Abstract
Brain glymphatic activity, as indicated by diffusion-tensor imaging analysis along the perivascular space (ALPS) index, is involved in developmental neuropsychiatric and neurodegenerative diseases, but its genetic architecture is poorly understood. Here, we identified 17 unique genome-wide significant loci and 161 candidate genes linked to the ALPS-indexes in a discovery sample of 31,021 individuals from the UK Biobank. Seven loci were replicated in two independent datasets. Genetic signals located at the 2p23.3 locus yielded significantly concordant effects in both young and aging cohorts. Genetic correlation and polygenic overlap analyses indicate a common underlying genetic mechanism between the ALPS-index, ventricular volumes, and cerebrospinal fluid tau levels, with GMNC (3q28) and C16orf95 (16q24.2) as the shared genetic basis. Our findings enhance the understanding of the genetics of the ALPS-index and provide insight for further research into the neurobiological mechanisms of glymphatic clearance activity across the lifespan and its relation to neuropsychiatric phenotypes.
Similar content being viewed by others
Introduction
The glymphatic system is a critical fluid drainage pathway in the exchange of cerebrospinal fluid (CSF) and interstitial fluid1. It plays a key role in maintaining brain homeostasis through brain fluid regulation and clearance of metabolic wastes (such as beta-amyloid[Aβ] and tau1,2), and potentially functions in brain immunity3. Above processes were described to be regulated by sleep and circadian rhythm4,5, which is important in the development of early brain and neurocognition in children and adolescents6,7,8. The glymphatic dysfunction is associated with developmental neuropsychiatric disorders, such as attention-deficit/hyperactivity disorder (ADHD)10 and autism spectrum disorder (ASD)11. In addition, accumulating evidence has suggested that glymphatic failure is an important pathological pathway in neurodegenerative diseases12, such as Alzheimer’s disease (AD)13.
Given the importance of the glymphatic system in human health and disease across the lifespan, it is crucial to identify its genetic factors throughout brain development and aging. Recently, a genome-wide association study (GWAS) identified multiple genetic loci associated with perivascular space (PVS) measured by magnetic resonance imaging (MRI)14, which is a structural component of the glymphatic system1. Another non-invasive indicator of conditions related to the glymphatic system in humans other than the PVS is the diffusion tensor image (DTI) analysis along the perivascular space (ALPS) index15,16. Compared to the PVS measuring the morphology of the glymphatic system, the ALPS-index measures the fluid diffusivity from the compartment along the PVS direction in those perpendicular projection fibers and association fibers15. Consistent with experimental findings1,17,18, mounting studies have observed a decrease of ALPS-index under conditions where glymphatic function is impaired, such as in neurodegenerative diseases and in the presence of Alzheimer’s pathologies19,20,21, demonstrating the potential of the ALPS-index in evaluating the glymphatic fluid dynamics22. However, to our knowledge, the genetic mechanisms of the ALPS-index have not been studied.
What’s more, the development of the glymphatic system is a long-lasting process that continues from gestation to adulthood9. In rodents, the glymphatic function gradually increases from neonatal to adult stages23,24 but declines with age25. Similarly, in the human brain, the ALPS-index increases with age, peaks in middle age (40s), and then gradually declines26,27. Accordingly, the causes and consequences of changes in the ALPS-index may differ between early and late life stages, which reflecting brain development and degeneration, respectively. We hypothesized that the genetic factors related to the ALPS-index in young brains are likely to differ significantly from those in aging brains. Discovering the genetic factors that explain the variation of the ALPS-index in young and aging brains may provide new insights into the development and aging process of the glymphatic system and may help identify disease risk and resilience.
Here, we conduct a GWAS to identify genetic variants, loci, and genes associated with the ALPS-index in the UK Biobank (UKB discovery cohort, N = 31,021, aged 45–82 years) and replicated the findings in two independent samples (UKB non-British white, N = 3470, aged 45–81 years; ADNI, N = 260, aged 55–75 years). Because it was uncertain whether the previously observed differences of bilateral ALPS-indexes in neurodegenerative diseases28,29 were due to measurement bias or were of clinical significance, we analyzed the mean ALPS-index (the primary phenotype), as well as the ALPS-indexes of the right and left hemispheres separately as a mutual validation or exploratory analysis. Second, we explore the concordance and discordance of genetic effects on the ALPS-index across the lifespan using the UKB discovery cohort, ADNI, and three other independent cohorts including ABCD (N = 4307, aged 9–11 years), IMAGEN (N = 1589, aged 13–18 years), HCP (N = 219, aged 22–36 years). We further investigate the genetic factors that are associated with longitudinal changes in ALPS-index. Functional annotation, as well as genetic correlation and shared genetic determinants with neuropsychiatric phenotypes, are subsequently examined to investigate the biological significance of identified loci.
