 (14.6)	1 (1.4)	0 (0)	2 (100)	1,531 (15.9)	0 (0)	40,596 (11.3)
Clinical relevance
Periodic angiography examination^d	85 (81.0)		53 (96.4)
Follow up, month ± S.D.	64.8 ± 42.9		55.0 ± 35.9
Angiographic progression, N (%)	3 (2.9)		7 (12.2)
Good neurologic outcome, N (%)	82 (78.1)		48 (87.3)
DF, disease-free; KARE, Korean Association Resource; MMD, moyamoya disease; UKB-ASN, Asian-ancestry population in UK Biobank data; UKB-EUR, European-ancestry population in UK Biobank data.
^a Data are shown as the numbers of subjects (percentage) for discrete variables and mean ± standard deviation (S.D.).
^b A total of 100 disease-free subjects was selected from 8840 participants of the Korean Association REsource (KARE), implemented by the Korean Genome Epidemiology Study (KoGES) consortium.
^c Asian- and European-ancestry individuals were selected from 454,787 participants of the UK Biobank composing whole-exome sequencing data.
^d A total of 138 patients with MMD have been regularly followed by cerebral angiography.

Whole exome-wide association study

The clustering of genetic ethnicity between PC1 and PC2 revealed a distribution that was nearly identical among 449 Korean and 504 East Asian subjects (Supplementary Figure S2). The coding variants, numbering 70,333, were categorized into 32,455 nonsynonymous SNV, 30,683 synonymous SNVs, 133 frameshift insertions and 260 nonframeshift insertions; 242 frameshift deletions; 454 nonframeshift deletions; 360 stop-gain and 30 stop-loss SNVs; 250 splicing sites; 4,382 ncRNA SNVs; and 1,084 sites of unknown function in two-stage WEWAS. Additionally, a comparison of 61,513 coding variants between 160 MMD patients and 100 disease-free healthy controls was conducted (Supplementary Table S2). During the discovery phase, two mutations, p.G576S (rs1800307-GAA) and p.R4810K (rs112735431-RNF213), achieved a genome-wide significance threshold (P = 2.63×10^− 8 with odds ratio [OR] = 4.36, 95% confidence interval [CI]: 2.60–7.32; P = 2.24×10^− 16 with OR = 122, 95% CI: 38.74–384.1, respectively) (Fig. 2a and 2b). In the validation phase, the p.R4810K mutation exhibited the most significant association with MMD (OR = 162.8, 95% CI: 26.83–987.4, P = 3.08×10^− 8) (Fig. 2c and 2d). Combining the discovery and validation sets revealed 14 mutations in the 17q25.3 region, including in GAA, RNF213, and CHMP6, with a genome-wide significance threshold (1×10^− 24 < P < 5×10^− 8) (Fig. 2e and 2f, Table 2). Among these, the minor allele "A" in p.R4810K (RNF213) exhibited the strongest association with the development of MMD (OR = 117.4, 95% CI: 46.39–297.3, P = 8.54×10^− 24). Furthermore, the p.G576S, p.E689K, and p.T711T mutations in GAA also significantly increased the risk of MMD (OR = 3.79, 2.78, and 2.72, respectively). Additionally, three mutations, p.T1070T, p.Q1133K, and p.A1550A, demonstrated significant associations with MMD development (P = 9.83×10^− 6, 6.40×10^− 6, and 5.58×10^− 8, respectively).

