nature medicine Article https://doi.org/10.1038/s41591-023-02268-w Genomics of perivascular space burden unravels early mechanisms of cerebral small vessel disease Received: 10 October 2021 A full list of authors and their affiliations appears at the end of the paper Accepted: 15 February 2023 Published online: 17 April 2023 Perivascular space (PVS) burden is an emerging, poorly understood, Check for updates magnetic resonance imaging marker of cerebral small vessel disease, a leading cause of stroke and dementia. Genome-wide association studies in up to 40,095 participants (18 population-based cohorts, 66.3 ± 8.6 yr, 96.9% European ancestry) revealed 24 genome-wide significant PVS risk loci, mainly in the white matter. These were associated with white matter PVS already in young adults (N = 1,748; 22.1 ± 2.3 yr) and were enriched in early-onset leukodystrophy genes and genes expressed in fetal brain endothelial cells, suggesting early-life mechanisms. In total, 53% of white matter PVS risk loci showed nominally significant associations (27% after multiple-testing correction) in a Japanese population-based cohort (N = 2,862; 68.3 ± 5.3 yr). Mendelian randomization supported causal associations of high blood pressure with basal ganglia and hippocampal PVS, and of basal ganglia PVS and hippocampal PVS with stroke, accounting for blood pressure. Our findings provide insight into the biology of PVS and cerebral small vessel disease, pointing to pathways involving extracellular matrix, membrane transport and developmental processes, and the potential for genetically informed prioritization of drug targets. PVS are physiological spaces surrounding small vessel walls as they run waste (for example, beta-amyloid peptide) and brain fluid regulation, from the subarachnoid space through the brain parenchyma1–3. Dilation as part of the ‘glymphatic system’4,7. These processes were described of PVS observed on brain magnetic resonance imaging (MRI) is thought to increase during sleep2,4,7. Mounting evidence suggests a major role to be a marker of PVS dysfunction and, speculated from preclinical of PVS in cerebral injury. Several studies suggested associations of PVS studies, to reflect impairment of brain fluid and waste clearance2,4. burden (number of visible PVS on brain MRI) with stroke2,6,10, Alzhei- PVS increase in number with age and vascular risk factors, espe- mer’s disease pathology2 and cerebral amyloid angiopathy (CAA)11–13. cially hypertension2. They are associated with white matter hyperin- Post-stroke edema has been linked to post-stroke PVS enlargement14, tensities (WMH) of presumed vascular origin, lacunes and cerebral and in amyotrophic lateral sclerosis PVS dilation was observed and microbleeds2, all MRI features of cerebral small vessel disease (cSVD), perivascular fibroblast proteins were associated with survival15. a leading cause of stroke and dementia with no specific mechanistic PVS burden is highly heritable16. Identifying genetic risk variants treatment to date5,6. PVS are detected on brain MRI much earlier than for PVS could be a powerful tool to decipher underlying biological WMH, lacunes or cerebral microbleeds7, and are described as the earli- pathways. We conducted genome-wide association study (GWAS) est stage of cSVD lesions on neuropathology8. Their pathophysiology meta-analyses and whole-exome/whole-genome sequencing (WES/ is poorly understood6,9. WGS) studies of extensive PVS burden in up to 40,095 and 19,010 In experimental models, PVS appear to be important conduits older community participants, respectively. Given differential asso- for substrate delivery, flushing interstitial fluid, clearing metabolic ciations with risk factors and neurological traits2,10,17 and anatomical e-mail: h.adams@erasmusmc.nl; stephanie.debette@u-bordeaux.fr Nature Medicine | Volume 29 | April 2023 | 950–962 950 Article https://doi.org/10.1038/s41591-023-02268-w differences18, we ran analyses separately for white matter (WM)-PVS, 3 (INS-IGF2/IGF2, PRKAG2, LRP4/CKAP5); BG-PVS: 4 (SH3PXD2A, WNT3, basal ganglia (BG)-PVS and hippocampal (HIP)-PVS. We followed up ZMYND15, KCNRG/TRIM13/SPRYD7); and HIP-PVS: 1 (PDZRN4); Fig. 1 and identified risk loci in independent samples of young healthy adults Supplementary Table 7). VEGAS identified one additional gene (NSF) and older Japanese community participants and examined shared for BG-PVS (same locus as WNT3; Supplementary Table 7). All were in genetic determinants with other vascular and neurological traits. suggestive GWAS loci (P < 5 × 10−6; Supplementary Table 8). Leveraging tissue and cell-specific gene expression databases and drug Using LD-score regression, we estimated heritability at 11% for target libraries, we conducted extensive bioinformatics exploration of WM-PVS, 5% for BG-PVS and 8% for HIP-PVS (Methods and Supplemen- identified PVS risk loci. tary Table 9). We found moderate genetic correlation between BG-PVS and HIP-PVS (rg (SE) = 0.63 (0.14), P = 7.23 × 10−6), and modest genetic Results correlation of WM-PVS with BG-PVS (rg (SE) = 0.24 (0.12), P = 0.055) Genetic discovery and HIP-PVS (rg (SE) = 0.27 (0.09), P = 0.003). The genetic correlation Twenty-one population-based cohorts were included, of which 18 were of PVS in CHARGE and UKB was moderate to high for WM-PVS and for GWAS and 8 for whole-exome association studies (Supplementary HIP-PVS and weaker for BG-PVS (Supplementary Table 10). Associa- Table 1 and Methods). We tested associations of extensive PVS burden tions with genome-wide significant PVS loci were highly consistent with ~8 million single-nucleotide polymorphisms (SNPs) (minor allele between the UKB and CHARGE contributions and between the two frequency (MAF) ≥ 1%) in GWAS meta-analyses, gathering up to 40,095 dichotomous and the continuous PVS measures in UKB (Methods and participants (66.3 ± 8.6 yr, 51.7% female, 66.7% with hypertension; Supplementary Tables 11 and 12). In sensitivity analyses in two repre- Supplementary Tables 1–3). We dichotomized PVS burden based on sentative cohorts (UKB and 3C-Dijon), continuous and dichotomous cut-offs closest to the top quartile of PVS distribution to account for PVS measures were strongly correlated (Spearman’s ρ, 0.61–0.80; differences in PVS quantification methods, image acquisition and par- Supplementary Table 13). ticipant characteristics (Supplementary Tables 1 and 2 and Methods). In To increase statistical power, we conducted secondary multi- total, 9,607 of 39,822, 9,189 of 40,000, and 9,339 of 40,095 participants variate association analyses using Multi-Trait Analysis of GWAS had extensive PVS burden in WM, BG and hippocampus. (MTAG) (Methods), including summary statistics from GWAS of other The GWAS meta-analysis comprised 17 cohorts from the Cohorts cSVD markers (WMH volume, lacunes; Supplementary Table 14). We for Heart and Aging Research in Genomic Epidemiology (CHARGE) con- observed the highest gain in power for BG-PVS: ten additional loci sortium (N ≤ 11,511)19, with PVS quantification primarily on visual rating reached genome-wide significance, of which two also for HIP-PVS scales, and UK Biobank (UKB, N ≤ 28,655), with computational PVS (STN1, DEGS2/EVL). Two MTAG BG-PVS loci (CACNB2, NSF/WNT3) and quantification (Methods). Participants were of European (N = 38,871), one MTAG WM-PVS locus (VWA2) were not described before with any Hispanic (N = 717), East-Asian (N = 339) and African-American (N = 168) MRI marker of cSVD. Six loci showed greater significance in MTAG ancestry. We identified 22 independent genome-wide significant risk than with PVS, WMH volume or lacunes alone: at VWA2 (WM-PVS); loci for extensive PVS burden (WM-PVS: 19; BG-PVS: 2; HIP-PVS: 3 SH3PXD2A/STN1, COL4A2, CACNB2 and NSF/WNT3 (BG-PVS); and (2 shared with WM-PVS)) and two additional risk loci for WM-PVS in DEGS2/EVL (BG-PVS, HIP-PVS). Europeans only, leading to 24 independent signals (Table 1, Fig. 1, Using WES data and exome content of WGS data in 19,010 par- Extended Data Fig. 1 and Supplementary Fig. 1). There was no system- ticipants from UKB and the Brain Imaging, cognition, dementia, atic inflation of association statistics (Supplementary Table 4 and and next-generation genomics (BRIDGET) consortium (Methods Extended Data Fig. 1). and Supplementary Table 1), of whom 4,531, 4,424 and 4,497 had We performed conditional logistic regression using Genome-wide extensive PVS in WM, BG and hippocampus, we identified 19 variants Complex Trait Analysis (GCTA)-COJO (Methods) to seek independent in the chr1q25.3 locus associated with HIP-PVS, including two mis- association signals within genome-wide significant loci. Consistent sense variants (rs20563 and rs20558) and one splice donor insertion with linkage disequilibrium (LD)-based clumping, this identified two (rs34133998) in LAMC1 at P < 5 × 10−8, in strong LD with the GWAS independent signals at chr3p25.1 (WNT7A) and six at chr20q13.12 sentinel variant (Supplementary Table 15a). Gene-based burden (SLC13A3; Supplementary Fig. 1 and Supplementary Table 5a), four of tests exploring protein-modifying rare variants (MAF < 0.01) did which with low-frequency variants (Table 1). The six polymorphisms at not identify any gene-wide significant association (Supplementary chr20q13.12 generated eight haplotypes with haplotypic R2 (percent- Table 15b). age of haplotypic variability explained by observed genotypes) > 0.7 in the Three-City Dijon Study (3C-Dijon) cohort, of European ancestry Follow-up of findings across the lifespan and ancestries (N = 1,500; Supplementary Table 5b). The two common rs2425881-A We explored associations of WM-PVS and BG-PVS risk variants with and rs2425884-C alleles, in very low LD with each other (r2 (a measure these phenotypes in young adults (Internet-based Students’ HeAlth of correlation of alleles for two genetic variants) ~ 0.05, D′ (a pairwise r2 Research Enterprise (i-Share) study, N = 1,748, 22.1 ± 2.3 yr) and in older standardized for allele frequencies) ~ 0.50), generated a common hap- Japanese community-dwelling people (Nagahama study, N = 2,862, lotype that was more frequent in individuals with extensive WM-PVS 68.3 ± 5.3 yr; Methods). We used an artificial intelligence-based method than in those without (0.50 versus 0.47, odds ratio (OR) = 1.19 (95% to derive quantitative WM-PVS and BG-PVS burden (HIP-PVS not avail- confidence interval (95% CI), 0.99–1.43)). The effect of this haplo- able) and dichotomized it (top quartile versus the rest; Supplementary type was amplified by 1.7 in the presence of the rs112407396-T allele Table 2). In total, 67% of WM-PVS loci reached nominally significant (MAF = 0.03), which has a high probability of being a regulatory variant associations in at least one of the two follow-up cohorts (P < 0.05 in (HaploReg, GTex, RegulomeDB). Next, to account for allelic heteroge- i-Share and/or Nagahama), 43% of which at P < 1.09 × 10−3 (correcting neity between ancestries, we conducted cross-ancestry meta-analyses for the number of loci tested), with consistent directionality of effect with MR-MEGA (Methods). There were no loci showing high hetero- (a binomial test showed significant concordance of risk alleles; Sup- geneity in allelic effects across ancestries (PHet < 0.01) and reaching plementary Table 12b). In i-Share, 52% of WM-PVS risk variants were genome-wide significance (Supplementary Table 6). associated with WM-PVS (P < 0.05, of which 4 at P < 1.09 × 10−3; Table 2 Using MAGMA and VEGAS, we performed gene-based association and Supplementary Table 12a). A WM-PVS rescaled weighted genetic analyses in participants of European ancestry, testing the combined risk score (rwGRS) derived from European GWAS loci was associated association of variants within a gene with PVS (Methods). MAGMA iden- with WM-PVS in i-Share (OR = 1.16 (95% CI, 1.08–1.24), P = 5.89 × 10−6 and tified 28 gene-wide significant associations (P < 2.63 × 10−6), of which 12 β (SE) = 0.064 (0.007), P = 2.06 × 10−19 for dichotomous and continuous in 8 loci did not reach genome-wide significance in the GWAS (WM-PVS: measures; Supplementary Fig. 2). Although meta-regression suggested Nature Medicine | Volume 29 | April 2023 | 950–962 951 Article https://doi.org/10.1038/s41591-023-02268-w Table 1 | Genetic variants associated with high PVS burden Region SNP ALL chr:position EA/OA EAF Function Nearest Effect SEa Z- Dirb N ext-PVS/N P value EUR P value All Het P gene(s) (β)a scoreb total value PVS in white matter (WM-PVS) 20q13.12 rs6011998 20:45269867 C/T 0.95 intronic SLC13A3 0.087 0.009 10.65 ++++ 9,502/39,128 1.90 × 10−24 1.80 × 10−26 0.11 3p25.1 rs13079464 3:13822439 C/G 0.46 intergenic WNT7A 0.026 0.004 8.70 ++++ 9,614/39,822 8.64 × 10−17 3.41 × 10−18 0.59 20q13.12 rs2425884 20:45258292 C/T 0.57 intronic SLC13A3 0.029 0.004 8.63 +−+− 9,614/39,822 2.60 × 10 −18 6.02 × 10 −18 0.14 9q31.3 rs10817108c 9:113658671 A/G 0.21 intronic LPAR1 0.029 0.004 8.20 +++? 9,550/39,516 1.07 × 10−15 2.46 × 10−16 0.75 20q13.12 rs2425881 20:45255618 A/G 0.83 intronic SLC13A3 0.033 0.005 7.68 +−+? 9,496/39,087 2.02 × 10−15 1.59 × 10−14 0.06 3q21.2 rs3772833 3:124518362 G/A 0.83 intronic ITGB5, 0.032 0.005 7.67 +++? 9,496/39,087 2.15 × 10−13 1.76 × 10−14 0.39 UMPS 20q13.12 rs112407396 20:45276381 T/A 0.03 intronic SLC13A3 0.078 0.012 6.91 +??? 8,426/34,530 4.81 × 10−12 4.81 × 10−12 1.00 1q41 rs10494988 1:215141570 C/T 0.63 intergenic CENPF, 0.021 0.004 6.54 ++++ 9,614/39,822 2.23 × 10 −10 6.03 × 10 −11 0.69 KCNK2 20q13.12 rs72485816d 20:45314435 T/C 0.96 UTR3 TP53RK, 0.059 0.010 6.45 ++?− 9,114/37,342 1.47 × 10−10 1.12 × 10−10 0.87 SLC13A3 15q25.3 rs8041189 15:85686327 G/A 0.70 intergenic PDE8A 0.022 0.004 6.44 +−?? 9,486/39,315 7.30 × 10−11 1.24 × 10−10 0.31 3p25 rs4685022 3:13832611 G/A 0.65 intergenic WNT7A 0.019 0.004 6.40 +++? 9,576/39,654 2.36 × 10−09 1.58 × 10−10 0.11 2p16.1 rs7596872 2:56128091 C/A 0.90 intronic EFEMP1 0.033 0.006 6.31 +−?? 9,333/38,442 1.00 × 10−10 2.80 × 10−10 0.11 17q21.31 rs1126642 17:42989063 C/T 0.96 exonic GFAP 0.051 0.009 6.23 +?+? 9,119/37,466 6.19 × 10−10 4.67 × 10−10 0.72 3q29 rs687610d 3:193515781 G/C 0.43 intergenic OPA1 0.021 0.004 6.20 +++− 9,614/39,822 2.99 × 10−10 5.81 × 10−10 0.76 6p25.2 rs4959689 6:2617122 C/A 0.58 intergenic C6orf195 0.020 0.004 6.03 ++++ 9,582/39,695 3.37 × 10−09 1.63 × 10−09 1.00 20q13.12 rs56104388 20:45302135 T/C 0.99 intronic SLC13A3 0.101 0.017 5.85 +??? 7,626/30,916 4.80 × 10−09 4.80 × 10−09 1.00 11q13.3 rs12417836 11:70089700 T/C 0.07 intergenic FADD, 0.034 0.007 5.58 +−+? 9,464/38,960 1.56 × 10 −08 2.47 × 10−08 0.40 PPFIA1 8p11.21 rs2923437d 8:42425399 A/C 0.41 intergenic SMIM19, 0.018 0.004 5.49 ++−− 9,614/39,822 4.73 × 10−08 4.08 × 10−08 0.14 CHRNB3, SLC20A2 6p25.3 rs1922930 6:1364691 C/A 0.12 intergenic FOXQ1, 0.029 0.006 5.47 ++?? 9,406/38,748 3.60 × 10−08 4.62 × 10−08 0.48 FOXF2 19p13.11 rs2385089 19:18550434 A/C 0.74 intergenic ISYNA1, 0.023 0.005 5.49 +++− 9,614/39,822 4.14 × 10−08 5.73 × 10−08 0.57 ELL, LRRC25c 7q33 rs10954468 7:134434661 C/A 0.40 intergenic BPGM, 0.019 0.004 5.52 +−?+ 9,524/39,483 3.39 × 10−08 8.79 × 10−08 0.29 CALD1c PVS in basal ganglia (BG-PVS) 2q33.2 rs4675310d 2:203880834 A/G 0.87 intronic NBEAL1, 0.027 0.005 5.92 ++?? 9,011/39,243 2.71 × 10−09 3.27 × 10−09 0.64 ICA1L 3q26.31 rs6769442 3:171565463 G/A 0.75 intronic TMEM212 0.020 0.004 5.74 ++?+ 9,101/39,788 1.68 × 10−08 9.34 × 10−09 0.96 PVS in hippocampus (HIP-PVS) 1q25.3 rs10797812d 1:182984597 A/G 0.54 intergenic SHCBP1L, 0.027 0.004 7.84 ++++ 9,399/40,095 1.67 × 10−14 4.39 × 10−15 0.68 LAMC1 2p16.1 rs78857879d 2:56135099 G/A 0.90 intronic EFEMP1 0.038 0.006 6.43 +??? 