Results
Extraction and evaluation of the ALPS-index
A total of 40,866 participants from five cohorts (UKB, ADNI, IMAGEN, HCP, and ABCD) ranging in age from children to the elderly (9–82 years, Table 1) were included in this study. The discovery GWAS analysis comprised data from 31,021 UKB brain imaging samples (53.0% female; age range 45–82 years; Table 1) of white British ancestry free of prevalent dementia, stroke, hydrocephalus, and intracranial tumors. A schematic overview of the study design and an illustration of the ALPS-index processes are shown in Supplementary Fig. 1 and Fig. 1. The ALPS-index was extracted from DTI in each cohort using a semi-automated and highly reliable pipeline15. The assessment of segmentation accuracy in the ALPS-index extraction prioritized the placement of regions of interest (ROIs), which was performed in the UKB and IMAGEN datasets (Methods). It was demonstrated that the ALPS-index derived from manual ROIs and group-average ROI templates are largely comparable in both elderly (e.g., the UKB participants; Pearson r = 0.77, p = 1.45 × 10−15) and young populations (e.g., the IMAGEN participants; Pearson r = 0.81, p = 1.08 × 10−12; Supplementary Fig. 2). Distribution of the ALPS-index in different cohorts are displayed in Table 1 and Supplementary Fig. 3. Our preceding research has demonstrated the plausibility and clinical relevance of the generated ALPS-index based on UKB and ADNI participants30, where the ALPS-index was associated with Alzheimer’s disease risk, Alzheimer’s disease biomarkers, and cognition. This current study further demonstrating that the lower ALPS-index was associated with older age, male sex, worse cognitive performance, larger white matter hyperintensities (WMH), less hand grip strength, lower body mass index, and higher blood pressure, but not with education, sleep duration, or handedness (Supplementary Data 1).
Analysis process from initial diffusion-weighted imaging images to calculate analysis along the perivascular space indexes. Horizontal slices at the ventricle level were selected to display the regions of projection fibers and association fibers, oriented perpendicular to the ventricular body. Color-coded FA map shows the projection fibers (blue) and association fibers (green). Four spherical ROIs, each with a radius of 4 mm, were placed in the bilateral area of the projection fibers (yellow solid circle) and the bilateral area of the association fibers (white solid circle). Then, we obtained ROI probability maps in a standard space, which were binarized into automated ROIs. Diffusivities along the x-, y-, and z-axes were extracted for each automatically binarized ROI. ALPS-index was calculated as the ratio of diffusivities perpendicular to fiber bundles and parallel to veins (Dxproj and Dxassoc) over diffusivities perpendicular to fiber bundles and perpendicular to veins (Dyproj and Dzassoc). ALPS analysis along the perivascular space, DWI diffusion-weighted images, FA fractional anisotropy, MNI Montreal Neurological Institute, ROIasso regions of interest in associative fibers, ROIproj regions of interest in projective fibers.
Genetic discovery of the ALPS-index
The quantile–quantile (QQ) plot of the genome-wide association analysis of the mean ALPS-index (mALPS-index) at 8,447,093 SNPs showed inflation (linkage disequilibrium score regression (LDSC): λGC = 1.159, Mean χ² = 1.195; Supplementary Fig. 4 and Supplementary Data 2), yet the LDSC intercept (1.005; standard error (SE) = 0.023) and ratio (2.7%) both indicated that the observed genomic inflation was due to the polygenic architecture rather than confounded by population stratification. Single nucleotide polymorphism (SNP)-based heritability (h2SNP) of the mALPS-index was 28.4% (SE = 0.023) using LDSC and 32.5% (SE = 0.014) using GCTA, demonstrating that mALPS is a heritable trait.