Table 2

Significant loci identified genome-wide or suggestively for adult moyamoya disease through a two-stage, whole exome-wide association study
Gene	Mutation	RS number	M/m^a	Discovery stage (2021–2022 years)^b				Validation stage (2022–2023 years)^b				Combined: 160 MMDs and 189 controls
(17q25.3)	Protein (p)	Variation	Allele	OR^c	L95^c	U95^c	P^c	OR^c	L95^c	U95^c	P^c	OR^c	L95^c	U95^c	P^c
Genome-wide significant associations
GAA	p.G576S	rs180 0307	G/A	4.36	2.60	7.32	2.63E-08	3.91	1.91	7.99	1.89E-04	3.79	2.56	5.63	3.65E-11
GAA	p.E689K	rs1800309	G/A	2.93	1.87	4.61	2.88E-06	3.08	1.58	6.00	9.69E-04	2.78	1.95	3.97	1.95E-08
GAA	p.T711T	rs1800310	A/G	2.53	1.67	3.83	1.32E-05	3.50	1.76	6.97	3.57E-04	2.72	1.92	3.84	1.65E-08
RNF213	p.K1034M	rs55996424	T/A	0.23	0.14	0.40	6.55E-08	0.14	0.06	0.36	3.81E-05	0.21	0.14	0.33	1.01E-11
RNF213	p.V1195M	rs10782008	G/A	0.32	0.19	0.52	3.92E-06	0.20	0.08	0.47	2.09E-04	0.29	0.19	0.44	4.78E-09
RNF213	p.E1272Q	rs9913636	G/C	0.34	0.21	0.56	2.23E-05	0.20	0.09	0.48	2.49E-04	0.31	0.21	0.47	2.69E-08
RNF213	p.D1331G	rs8074015	A/G	0.35	0.22	0.57	2.11E-05	0.20	0.08	0.47	2.09E-04	0.31	0.21	0.47	2.19E-08
RNF213	p.V1340V	rs9908287	C/G	0.33	0.20	0.54	1.03E-05	0.20	0.08	0.47	2.09E-04	0.30	0.20	0.45	9.86E-09
RNF213	p.S2334N	rs9674961	G/A	0.33	0.20	0.55	1.80E-05	0.21	0.09	0.50	4.20E-04	0.30	0.19	0.45	2.45E-08
RNF213	p.P2415P	rs4890012	G/C	0.32	0.19	0.52	5.46E-06	0.20	0.08	0.47	2.09E-04	0.28	0.19	0.43	4.42E-09
RNF213	p.E3490E	rs7216493	G/A	0.27	0.17	0.44	6.23E-08	0.15	0.06	0.36	1.79E-05	0.24	0.16	0.36	8.01E-12
RNF213	p.H4557H	rs4889848	C/T	0.26	0.16	0.42	9.14E-08	0.23	0.10	0.49	1.46E-04	0.26	0.18	0.39	6.36E-11
RNF213	p.R4810K	rs112735431	G/A	122	38.74	384.1	2.24E-16	162.8	26.83	987.4	3.08E-08	117.4	46.39	297.3	8.54E-24
CHMP6	p.D54N	rs572398921	G/A	23.11	6.78	78.74	5.15E-07	NA	NA	NA	NA	31.53	9.41	105.70	2.25E-08
Suggestive significant associations
CCDC40	p.T1070T	rs56407805	G/A	0.51	0.35	0.75	6.92E-04	2.39	1.27	4.53	7.35E-03	0.48	0.35	0.67	9.83E-06
RNF213	p.Q1133K	rs8082521	C/A	0.39	0.23	0.68	9.33E-04	0.17	0.06	0.49	9.29E-04	0.33	0.20	0.53	6.40E-06
RNF213	p.A1550A	rs4890009	G/A	0.35	0.21	0.58	3.72E-05	0.21	0.09	0.50	4.20E-04	0.31	0.20	0.47	5.58E-08
M, major; m, minor; OR, odds ratio; L95, lower 95% confidence interval; U95, upper 95% confidence interval; NA, Not available.
^a M/m indicates Major and minor allele type, respectively.
^b Two independent stages, discovery (2021–2022 years) and validation (2022–2023 years) sets composed 105 patients and 115 controls and 55 patients and 74 controls, respectively.
^c Odds ratio (OR), 95% confidence interval (Lower 95% CI [L95] and Upper 95 CI [U95]), and P values were estimated under the multivariate logistic regression model after adjustment for age, gender, and four genetic ancestry values of principal component analysis.

A comparison between 160 MMD patients and 100 disease-free controls highlighted six WEWAS-driven mutations in MUC4, FGFRL1, PRIM2 (Supplementary Figure S3); and MAP3K9, with 17 suggestive loci associated with MMD (Supplementary Table S3). A rare nonframeshift insertion (p.S2026ins:rs112774151-MUC4) had the most significant association (OR = 64.7, 95% CI = 23.58–177.5, P = 5.65×10^− 16) (Supplementary Figure S4). Nine mutations within the 17q25.3 region presented nominal or suggestive associations according to comparative analysis (i.e., p.T1070T, p.G576S, p.E689K, p.T711T, p.K1034M, p.S2334N, p.P2415P, p.E3490E, and p.H4557H; 5×10^− 8 < P < 8×10^− 4) (Supplementary Table S4).