9,033/38,008 8.20 × 10−11 1.31 × 10−10 1.00 1q41 rs6540873 1:215137222 A/C 0.62 intergenic CENPF, 0.020 0.004 5.95 +−−+ 9,399/40,095 1.38 × 10−09 2.72 × 10−09 0.11 KCNK2 EA, effect allele; OA, other allele; EAF, effect allele frequency; Z-scores of the sample size-weighted cross-ancestry meta-analysis are represented, except for the two SNPs reaching genome-wide significance in Europeans only (rs2385089, rs10954468) for which the European meta-analysis Z-score is reported; dir, the association direction of the EA with the phenotype (extensive PVS burden versus the rest) for European, Hispanic, Asian and African-American ancestry studies, in this order; N ext-PVS, the number of participants with extensive PVS burden in the cross-ancestry meta-analysis, in each location (WM-PVS, BG-PVS, HIP-PVS); N total, the total number of participants in the cross-ancestry meta-analysis; P value EUR, P value in the European meta-analysis; P value All, P value in the cross-ancestry meta-analysis; Het P value corresponds to the heterogeneity P value in the meta-analysis (except for rs2385089 and rs10954468 for which the European meta-analysis Het P value is reported); P values for genome-wide significant loci (P < 5 × 10−8) are in bold; PVS GWAS analyses in individual cohorts were adjusted for age, sex and intracranial volume (or brain parenchymal fraction for ASPS), principal components of population stratification, and study site. aFrom cross-ancestry inverse variance-weighted meta-analysis. bFrom cross-ancestry Z-score-based meta-analysis. cGenome-wide significant association in Europeans only. dFor these loci, the lead SNP was different in the European meta-analysis (Cross-ancestry lead SNP → European lead SNP: rs72485816→rs6094423; rs687610→rs6444747; rs2923437→rs62509329; rs4675310→rs140244541; rs10797812→rs2022392; rs78857879→rs7596872); the P value corresponding to the European lead SNP is reported under "P value EUR" (r² > 0.50 between the European and cross-ancestry lead SNPs at these loci). larger effect sizes at younger ages for lead variants at OPA1 and SLC13A3, risk score (wGRS) combining 14 independent loci (1000 Genomes pro- differences were not significant after removing the much younger ject (1000G) Japanese reference panel) was associated with WM-PVS in i-Share cohort (Supplementary Fig. 3). In Nagahama, out of 15 available Nagahama (OR = 1.18 (95% CI, 1.13–1.24), P = 5.68 × 10−13 and β (SE) = 0.01 WM-PVS risk loci (six were rare or monomorphic), eight loci (53%) were (0.001), P = 7.18 × 10−18 for dichotomous and continuous measures). associated with continuous PVS burden at P < 0.05, of which four at Although HIP-PVS data were not available in the follow-up cohorts, P < 1.09 × 10−3 and one at genome-wide significance (at SLC13A3; Table 2 two of the three HIP-PVS loci were shared with WM-PVS and replicated and Supplementary Table 12a). A European WM-PVS weighted genetic with that phenotype. Nature Medicine | Volume 29 | April 2023 | 950–962 952 Article https://doi.org/10.1038/s41591-023-02268-w a b KCNRG SPRYD7 TRIM13 WM-PVS PD ZRN chr13 chr14 4 12 chr1 chr 5 5 D1 ch YN r16 LR AP5 CK ZM r11 T3 P4 IN N S- SH IG ch W IG 3P F2 F2 XD ch 2A r 17 ch 10 r18 1 r NA ch ISY ch LPA r1 R1 9 3 13A SLC RK chr TP532A10 chr9 SLC 2 0 BG-PVS chr21 chr22 chr8 SMIM19 SLC20A2 2 PRKAG 1 chr1 CALD chr 7 LAM C1 HIP-PVS ch r2 r6 ch 5 chr f19 IC DR F 5 W AR L1 A1 12 or C BEA L C6 N 3 chr chr4 UMP 5 ITGB WM-PVS S BG-PVS HIP-PVS Fig. 1 | Illustration of extensive PVS burden and results of the cross-ancestry analyses results, the middle circle to the results of the MTAG analysis and the PVS GWAS meta-analysis, MTAG analysis and gene-based test. a, Extensive PVS outer circle to gene-based test results. Results for WM-PVS are in blue, for BG-PVS burden (arrows) in WM (top, WM-PVS), BG (middle, BG-PVS) and hippocampus in purple and for HIP-PVS in green. The gray line corresponds to the genome-wide (bottom, HIP-PVS) on T1-weighted axial magnetic resonance images. b, Circular significance threshold (P = 5 × 10−8, two-sided, correcting for multiple testing at Manhattan plot. The inner circle corresponds to the cross-ancestry GWAS meta- the genome-wide level). Clinical correlates of identified PVS loci intracerebral hemorrhage (ICH). Genetic correlations were consist- We examined whether PVS risk loci (lead and proxy variants with r2 > 0.9) ent in secondary analyses conducted separately in CHARGE and UKB were associated with MRI markers of brain aging, putative risk factors (Supplementary Table 9). (vascular risk factors and sleep patterns) and common neurological Third, we used two-sample Mendelian randomization (MR) to seek diseases (stroke, Alzheimer’s disease, Parkinson’s disease), using the evidence for a causal association of putative risk factors with PVS bur- largest published GWAS (Methods). Of 24 independent PVS risk loci, den and of PVS burden with neurological diseases, using generalized five (21%) were significantly (P < 3.3 × 10−5) associated with WMH volume summary-data-based MR (GSMR), and confirming significant associa- and five (21%) with blood pressure traits (in the same and opposite direc- tions (P < 1.19 × 10−3) with RadialMR, TwoSampleMR and MR-CAUSE tions; Fig. 2). Colocalization analyses suggested a shared causal variant (Methods). Genetically determined higher SBP and DBP were consist- for two-thirds of these associations (posterior probability for a shared ently associated with BG-PVS, HIP-PVS and WM-PVS, although for causal variant, PP4 > 0.75; Supplementary Table 16). Sixteen PVS loci WM-PVS the association with SBP was only nominally significant in (67%) did not show any association with vascular or neurological traits, RadialMR (Supplementary Table 18 and Extended Data Fig. 2). There thus pointing to pathways that do not seem mediated by established was no evidence for reverse causation using MR-Steiger, but some risk factors (Methods and Supplementary Tables 16 and 17). evidence of residual pleiotropy after removal of outlier variants for Second, we explored genetic correlations of PVS burden with SBP and DBP (RadialMR), with significant evidence for a causal model in the same traits using LD-score regression (Methods, Fig. 3 and Sup- MR-CAUSE for BG-PVS. Genetic liability to BG-PVS and HIP-PVS derived plementary Table 9). We observed significant (P < 7.9 × 10−4) genetic from a multi-trait analysis accounting for other MRI markers of cSVD correlation of BG-PVS with larger WMH and caudate nucleus volumes, (MTAG) was associated with an increased risk of any stroke, ischemic and of HIP-PVS with larger hippocampal volume. BG-PVS and HIP-PVS stroke and small vessel stroke (SVS) for BG-PVS, and SVS for HIP-PVS, showed significant genetic correlation with higher systolic blood pres- suggesting that shared pathways between PVS, WMH and lacunes sure (SBP), diastolic blood pressure (DBP), any stroke and ischemic may be causally associated with stroke (Supplementary Table 18 and stroke, and nominally significant genetic correlation with (deep) Extended Data Fig. 3). In multivariable MR analyses accounting for Nature Medicine | Volume 29 | April 2023 | 950–962 953 Article https://doi.org/10.1038/s41591-023-02268-w Table 2 | Association of genome-wide significant WM- and BG-PVS risk loci with PVS burden across the lifespan (i-Share study, N = 1,748) and across ancestries (Nagahama study, N = 2,862) GWAS meta-analysis i-Share (dichotomous) i-Share (continuous) Nagahama (dichotomous) Nagahama (continuous) SNP chr:position EA/OA Nearest OR (95% P value β (SE) P value OR (95% P value β (SE) P value gene(s) CI) CI) PVS in white matter (WM-PVS) rs6011998 20:45269867 C/T SLC13A3 1.26 0.28 0.164 4.20 × 10−05a 1.69 1.22 × 10−05a 0.037 6.21 × 10−07a (0.83–1.92) (0.04) (1.33–2.13) (0.008) rs13079464 3:13822439 C/G WNT7A 1.12 0.29 0.014 0.50 1.16 0.11 0.015 1.50 × 10−02 (0.91–1.40) (0.02) (0.97–1.40) (0.006) rs2425884 20:45258292 C/T SLC13A3 1.18 0.13 0.077 2.98 × 10−04a 1.29 3.48 × 10−03 0.026 1.77 × 10−06a (0.95–1.45) (0.02) (1.09–1.52) (0.005) rs10817108 9:113658671 A/G LPAR1 0.90 0.44 0.058 2.23 × 10−02 1.18 0.07 0.017 4.10 × 10−03 (0.69–1.17) (0.03) (0.98–1.43) (0.006) rs2425881 20:45255618 A/G SLC13A3 1.47 1.40 × 10−02 0.063 2.62 × 10−02 1.18 3.66 × 10−02 0.014 4.68 × 10−03 (1.03–2.01) (0.03) (1.01–1.37) (0.005) rs3772833 3:124518362 G/A ITGB5, 1.22 0.21 0.006 0.85 1.06 0.51 0.008 0.16 UMPS (0.89–1.66) (0.03) (0.88–1.29) (0.006) rs112407396 20:45276381 T/A SLC13A3 1.47 0.24 0.147 3.13 × 10−02 NA NA NA NA (0.77–2.78) (0.07) rs10494988 1:215141570 C/T CENPF, 1.18 0.14 0.079 1.94 × 10−04a 1.01 0.90 −0.002 0.67 KCNK2 (0.95–1.47) (0.02) (0.86–1.18) (0.005) rs72485816 20:45314435 T/C TP53RK, 1.01 0.98 0.093 0.095 1.32 2.83 × 10−03 0.033 1.91 × 10−08b SLC13A3 (0.56–1.80) (0.06) (1.10–1.59) (0.006) rs8041189 15:85686327 G/A PDE8A 1.14 0.30 0.041 0.073 1.67 0.14 0.046 2.40 × 10−02 (0.89–1.44) (0.02) (0.84–3.33) (0.021) rs4685022 3:13832611 G/A WNT7A 1.12 0.34 0.023 0.31 1.15 0.10 0.010 0.075c (0.88–1.42) (0.02) (0.97–1.36) (0.005) rs7596872 2:56128091 C/A EFEMP1 1.65 1.14 × 10−02 0.089 1.10 × 10−02 NA NA NA NA (1.10–2.46) (0.03) rs1126642 17:42989063 C/T GFAP 1.13 0.67 0.127 1.27 × 10−02 1.35 0.11 0.033 9.88 × 10−07a (0.65–1.97) (0.05) (1.09–1.67) (0.007) rs687610 3:193515781 G/C OPA1 1.46 4.88 × 10−04a 0.109 1.29 × 10−07a 0.95 0.59 0.006 0.28 (1.18–1.80) (0.02) (0.81–1.13) (0.005) rs4959689 6:2617122 C/A C6orf195 1.10 0.37 0.022 0.30 NA NA 0.024 0.34c (0.89–1.37) (0.02) (0.026) rs56104388 20:45302135 T/C SLC13A3 1.30 0.67 0.274 1.47 × 10−02 NA NA NA NA (0.40–4.24) (0.11) rs12417836 11:70089700 T/C FADD, 0.99 0.99 0.045 0.29 0.87 0.40 −0.002 0.84 PPFIA1 (0.64–1.56) (0.04) (0.62–1.21) (0.011) rs2923437 8:42425399 A/C SMIM19, 0.98 0.88 0.047 2.60 × 10−02 1.11 0.23 0.008 0.11 CHRNB3, (0.78–1.23) (0.02) (0.94–1.31) (0.005) SLC20A2 rs1922930 6:1364691 C/A FOXQ1, 0.93 0.70 0.035 0.33 NA NA NA NA FOXF2 (0.65–1.33) (0.04) rs2385089 19:18550434 A/C ISYNA1, 1.20 0.14 0.049 0.057 NA NA NA NA ELL, (0.94–1.53) (0.03) LRRC25 rs10954468 7:134434661 C/A BPGM, 1.08 0.50 0.033 0.13 NA NA NA NA CALD1 (0.86–1.36) (0.02) PVS in basal ganglia (BG-PVS) rs4675310d 2:203880834 A/G NBEAL1, 1.07 0.61 0.01 0.78 1.88 0.26 0.046 0.11 ICA1L (0.81–1.41) (0.04) (0.63-5.60) (0.03) rs6769442 3:171565463 G/A TMEM212 1.10 0.37 0.03 0.33 1.04 0.82 0.008 0.35 (0.87–1.40) (0.03) (0.71-1.53) (0.01) NA in the Nagahama Study correspond to variants that are rare (MAF < 1%: rs7596872; rs1922930; rs10954468) or monomorphic (rs112407396; rs56104388) in East Asians, or not available including in EAS 1000G data (rs2385089). Analyses were adjusted for age, sex and intracranial volume, principal components of population stratification in the i-Share and Nagahama studies, and additionally adjusted for study center in the Nagahama study. In the Nagahama study, when the lead SNP from the PVS GWAS meta-analysis was not present, we used a tag SNP with r2 > 0.80 using the 1000G Japanese reference panel. SNPs or tag SNPs (r² > 0.80, 1000G EAS) with a P < 0.05 are in bold. aSNPs with a P < 1.09 × 10−3 (Bonferroni correction for 23 independent loci and two PVS locations). bSNPs reaching genome-wide significance. cThe tag SNP (r² > 0.80) is nominally significant: rs4685022 (r² = 0.81 with rs934448, 1000G EAS), P = 0.048; rs4959689 (r² = 0.83 with rs1772953, 1000G EAS), P = 0.02. dThe lead SNP for this locus is not present in the Nagahama study; we used a tag SNP (rs150788469, r² = 1.0 with rs4675310) where the A allele of rs4675310 is in phase with the G allele of rs150788469. Nature Medicine | Volume 29 | April 2023 | 950–962 954 Article https://doi.org/10.1038/s41591-023-02268-w a PVS burden SLC13A3* PDE8A C6orf195 WNT7A** GFAP PPFIA1 ITGB5 OPA1 CALD1† TMEM212 SMIM19-CHRNB3 LPAR1‡ ISYNA1†‡ FOXF2‡ LAMC1 EFEMP1‡ KCNK2‡ NBEAL1- ICA1L‡ Blood pressure WMH traits volume BMI, LDL b B3 1 C RN K2 2 AM L XF A1 N H LL -L FO IC C -C 1L -E -K 1- 19 1- P1 A1 BP PF AL XQ 1 IM EM AR YN C EN BE SM FO SH LP EF IS C N 1.10 × 10–5 2.81 × 10–5 SBP DBP 4,53 × 10–6 2.12 × 10–5 –15 PP 5.43 × 10–9 3.17 × 10–12 1.31 × 10–14 1.56 × 10 1.50 × 10–5 BMI 8.29 × 10–9 LDL –24 1.72 × 10–5 5.44 × 10–6 1.73 × 10–6 3.81 × 10 2.23 × 10–13 WMH volume Fig. 2 | Association of PVS loci with vascular risk factors and other MRI P values are provided in Supplementary Table 16. b, Direction of association and markers of cSVD. a, Venn diagram displaying significant association of genome- level of significance of pleiotropic SNPs displayed in a: in red when the risk allele wide significant risk loci for PVS burden with vascular risk factors and other MRI for extensive PVS burden is positively associated with the trait, in blue when the markers of cSVD: in italics for BG-PVS; underlined for HIP-PVS; underlined and in PVS risk allele is negatively associated with the trait (unexpected direction), in bold for HIP- and WM-PVS; all others for WM-PVS only (P < 3.3 × 10−5, two-sided, dark red and dark blue for genome-wide significant associations and in light correcting for multiple testing (21 independent phenotypes, 3 PVS locations and red and light blue for significant association after multiple-testing correction 24 independent loci)); *6 independent loci; **2 independent loci; †genome-wide (P < 3.3 × 10−5, two-sided, correcting for multiple testing (21 independent significant in Europeans only; ‡in colocalization analyses the posterior probability phenotypes, 3 PVS locations and 24 independent loci)). PP, pulse pressure; PP4 was higher than 75% for these loci (only with WMH at NBEAL1-ICA1L). Exact BMI, body mass index; LDL, LDL cholesterol. Nature Medicine | Volume 29 | April 2023 | 950–962 955 Article https://doi.org/10.1038/s41591-023-02268-w a l es l ro ro n et tio te te e ab ep es p es ra id ee ol di du le er ol sl ch ts yc ch 2 p ng pe or l ee DL P P ig L I BM DB SB LD Sh PP Lo Ty Tr Sl H rg WM – PVS 1.0 0.5 BG – PVS 1.0 × 10–4 4.1 × 10–5 0.02 ** ** * 0 –0.5 HIP – PVS 7.0 × 10–4 1.0 × 10–4 0.02 ** ** * –1.0 b ke ro st ke se e e ro ke as ea as st ro se is se ke lic st ld di di ro bo ry se ’s st s ke n’ te er H m es H ic C so ro ar m IC oe lv em rI st ei in e di ep al ba rg zh rk ch y H ar Sm An De Lo Pa La IC Al Is C rg WM – PVS 1.0 0.5 BG – PVS 3.9 × 10–7 2.6 × 10–6 0.002 0.024 0.004 ** ** * * * 0 –0.5 HIP – PVS 2.0 × 10–4 0.002 0.02 ** * * –1.0 c ts rc fa e s um in ne n ol cu ai lv br la e pa m ed ed ns m lu a be in in al en m ca e vo ef ef t d m du da m po yg -d -d H cu ta lli au M RI RI ip Am Ac Pu Pa W M M H C rg WM – PVS 1.0 0.5 –5 –4 BG – PVS 0.028 0.02 3.5 × 10 4.0 × 10 0.005 * * ** ** * 0 –0.5 HIP – PVS 0.021 7.4 × 10–6 * ** –1.0 Fig. 3 | Genetic correlations of extensive PVS burden with risk factors, correcting for 21 independent phenotypes and the three PVS locations); neurological diseases and other MRI markers of brain aging. a–c, Genetic full results are provided in Supplementary Table 9. Larger colored squares correlation using LD-score regression of extensive PVS burden with putative risk correspond to more significant P values and the colors represent the direction factors (a), neurological diseases (b) and other MRI markers of brain aging (c); of the genetic correlation (positive in red, negative in blue). HDL, high-density two-sided exact P values are provided for nominally significant results (*P < 0.05) lipoprotein; amygdala, accumbens (nucleus), caudate (nucleus), pallidum, and and significant results after multiple-testing correction (**P < 7.9 × 10−4, putamen correspond to the volumes of these subcortical structures. Nature Medicine | Volume 29 | April 2023 | 950–962 956 Article https://doi.org/10.1038/s41591-023-02268-w SBP and DBP, genetic liability to BG-PVS and HIP-PVS was significantly genes in vascular disease drugs, including simvastatin, vincamine and associated with an increased risk of any stroke, ischemic stroke and SVS macitentan (Extended Data Fig. 8). (Supplementary Table 19). Discussion Functional exploration of