Genome-wide association analysis for the mALPS-index identified 554 genome-wide significant SNPs (p < 5 × 10−8) in the UKB discovery cohort (Supplementary Data 3). We then utilized the FUMA pipeline31 to discovered 25 independent significant variants (r2 < 0.6) and 15 lead variants (r2 < 0.1) of mALPS-index, which mapped to 15 unique genomic loci (Table 2, Fig. 2A, and Supplementary Data 4, 5), spanning from 14.98 to 526.18 kb (Fig. 2B and Supplementary Fig. 5). All lead variants were successfully confirmed by conditional and joint (COJO) analysis32 (Supplementary Data 6). We then used FINEMAP33 to identify variants with a high posterior probability of causality (>0.6) (Supplementary Data 7). The lead SNP for the NUAK family SNF1-like kinase 1 (NUAK1), rs12146713, was the variant with the highest posterior probability (0.99) across all loci, followed by the SNP rs6012259 (posterior probability 0.69), mapped in the gene of SULF2. Detailed results of candidate SNPs are summarized in Fig. 2C, Supplementary Results, and Supplementary Data 8–12. To explore the effect of rare variants on ALPS-indexes, rare variant association studies (RVAS) were performed using the SKAT-O test on whole-exome sequencing data from the UKB. Gene-based burden tests and single-variant analyses did not identify any significant association after Bonferroni correction, whereas several subthreshold associations were observed (Supplementary Data 13, 14). The strongest associations were observed for the gene HPS5 and the mALPS-index (pSKAT-O = 9.74 × 10−5).
A Manhattan plot of the genome-wide association study (GWAS) results for the mean ALPS-index in the UKB discovery cohort. The plot shows the −log10-transformed P value of each SNP on the y-axis and base-pair positions along the chromosomes on the x-axis (n = 37,327 individuals). P values shown are two-sided. The dashed line indicates genome-wide significance for the GWAS performed in linear additive regression model (P < 5 × 10−8). Significant loci are marked as yellow squares and labeled by their lead SNP. B Overview of the sizes of genomic risk loci and the number of variants associated with the mean ALPS-index. C Pie charts showing the distribution of functional consequences of SNPs in LD with genome-wide significant lead SNPs in the discovery GWAS of the mean ALPS-index, the minimum chromatin state across 127 tissue and cell types, and the distribution of RegulomeDB score (a categorical score between 1a and 7, indicating biological evidence of a SNP being a regulatory element, with a low score denoting a higher likelihood of a SNP being regulatory). SNP single nucleotide polymorphism.
Stratified LDSC34 further showed that significant enrichment for h2 resulted from background selection (enrichment = 1.47, p = 5.21 × 10−12), CpG content (enrichment = 1.15, p = 3.70 × 10−5), histone marks (H3K27ac, H3K4me1, and H3K9ac), and SNPs located in conserved regions (GERP NS; enrichment = 1.86, p = 8.01 × 10−6) and super-enhancers (enrichment = 2.29, p = 5.55 × 10−6; Supplementary Data 15), indicating epigenetic gene regulation.
Since the mALPS-index was averaged from the ALPS-indexes in left and right hemisphere, we further examined the asymmetry of the ALPS index by comparing the genetic signal of the mALPS-index with the left and right ALPS-indexes. In the analyses of the left and right ALPS-indexes, the results were strongly correlated with those of the mALPS-index (Supplementary Data 2–5). Two additional risk loci, 3q28 (GMNC) and 12q23.1 (ANKS1B), were identified as significant for the right ALPS-index (Table 2, Supplementary Figs. 4–6, and Supplementary Data 4). The phenotypic and genetic correlations of the three ALPS-indexes were all significant (Supplementary Data 16). Sensitivity analyses were performed to validate the robustness of independent significant SNPs in the genetic association analysis. As shown in Supplementary Data 17, 85.7% of SNPs remained genome-wide significant when mean FA and MD were used as additional covariates, indicating minimal potential confounding by white matter integrity.
Genetic architecture of the ALPS-index across the lifespan
Replication was conducted in two independent samples: 3470 UKB participants of non-British white descent who were excluded from the discovery cohort and 260 non-demented participants of non-Hispanic white descent from the ADNI cohort (Table 1 and Supplementary Fig. 7). Of the 25 independent significant risk variants identified for the mALPS-index, 14 (60.0%) variants were confirmed in at least one replication cohort (p < 0.05). Eight of these variants reached significance after FDR correction, and their effect directionality was consistent with that observed in the discovery analysis (Supplementary Data 5).
As the ALPS-index demonstrated significant associations with age (Supplementary Figs. 8, 9 and Supplementary Data 18), we hypothesized that certain genetic loci may have varying effects on the ALPS-index across the lifespan. To test this hypothesis, we examined the associations of the identified independent significant variants with ALPS-indexes in independent cohorts of young adults as generalization analyses. The study involved three generalization cohorts: ABCD (N = 4307, mean age 9.91 ± 0.62 years), IMAGEN (N = 1589, 13.97 ± 0.40 years), and HCP (N = 219, 28.89 ± 3.68 years) (Table 1). Seven risk variants of the mALPS-index demonstrated nominally significant associations with consistent direction of effect with the discovery results in at least one cohort, four of which reached significance after FDR correction (Supplementary Data 5). These consistent findings suggest that the genetic associations are robust throughout the lifespan. For instance, variants in the 2p23.3 locus were consistently associated with the mALPS-index across different age cohorts, except for HCP and ADNI, which had the smallest sample size (Fig. 3A). The polygenic risk score (PRS) derived from the UKB discovery GWAS variants was significantly associated with the mALPS-index (β = 0.19, p = 1.89 × 10−31 in UKB non-British white subgroup, and β = 0.14, p = 0.02 in ADNI) in both replication cohorts, but not in generalization cohorts.