A comparison of 14 genome-wide significant mutations in WEWAS patients with those in UKB WES data was also performed (Table 3). All the mutations exhibited a marked difference in effect allele frequencies between the patient and control groups across both Asian- and European-ancestry individuals (P < 1×10^− 8). The effect allele frequency (EAF) in p.G576S (GAA) was less common in Europeans (0.03%) and rare in Asians (4.49%) than in Koreans (16.4%). Two mutations, p.R4810K (RNF213) and p.D54N (CHMP6), had no mutation in European-ancestry populations, in contrast to Korean and Asian-ancestry individuals in the UKB (0.03–1.59% in EAF). No mutant allele for these two mutations was observed in three MMD patients via UKB WES; however, these mutant alleles are highly prevalent among Korean carriers (126 MMDs in p.R4810K and 52 MMDs in p.D54N) (Supplementary Table S5).

Table 3

Comparison of allele frequencies for 14 mutations within the 17q25.3 region between patients with moyamoya disease and controls
Gene	Exon	Mutation	N/E^a	KOR	KOR-MMD vs. KOR Control		KOR-MMD vs. UKB-ASN		KOR-MMD vs..UKB-EUR
(17q25.3)	Number	Protein (p)	Allele	EAF-MMD^a	EAF-Control^a	P^b	EAF-Control^a	P^b	EAF-Control^a	P^b
GAA	exon12	p.G576S	G/A	0.4156	0.1640	1.69E-13	0.0449	3.78E-87	0.0003	< 1E-300
GAA	exon15	p.E689K	G/A	0.4625	0.2434	1.61E-09	0.0879	2.77E-67	0.0365	1.69E-121
GAA	exon15	p.T711T	A/G	0.5875	0.3598	2.43E-09	0.2983	3.06E-26	0.2787	9.99E-31
RNF213	exon17	p.K1034M	T/A	0.1656	0.4021	5.00E-12	0.3185	1.22E-79	0.1464	6.64E-166
RNF213	exon21	p.V1195M	G/A	0.1500	0.3571	3.59E-10	0.4973	3.56E-38	0.6377	2.03E-72
RNF213	exon21	p.E1272Q	G/C	0.1469	0.3386	5.24E-09	0.4154	6.61E-25	0.5314	3.58E-34
RNF213	exon21	p.D1331G	A/G	0.1562	0.3519	3.35E-09	0.5515	8.45E-48	0.6774	9.51E-83
RNF213	exon21	p.V1340V	C/G	0.1500	0.3492	1.50E-09	0.4969	3.80E-38	0.6378	1.92E-72
RNF213	exon29	p.S2334N	G/A	0.1312	0.3201	3.64E-09	0.5479	9.15E-54	0.6601	1.18E-85
RNF213	exon29	p.P2415P	G/C	0.1375	0.3439	1.82E-10	0.5063	4.19E-43	0.6379	1.50E-76
RNF213	exon34	p.E3490E	G/A	0.2031	0.4630	3.92E-13	0.5989	1.96E-46	0.7076	1.07E-77
RNF213	exon56	p.H4557H	C/T	0.2531	0.4947	4.61E-11	0.6642	5.68E-50	0.8417	1.13E-120
RNF213	exon60	p.R4810K	G/A	0.4000	0.0159	8.89E-43	0.0004	7.57E-230	0.0000	< 1E-300
CHMP6	exon2	p.D54N	G/A	0.1656	0.0079	6.72E-16	0.0003	8.88E-90	0.0000	2.38E-180
E, effect; N, non-effect; EAF, effect allele frequency; KOR, Korean; MMD, moyamoya disease; UKB-ASN, Asian-ancestry population in UK Biobank; UKB-EUR, European-ancestry population in UK Biobank; vs., versus.
^a N/E indicates a non-effect and effect allele type, respectively; EAF was computed in either patient or control groups.
^b P value for difference in an effect allele frequency between patients and controls.