identified PVS loci In up to 40,095 participants from older population-based cohorts, we Using MAGMA and VEGAS2Pathway (Methods), we identified signifi- identified 24 genome-wide significant risk loci for extensive PVS bur- cant enrichment of PVS loci in pathways involved in extracellular matrix den, predominantly for WM-PVS, and six additional loci after account- (ECM) structure and function, lymphatic endothelial cell differen- ing for other MRI markers of cSVD. Consistent with distinct risk factor tiation, cell motility and thyroid hormone transport (Supplementary profiles2,10, the genetic architecture of PVS differed across PVS loca- Tables 20 and 21). tions, with WM-PVS showing the highest heritability and low genetic Genes closest to PVS lead risk variants were significantly enriched correlation with BG-PVS and HIP-PVS1,2,16. In line with the hypothesis in genes mutated in Online Mendelian Inheritance in Man (OMIM) that PVS is a marker of cSVD, moderate to high genetic correlation syndromes associated with leukodystrophy, leukoencephalopathy was observed with other MRI markers of cSVD, primarily for BG- and or WMH, with a 20-fold enrichment in genes containing an intragenic HIP-PVS. Pathway analyses highlight ECM structure and function, lead variant. This enrichment was 30-fold when focusing on WM-PVS known to play an important role in cSVD5,20,21, and several loci include loci only, comprising several genes involved in early-onset leukod- genes involved in the matrisome (ECM and associated proteins), per- ystrophies: GFAP (chr17q21.31), mutations of which cause Alexander turbations of which were proposed as a convergent pathologic pathway disease, a rare neurodegenerative disorder of astrocytes leading to in cSVD (LAMC1, EFEMP1, COL4A2, SH3PXD2A, VWA2)5,21. Several PVS psychomotor regression and death; SLC13A3 (chr20q13.12), caus- risk loci (at FOXF2, EFEMP1, KCNK2 and NBEAL1-ICA1L) are known risk ing acute reversible leukoencephalopathy with increased urinary loci for other cSVD features (WMH, SVS)5,22,23, and mutations in two alpha-ketoglutarate; and PNPT1 (chr2p16.1), causing Aicardi–Goutières MTAG genes cause monogenic SVD (at COL4A1-COL4A2 and STN1)24,25. syndrome and cystic leukoencephalopathy (Methods, Extended Data PVS have been described early in life7,26, but their clinical signifi- Fig. 4 and Supplementary Table 22). Although several genes near PVS cance at young ages is unknown. Our results suggest shared molecular lead risk variants were described to be involved in glioma, we found no mechanisms underlying PVS in young and older age. This corroborates significant enrichment for glioma genes (Methods). recently described associations of WMH risk variants with changes in To seek evidence for a causal implication of specific genes and MRI-detected WM microstructure at age 20 yr (ref. 5). The significant variants, we performed transcriptome-wide association studies enrichment of PVS risk loci in genes involved in early-onset leukodys- (TWAS) using TWAS-Fusion (Methods), with European PVS GWAS trophies and expressed in fetal brain vascular endothelial cells supports summary statistics and the GTEx v7 multi-tissue (RNA sequencing) involvement of developmental processes. In spontaneously hyperten- database, focusing on brain, vascular and blood tissues. We found sive stroke-prone rats, closely modeling cSVD, intrinsic endothelial cell 36 transcriptome-wide significant expression–trait associations for dysfunction was observed at birth, including reduced tight junctions, WM-PVS, 25 for BG-PVS and seven for HIP-PVS that were significant in as well as altered oligodendrocyte maturation and myelination27. At the colocalization analyses (TWAS-COLOC), providing evidence of a shared most significant WM-PVS locus in young adults, OPA1 harbors muta- causal variant between the corresponding gene expression and PVS tions causing autosomal-dominant optical atrophy, sometimes associ- (Supplementary Table 23). Most genes with significant expression–trait ated with multiple sclerosis-like illness, parkinsonism and dementia28, associations (12) were in genome-wide significant PVS risk loci: eight and endothelial OPA1 plays an important role in developmental angio- genes in five WM-PVS GWAS loci (C6orf195, ITGB5, LPAR1, LRRC25, genesis29. These observations corroborate epidemiological associa- RP11-71H17.9, SLC20A2, SMIM19, UMPS), two genes in one BG-PVS GWAS tions of early-life factors with cSVD severity in older age30. locus (ICA1L, NBEAL1) and two genes in an HIP-PVS GWAS locus (LAMC1 The present effort has the largest East-Asian contribution com- and RP11-181K3.4), while nine were outside GWAS loci, requiring con- pared with other large GWAS of MRI-defined phenotypes31,32, with firmation (Fig. 4). TWAS-COLOC signals were mostly observed in brain over half of available WM-PVS loci reaching nominally significant, tissues (17 genes), but also in vascular tissues (ten genes) and blood directionally consistent associations in the Japanese follow-up study. (two genes). The prevalence of cSVD is higher in East-Asian than European popula- To identify enrichment in specific brain cell types, we used a tions33. Our results are an important initial step to establish the gen- recently developed pipeline combining three cell type enrichment eralizability of cSVD genetic associations across ancestries. Efforts to methods, stratified LD-score, MAGMA and H-MAGMA (Supplemen- further enhance the non-European contribution to MRI cSVD genomic tary Table 24). We observed significant enrichment in brain vascular studies, including in populations of African-ancestry in whom cSVD is endothelial cells for all PVS locations, based on a human single-cell atlas also more frequent34, are of paramount importance. of fetal gene expression, and in pericytes and astrocytes for WM-PVS The combination of PVS GWAS findings with TWAS and WES/WGS (Supplementary Tables 24 and 25). strongly supports putative causal genes. WM-PVS associates with lower We explored brain expression patterns from development to adult- LPAR1 expression in vascular tissues. LPAR1 (chr9q31.3), expressed in hood of genes nearest to PVS loci, prioritizing TWAS-COLOC genes oligodendrocytes, encodes a receptor for lysophosphatidic acid, an (Methods). Several genes showed important variations in expression extracellular signaling small lipid, and is involved in postnatal myeli- levels throughout the life course, some peaking in the prenatal period nation and functional connectivity across brain regions35. An LPAR1 (for example, LAMC1, UMPS), suggestive of developmental mechanisms antagonist was found to attenuate brain damage after transient arterial (Extended Data Fig. 5 and Supplementary Fig. 4). occlusion, by decreasing inflammation36, and LPAR1 modulation may Finally, we conducted an exploratory search for enrichment of also impact neural regeneration37. Several drugs targeting LPAR1 are PVS genes in targets of drugs validated in other indications (Meth- available (for example, the antidepressant mirtazapine38) or in devel- ods). We found significant enrichment of BG-PVS genes in targets for opment39. WNT7A (chr3p25.1) encodes a secreted signaling protein anti-infectives, driven by CRHR1 (chr17q21.31, target for telavancin), that targets the vascular endothelium, and was implicated in brain and for diseases of the nervous system, driven by MAPT (chr17q21.31, angiogenesis and blood brain barrier regulation40. Loss of Wnt7a/b target for davunetide); and of HIP-PVS genes in targets for ear disease function in mice results in severe WM damage41. drugs, driven by SERPIND1 (chr22q11.21, target for sulodexide, also WM-PVS was associated with lower ITGB5 (chr3q21.2) expression used for venous thrombosis prevention; Extended Data Figs. 6 and 7). in whole blood. ITGB5 encodes a beta subunit of integrin, and plays We also observed significant enrichment of TWAS-significant HIP-PVS a central role in monogenic SVD42. Higher ITGB5 plasma levels were Nature Medicine | Volume 29 | April 2023 | 950–962 957 Article https://doi.org/10.1038/s41591-023-02268-w Genes t t 3 –2 18.2 –2 18.1 8. .4 20 .9 § 5 § IM G1 K3 7 G G UM 9 § 1 § BE A2 RP 5 t S L 1 H1 SL A 2 59 TR 5 9 § 59 ŧ SM A 2 1 § § § L § 81 ŧ 19 37 13 1 –2 AL –7 1 2 1 –1 ŧ B5 C T3 PS rf O IM RC RC AR 11 11 11 11 11 A1 M DN 6o C C SF N G RP RP RP RP LA LR LR LP IC W IT C N N N GTEX – nerve tibial * * ** ** ** ** ** ** GTEX – brain amygdala ** ** GTEX – brain anterior cingulate cortex BA24 ** GTEX – brain caudate BG ** ** ** GTEX – brain cerebellar hemisphere ** ** ** GTEX – brain cerebellum ** ** ** GTEX – brain cortex ** ** ** ** ** GTEX – brain hippocampus ** ** ** GTEX – brain hypothalamus ** ** GTEX – brain nucleus accumbens BG ** ** ** TWAS Z–scores GTEX – brain putamen BG ** ** 4 Tissues 0 GTEX – brain spinal cord cervical c1 ** ** –4 GTEX – brain substantia nigra ** GTEX – brain frontal cortex BA9 ** ** ** ** CMC – brain DLPFC RNA–seq splicing ** ** ** GTEX – heart left ventricle ** ** ** ** GTEX – heart atrial appendage ** ** ** ** ** GTEX – artery tibial * ** ** ** ** ** GTEX – artery coronary ** ** ** ** GTEX – artery aorta ** ** ** ** ** NTR – peripheral blood * YFS – whole blood ** ** * WM BG HIP Fig. 4 | Transcriptome-wide significant genes with extensive PVS burden. We using Bonferroni correction, based on the average number of features (4,235 used precomputed functional weights from 22 publicly available gene expression genes) tested across all tissues and correcting for the three independent PVS reference panels from brain (GTEx v7, CommonMind Consortium (CMC)), locations (P < 3.93 × 10−6). *Significant result in the TWAS and conditional peripheral nerve tissues (GTEx v7), heart and arteries (GTEx v7), and blood analyses; **significant result in the TWAS and conditional analyses, and with a (Netherlands Twin Registry (NTR) and Young Finns Study (YFS)). Transcriptome- COLOC PP4 > 0.75; eGenes for loci identified in the GWAS (†), gene-based test (‡) wide significant genes (eGenes) and the corresponding eQTLs were determined or both GWAS and gene-based test (§). associated with decreased odds of cognitive impairment or dementia, causing juvenile amyotrophic lateral sclerosis48, while NBEAL1 (encod- lower brain amyloid burden and slower brain atrophy rates43. HIP-PVS ing neurobeachin-like 1 protein) modulates low-density lipoprotein was associated with lower expression of LAMC1 (chr1q25.3, encoding (LDL)-receptor expression49. Laminin gamma-1) in brain and higher expression in vascular tissues, Our study points to an important involvement of solute carriers while WES/WGS identified a splice donor variant at LAMC1. Laminins (SLCs), the largest family of transporters and candidates for drug target are ECM glycoproteins, and the major noncollagenous constituent of development50, in PVS pathophysiology. The most significant PVS risk basement membranes. Genes encoding other basement membrane variants involve an intronic haplotype of SCL13A3, encoding a plasma proteins (NID2, COL4A1/2) are implicated in cSVD5,22. Laminin regulates membrane Na+/dicarboxylate cotransporter expressed in kidney, astro- blood vessel diameter44 and blood brain barrier integrity and function45, cytes and choroid plexus51. Mutations in SLC13A3 cause acute reversible and astrocytic laminin loss decreases expression of tight junction pro- leukoencephalopathy with increased urinary alpha-ketoglutarate51, teins and aquaporin-4 (AQP4)45, a key modulator of glymphatic flow in where SLC13A3 loss-of-function may affect elimination of organic experimental models7. anions and xenobiotics from the cerebrospinal fluid (CSF)51. At the Some genes point to complex pleiotropic mechanisms. At same locus (Supplementary Fig. 1), other genome-wide significant chr2q33.2, also associated with WMH, SVS, Alzheimer’s disease and variants are located near SLC2A10, harboring mutations causing arte- caudate volume5,23,46,47, BG-PVS was associated with higher expression rial tortuosity syndrome52, described to be associated with PVS burden of ICA1L in brain tissues and of NBEAL1 in vascular tissues, similar to and cSVD53. WM-PVS was associated with lower SLC20A2 expression TWAS of WMH and SVS5,22. ICA1L (encoding islet cell autoantigen-1-like in brain tissue. SLC20A2, involved in phosphate transport, harbors and predominantly expressed in endothelial cells) harbors mutations loss-of-function mutations causing idiopathic familial BG calcification, Nature Medicine | Volume 29 | April 2023 | 950–962 958 Article https://doi.org/10.1038/s41591-023-02268-w a neurodegenerative disorder with inorganic phosphate accumulation of drug targets for prevention trials of cSVD, a major cause of stroke in the ECM54. Given their role in CSF secretion and substance transport and dementia worldwide. at the blood–CSF barrier55, SLCs could be involved in interstitial fluid accumulation adjacent to the PVS56. Online content Consistent with other SVD phenotypes, we observed evidence Any methods, additional references, Nature Portfolio reporting sum- for a causal association of blood pressure with PVS. Experimental maries, source data, extended data, supplementary information, work suggests that the perivascular pump becomes less efficient with acknowledgements, peer review information; details of author contri- increasing blood pressure, reducing net forward flow in the PVS. These butions and competing interests; and statements of data and code avail- effects were found to be larger at more distal locations, where arteries ability are available at https://doi.org/10.1038/s41591-023-02268-w. have thinner and less muscular walls57. Such hemodynamic and ana- tomic differences1,2,18 could, perhaps, at least partly explain the more References significant association of blood pressure with BG-PVS and HIP-PVS 1. Pollock, H., Hutchings, M., Weller, R. O. & Zhang, E. T. Perivascular compared with WM-PVS. In contrast, WM-PVS were previously found to spaces in the basal ganglia of the human brain: their relationship be associated with CAA11 and with higher brain amyloid deposition on to lacunes. J. Anat. 191, 337–346 (1997). positron emission tomography, across the clinical spectrum of CAA12. 2. Wardlaw, J. M. et al. Perivascular spaces in the brain: The updated Boston Criteria (v.2.0) for CAA include severe WM-PVS as anatomy, physiology and pathology. Nat. Rev. Neurol. 16, a new diagnostic criterion13. 137–153 (2020). The clinical relevance of PVS is strongly supported by the signifi- 3. Wardlaw, J. M. et al. Neuroimaging standards for research cant genetic correlation of BG-PVS and HIP-PVS with any stroke and into small vessel disease and its contribution to ageing and ischemic stroke and robust evidence for a possible causal association of neurodegeneration. Lancet Neurol. 12, 822–838 (2013). BG-PVS and HIP-PVS with any stroke, ischemic stroke and SVS, account- 4. Jessen, N. A., Munk, A. S. F., Lundgaard, I. & Nedergaard, M. The ing for blood pressure. The nominally significant genetic correlation of glymphatic system: a beginner’s guide. Neurochem. Res. 40, BG-PVS and HIP-PVS with (deep) ICH, based on smaller GWAS and thus 2583–2599 (2015). less statistical power, is also consistent with epidemiological findings10. 5. Sargurupremraj, M. et al. Cerebral small vessel disease genomics Considering the association of HIP-PVS with lower LAMC1 expression and its implications across the lifespan. Nat. Commun. 11, 6285 in brain, it is striking to note that conditional knock-out of laminin in (2020). astrocytes leads to deep ICH in mice58. This is reminiscent of known 6. Debette, S., Schilling, S., Duperron, M. G., Larsson, S. C. & associations of variants in COL4A1/A2, encoding another basement Markus, H. S. Clinical significance of magnetic resonance membrane protein, with monogenic and multifactorial deep ICH46,59. imaging markers of vascular brain injury: a systematic review and Significant enrichment of PVS genes in targets of drugs validated meta-analysis. JAMA Neurol. 