Integrative allele-based association of the mutant allele risk score in MMD

Descriptions of the distributions of summed MARS per individual for predicting the development of MMD were based on WEWAS-driven mutations within the 17q25.3 region (Fig. 3). The individual AUROC for each mutation is detailed in elsewhere (Supplementary Table S6). Among the 14 mutations assessed, RNF213-p. R4810K exhibited the highest AUROC (AUROC = 0.8781, 95% CI: 0.8439–0.9122). Two mutations, namely, GAA-p. G576S and RNF213-p.E3490E, demonstrated AUROCs of approximately 0.7. The AUROCs of other mutations ranged between 0.65 and 0.70. The predictability of MMD using MARS14, which encompasses all 14 mutations, was assessed, and the results showed an AUROC of 0.7998 (95% CI: 0.7427–0.8569) in the discovery stage (Fig. 3a), an AUROC of 0.8167 (95% CI: 0.7429–0.8905) in the validation stage (Fig. 3b), and an AUROC of 0.8055 (95% CI: 0.7606–0.8503) in the combined stage (Fig. 3c). A MARS9 incorporating only nine missense mutations resulted in an enhanced predictability for MMD (AUROC = 0.8323, 95% CI: 0.7907–0.8739 in combined stage), compared to that of MARS14. MARS9 demonstrated a markedly stronger association with the development of MMD (OR = 1.59, 95% CI: 1.43–1.76, P = 2.0×10^− 18) than did MARS14 (OR = 1.22, 95% CI: 1.16–1.27, P = 1.7×10^− 16). Although the effects of the two polygenic MARS models did not surpass that of the single mutation in p.R4810K, it appeared that the collective effects of the 14 mutations may have synergistically contributed to the development of MMD, potentially through interactions or correlations among them, thereby mitigating the overlap in effect sizes in the MARS models.

Gene-based association study

A comprehensive analysis of 14,225 genes was conducted to explore gene-based associations with MMD using a two-stage WEWAS approach (Supplementary Table S7). In the discovery phase, four genes, namely, GAA, RNF213, CHMP6, and SLC44A2, demonstrated suggestive or substantial associations with MMD (1×10^− 13 < P < 1×10^− 5) (Fig. 4a); remarkably, RNF213 was independently confirmed (Fig. 4b). Moreover, in the combined stage, four genes including GAA, RNF213, CHMP6, and CARD14 were intimately linked to MMD (5×10^− 19 < P < 5×10^− 7; Fig. 4c).

In an ensuing comparison involving 13,975 genes, 160 MMD patients were compared with 100 disease-free controls (Supplementary Table S8). Six genes, FGFRL1, PRIM2, TBP, CASP5, MAP3K9, and RNF213, exhibited strong associations with MMD. Among these genes, PRIM2 was the most significantly correlated with the development of MMD (Q = 580.47, P = 5.03×10^− 19; Fig. 4d). Notably, RNF213 was consistently associated with MMD across the different control groups (P < 1×10^− 8, Supplementary Table S9).

A pooled analysis of RNF213 in East Asian populations

An inverse variance analysis employing WES datasets from 263 MMD patients and 473 controls—including current WEWAS data (160 MMD patients and 189 controls) and Japanese cohorts (103 MMD patients and 184 controls) (Supplementary Table S10)— showed that among 11 mutations,²² four specific mutations in RNF213 (p.V1195M, p.D1331G, p.S2334N, and p.R4810K) exhibited a strong association with MMD, without significant heterogeneity across both fixed- and random-effect models (P ranged from 1×10^− 38 to 3×10^− 8, Supplementary Table S11). Notably, the p.R4810K mutation was identified as having the most pronounced association with MMD risk (OR = 119.35, 95% CI: 57.93-245.87, P = 2.05×10^− 38). Furthermore, another mutation, p.Q1133K, demonstrated statistical significance in pooled analyses under both fixed- and random-effect models (P = 8.91×10^− 8 and 8.60×10^− 7, respectively), signifying enhanced significance in comparison to P values from individual studies within the Japanese and Korean populations (P = 0.0023 and 6.40×10^− 6, respectively).

For the first time, we conducted a two-phases WEWAS on Korean adults with MMD and compared the WES data of these patients with those of disease-free healthy subjects, validating the mutational signal of RNF213. Additionally, potential mutations in GAA and CHMP6 were identified during the discovery and validation phases, as were de novo mutations in MUC4, FGFRL1, PRIM2, and MAP3K9. Intriguingly, WEWAS-driven mutations within the 17q25.3 region, referred to as MARS9, demonstrated an area under the curve between 0.8296 and 0.8380 in predicting MMD development. Nonetheless, no significant mutations associated with angiographic progression were discovered, despite MMD patients with angiographic progression harboring at least five missense mutations in RNF213.