76, 81–94 (2019). or under investigation for vascular and cognitive disorders (for exam- 7. Mestre, H., Kostrikov, S., Mehta, R. I. & Nedergaard, M. Perivascular ple, telavancin and davunetide) highlights the potential of PVS genetics spaces, glymphatic dysfunction, and small vessel disease. Clin. for cSVD drug discovery. Sci. (Lond.) 131, 2257–2274 (2017). To our knowledge, this is the first study exploring the genetic 8. Deramecourt, V. et al. Staging and natural history of determinants of PVS, using a comprehensive gene-mapping strategy cerebrovascular pathology in dementia. Neurology 78, and extensive bioinformatics follow-up. We acknowledge limitations. 1043–1050 (2012). To account for heterogeneity in PVS quantification methods, we prag- 9. Bacyinski, A., Xu, M., Wang, W. & Hu, J. The paravascular pathway matically dichotomized PVS variables based on the top quartile of for brain waste clearance: current understanding, significance the distribution, which may be less powerful than continuous meas- and controversy. Front. Neuroanat. 11, 101 (2017). ures. This may have been most prominent for BG-PVS, for which the 10. Duperron, M. G. et al. High dilated perivascular space burden: a genetic correlation pattern between CHARGE and UKB was low, in new MRI marker for risk of intracerebral hemorrhage. Neurobiol. contrast with WM-PVS and HIP-PVS. Reassuringly, loci identified using Aging 84, 158–165 (2019). dichotomous PVS phenotypes were also associated with continuous 11. Charidimou, A. et al. MRI-visible perivascular spaces in cerebral PVS burden in studies where computational methods were available amyloid angiopathy and hypertensive arteriopathy. Neurology 88, (UKB, i-Share, Nagahama), mostly with more significant P values. A 1157–1164 (2017). conservative approach will also have helped minimize the effect of 12. Tsai, H. H. et al. Centrum semiovale perivascular space and accidentally including WMH in the PVS measures, a problem which amyloid deposition in spontaneous intracerebral hemorrhage. some computational PVS methods have not yet overcome. Strikingly, Stroke 52, 2356–2362 (2021). 67% of WM-PVS loci were associated at least nominally with WM-PVS in 13. Charidimou, A. et al. The Boston criteria version 2.0 for one or both follow-up cohorts, despite considerably smaller samples cerebral amyloid angiopathy: a multicentre, retrospective, and distinct age and ancestry, with consistent directionality. This sug- MRI-neuropathology diagnostic accuracy study. Lancet Neurol. gests that our genomic discovery approach, although likely conserva- 21, 714–725 (2022). tive, led to robust findings. With increasing development of artificial 14. Mestre, H. et al. Cerebrospinal fluid influx drives acute ischemic intelligence-based computational methods for PVS quantification, tissue swelling. Science 367, eaax7171 (2020). future genomic studies will likely have even greater power to detect 15. Månberg, A. et al. Altered perivascular fibroblast activity genetic associations, to enable studying the genomics of total PVS precedes ALS disease onset. Nat. Med 27, 640–646 (2021). volume, accounting for differences in individual PVS volume, width, 16. Duperron, M. G. et al. Burden of dilated perivascular spaces, length, shape60, density, location and anatomical predominance, and an emerging marker of cerebral small vessel disease, is highly to run sex-specific analyses. heritable. Stroke 49, 282–287 (2018). In conclusion, in this gene-mapping study of PVS, one of the ear- 17. Yao, M. et al. Hippocampal perivascular spaces are related to liest MRI markers of cSVD, we describe 24 genome-wide significant aging and blood pressure but not to cognition. Neurobiol. Aging risk loci, with six additional loci in secondary multivariate analyses 35, 2118–2125 (2014). accounting for other cSVD markers. Our findings provide insight into 18. Bouvy, W. H. et al. Visualization of perivascular spaces and the biology of PVS across the adult lifespan and its contribution to cSVD perforating arteries with 7 T magnetic resonance imaging. Invest. pathophysiology, with potential for genetically informed prioritization Radiol. 49, 307–313 (2014). Nature Medicine | Volume 29 | April 2023 | 950–962 959 Article https://doi.org/10.1038/s41591-023-02268-w 19. Psaty, B. M. et al. Cohorts for Heart and Aging Research in 39. Allanore, Y. et al. Lysophosphatidic acid receptor 1 Genomic Epidemiology (CHARGE) Consortium: design of antagonist SAR100842 for patients with diffuse cutaneous prospective meta-analyses of genome-wide association systemic sclerosis: a double-blind, randomized, eight-week studies from 5 cohorts. Circ. Cardiovasc. Genet. 2, 73–80 placebo-controlled study followed by a sixteen-week open-label (2009). extension study. Arthritis Rheumatol. 70, 1634–1643 (2018). 20. Bordes, C., Sargurupremraj, M., Mishra, A. & Debette, S. Genetics 40. Stenman, J. M. et al. Canonical Wnt signaling regulates of common cerebral small vessel disease. Nat. Rev. Neurol. 18, organ-specific assembly and differentiation of CNS vasculature. 84–101 (2022). Science 322, 1247–1250 (2008). 21. Joutel, A., Haddad, I., Ratelade, J. & Nelson, M. T. Perturbations 41. Chavali, M. et al. Wnt-dependent oligodendroglial-endothelial of the cerebrovascular matrisome: a convergent mechanism in interactions regulate white matter vascularization and attenuate small vessel disease of the brain? J. Cereb. Blood Flow. Metab. 36, injury. Neuron 108, 1130–1145.e1135 (2020). 143–157 (2016). 42. Capone, C. et al. Reducing Timp3 or vitronectin ameliorates 22. Traylor, M. et al. Genetic basis of lacunar stroke: a pooled analysis disease manifestations in CADASIL mice. Ann. Neurol. 79, of individual patient data and genome-wide association studies. 387–403 (2016). 20(5):351-361 Lancet Neurol. 20, 351–361 (2021). 43. Tanaka, T. et al. Plasma proteomic signatures predict dementia 23. Persyn, E. et al. Genome-wide association study of MRI markers and cognitive impairment. Alzheimers Dement. (N. Y.) 6, e12018 of cerebral small vessel disease in 42,310 participants. Nat. (2020). Commun. 11, 2175 (2020). 44. Jakobsson, L., Domogatskaya, A., Tryggvason, K., Edgar, D. 24. Simon, A. J. et al. Mutations in STN1 cause Coats plus syndrome & Claesson-Welsh, L. Laminin deposition is dispensable for and are associated with genomic and telomere defects. J. Exp. vasculogenesis but regulates blood vessel diameter independent Med. 213, 1429–1440 (2016). of flow. FASEB J. 22, 1530–1539 (2008). 25. Whittaker, E. et al. Systematic review of cerebral phenotypes 45. Yao, Y., Chen, Z. L., Norris, E. H. & Strickland, S. Astrocytic laminin associated with monogenic cerebral small-vessel disease. J. Am. regulates pericyte differentiation and maintains blood brain Heart Assoc. 11, e025629 (2022). barrier integrity. Nat. Commun. 5, 3413 (2014). 26. Piantino, J. et al. Characterization of MR imaging- 46. Chung, J. et al. Genome-wide association study of cerebral small visible perivascular spaces in the white matter of healthy vessel disease reveals established and novel loci. Brain 142, adolescents at 3T. AJNR Am. J. Neuroradiol. 41, 2139–2145 3176–3189 (2019). (2020). 47. Armstrong, N. J. et al. Common genetic variation indicates 27. Rajani, R. M. et al. Reversal of endothelial dysfunction reduces separate causes for periventricular and deep white matter white matter vulnerability in cerebral small vessel disease in rats. hyperintensities. Stroke 51, 2111–2121 (2020). Sci. Transl. Med. 10, eaam9507 (2018). 48. Hadano, S. et al. A gene encoding a putative GTPase regulator is 28. Carelli, V. et al. Syndromic parkinsonism and dementia mutated in familial amyotrophic lateral sclerosis 2. Nat. Genet. 29, associated with OPA1 missense mutations. Ann. Neurol. 78, 166–173 (2001). 21–38 (2015). 49. Bindesbøll, C. et al. NBEAL1 controls SREBP2 processing and 29. Herkenne, S. et al. Developmental and tumor angiogenesis cholesterol metabolism and is a susceptibility locus for coronary requires the mitochondria-shaping protein Opa1. Cell Metab. 31, artery disease. Sci. Rep. 10, 4528 (2020). 987–1003.e1008 (2020). 50. Wang, W. W., Gallo, L., Jadhav, A., Hawkins, R. & Parker, C. G. The 30. Backhouse, E. V. et al. Early life predictors of late life cerebral druggability of solute carriers. J. Med. Chem. 63, 3834–3867 small vessel disease in four prospective cohort studies. Brain 144, (2020). 3769–3778 (2021). 51. Dewulf, J. P. et al. SLC13A3 variants cause acute reversible 31. Zhao, B. et al. Common genetic variation influencing leukoencephalopathy and α-ketoglutarate accumulation. Ann. human white matter microstructure. Science 372, Neurol. 85, 385–395 (2019). eabf3736 (2021). 52. Beyens, A. et al. Arterial tortuosity syndrome: 40 new families and 32. Grasby, K. L. et al. The genetic architecture of the human cerebral literature review. Genet. Med. 20, 1236–1245 (2018). cortex. Science 367, eaay6690 (2020). 53. Chen, Y. C. et al. Correlation between internal carotid artery 33. Mok, V. et al. Race-ethnicity and cerebral small vessel disease tortuosity and imaging of cerebral small vessel disease. Front. – comparison between Chinese and White populations. Int. J. Neurol. 11, 567232 (2020). Stroke 9, 36–42 (2014). 54. Wang, C. et al. Mutations in SLC20A2 link familial idiopathic basal 34. Akinyemi, R. O. et al. Stroke in Africa: profile, progress, prospects ganglia calcification with phosphate homeostasis. Nat. Genet. 44, and priorities. Nat. Rev. Neurol. 17, 634–656 (2021). 254–256 (2012). 35. Mollink, J. et al. The spatial correspondence and genetic 55. Ho, H. T., Dahlin, A. & Wang, J. Expression profiling of solute influence of interhemispheric connectivity with white matter carrier gene families at the blood-CSF barrier. Front. Pharm. 3, 154 microstructure. Nat. Neurosci. 22, 809–819 (2019). (2012). 36. Gaire, B. P., Sapkota, A., Song, M. R. & Choi, J. W. 56. Wang, H. et al. Structure, function, and genomic organization of Lysophosphatidic acid receptor 1 (LPA1) plays critical roles in human Na+-dependent high-affinity dicarboxylate transporter. microglial activation and brain damage after transient focal Am. J. Physiol. Cell Physiol. 278, C1019–1030 (2000). cerebral ischemia. J. Neuroinflammation 16, 170 (2019). 57. Mestre, H. et al. Flow of cerebrospinal fluid is driven by arterial 37. Gross, I. & Brauer, A. U. Modulation of lysophosphatidic acid (LPA) pulsations and is reduced in hypertension. Nat. Commun. 9, 4878 receptor activity: the key to successful neural regeneration? (2018). Neural Regen. Res. 15, 53–54 (2020). 58. Chen, Z. L. et al. Ablation of astrocytic laminin impairs vascular 38. Hisaoka-Nakashima, K. et al. Mirtazapine increases glial smooth muscle cell function and leads to hemorrhagic stroke. cell line-derived neurotrophic factor production through J. Cell Biol. 202, 381–395 (2013). lysophosphatidic acid 1 receptor-mediated extracellular 59. Rannikmäe, K. et al. COL4A2 is associated with lacunar ischemic signal-regulated kinase signaling in astrocytes. Eur. J. Pharm. 860, stroke and deep ICH: meta-analyses among 21,500 cases and 172539 (2019). 40,600 controls. Neurology 89, 1829–1839 (2017). Nature Medicine | Volume 29 | April 2023 | 950–962 960 Article https://doi.org/10.1038/s41591-023-02268-w 60. Ballerini, L. et al. Computational quantification of brain as long as you give appropriate credit to the original author(s) and the perivascular space morphologies: associations with vascular source, provide a link to the Creative Commons license, and indicate risk factors and white matter hyperintensities. A study if changes were made. The images or other third party material in this in the Lothian Birth Cohort 1936. Neuroimage Clin. 25, article are included in the article’s Creative Commons license, unless 102120 (2020). indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended Publisher’s note Springer Nature remains neutral with regard to use is not permitted by statutory regulation or exceeds the permitted jurisdictional claims in published maps and institutional affiliations. use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. Open Access This article is licensed under a Creative Commons org/licenses/by/4.0/. Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, © The Author(s) 2023 Marie-Gabrielle Duperron 1,2,80, Maria J. Knol 3,80, Quentin Le Grand 1,80, Tavia E. Evans 4,5,80, Aniket Mishra1,80, Ami Tsuchida 1,6,80, Gennady Roshchupkin 3,5, Takahiro Konuma7, David-Alexandre Trégouët1, Jose Rafael Romero 8,9, Stefan Frenzel10, Michelle Luciano 11, Edith Hofer12,13, Mathieu Bourgey 14,15,16, Nicole D. Dueker17, Pilar Delgado18,19, Saima Hilal20,21,22, Rick M. Tankard 23, Florian Dubost5,24, Jean Shin 25,26, Yasaman Saba 1,27, Nicola J. Armstrong 23, Constance Bordes 1, Mark E. Bastin28, Alexa Beiser8,9,29, Henry Brodaty 30,31, Robin Bülow 32, Caty Carrera33, Christopher Chen20,21,34, Ching-Yu Cheng 35,36,37,38, Ian J. Deary11, Piyush G. Gampawar27, Jayandra J. Himali 8,9,29,39,40, Jiyang Jiang 30, Takahisa Kawaguchi41, Shuo Li 9,29, Melissa Macalli 1, Pascale Marquis14,15,16, Zoe Morris42, Susana Muñoz Maniega 28,43, Susumu Miyamoto44, Masakazu Okawa 45, Matthew Paradise30, Pedram Parva9,46,47, Tatjana Rundek 48,49, Muralidharan Sargurupremraj1, Sabrina Schilling1, Kazuya Setoh41,50, Omar Soukarieh1, Yasuharu Tabara41,50, Alexander Teumer 51, Anbupalam Thalamuthu30, Julian N. Trollor 30,52, Maria C. Valdés Hernández28,53, Meike W. Vernooij3,5, Uwe Völker54, Katharina Wittfeld 10,55, Tien Yin Wong35,56, Margaret J. Wright 57,58, Junyi Zhang59, Wanting Zhao35,60, Yi-Cheng Zhu59, Helena Schmidt 27, Perminder S. Sachdev 30,61, Wei Wen 30, Kazumichi Yoshida45, Anne Joutel62, Claudia L. Satizabal8,9,39,40, Ralph L. Sacco48,49,63,64,65, Guillaume Bourque 14,15,16, the CHARGE consortium*, Mark Lathrop14,15, Tomas Paus 66,67,68,69, Israel Fernandez-Cadenas33,70, Qiong Yang9,29, Bernard Mazoyer 6,71, Philippe Boutinaud72, Yukinori Okada 7,73,74,75,76, Hans J. Grabe 10,55, Karen A. Mather 30,77, Reinhold Schmidt12, Marc Joliot 6, M. Arfan Ikram 3, Fumihiko Matsuda41,81, Christophe Tzourio1,78,81, Joanna M. Wardlaw 28,43,53,81, Sudha Seshadri 8,9,39,40,81, Hieab H. H. Adams 4,5,79,81 & Stéphanie Debette 1,2 1 Bordeaux Population Health Research Center, UMR 1219, University of Bordeaux, Inserm, Bordeaux, France. 2Department of Neurology, Institute of Neurodegenerative Diseases, Bordeaux University Hospital, Bordeaux, France. 3Department of Epidemiology, Erasmus MC University Medical Center, Rotterdam, the Netherlands. 4Department of Clinical Genetics, Erasmus MC University Medical Center, Rotterdam, the Netherlands. 5Department of Radiology and Nuclear Medicine, Erasmus MC University Medical Center, Rotterdam, the Netherlands. 6Groupe d’Imagerie Neurofonctionelle - Institut des maladies neurodégénératives (GIN-IMN), UMR 5293, University of Bordeaux, CNRS, CEA, Bordeaux, France. 7Department of Statistical Genetics, Osaka University Graduate School of Medicine, Suita, Japan. 8Department of Neurology, Boston University School of Medicine, Boston, MA, USA. 9The Framingham Heart Study, Framingham, MA, USA. 10Department of Psychiatry and Psychotherapy, University Medicine Greifswald, Greifswald, Germany. 11 School of Psychology, University of Edinburgh, Edinburgh, UK. 12Clinical Division of Neurogeriatrics, Department of Neurology, Medical University of Graz, Graz, Austria. 13Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Graz, Austria. 14Department of Human Genetics, McGill University, Montreal, Quebec, Canada. 