The RNF213 polymorphism within the 17q25 region, notably the p.R4810K (rs112735431) mutation, which has been associated with MMD risk, has been acknowledged as a potential factor for MMD.^23–25 Wang et al. (2020) reported that mutation of the “A” allele in p.R4810K was closely associated with early onset of MMD and posterior cerebral artery involvement.²⁵ In the contrary of East-Asia, the R4810K variant of RNF213, which encodes a RING finger and two AAA + domains, was not often found in Caucasian MMD patients. Studies have indicated that carrying the “A” allele for p.R4810K is correlated with early onset of MMD and involvement of the posterior cerebral artery. This variation is less common in Caucasian MMD patients than in East-Asian patients, suggesting a diverse genetic predisposition based on ethnicity.²⁶ In Northern European adults with MMD, WES analysis revealed de novo mutations affecting nitric oxide metabolism but did not reveal the p.R4810K mutation.⁷ In our study, six out of the ten RNF213 gene missense mutations, p.K1034M, p.V1195M, p.E1272Q, p.D1331G, p.S2334N, and p.R4810K, were found to be genome-wide significant for adult MMD in Koreans. These findings imply that MMD may result from multiple mutations or interactions rather than from a single point mutation, such as p.R4810K. Although certain mutations, such as p.R4810K, were validated in our analysis, further studies are needed to understand the structural role and overall impact of other functional mutations in RNF213.

In addition to RNF213, our research points to significant mutations in two loci, GAA and CHMP6, as additional major contributors to MMD. The missense mutations p.G576S and p.E689K in GAA, identified for the first time in this WEWAS, play a crucial role in producing the acid alpha-glucosidase enzyme, which is pivotal for converting glycogen polymers to glucose in lysosomes. The absence or deficiency of GAA products causes glycogen accumulation and subsequent cellular damage.²⁷ Jia et al. reported that heterozygous mutations in GAA were related to cerebral inflammation in patients with Pompe disease and glycogen storage disease type II.²⁸ However, studies on how GAA affects MMD are rare. Glycogen accumulation caused by GAA deficiency leads to abnormal lysosomes, which impair autophagy.²⁹ Although the relationship between RNF213 variants and autophagic dysfunction is known, autophagic inhibition has been shown to impair endothelial function in human endothelial cells under oxygen–glucose deprivation conditions.³⁰ Youn et al. also reported that autophagic and mitophagic dysfunction in cerebrospinal fluid cells may be related to neurological outcomes in adult MMD patients.³¹ Based on these findings, we hypothesized that dysfunction due to GAA mutations may damage the cerebral endothelium and cause arterial abnormalities via autophagic impairment, resulting in an abnormal vascular network.

Charged multivesicular body protein 6, CHMP6 is responsible for encoding a chromatin-modifying protein and charged multivesicular body protein. Overexpression of CHMP6–GFP (green fluorescent protein) leads to the accumulation of transferrin receptors (TfRs) in the cytoplasm; TfRs play a crucial role in regulating the cellular iron supply by binding to transferrin.³² Furthermore, elevated levels of ubiquitinated membrane proteins, including the epidermal growth factor receptor, can be observed in cells expressing CHMP6–GFP. An increase in ubiquitinated proteins during endocytosis can potentially harm vascular smooth muscle cells (VSMCs). Dziewulska and Rafalowska (2008) suggested the potential involvement of aberrant ubiquitin-dependent endocytosis of the Notch 3 ligand in the injury of VSMCs in patients with cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy.³³ Additionally, RNF213 contains a unique class of E3 ubiquitin ligases. Pathogenic mutations in RNF213, which cluster within the composite E3 domain, could lead to abnormal ubiquitination interactions with substrates.³⁴ However, the role of CHMP6 mutations in the development of MMD has not been determined, necessitating further in vivo or in vitro research to elucidate the pathogenesis of CHMP6.

Our WEWAS-drivee models revealed that the MARS of polygenic architecture indicate multiple mutational effects or interactions at regional or chromosomal levels, such as 17q25.3, yielding a diagnostic accuracy exceeding 80% for MMD. Notably, the MARS9 model, which includes nine missense mutations, emphasizes that MMD is closed to a complex disorder rather than a Mendelian disorder. Polygenic risk assessment could prove beneficial for early disease detection and timely intervention, despite inherent challenges such as reduced effectiveness in the general population and reliance on statistical algorithms or existing knowledge.^35–37 In addition, polygenic risk contributes to be cost- and time-effective than other alternatives if direct functional mutations can be validated in the other independent group, reducing a cost of unnecessary screening for MMD. Furthermore, drive mutations can contribute to improving an informative strategy by avoiding unnecessary genotyping tests.³⁸ Our MARS models will definitely support a better information in clinical and precision medicines. In addition, the profits of pre-screening by a reliable MARS-based panel is likely to reduce potential side effects such as, for example, unnecessary screening, radiation exposure, overdiagnosis, and neuro-degenerative distress after onset symptoms.³⁹