15Victor Phillip Dahdaleh Institute of Genomic Medicine at McGill University, Montreal, Quebec, Canada. 16Canadian Centre for Computational Genomics, McGill University, Montreal, Quebec, Canada. 17John P. Hussman Institute for Human Genomics, University of Miami, Miami, FL, USA. 18Institut de Recerca Vall d’hebron, Neurovascular Research Lab, Universitat Autònoma de Barcelona, Barcelona, Spain. 19Hospital Universitari Vall d’Hebron, Neurology Department, Universitat Autònoma de Barcelona, Barcelona, Spain. 20Memory Aging and Cognition Center, National University Health System, Singapore, Singapore. 21Department of Pharmacology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore. 22Saw Swee Hock School of Public Health, National University of Singapore and National University Health System, Singapore, Singapore. 23Department of Mathematics and Statistics, Curtin University, Perth, Western Australia, Australia. 24Department of Medical Informatics, Erasmus MC University Medical Center, Rotterdam, the Netherlands. 25The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada. 26Departments of Physiology and Nutritional Sciences, University of Toronto, Toronto, Ontario, Canada. 27Institute for Molecular Biology & Biochemistry, Gottfried Schatz Research Center (for Cell Signaling, Metabolism and Aging), Medical University of Graz, Graz, Austria. 28Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, UK. 29Department of Biostatistics, Boston University School of Public Health, Boston, MA, USA. 30Centre for Healthy Brain Ageing (CHeBA), Discipline of Psychiatry & Mental Health, University of New South Wales, Sydney, New South Wales, Australia. 31Dementia Collaborative Research Centre Assessment and Better Care, UNSW, Sydney, New South Wales, Australia. 32Institute for Radiology and Neuroradiology, University Medicine Greifswald, Greifswald, Germany. 33Stroke Pharmacogenomics and Genetics Group, Biomedical Research Institute Sant Pau (IIB Sant Pau), Barcelona, Spain. 34Department of Psychological Medicine, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore. 35Singapore Eye Research Institute, Singapore National Eye Centre, Singapore, Singapore. 36Center for Innovation and Precision Eye Health, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore. 37Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore. 38Ophthalmology & Visual Sciences Academic Clinical Program, Nature Medicine | Volume 29 | April 2023 | 950–962 961 Article https://doi.org/10.1038/s41591-023-02268-w Duke-NUS Medical School, Singapore, Singapore. 39Glenn Biggs Institute for Alzheimer’s and Neurodegenerative Diseases, UT Health San Antonio, San Antonio, TX, USA. 40Department of Population Health Sciences, UT Health San Antonio, San Antonio, TX, USA. 41Center for Genomic Medicine, Kyoto University Graduate School of Medicine, Kyoto, Japan. 42Neuroimaging, Department of Clinical Neurosciences, Royal Infirmary of Edinburgh, Edinburgh, UK. 43UK Dementia Research Institute Centre at the University of Edinburgh, Edinburgh, UK. 44Kyoto University Hospital, Kyoto, Japan. 45Department of Neurosurgery, Kyoto University Graduate School of Medicine, Kyoto, Japan. 46Radiology Department, Boston University School of Medicine, Boston, MA, USA. 47Department of Radiology, Harvard Medical School, Boston, MA, USA. 48Department of Neurology, Miller School of Medicine, University of Miami, Miami, FL, USA. 49Evelyn F. McKnight Brain Institute, Department of Neurology, University of Miami, Miami, FL, USA. 50Graduate School of Public Health, Shizuoka Graduate University of Public Health, Shizuoka, Japan. 51Institute for Community Medicine, University Medicine Greifswald, Greifswald, Germany. 52Department of Developmental Disability Neuropsychiatry, UNSW, Sydney, New South Wales, Australia. 53Row Fogo Centre for Research into Ageing and the Brain, University of Edinburgh, Edinburgh, UK. 54Interfaculty Institute for Genetics and Functional Genomics, University Medicine Greifswald, Greifswald, Germany. 55German Center for Neurodegenerative Diseases (DZNE), Site Rostock/Greifswald, Greifswald, Germany. 56Tsinghua Medicine, Tsinghua University, Beijing, China. 57Queensland Brain Institute, University of Queensland, Brisbane, Queensland, Australia. 58Centre for Advanced Imaging, University of Queensland, Brisbane, Queensland, Australia. 59Department of Neurology, Peking Union Medical College Hospital, Beijing, China. 60The Centre for Quantitative Medicine, Duke-NUS Medical School, Singapore, Singapore. 61Neuropsychiatric Institute, the Prince of Wales Hospital, Sydney, New South Wales, Australia. 62Institut de Psychiatrie et Neurosciences de Paris, Université Paris Cité, Inserm, France. 63Department of Public Health Sciences, Miller School of Medicine, University of Miami, Miami, FL, USA. 64Department of Human Genomics, Miller School of Medicine, University of Miami, Miami, FL, USA. 65Department of Neurosurgery, Miller School of Medicine, University of Miami, Miami, FL, USA. 66University of Montreal, Faculty of Medicine, Departments of Psychiatry and Neuroscience, Montreal, Quebec, Canada. 67Department of Psychology, University of Toronto, Toronto, Ontario, Canada. 68Centre Hospitalier Universitaire Sainte Justine, Montreal, Quebec, Canada. 69Department of Psychiatry, University of Toronto, Toronto, Ontario, Canada. 70Stroke Pharmacogenomics and Genetics Group, Fundació per la Docència i la Recerca Mutua Terrassa, Terrassa, Spain. 71Bordeaux University Hospital, Bordeaux, France. 72Fealinx, Lyon, France. 73Laboratory of Statistical Immunology, Immunology Frontier Research Center (WPI-IFReC), Osaka University, Suita, Japan. 74Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives, Osaka University, Suita, Japan. 75Center for Infectious Disease Education and Research (CiDER), Osaka University, Suita, Japan. 76 Department of Genome Informatics, Graduate School of Medicine, University of Tokyo, Tokyo, Japan. 77Neuroscience Research Australia, Sydney, New South Wales, Australia. 78Department of Medical Informatics, Bordeaux University Hospital, Bordeaux, France. 79Latin American Brain Health (BrainLat), Universidad Adolfo Ibáñez, Santiago, Chile. 80These authors contributed equally: Marie-Gabrielle Duperron, Maria J. Knol, Quentin Le Grand, Tavia E. Evans, Aniket Mishra, Ami Tsuchida. 81These authors jointly supervised this work: Fumihiko Matsuda, Christophe Tzourio, Joanna M. Wardlaw, Sudha Seshadri, Hieab H. H. Adams, Stéphanie Debette. *A list of authors and their affiliations appears at the end of the paper. e-mail: h.adams@erasmusmc.nl; stephanie.debette@u-bordeaux.fr the CHARGE consortium Marie-Gabrielle Duperron1,2,80, Maria J. Knol3,80, Quentin Le Grand1, Tavia E. Evans4,5,80, Aniket Mishra1,80, Ami Tsuchida1,6,80, Gennady Roshchupkin3,5, Takahiro Konuma7, David-Alexandre Trégouët1, Jose Rafael Romero8,9, Stefan Frenzel10, Michelle Luciano11, Edith Hofer12,13, Mathieu Bourgey14,15,16, Nicole D. Dueker17, Pilar Delgado18,19, Saima Hilal20,21,22, Rick M. Tankard23, Florian Dubost5,24, Jean Shin25,26, Yasaman Saba1,27, Nicola J. Armstrong23, Constance Bordes1, Mark E. Bastin28, Alexa Beiser8,9,29, Henry Brodaty30,31, Robin Bülow32, Caty Carrera33, Christopher Chen20,21,34, Ching-Yu Cheng35,36,37,38, Ian J. Deary11, Piyush G. Gampawar27, Jayandra J. Himali8,9,29,39,40, Jiyang Jiang30, Takahisa Kawaguchi41, Shuo Li9,29, Melissa Macalli1, Pascale Marquis14,15,16, Zoe Morris42, Susana Muñoz Maniega28,43, Susumu Miyamoto44, Masakazu Okawa45, Matthew Paradise30, Pedram Parva9,46,47, Tatjana Rundek48,49, Muralidharan Sargurupremraj1, Sabrina Schilling1, Kazuya Setoh41,50, Omar Soukarieh1, Yasuharu Tabara41,50, Alexander Teumer51, Anbupalam Thalamuthu30, Julian N. Trollor30,52, Maria C. Valdés Hernández28,53, Meike W. Vernooij3,5, Uwe Völker54, Katharina Wittfeld10,55, Tien Yin Wong35,56, Margaret J. Wright57,58, Junyi Zhang59, Wanting Zhao35,60, Yi-Cheng Zhu59, Helena Schmidt27, Perminder S. Sachdev30,61, Wei Wen30, Kazumichi Yoshida45, Anne Joutel62, Claudia L. Satizabal8,9,39,40, Ralph L. Sacco48,49,63,64,65, Guillaume Bourque14,15,16, Mark Lathrop14,15, Tomas Paus66,67,68,69, Israel Fernandez-Cadenas33,70, Qiong Yang9,29, Bernard Mazoyer6,71, Philippe Boutinaud72, Yukinori Okada7,73,74,75,76, Hans J. Grabe10,55, Karen A. Mather30,77, Reinhold Schmidt12, Marc Joliot6, M. Arfan Ikram3, Fumihiko Matsuda41,81, Christophe Tzourio1,78,81, Joanna M. Wardlaw28,43,53,81, Sudha Seshadri8,9,39,40,81, Hieab H. H. Adams4,5,79,81 & Stéphanie Debette1,2 Nature Medicine | Volume 29 | April 2023 | 950–962 962 Article https://doi.org/10.1038/s41591-023-02268-w Methods was used (ratio of brain parenchymal tissue volume to total volume Study design within the surface contour of the whole brain). Other covariates are This study complies with all relevant ethical regulations, and all par- described in Supplementary Table 1. ticipants gave written, informed consent. Analyses were performed on stroke-free participants from 22 population-based cohorts (18 for the Genotyping and imputation GWAS meta-analysis), taking part in UKB, the CHARGE consortium and Genome-wide genotypes were imputed to the 1000G project (1000G the BRIDGET initiative. Institutional review boards approved individual pIv3) or the Haplotype Reference Consortium reference panels (Sup- studies: UKB (National Research Ethics Service Committee North West– plementary Table 3). Haydock), 3C-Dijon (Ethical Committee of the University Hospital of Kremlin-Bicêtre), Austrian Stroke Prevention Study and Austrian Stroke PVS genome-wide association analyses in individual cohorts Prevention Family Study (ASPS/ASPS-Fam) (Ethics Committee of the Med- Ancestry-specific logistic regression analyses with an additive genetic ical University of Graz), Epidemiology of Dementia in Singapore Study model were performed, adjusting for age, sex (genetically determined) (EDIS) (the Singapore Chinese Eye Study/Singapore Malay Eye Study-2, and intracranial volume (or brain parenchymal fraction for ASPS), Singapore Eye Research Institute and the National Healthcare Group principal components of population stratification, and study site. Domain-Specific Review Board), Framingham Heart Study (FHS) (Institu- As sensitivity analyses, we ran linear mixed models in UKB, (1) using tional Review Board of Boston University Medical Center), Investigating the log-transformed (log(variable + 1)) continuous PVS measurements, Silent Strokes in Hypertensives Study (ISSYS) (Comité de ética de investi- adjusting for the same covariates as above; (2) generating residuals gacion con medicamentos, Hospital Universitari Vall d’Hebron), Lothian adjusting for the same covariates and then dichotomizing the residuals Birth Cohort 1936 (LBC1936) (Lothian and Scottish Multicentre Research (instead of adjusting for covariates after dichotomization). Ethics Committees), Northern Manhattan Study (NOMAS) (Columbia University Medical Center Institutional Review Board and the University PVS genome-wide association meta-analyses of Miami Institutional Review Board), Rotterdam Study I, II and III (RS-I, We performed quality control in each study following the recommen- RS-II and RS-III) (Ministry of Health, Welfare, and Sport of the Nether- dations of Winkler et al.63. Analyses were done on autosomal biallelic lands), Study of Health in Pomerania (SHIP) (SHIP-2, SHIP-Trend Batch 1 markers. Duplicate markers were removed, marker names and alleles and 2, Ethics Commission of the University of Greifswald), i-Share study were harmonized across studies, and PZ-plots, quantile–quantile (Comités de Protection des Personnes (CPP) Sud-Ouest Outre-Mer III, plots and allele frequency plots were constructed63. In each study, rare Sydney Memory and Ageing Study (MAS) (Ethics Committees of the Uni- variants (MAF < 0.01) and variants with low imputation accuracy (R², versity of New South Wales, South-Eastern Sydney, and the Illawarra Area oevar_imp or info score < 0.5) or extensive effect size values (β > 5 or Health Service), Older Australian Twins Study (OATS) (Ethical Committees β < −5) were removed. We reported the number of SNPs passing quality of the Australian Twin Registry, the University of New South Wales, the control for each study (Supplementary Table 4). GWAS were run within University of Melbourne, the Queensland Institute of Medical Research, each cohort using logistic regression (or linear regression for sensitivity and the South-Eastern Sydney and Illawarra Area Health Service) and analyses), using software described in Supplementary Table 3. We then the Nagahama Study (Ethics Committee of Kyoto University Graduate conducted GWAS meta-analyses across participating cohorts in METAL, School of Medicine and the Nagahama Municipal Review Board) (Sup- using sample size-weighted meta-analysis as PVS were measured on plementary Table 1). Characteristics of study participants are provided different scales. Meta-analyses were conducted within each ances- in Supplementary Tables 1–3 and 26 and Supplementary Fig. 5. try (European (EUR), Asian (ASN), African-American (AA), Hispanic (HISP)) using METAL (https://github.com/statgen/METAL), followed PVS burden definition by meta-analyses across ancestries. Ancestry was genetically inferred PVS were defined as fluid-filled spaces with a signal identical to that of using principal components of population stratification (Supplemen- CSF, of round, ovoid or linear shape depending on the slice direction, tary Tables 1 and 3). Genomic control was applied to each study-specific with usually a maximum diameter smaller than 3 mm, no hyperintense GWAS with a genomic inflation factor greater than 1.00. Variants with an rim on T2-weighted or FLAIR sequences, and located in areas supplied effective allele count (twice the product of MAF, imputation accuracy by perforating arteries3. In most CHARGE cohorts, visual semiquantita- and number of participants with extensive PVS) < 10 and significant tive rating scales were used to quantify PVS burden. As different scales heterogeneity (PHet < 5.0 × 10−8) were excluded from the meta-analysis. were used across studies, we dichotomized PVS burden into ‘extensive We performed LD-clumping, sorting the genome-wide significant SNPs PVS burden’ versus the rest, defined by a cut-off closest to the top by P value, keeping the most significant SNP and removing SNPs with quartile of the semiquantitative scale distribution within each cohort an r² > 0.1 within 1 megabase (Mb). Only variants present in at least (Supplementary Tables 2, 27 and 28). This cohort-specific threshold half of participants of the final meta-analysis were used to construct definition was chosen because (1) small PVS counts are very sensitive quantile–quantile and Manhattan plots. In secondary analyses, we ran to MRI field strength and less prominently associated with age and inverse variance-weighted meta-analyses to obtain effect estimates and vascular risk factors61; (2) extreme burden of other MRI markers of standard errors for follow-up bioinformatics analyses. cSVD (for example, extensive WMH burden within the top quartile of the distribution) was previously shown to facilitate the identification Conditional and joint multiple-SNP analysis of genetic variants underlying cSVD62; and (3) PVS burden is highly We used GCTA-COJO64 to perform conditional and joint multiple-SNP dependent on participant characteristics, especially age, PVS quan- analysis of PVS GWAS summary statistics, to identify secondary associa- tification methods and image acquisition parameters. In RS-III and in tion signals at each of the genome-wide significant loci within 1 Mb of UKB, a recently developed automated method was used to quantify the the lead SNP. We used European GWAS summary statistics as recom- number of PVS (Supplementary Table 27), dichotomized according to mended to avoid population stratification. This method relied on a the same cut-off (top quartile). For sensitivity analyses, we also com- stepwise selection procedure to select SNPs based on the conditional pared results obtained in UKB with the dichotomized and continuous P values, and the joint effects of all selected SNPs after optimization (log-transformed) PVS variables. of the model were estimated64. We used genotypes of 6,489 unre- lated participants of European ancestry from the 1000G-imputed Covariates and descriptive variables 3C-Dijon study data for LD correction. We performed haplotype asso- Intracranial volume (sum of gray matter, WM and CSF volumes) was ciation analyses on the six independent lead variants at chr20q13.12 available in all studies except ASPS, where brain parenchymal fraction (Supplementary Table 5b). Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Cross-ancestry meta-regression of GWAS for significant modifying effects of age on associations with WM-PVS We conducted cross-ancestry meta-analyses using MR-MEGA65, which for the three genome-wide significant WM-PVS loci in young adults (at uses meta-regression to model allelic effects, including axes of genetic chr2p16.1, chr3q29 and chr20q13.12). We collected effect estimates and variation as covariates in the model. standard errors for the lead SNPs at these three loci in each individual cohort, and fitted a meta-regression of the lead SNPs’ effect sizes onto Gene-based analyses an intercept and age. Meta-regression analysis was performed using We performed gene-based analyses on European PVS GWAS Metafor69, and any statistical evidence of linear association was cor- meta-analyses. We included variants within 10 kilobase (kb) of the rected for multiple testing (P < 0.05/3 = 1.7 × 10−2). 3′ and 5′ untranslated regions (UTRs) of a gene to capture regulatory In Nagahama we explored the association of WM-PVS with a rwGRS variants. We used the MAGMA software implemented in FUMA66 to per- of WM-PVS burden, including the 14 available independent SNPs iden- form a gene-based association study, including 19,037 protein-coding tified in the European GWAS meta-analysis (r² < 0.10 based on 1000G genes. This method is based on a multiple linear principal components Japanese reference panel); SNPs were weighted by the SNP effect sizes regression model. Gene-wide significance was defined at P < 2.63 × 10−6. in the European GWAS meta-analysis. We also performed gene-based tests using VEGAS2 (ref. 67), includ- ing 18,371 autosomal genes, leading to a gene-wide significance at Shared genetic variation with other phenotypes P < 2.72 × 10−6. Genes were considered in the same locus if they were In the European ancestry meta-analysis, we explored shared genetic within 200 kb of each other. variation with vascular and neurological phenotypes: (1) putative risk factors (SBP, DBP, pulse pressure, body mass index, high-density PVS heritability estimates lipoprotein cholesterol, LDL cholesterol, triglycerides, type 2 diabetes We used LD-score regression (ldsc package https://github.com/bulik/ and sleep patterns); (2) other MRI markers of brain aging (WMH bur- ldsc/) to estimate the heritability of extensive PVS burden in each den, covert MRI-defined brain infarcts and lacunes, and hippocampal, location, overall and, in secondary analyses, separately, in CHARGE nucleus accumbens, amygdala, caudate nucleus, pallidum and puta- and UKB. men volumes); and (3) the most common neurological conditions previ- ously reported to be associated with PVS, namely stroke (any stroke, any Multi-trait GWAS with PVS and other MRI markers of cSVD ischemic stroke, large artery stroke, cardio-embolic stroke, SVS, ICH), We conducted a joint analysis of summary statistics from GWAS of Alzheimer’s disease and Parkinson’s disease (Supplementary Table 30). PVS, WMH and lacunes using MTAG68, with the expectation to gain in We explored whether genome-wide significant PVS risk loci (lead power because of the genetic correlation between these MRI markers variants or in LD with r² > 0.9, based on the 1000G European reference of cSVD. MTAG is a generalization of inverse variance-weighted GWAS panel) were associated with these traits. A P value threshold <3.3 × 10−5, meta-analysis of two or more traits, which accounts for sample over- correcting for 21 independent phenotypes, three PVS locations and 24 lap between GWAS results for different traits by employing LD-score independent loci tested, was used (Supplementary Table 30). We per- regression. MTAG is based on the assumption that all SNPs share the formed a colocalization analysis using COLOC to search for evidence same variance–covariance matrix of effect sizes across traits. We prior- for a single causal variant between PVS and the other phenotypes, a itized variants with a P < 5 × 10−8 in the PVS MTAG analysis and P < 0.05 posterior probability (PP4) > 75% supporting a single causal variant in the univariate PVS GWAS, which showed greater significance for for both traits70. association with PVS in MTAG than in univariate analyses for PVS, Second, we used LD-score regression (ldsc package: https://github. WMH and lacunes. com/bulik/ldsc/) to estimate the genetic correlation of extensive PVS burden with these phenotypes (P < 7.9 × 10−4 was used as a significance PVS next-generation sequencing association analyses threshold, correcting for 21 phenotypes and three PVS locations). To Using WES data and exome content of WGS data in 19,010 participants decrease potential bias due to poor imputation quality, the summary from UKB and BRIDGET, of whom 4,531, 4,424 and 4,497 had extensive statistics were filtered to the subset of HapMap3 SNPs for each trait. In PVS in WM, BG and HIP, respectively, we performed a whole-exome secondary analyses, we estimated genetic correlation of PVS burden association study to identify (rare) exonic variants associated with with the same traits separately in CHARGE and UKB. extensive PVS (Supplementary Tables 1 and 29). We used FUMA to obtain extensive functional annotation for genome-wide significant SNPs and to identify SNPs associated with Follow-up of findings across lifespan and ancestries other traits at genome-wide significance from the GWAS catalog66. We explored associations of WM-PVS and BG-PVS risk variants iden- tified in the GWAS meta-analysis with these phenotypes in young MR adults (i-Share study, N = 1,748, mean age 22.1 ± 2.3 yr) and in older We used an MR approach to explore the possible causal relation of Japanese population-based cohort participants (Nagahama study, putative risk factors (vascular risk factors and sleep patterns) with N = 2,862, 68.3 ± 5.3 yr; Supplementary Tables 1 and 3). In each study, extensive PVS burden, and of extensive PVS burden with neurological we used both quantitative PVS measurements derived from a com- traits (stroke, Alzheimer’s disease and Parkinson’s disease). putational artificial intelligence-based method (Supplementary We used the GSMR method implemented in GCTA71. Summary Tables 27 and 28) and dichotomized PVS burden (top quartile of statistics were clumped using 1000G-imputed 3C-Dijon study data PVS distribution; Supplementary Table 2). HIP-PVS data were not (r² < 0.05 and P < 5 × 10−8) using only SNPs with MAF > 0.01. The het- available. Continuous PVS measurements were log-transformed erogeneity in independent instrument (HEIDI)-outlier method was (log(variable + 1)) to obtain a normal distribution. used to remove genetic instruments that showed pleiotropic effects In i-Share participants of European ancestry, we also explored the on both the exposure and the outcome. association of WM-PVS with a wGRS of WM-PVS burden derived from For (at least) nominally significant GSMR associations, we the 21 independent genome-wide significant SNPs identified in the conducted secondary MR analyses using both TwoSampleMR and European GWAS meta-analysis (r² < 0.10 based on the 1000G European RadialMR72,73. Only independent SNPs (r2 < 0.01 based on 1000G reference panel). SNPs were weighted by the SNP effect sizes in the European, window size = 1 Mb) reaching P < 5 × 10−8 in the primary European GWAS meta-analysis (for the allele associated with larger meta-analysis were included as recommended. Effect estimates PVS burden); the wGRS was rescaled (rwGRS) so that one unit of the (β values) and SE values were derived from the inverse variance- wGRS corresponds to one additional WM-PVS risk allele. We tested weighted GWAS meta-analyses. With TwoSampleMR, we estimated Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w the effect of each exposure on each outcome using weighted median, (P < 3.93 × 10−6, correcting for 4,235 genes tested and three PVS loca- random-effect inverse variance weighting (IVW) and MR-Egger. In tions). eGenes were then tested in conditional analyses as implemented addition, we confirmed the directionality of the observed associations in TWAS-Fusion. Next, we performed a genetic colocalization analysis with the Steiger test74. With RadialMR (https://github.com/WSpiller/ of gene expression and PVS burden for each conditionally significant RadialMR), the putative causal effect of each exposure on each outcome gene (P < 0.05) using COLOC70, to estimate the posterior probability was estimated using the fixed-effect IVW method using the modified of a shared causal variant between the gene expression and the trait second-order inverse variance weight73. Cochran’s Q statistic was used (PP4 ≥ 0.75). Gene regions with eQTLs not reaching genome-wide to test for the heterogeneity (P < 0.05) due to horizontal pleiotropy73. significance in association with PVS, and not in LD (r2 < 0.01) with the We excluded outlier SNPs, identified by regressing the predicted causal lead SNP for genome-wide significant PVS risk loci, were considered estimate against the inverse variance weights73, and re-ran IVW tests, as novel. as well as MR-Egger regression, assessing heterogeneity with Rücker’s Q′ statistic73. When the ratio of Q′ (Egger) on Q (IVW) (QR) was close to Cell type enrichment analysis 1, indicating that both IVW and MR-Egger models fit the data equally, We conducted a cell type enrichment analysis using Single cell Type we selected the IVW model. We formally ruled out horizontal plei- Enrichment Analysis for Phenotypes (https://github.com/erwinerdem/ otropy when the MR-Egger intercept after exclusion of outliers was STEAP/). This is an extension to CELLECT and uses S-LDSC, MAGMA and nonsignificant (P ≥ 0.05). To account for potential residual correlated H-MAGMA for enrichment analysis. PVS GWAS summary statistics were pleiotropy, we used MR-CAUSE75. Finally, we explored the association munged. Then, expression specificity profiles were calculated using between genetic liability to PVS and stroke, conditioning on blood pres- human and mouse single-cell RNA sequencing databases (PsychEN- sure (SBP and DBP separately), by running multivariable MR analyses CODE DER-22, GSE67835, GSE101601, DroNc Human Hippocampus, using TwoSampleMR72. A P < 1.19 × 10−3, correcting for 14 independent Allen Brain Atlas MTG and LNG, Mousebrain, Tabula Muris, Descartes phenotypes and the three PVS locations, was considered significant. Human Cerebrum and Cerebellum; Supplementary Table 24). Cell type enrichment was calculated with MAGMA, H-MAGMA (incorporating Pathway analyses chromatin interaction profiles from human brain tissues in MAGMA) We used MAGMA gene set analyses (in FUMA66) to identify pathways and stratified LD-score regression. P values were corrected for the overrepresented in the associations. We identified genes associated number of independent cell types in each database. with extensive PVS burden and estimated the correlation between genes. The P values and gene correlation matrix were used in a gen- Lifetime brain gene expression profile eralized least squares model. A P < 3.2 × 10−6 correction for 15,496 We studied the lifetime expression of genes identified in the gene sets was considered significant. As a sensitivity analysis, we TWAS-COLOC analysis, and the three genes associated with WM-PVS used VEGAS2Pathway76, which aggregates association strengths burden in both the old and young populations, to search for develop- of individual markers into prespecified biological pathways using mental processes. We used a public database (https://hbatlas.org/) VEGAS-derived gene association P values for extensive PVS burden, comprising genome-wide exon-level transcriptome data from 1,340 with an empirical significance threshold of P < 1 × 10−5 (accounting tissue samples from 16 brain regions (cerebellar cortex, mediodorsal for 6,213 correlated pathways). nucleus of the thalamus, striatum, amygdala, hippocampus and 11 neocortex areas) of 57 postmortem human brains, from embryonic Enrichment analyses in OMIM and COSMIC genes development to older adults of different ancestries. Using hypergeometric tests, we performed enrichment analyses of genes within 1 Mb, 100 kb or 10 kb of the lead variants, but also of Enrichment in drug target genes genes within 10 kb of the lead variants with intragenic variants, and We used the GREP (Genome for Repositioning)79 software tool, which genes within 10 kb of the genetic loci with intragenic lead variants. quantifies an enrichment of gene sets from GWAS summary statistics in We used the rest of the protein-coding genome as a reference. We drugs of certain Anatomical Therapeutic Chemical Classification (ATC) performed the analysis first combining loci of all PVS locations, and classes, or indicated for some ICD10 (10th revision of the International second including only WM-PVS loci. We searched for an enrichment Statistical Classification of Diseases and Related Health Problems) in different gene groups from the OMIM database77, including PVS disease categories, and captures potentially repositionable drugs (‘perivascular space’ OR ‘virchow-robin space’), WMH (‘leukoaraio- targeting the gene set. Genes with false discovery rate FDR q < 0.1 in sis’ OR ‘white matter lesion’ OR ‘white matter hyperintensities’) and MAGMA were used for enrichment analyses (in GREP) of target genes leukodystrophy (‘leukodystrophy’ OR ‘leukoencephalopathy’) genes. for approved or investigated drugs curated in DrugBank and the Thera- We also searched for an enrichment of genes involved in glioma and peutic Target Database. glioblastoma, identified in the Catalog Of Somatic Mutations In Cancer We used the Trans-Phar (integration of TWAS and Pharmacological (COSMIC) (https://cancer.sanger.ac.uk). database) software to identify drug target candidates in a specific tis- sue or cell type80, using first FOCUS to identify up- and downregulated TWAS genes in participants with extensive PVS burden, followed by a negative We performed TWAS using TWAS-Fusion78, to identify genes whose Spearman’s rank correlation analysis between the gene expression expression is significantly associated with PVS burden without directly (Z-score) of the top 10% genes with the highest expression variation measuring expression levels. We restricted the analysis to tissues con- and the LINCS CMap L1000 library database (Extended Data Fig. 8). sidered relevant for cerebrovascular disease, and used precomputed functional weights from 22 publicly available gene expression refer- Reporting summary ence panels from blood, arterial, brain and peripheral nerve tissues Further information on research design is available in the Nature Port- (Fig. 4). TWAS-Fusion was then used to estimate the TWAS association folio Reporting Summary linked to this article. statistics between predicted gene expression and PVS burden by inte- grating information from expression reference panels (SNP expression Data availability weights), GWAS summary statistics (SNP PVS effect estimates) and Genome-wide summary statistics for the European and cross-ancestry LD reference panels (SNP correlation matrix). Transcriptome-wide meta-analysis generated and analyzed during the current study are significant genes (eGenes) and the corresponding expression quanti- deposited on the GWAS Catalog (study code GCST90244151- tative trait loci (eQTLs) were determined using Bonferroni correction GCST90244156). As for other meta-analyses of GWAS or sequencing Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w data, individual cohort data are subject to controlled access, for pri- 67. Mishra, A. & Macgregor, S. VEGAS2: software for more flexible vacy and legal issues (national and European regulations, including gene-based testing. Twin Res. Hum. Genet. 