Our study has several limitations. Primarily, the relatively small size of the datasets may impede the ability to draw definitive conclusions, even though our work represents the largest WES effort in MMD featuring discovery and validation phases, along with comparisons to genetic data from open cohorts. Second, while we identified several mutations in RNF213, GAA, and CHMP6 localized at 17q25.3, unlike the point mutations p.R4810K (RNF213) and p.D54K (CHMP6), LD-associated mutations are thought to influence gene expression rather than cause structural dysfunctions. Consequently, extensive research is needed to explore not only the individual impact of specific mutations but also their interactions and the resultant overall effect on MMD development, which could involve gene or allele-specific expression analysis, transcriptome-wide association studies, or in vivo investigations. Third, while we introduced Korean-specific MARS models for MMD diagnosis prediction, their broad application in the general clinical domain of neurodegenerative diseases necessitates further validation across independent cohorts and various ethnic groups. Finally, the absence of mutations specifically associated with angiographic progression in our study might reflect the limited number of documented patients. Therefore, expanded studies based on a larger cohort of MMD patients experiencing angiographic progression are essential.

We shed light on coding modifiers and mutations associated with adult MMD in the two-stage WES-based study, offering valuable insights into its genetic underpinnings. Our best-fitting MARS approach proposed herein holds promise for improving MMD diagnosis and understanding the potential for disease development. Further in-depth applications will be addressed to confirm our findings, to evaluate gene-gene or gene-environment interactions, and to determine the clinical relevance to genetic mechanisms for understanding MMD etiology.

Data and resource availability

The data that support the findings of this study was submitted as online supplemental material, and further detailed information is available upon request to the corresponding author. All genotype and phenotype resources of Korean populations are managed by “The First Korean Stroke Genetics Association Research” (The FirstKSGAR) Study constructed from the multi-centered hospital database (https://1ksgh.org/resources/data_accessibility). Futhermore, the original raw data contains potentially sensitive data including patient’s genomic profile and is access is restricted by the Ethics Committee and Institutional Review Board (IRB) of Hallym University Chuncheon Scared Hospital (https://chuncheon.hallym.or.kr/irb/index.asp). Interested researchers can submit an “The Agreement Form of Controlled Data Usage” to the Ethics Committee and IRB to request access to the data. Another WES and epidemiology data from disease-free 100 subjects (KARE study) was approved by the National Biobank of Korea, the Korea Disease Control and Prevention Agency, Republic of Korea (NBK-2022-068) (https://biobank.nih.go.kr/eng/cmm/main/mainPage.do). Access to all UKB data including sequencing and other information was acquired following project approval (application number 80385) (https://ams.ukbiobank.ac.uk/ams/resProjects).

Acknowledgments

This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health and Welfare, Republic of Korea (HR21C0198); a grant from the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education, Republic of Korea (2022R1l1A1A01053742); and the Hallym University Research Fund (HURF). The analysis of WES data from disease-free 100 subjects (KARE study) was approved by the National Biobank of Korea, the Korea Disease Control and Prevention Agency, Republic of Korea (NBK-2022-068). A license from the UK Biobank Resource was approved by Application Number 80385. The authors thank all the members of the FirstKSGAR consortium who has collected the biospecimens and clinical datasets.

Author contributions

E.P.H., J.P.J., and J.E.K. designed and managed this study. Analysis was done by E.P.H. and E.J.H. Sample preparation and data collection were done by D.H.Y., Y.C., K.M.K., S.H.L., W.S.C., and H.S.K. Drafting and reviewing of manuscript were done by E.P.H., E.J.H., D.H.Y., Y.C., K.M.K., S.H.L., W.S.C., H.S.K., J.P.J., and J.E.K.

Competing interests

The authors declare no competing interests.

Additional information

Supplementary information is available for this paper. Correspondence and requests for materials should be addressed to J.P.J.