18, 86–91 (2015). GDPR). This applies to all participating cohorts (cohorts included 68. Turley, P. et al. Multi-trait analysis of genome-wide association in the meta-analyses and follow-up cohorts). UKB data (GWAS summary statistics using MTAG. Nat. Genet. 50, 229–237 (2018). and sequencing) are accessible by submitting an application to 69. Viechtenbauer, W. Conducting meta-analysis in R with the the UKB portal (this research has been conducted under Applica- metafor package. J. Stat. Softw. 36(3), 1–48 (2010). tion Number 23509). We used publicly available data for analyses 70. Giambartolomei, C. et al. Bayesian test for colocalisation between described in this manuscript, including data from GTEx (https:// pairs of genetic association studies using summary statistics. gtexportal.org/home/), the Gusev laboratory (http://gusevlab. PLoS Genet. 10, e1004383 (2014). org/projects/fusion/), the CommonMind Consortium (https:// 71. Zhu, Z. et al. Causal associations between risk factors and www.nimhgenetics.org/resources/commonmind), the Nether- common diseases inferred from GWAS summary data. Nat. lands Twin Registry (https://tweelingenregister.vu.nl/), the Young Commun. 9, 224 (2018). Finns Study (https://youngfinnsstudy.utu.fi/), OMIM (https://www. 72. Hemani, G. et al. The MR-Base platform supports systematic omim.org/), OMIM genes description are publicly available: GFAP causal inference across the human phenome. eLife 7, e34408 (https://www.omim.org/entry/137780); SLC13A3 (https://www.omim. (2018). org/entry/618384); PNPT1 (https://www.omim.org/entry/610316), 73. Bowden, J. et al. Improving the visualization, interpretation and COSMIC (https://cancer.sanger.ac.uk), RNA sequencing datasets: analysis of two-sample summary data Mendelian randomization PsychENCODE DER-22 (www.ncbi.nlm.nih.gov/geo/, accession code via the Radial plot and Radial regression. Int. J. Epidemiol. 47, GSE97942), GSE67835 (www.ncbi.nlm.nih.gov/geo/, accession code 1264–1278 (2018). GSE67835), GSE101601, DroNc_Human Hippocampus (https://www. 74. Hemani, G., Tilling, K. & Davey Smith, G. Orienting the causal gtexportal.org/home/datasets), Allen Brain Atlas (http://portal.brain- relationship between imprecisely measured traits using GWAS map.org/), Descartes_Human (https://descartes.brotmanbaty.org/), summary data. PLoS Genet. 13, e1007081 (2017). Mousebrain (http://mousebrain.org/), Tabula Muris (https://tabula- 75. Morrison, J., Knoblauch, N., Marcus, J. H., Stephens, M. & He, muris.ds.czbiohub.org/). All other data supporting the findings of this X. Mendelian randomization accounting for correlated and study are available within the article, the supplementary information uncorrelated pleiotropic effects using genome-wide summary or the supplementary data files. statistics. Nat. Genet. 52, 740–747 (2020). 76. Mishra, A. & MacGregor, S. A novel approach for pathway analysis Code availability of GWAS data highlights role of BMP signaling and muscle cell The PVS quantification method used in the Nagahama Study is avail- differentiation in colorectal cancer susceptibility. Twin Res. Hum. able using this link: https://github.com/pboutinaud/SHIVA_PVS. We Genet. 20, 1–9 (2017). used publicly available data from METAL (https://github.com/stat 77. Amberger, J. S., Bocchini, C. A., Schiettecatte, F., Scott, A. F. & gen/METAL), GCTA-cojo (https://yanglab.westlake.edu.cn/software/ Hamosh, A. OMIM.org: Online Mendelian Inheritance in Man gcta/#COJO), FUMA (https://fuma.ctglab.nl/), MAGMA (https://ctg. (OMIM(R)), an online catalog of human genes and genetic cncr.nl/software/magma), LD-Score Regression (https://github.com/ disorders. Nucleic Acids Res. 43, D789–798 (2015). bulik/ldsc/), MTAG (https://github.com/JonJala/mtag), coloc (https:// 78. Gusev, A. et al. Integrative approaches for large-scale chr1swallace.github.io/coloc/) R package, RadialMR (https://github. transcriptome-wide association studies. Nat. Genet. 48, 245–252 com/WSpiller/RadialMR), TwoSampleMR (https://mrcieu.github.io/ (2016). TwoSampleMR/), STEAP (https://github.com/erwinerdem/STEAP/), 79. Sakaue, S. & Okada, Y. GREP: Genome for REPositioning drugs. CELLECT (https://github.com/perslab/CELLECT), S-LSC (https:// Bioinformatics 35, 3821–3823 (2019). github.com/bulik/ldsc), H-MAGMA (https://github.com/thewonlab/H- 80. Konuma, T., Ogawa, K. & Okada, Y. Integration of genetically MAGMA), HBT (https://hbatlas.org/). Drug discovery analysis was regulated gene expression and pharmacological library provides conducted using the following publicly available tools: GREP (https:// therapeutic drug candidates. Hum. Mol. Genet. 30, 294–304 github.com/saorisakaue/GREP), Trans-Phar (https://github.com/ (2021). konumat/Trans-Phar). Acknowledgements References Austrian Stroke Prevention Study (ASPS)/Austrian Stroke Prevention 61. Bouvy, W. H. et al. Perivascular spaces on 7 Tesla brain MRI Family Study (ASPS-Fam) (E.H., P.G.G., H.S. and R.S.): We thank the staff are related to markers of small vessel disease but not to age and the participants for their valuable contributions. We thank B. or cardiovascular risk factors. J. Cereb. Blood Flow. Metab. 36, Reinhart for her long-term administrative commitment, E. Hofer for the 1708–1717 (2016). technical assistance in creating the DNA bank, J. Semmler and A. Harb 62. Mishra, A. et al. Gene-mapping study of extremes of cerebral for DNA sequencing and DNA analyses by TaqMan assays, and I. Poelzl small vessel disease reveals TRIM47 as a strong candidate. Brain for supervising the quality management processes after ISO9001 in 30, 1992–2007 (2022). the biobanking and DNA analyses. The Medical University of Graz and 63. Winkler, T. W. et al. Quality control and conduct of genome-wide the Steiermärkische Krankenanstaltengesellschaft support the association meta-analyses. Nat. Protoc. 9, 1192–1212 (2014). databank of the ASPS/ASPS-Fam. The research reported in this article 64. Yang, J. et al. Conditional and joint multiple-SNP analysis of GWAS was funded by the Austrian Science Fund (FWF) (grant nos. PI904, summary statistics identifies additional variants influencing P20545-P05 and P13180) and supported by the Austrian National Bank complex traits. Nat. Genet. 44, S361–363 (2012). Anniversary Fund (grant no. P15435) and the Austrian Ministry of 65. Magi, R. et al. Trans-ethnic meta-regression of genome-wide Science under the aegis of the EU Joint Programme– association studies accounting for ancestry increases power Neurodegenerative Disease Research (JPND): www.jpnd.eu. for discovery and improves fine-mapping resolution. Hum. Mol. Epidemiology of Dementia in Singapore (EDIS) (S.H., C.Chen, C.-Y.C., Genet. 26, 3639–3650 (2017). T.Y.W. and W.Z.): The EDIS study is supported by the National Medical 66. Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Research Council (NMRC), Singapore (NMRC/CG/NUHS/2010 (grant Functional mapping and annotation of genetic associations with no. R-184-006-184-511), NMRC/CSA/038/2013) and a Ministry of FUMA. Nat. Commun. 8, 1826 (2017). Education Tier 1 grant (no. A-0006106-00-00). Framingham Heart Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Study (FHS) (J.R.R., A.B., J.J.H., S.L., P.P., C.L.S., Q.Y. and S.Seshadri): Evelyn F. McKnight Brain Institute. Rotterdam Study (M.J.K., F.D., This work was supported by the National Heart, Lung and Blood M.W.V., M.A.I. and H.H.H.A.): The Rotterdam Study is funded by Institute’s FHS Contract (no. N01-HC-25195, no. HHSN268201500001I Erasmus Medical Center and Erasmus University, Rotterdam, the and no. 75N92019D00031). This study was also supported by grants Netherlands Organization for Health Research and Development from the National Institute of Aging (R01 grant nos. AG031287, (ZonMw), the Research Institute for Diseases in the Elderly (RIDE), the AG054076, AG049607, AG059421, AG059725; U01 grant nos. Ministry of Education, Culture and Science, the Ministry for Health, AG049505, AG052409) and the National Institute of Neurological Welfare and Sports, the European Commission (DG XII), and the Disorders and Stroke (R01 grant no. NS017950). Funding for SHARe Municipality of Rotterdam. The authors are grateful to the study Affymetrix genotyping was provided by NHLBI Contract no. participants, the staff from the Rotterdam Study and the participating N02-HL64278. The computational work reported in this paper was general practitioners and pharmacists. The generation and performed on the Shared Computing Cluster which is administered by management of GWAS genotype data for the Rotterdam Study (RS I, Boston University’s Research Computing Services. We also thank all RS II, RS III) were executed by the Human Genotyping Facility of the the FHS study participants. Internet-based Students’ Health Research Genetic Laboratory of the Department of Internal Medicine, Erasmus Enterprise (i-Share) study (C.B., J.Z., M.M., Q.LG., S. Schilling, Y.-C.Z., MC, Rotterdam, the Netherlands. The GWAS datasets are supported by A.Tsuchida, M.-G.D., B.M., S.D. and C.T.): The i-Share study is conducted the Netherlands Organisation for Scientific Research (NWO) by the Universities of Bordeaux and Versailles Investments (no. 175.010.2005.011, 911-03-012), the Genetic Saint-Quentin-en-Yvelines (France). The i-Share study has received Laboratory of the Department of Internal Medicine, Erasmus MC, the funding by the French National Agency (Agence Nationale de la Research Institute for Diseases in the Elderly (grant no. 014-93-015; Recherche, ANR), via the Investment for the Future program (grant RIDE2), the Netherlands Genomics Initiative/NWO, the Netherlands nos. ANR-10-COHO-05 and ANR-18-RHUS-0002) and from the Consortium for Healthy Aging, project no. 050-060-810. We thank P. University of Bordeaux Initiative of Exellence (IdEX). This project has Arp, M. Jhamai, M. Verkerk, L. Herrera, M. Peters and C. Medina-Gomez also received funding from the European Research Council under the for their help in creating the GWAS database; and K. Estrada, Y. European Union’s Horizon 2020 research and innovation program Aulchenko and C. Medina-Gomez for the creation and analysis of under grant agreement no. 640643 and from the Fondation pour la imputed data. H.H.H.A. is supported by ZonMW grant no. 916.19.151. Recherche Médicale (grant no. DIC202161236446). Q.L.G. was Study of Health in Pomerania (SHIP) (S.F., R.B., A.T., K.W., H.J.G. and supported by the Digital Public Health Graduate Program (DPH), a PhD U.V.): SHIP is part of the Community Medicine Research net (CMR) program supported by the French Investment for the Future Program (http://www.medizin.uni-greifswald.de/icm) of the University Medicine (grant no. 17-EURE-0019). Investigating Silent Strokes in Greifswald, which is funded by the Federal Ministry of Education and Hypertensives: a Magnetic Resonance Imaging Study (ISSYS) (P.D., Research (grant nos. 01ZZ9603, 01ZZ0103 and 01ZZ0403), the C.C. and I.F.-C.): This research was funded by the Instituto de Salud Ministry of Cultural Affairs as well as the Social Ministry of the Federal Carlos III (grant nos. PI10/0705, PI14/01535, PI17/02222), cofinanced by State of Mecklenburg-West Pomerania, and the network ‘Greifswald the European Regional Development Fund. Lothian Birth Cohort 1936 Approach to Individualized Medicine (GANI_MED)’ funded by the (LBC1936) (M.L., M.E.B., I.J.D., Z.M., S.M.M., M.C.V.H. and J.M.W.): We Federal Ministry of Education and Research (grant no. 03IS2061A). thank the LBC1936 cohort members and research staff involved in Genome-wide data have been supported by the Federal Ministry of data collection, processing and preparation. The LBC1936 is Education and Research (grant no. 03ZIK012) and a joint grant from supported by Age UK (Disconnected Mind program grant). The work Siemens Healthineers, Erlangen, Germany and the Federal State of was undertaken by The University of Edinburgh Centre for Cognitive Mecklenburg-West Pomerania. The University of Greifswald is a Ageing and Cognitive Epidemiology, part of the cross-council Lifelong member of the Caché Campus program of the InterSystems GmbH. Health and Wellbeing Initiative (grant no. MR/K026992/1). The brain This study was further supported by the EU-JPND Funding for BRIDGET imaging was performed in the Brain Research Imaging Centre (www. (grant no. FKZ:01ED1615). H.J.G. has received travel grants and bric.ed.ac.uk), a center in the SINAPSE Collaboration (www.sinapse. speakers’ honoraria from Fresenius Medical Care, Servier, ac.uk) supported by the Scottish Funding Council and Chief Scientist Neuraxpharm and Janssen Cilag, as well as research funding from Office. Funding from the UK Biotechnology and Biological Sciences Fresenius Medical Care. Sydney Memory and Ageing Study (MAS) & Research Council (BBSRC), the UK Medical Research Council (MRC), Older Australian Twins Study (OATS) (R.M.T., N.J.A., H.B., J.J., M.P., A.T., the Row Fogo Charitable Trust (M.C.V.H.) and the UK Dementia J.N.T., P.S.S., W.W., K.A.M. and M.J.W.): Sydney MAS: The Sydney MAS Research Institute, which receives its funding from the UK Medical has been funded by three National Health & Medical Research Council Research Council, Alzheimer’s Society and Alzheimer’s Research UK (NHMRC) Program Grants (grant nos. ID350833, ID568969 and (J.M.W.), is gratefully acknowledged. Genotyping was supported by a APP1093083). Collection of WGS data was supported by the NHMRC grant from the BBSRC (no. BB/F019394/1). The Nagahama Study (T.K., National Institute for Dementia Research Grants no. APP1115575 and S.M., M.O., K.S., Y.T., K.Y., A.Tsuchida, P.B., B.M., M.J., M.