Kim, S. K. et al. Elevation of CRABP-I in the cerebrospinal fluid of patients with Moyamoya disease. Stroke 34, 2835-2841, doi:10.1161/01.STR.0000100159.43123.D7 (2003).
Kleinloog, R., Regli, L., Rinkel, G. J. & Klijn, C. J. Regional differences in incidence and patient characteristics of moyamoya disease: a systematic review. J Neurol Neurosurg Psychiatry 83, 531-536, doi:10.1136/jnnp-2011-301387 (2012).
Cecchi, A. C. et al. RNF213 rare variants in an ethnically diverse population with Moyamoya disease. Stroke 45, 3200-3207, doi:10.1161/STROKEAHA.114.006244 (2014).
Wang, X. et al. Association of Genetic Variants With Moyamoya Disease in 13 000 Individuals: A Meta-Analysis. Stroke 51, 1647-1655, doi:10.1161/STROKEAHA.120.029527 (2020).
Jeon, J. P. et al. Genome-wide association study identifies novel susceptibilities to adult moyamoya disease. J Hum Genet 68, 713-720, doi:10.1038/s10038-023-01167-9 (2023).
Mukawa, M. et al. Exome Sequencing Identified CCER2 as a Novel Candidate Gene for Moyamoya Disease. J Stroke Cerebrovasc Dis 26, 150-161, doi:10.1016/j.jstrokecerebrovasdis.2016.09.003 (2017).
Wiedmann, M. K. H. et al. Whole-exome sequencing in moyamoya patients of Northern-European origin identifies gene variants involved in Nitric Oxide metabolism: A pilot study. Brain Spine 3, 101745, doi:10.1016/j.bas.2023.101745 (2023).
Backman, J. D. et al. Exome sequencing and analysis of 454,787 UK Biobank participants. Nature 599, 628-634, doi:10.1038/s41586-021-04103-z (2021).
Kim, J. E. et al. Clinical features of adult moyamoya disease with special reference to the diagnosis. Neurol Med Chir (Tokyo) 52, 311-317, doi:10.2176/nmc.52.311 (2012).
Jeon, J. P. & Kim, J. E. A Recent Update of Clinical and Research Topics Concerning Adult Moyamoya Disease. J Korean Neurosurg Soc 59, 537-543, doi:10.3340/jkns.2016.59.6.537 (2016).
Park, J. J., Kim, B. J., Youn, D. H., Choi, H. J. & Jeon, J. P. A Preliminary Study of the Association between SOX17 Gene Variants and Intracranial Aneurysms Using Exome Sequencing. J Korean Neurosurg Soc 63, 559-565, doi:10.3340/jkns.2019.0225 (2020).
Moon, S. et al. The Korea Biobank Array: Design and Identification of Coding Variants Associated with Blood Biochemical Traits. Sci Rep 9, 1382, doi:10.1038/s41598-018-37832-9 (2019).
Patel, R. K. & Jain, M. NGS QC Toolkit: a toolkit for quality control of next generation sequencing data. PLoS One 7, e30619, doi:10.1371/journal.pone.0030619 (2012).
Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754-1760, doi:10.1093/bioinformatics/btp324 (2009).
Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078-2079, doi:10.1093/bioinformatics/btp352 (2009).
DePristo, M. A. et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet 43, 491-498, doi:10.1038/ng.806 (2011).
Wang, K., Li, M. & Hakonarson, H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res 38, e164, doi:10.1093/nar/gkq603 (2010).
Chang, C. C. et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 4, 7, doi:10.1186/s13742-015-0047-8. (2015).
Pruim, R. J. et al. LocusZoom: regional visualization of genome-wide association scan results. Bioinformatics 26, 2336-2337, doi:10.1093/bioinformatics/btq419 (2010).
Ionita-Laza, I., Lee, S., Makarov, V., Buxbaum, J. D. & Lin, X. Sequence kernel association tests for the combined effect of rare and common variants. Am J Hum Genet 92, 841-853, doi:10.1016/j.ajhg.2013.04.015 (2013).
Magi, R. & Morris, A. P. GWAMA: software for genome-wide association meta-analysis. BMC Bioinformatics 11, 288, doi:10.1186/1471-2105-11-288 (2010).
Moteki, Y. et al. Systematic Validation of RNF213 Coding Variants in Japanese Patients With Moyamoya Disease. J Am Heart Assoc 4, e001862, doi:10.1161/JAHA.115.001862 (2015).
Liu, W. et al. Identification of RNF213 as a susceptibility gene for moyamoya disease and its possible role in vascular development. PLoS One 6, e22542, doi:10.1371/journal.pone.0022542 (2011).