-G.D. and F.M.): no. APP1115462. MRI scans were processed with the support of We are grateful to the Nagahama City Office and nonprofit NHMRC Project Grants (grant nos. 510175 and 1025243) and an organization Zeroji Club for their help in conducting the study. This Australian Research Council (ARC) Discovery Project Grant (no. project is supported by operational funds of Kyoto University and the DP0774213) and the John Holden Family Foundation. We also thank the Top Global University Project of the Ministry of Education, Culture, MRI Facility at NeuRA. We thank the participants and their informants Sports, Science and Technology (MEXT) in Japan. We also received a for their time and generosity in contributing to this research. We also Grant-in-Aid for Scientific Research from the Japan Society for the acknowledge the MAS research team: https://cheba.unsw.edu.au/ Promotion of Science, research grants from the Japan Agency for research-projects/sydney-memory-and-ageing-study. OATS: The OATS Medical Research and Development for the Practical Research Project study has been funded by an NHMRC and ARC Strategic Award Grant for Rare/Intractable Diseases, and the Comprehensive Research on of the Ageing Well, Ageing Productively Program (grant no. 401162); Aging and Health Science for Dementia R&D. We thank C. Galmiche for NHMRC Project (seed) Grants (grant nos. 1024224 and 1025243); rating PVS in the validation dataset for the artificial intelligence-based NHMRC Project Grants (grant nos. 1045325 and 1085606); and method. The Northern Manhattan Study (NOMAS) (N.D.D., T.J. and NHMRC Program Grants (grant nos. 568969 and 1093083). Collection R.L.S.): We gratefully acknowledge and thank the NOMAS participants. of WGS data was supported by the NHMRC National Institute for Funding was awarded through grants from the National Institute of Dementia Research Grants no. APP1115575 and no. APP1115462. This Neurological Disorders and Stroke (R01 grant no. NS 29993) and the research was facilitated through access to Twins Research Australia, Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w a national resource supported by a Centre of Research Excellence Genome Center (M.B., P.M., G.B. and M.Lathrop): Work done at the Grant (no. 1079102) from the National Health and Medical Research Canadian Center for Computational Genomics was supported by Council. We thank the participants for their time and generosity in Genome Canada. Data analyses were enabled by computing and contributing to this research. We acknowledge the contribution of the storage resources provided by Compute Canada and Calcul Québec. OATS research team (https://cheba.unsw.edu.au/project/ G.B. is supported by the Fonds de Recherche Santé Québec and the older-australian-twins-study) to this study. Three-City Dijon Study Canada Research Chair program. We thank all the participating (3C-Dijon) (S.D., M.-G.D., S. Schilling, C.T., B.M. and A.M.): This project cohorts for contributing to this study. We thank H. Jacqmin-Gadda, is supported by a grant overseen by the French National Research Bordeaux Population Health research center, University of Bordeaux/ Agency (ANR) as part of the ‘Investment for the Future Program’ no. Inserm U1219 for statistical advice. We thank J. Thomas-Crusells, ANR-18-RHUS-0002. It is also supported by a JPND project through the Bordeaux Population Health Research Center, University of Bordeaux/ following funding organizations under the aegis of JPND: www.jpnd. Inserm U1219, for editorial assistance. The funders had no role in study eu: Australia, National Health and Medical Research Council; Austria, design, data collection and analysis, decision to publish or preparation Federal Ministry of Science, Research and Economy; Canada, of the manuscript. Canadian Institutes of Health Research; France, French National Research Agency; Germany, Federal Ministry of Education and Author contributions Research; the Netherlands, the Netherlands Organisation for Health F.M., C.T., J.M.W., S.Seshadri, H.H.H.A. and S.D. jointly supervised Research and Development; United Kingdom, Medical Research research. M.-G.D., M.J.K., Q.LG., T.E.E., A.M. and A.Tsuchida contributed Council. This project has received funding from the European Union’s equally. M.-G.D., M.J.K., H.H.H.A. and S.D. designed and conceived the Horizon 2020 research and innovation program under grant study. J.R.R., S.F., M.L., E.H., M.B., N.D.D., P.D., S.H., R.M.T., F.D., J.S., Y.S., agreement nos. 643417, 640643, 667375 and 754517. The project also N.J.A., C.B., M.E.B., A.B., H.B., R.B., C. Carrera, C.Chen, C.-Y.C., I.J.D., received funding from the French National Research Agency (ANR) P.G.G., J.J.H., J.J., T.K., S.L., M.M., P.M., Z.M., S.M.M., S.M., M.O., M.P., P.P., through the VASCOGENE and SHIVA projects, and from the Initiative of T.R., M.S., S. Schilling, K.S., O.S., Y.T., A.T., A.Thalamuthu, J.N.T., M.C.V.H., Excellence of the University of Bordeaux (C-SMART project). M.W.V., U.V., K.W., T.Y.W., M.J.W., J.Z., W.Z., Y.-C.Z., H.S., P.S.S., W.W., K.Y., Computations were performed on the Bordeaux Bioinformatics Center C.L.S., R.L.S., G.B., M.Lathrop, I.F.-C., Q.Y., B.M., P.B., H.J.G., K.A.M., (CBiB) computer resources, University of Bordeaux. Funding support R.S., M.J. and M.A.I. generated the PVS phenotype and genomic data for additional computer resources at the CREDIM (Centre de and conducted cohort-wise GWAS analyses. G.R., T.K., D.-A.T., A.J., Recherche et Développement en Informatique Médicale, University of T.P. and Y.O. contributed to bioinformatics analyses. M.-G.D., S.D., Bordeaux) has been provided to S.D. by the Fondation Claude M.J.K., Q.L.G., T.E.E., A.M., A.Tsuchida, F.M., C.T., J.M.W., S.Seshadri and Pompidou. The Three-City (3C) Study: The 3C Study is conducted H.H.H.A. wrote and edited the manuscript. under a partnership agreement among the Institut National de la Santé et de la Recherche Médicale (INSERM), the University of Bordeaux and Competing interests Sanofi-Aventis. The Fondation pour la Recherche Médicale funded the The authors declare no competing interests. preparation and initiation of the study. The 3C Study is also supported by the Caisse Nationale Maladie des Travailleurs Salariés, Direction Additional information Générale de la Santé, Mutuelle Générale de l’Education Nationale Extended data is available for this paper at (MGEN), Institut de la Longévité, Conseils Régionaux of Aquitaine and https://doi.org/10.1038/s41591-023-02268-w. Bourgogne, Fondation de France and Ministry of Research–INSERM program ‘Cohortes et collections de données biologiques.’ C.T. and Supplementary information The online version contains supplementary S.D. have received investigator-initiated research funding from the material available at https://doi.org/10.1038/s41591-023-02268-w. French National Research Agency (ANR) and from the Fondation Leducq. M.-G.D. received a grant from the ‘Fondation Bettencourt Correspondence and requests for materials should be addressed to Schueller’. We thank P. Amouyel, U1167 Institut Pasteur de Lille - Hieab H. H. Adams or Stéphanie Debette. University of Lille - Inserm, for supporting funding of genome-wide genotyping of the 3C Study. This work was supported by the National Peer review information Nature Medicine thanks Andreas Charidimou, Foundation for Alzheimer’s disease and related disorders, the Institut Timothy Frayling and the other, anonymous, reviewer(s) for their Pasteur de Lille, the labex DISTALZ and the Centre National de contribution to the peer review of this work. Primary Handling Editor: Génotypage. We thank A. Boland (CNG) for her technical help in Anna Maria Ranzoni, in collaboration with the Nature Medicine team. preparing the DNA samples for analyses. UK Biobank (UKB) (M.J.K., F.D., M.W.V., M.A.I., H.H.H.A., A.M. and T.E.): This research has been Reprints and permissions information is available at conducted using the UK Resource under application no. 23509. McGill www.nature.com/reprints. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 1 | Manhattan and QQ plots of extensive PVS burden in genomic inflation factor (λ). For QQ plots the observed -log10(p) is represented the cross-ancestry and European meta-analyses. Manhattan and Quantile- in the y-axis and the expected -log10(p) in x-axis. The dotted line corresponds to Quantile (QQ) plots of the p-values (observed versus expected) in the cross- the genome-wide significance threshold (p = 5×10−8, two sided). PVS indicates ancestry (A) and European (B) GWAS meta-analyses are presented along with the perivascular spaces; WM, white matter; BG, basal ganglia; HIP, hippocampus. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 2 | Two-sample Mendelian randomization analysis are shown. Each dot (or triangle if p < 1.19×10−3) represents the beta estimate from between putative risk factors and extensive PVS burden. Two-sample Mendelian randomization with lines representing the 95% confidence interval. Mendelian randomization was conducted using European PVS GWAS summary Two-sided p-values of GSMR are reported. PVS indicates perivascular spaces; statistics (N = 38,598 (WM-PVS), N = 38,903 (BG-PVS) and N = 38,871 (HIP-PVS)), WM, white matter; BG, basal ganglia; HIP, hippocampus; MTAG, multi-trait combined with summary statistics for lacunes (N = 1,715 cases / N = 15,096 analysis of genome-wide association summary statistics; DBP, diastolic blood controls) and WMH volume (N = 48,454) in MTAG, for the outcomes and pressure; PP, pulse pressure; SBP, systolic blood pressure; NS, not significant; European GWAS summary statistics for blood pressure traits (N = 757,601) to GSMR, Generalised Summary-data-based Mendelian Randomisation; IVW, generate instruments for exposures (Supplementary Tables 1 and 29). Only Inverse variance weighted; 2SMR, Two-SampleMR. significant associations after multiple testing correction (p < 1.19×10−3) in GSMR Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 3 | Two-sample Mendelian randomization analysis (Supplementary Tables 1 and 29). Only significant associations after multiple between extensive PVS burden and neurological traits. Two-sample testing correction (p < 1.19×10−3) in GSMR are shown. Each dot (or triangle if Mendelian randomization was conducted using European GWAS summary p < 1.19×10−3) represents the beta estimate from Mendelian randomization with statistics for any stroke (N = 40,585 cases / N = 406,111 controls), ischemic stroke lines representing the 95% confidence interval. Two-sided p-values of GSMR are (N = 34,217 cases / N = 400,201 controls) and small vessel stroke (N = 5,386 cases / reported. PVS indicates perivascular spaces; WM, white matter; BG, basal ganglia; N = 254,558 controls) for outcomes and European PVS GWAS summary statistics HIP, hippocampus; MTAG, multi-trait analysis of genome-wide association (N = 38,598 (WM-PVS), N = 38,903 (BG-PVS) and N = 38,871 (HIP-PVS)), combined summary statistics; NS, not significant; GSMR, Generalised Summary-data- with summary statistics for lacunes (N = 1,715 cases / N = 15,096 controls) and based Mendelian Randomisation; IVW, Inverse variance weighted; 2SMR, WMH volume (N = 48,454) in MTAG, to generate instruments for exposures Two-SampleMR. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 4 | Enrichment of PVS risk loci in genes mutated in OMIM p-values for all PVS loci: p = 0.041 (genes with intragenic SNPs), p = 9,27×10−06 syndromes. Enrichment of all perivascular spaces (PVS) loci (left) and WM-PVS (genes with intragenic lead SNPs); exact p-values for WM-PVS loci only: loci only (right) in genes mutated in OMIM syndromes associated with white p = 0,018 (genes within 10 kb distance), p = 0.008 (genes with intragenic SNPs), matter hyperintensities, leukodystrophy, leukoencephalopathy, according to p = 8,61×10−08 (genes with intragenic lead SNPs), full results are provided in distance from the lead variant; * p < 0.05; ** p < (0.05/5); *** p < (0.05/5/2). Exact Supplementary Table 22. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 5 | Brain expression pattern across the lifespan of genes 1-6 years (period 10), 6-12 years (period 11), 12-20 years (period 12), 20-40 years near genome-wide significant PVS loci that are also identified in the TWAS (period 13), 40-60 years (period 14), and 60 years + (period 15). The boundary and peaking in the pre-natal period (UMPS and LAMC1). The figure displays between pre- and postnatal periods is indicated by the solid vertical line. Each brain expression patterns across the lifespan of the 2 significant PVS TWAS- colored point represents the expression level of each gene across 16 anatomical COLOC genes from genome-wide significant PVS GWAS loci peaking in the brain regions and ages. Brain structure includes 11 neocortical areas (NCX, blue), pre-natal period (UMPS for WM-PVS and LAMC1 for HIP-PVS) (see Supplementary and 5 subcortical regions: hippocampus (HIP, cyan), amygdala (AMY, orange), Table 4 for brain expression pattern across the lifespan of other genes). The striatum (STR, black), mediodorsal nucleus of thalamus (MD, dark green), and spatio-temporal gene expression level is plotted as log2-transformed exon cerebellar cortex (CBC, red). Neocortical areas include orbital prefrontal cortex array signal intensity (y-axis) against the post conception days (x-axis) as (OFC), dorsolateral prefrontal cortex (DFC), ventrolateral prefrontal cortex provided by the Human Brain Transcriptome project database. Periods of human (VFC), medial prefrontal cortex (MFC), primary motor cortex (M1C), primary development and adulthood are indicated by vertical dashed lines: 4-8 post somatosensory cortex (S1C), posterior inferior parietal cortex (IPC), primary conception weeks [PCW] (period 1), 8-10 PCW (period 2), 10-13 PCW (period 3), auditory cortex (A1C), posterior superior temporal cortex (STC), inferior 13-16 PCW (period 4), 16-19 PCW (period 5), 19-24 PCW (period 6), 24-38 PCW temporal cortex (ITC), and primary visual cortex (V1C). (period 7), birth-6 postnatal months (period 8), 6-12 postnatal months (period 9), Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 6 | See next page for caption. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 6 | Enrichment of PVS genes in targets of drugs validated for davunetide (B). We also found a significant enrichment of HIP-PVS genes in other indications (Genome for REPositioning drugs, using ICD10 codes). in targets of drugs for diseases of the ear (ICD10 codes H90-H95, OR = 57.6, Using the Genome for REPositioning drugs (GREP) software (A), we found a p = 2.34×10−2), driven by SERPIND1 (chr22q11.21) a target for sulodexide, also significant enrichment of BG-PVS genes in targets of drugs for diseases of the used for the prophylaxis and treatment of vascular diseases with increased risk of nervous system (ICD10 codes G50-G59, G70-G73, and G80-G83, OR = 51.0, 49.7, thrombosis (B). PVS indicates perivascular spaces; ATC, Anatomical Therapeutic and 60.4, p = 3.90×10−2, 4.00×10−2, and 3.32×10−2) and for symptoms and signs Chemical classification; ICD10, the 10th revision of the International Statistical involving cognition, perception, emotional state and behaviour (ICD10 codes Classification of Diseases and Related Health Problems; TTD, Therapeutic R40-R46, OR = 48.4, p = 4.10×10−2), driven by MAPT (chr17q21.31), a target Target Database. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 7 | Enrichment of PVS genes in targets of drugs CRHR1 (chr17q21.31), a target for telavancin (A, B); of note, CRHR1 antagonists validated in other indications (Genome for REPositioning drugs, using were shown to attenuate blood brain barrier permeability, cortical vascular ATC codes). We found a significant enrichment of BG-PVS genes in targets for hyperpermeability and tight junction disruption. ATC indicates Anatomical antiinfectives for systemic use (ATC J01, OR = 252.4, P = 8.85×10−3), driven by Therapeutic Chemical classification. Nature Medicine Article https://doi.org/10.1038/s41591-023-02268-w Extended Data Fig. 8 | Enrichment of PVS genes in targets of drugs validated L1000 library cell types) was performed to identify up- and down-regulated in other indications (Trans-Phar). We used the Integration of Transcriptome- genes in participants with extensive PVS burden, and select the top 10% genes wide Association Study and Pharmacological Database (Trans-Phar) software, with the highest expression variation. We observed significant enrichment of leveraging TWAS on all GTEx v7 tissues and a database of compound- HIP-PVS in drugs for vascular diseases, including simvastatin (lipid-lowering regulated gene expression (C-map) (Methods). A TWAS using FOCUS, which drug, p = 1.64×10−4), vincamine (vasodilator increasing cerebral blood flow, demonstrates fine-mapping of causal gene sets from TWAS results, and 27 p = 2.87×10−4), and macitentan (used for pulmonary arterial hypertension, tissues in the GTEx v7 database (corresponding to defined 13 tissue-cell-type p = 7.46×10−4). PVS indicates perivascular spaces. categories assigned by the 27 tissues in GTEx v7 database and 77 LINCS CMap Nature Medicine