Mertens, R. et al. The Genetic Basis of Moyamoya Disease. Transl Stroke Res 13, 25-45, doi:10.1007/s12975-021-00940-2 (2022).
Wang, Y. et al. Predictive role of heterozygous p.R4810K of RNF213 in the phenotype of Chinese moyamoya disease. Neurology 94, e678-e686, doi:10.1212/WNL.0000000000008901 (2020).
Guey, S. et al. Rare RNF213 variants in the C-terminal region encompassing the RING-finger domain are associated with moyamoya angiopathy in Caucasians. Eur J Hum Genet 25, 995-1003, doi:10.1038/ejhg.2017.92 (2017).
Roig-Zamboni, V. et al. Structure of human lysosomal acid alpha-glucosidase-a guide for the treatment of Pompe disease. Nat Commun 8, 1111, doi:10.1038/s41467-017-01263-3 (2017).
Jia, X. et al. GAA compound heterozygous mutations associated with autophagic impairment cause cerebral infarction in Pompe disease. Aging (Albany NY) 12, 4268-4282, doi:10.18632/aging.102879 (2020).
Taylor, K. M. et al. Dysregulation of multiple facets of glycogen metabolism in a murine model of Pompe disease. PLoS One 8, e56181, doi:10.1371/journal.pone.0056181 (2013).
Shin, H. S. et al. RNF213 variant and autophagic impairment: A pivotal link to endothelial dysfunction in moyamoya disease. J Cereb Blood Flow Metab, 271678X241245557, doi:10.1177/0271678X241245557 (2024).
Youn, D. H. et al. Autophagy and mitophagy-related extracellular mitochondrial dysfunction of cerebrospinal fluid cells in patients with hemorrhagic moyamoya disease. Sci Rep 13, 13753, doi:10.1038/s41598-023-40747-9 (2023).
Yorikawa, C. et al. Human CHMP6, a myristoylated ESCRT-III protein, interacts directly with an ESCRT-II component EAP20 and regulates endosomal cargo sorting. Biochem J 387, 17-26, doi:10.1042/BJ20041227 (2005).
Dziewulska, D. & Rafalowska, J. Is the increased expression of ubiquitin in CADASIL syndrome a manifestation of aberrant endocytosis in the vascular smooth muscle cells? J Clin Neurosci 15, 535-540, doi:10.1016/j.jocn.2007.06.022 (2008).
Ahel, J. et al. Moyamoya disease factor RNF213 is a giant E3 ligase with a dynein-like core and a distinct ubiquitin-transfer mechanism. Elife 9, e56185, doi:10.7554/eLife.56185. (2020).
Wald, N. J. & Old, R. The illusion of polygenic disease risk prediction. Genet Med. 21, 1705-1707, doi:10.1038/s41436-018-0418-5 (2019).
Lewis, C. M. & Vassos, E. Polygenic risk scores: from research tools to clinical instruments. Genome Med. 12, 44, doi:10.1186/s13073-020-00742-5. (2020).
McGrath, I. M., Consortium., I. E. G., Montgomery, G. W. & Mortlock, S. Polygenic risk score phenome-wide association study reveals an association between endometriosis and testosterone. BMC Med 21, 482, doi:10.1186/s12916-023-03184-z. (2023).
Mujwara, D., Kintzle, J., Di, D., P., Busby, G. B. & Bottà, G. Cost-effectiveness analysis of implementing polygenic risk score in a workplace cardiovascular disease prevention program. Front Public Health 11, 1139496, doi:10.3389/fpubh.2023.1139496 (2023).
Slunecka, J. L. et al. Implementation and implications for polygenic risk scores in healthcare. Hum Genomics 15, 46, doi:10.1186/s40246-021-00339-y. (2021).

No competing interests reported.

OnlineSupplementalMaterials.zip

Download PDF

Editor invited by journal
22 Aug, 2024
Submission checks completed at journal
21 Aug, 2024
First submitted to journal
21 Aug, 2024

You are reading this latest preprint version

Two-Stage Whole-Exome Sequencing Improves to Predict a Risk of Adult Moyamoya Disease in 369,570 Individuals

Status:

Version 1

Abstract

Figures

Introduction

Materials and Methods

Study subjects and UK Biobank resources

Whole-exome sequencing

Read alignment, base recalibration, and variant calling

Variant annotation and quality controls

Statistical analyses

Results

Clinical characteristics of study populations

Whole exome-wide association study

Integrative allele-based association of the mutant allele risk score in MMD

Gene-based association study

Discussion

Data and resource availability

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1