A Fixed-effects meta-analysis of rs4665946 (2:27347931) in locus 2p23.3 for the mean ALPS-index. B and C Meta-regression of SNP effects according to age was conducted using linear and quadratic models. P values shown are two-sided and uncorrected for multiple comparisons. Plots show the results for two variants that passed the FDR correction. The x-axis represents the median age of each cohort and the y-axis denotes the genetic effect on the mean ALPS-index. The dot size indicates the −log10 p-value of the genetic effect on the mean ALPS-index. The error band around the regression line indicates the 95% CI. D Forest plots show genetic effects of rs6012259 (20:46374885) at locus 20q13.12 and rs12146713 (12:106476805) at locus 12q23.3. CI confidence interval, SNP single nucleotide polymorphism, mALPS mean ALPS-index.
The genetic effects of independent significant variants were subsequently tested in seven subgroups from four independent cohorts: ABCD (age range, 9–11 years), IMAGEN (13–18 years), HCP (22–36 years), ADNI (55–75 years), and three UKB subgroups (age range after dividing the sample size by tertiles: 45–61, 62–68, 69–82; Supplementary Data 19). A meta-regression of genetic effects by age was conducted using linear and quadratic models after aligning the alleles. The analysis identified six variants that showed linear or quadratic associations with age after FDR correction (Supplementary Data 20). Notably, the lead variant rs6012259 in 20q13.12 of mALPS-index may be suggested to be involved in neurodegenerative processes, as its negative effect was more pronounced in older adults (Fig. 3B). The genetic effect of variants in the 16q24.2 locus yielded quadratic associations with age (Fig. 3C).
Leveraging the longitudinal DTI data from young (ABCD: N = 2152, baseline age = 9.91 ± 0.61 years) and aging cohorts (UKB: N = 3580, baseline age = 62.07 ± 7.35 years), we observed that the mALPS-index increased over a mean (SD) follow-up of 1.99 (0.14) years in ABCD (paired Wilcoxon p = 6.7 × 10−4), and decreased over a mean (SD) follow-up of 2.60 (0.99) years in UKB (paired Wilcoxon p < 2.22 × 10−16) (Supplementary Fig. 10). Glymphatic function measured by ALPS-index accumulates overtime in young brains (mALPS-index annual change rate per year: 0.004 ± 0.06), but declines overtime in aged brains (−0.005 ± 0.02) (Table 1).
We examined whether the independent significant variants identified for the mALPS-index in the discovery of GWAS could be validated in the analysis of its annual change rate. Two variants in ABCD (rs4757810 and rs12146713) and one in UKB (rs6012259) were significant (P < 0.05) and showed consistent direction with those in the discovery GWAS (Fig. 3D and Supplementary Data 21). In both the longitudinal analysis and the meta-regression by age, the lead SNP at locus 20q13.12 (rs6012259) consistently exerted much stronger genetic influences on the mALPS-index in neurodegeneration than in neurodevelopment (Fig. 3B, D and Supplementary Data 20, 21).
Genetic associations with neuropsychiatric phenotypes and vascular risk factors
We examined whether variants related to ALPS-indexes were associated with other phenotypes by performing association lookups in the NHGRI-EBI GWAS Catalog35 and previously published PVS study14. Out of the 17 loci associated with ALPS-indexes, four (12q23.1, 14q23.2, 2q23.3, and 3q26.2) were not previously reported, while the remaining 13 loci have all been previously reported to be related to structural MRI-derived metrics (Supplementary Data 22). Our results also tagged genetic loci that shared genetic influences with cognitive measures (2p22.2, 2p23.3, 5q15, 17q21.31, and 22q13.1), neuropsychiatric traits (2p23.3, 3q28, 11q12.2, and 17q21.31), and cardiovascular traits (2q22.2, 2p23.3, 11q12.2, 14q32.2, and 17q21.31) (Supplementary Data 22). In particular, the lead variants (rs10817103) of the LPAR1 (9q31.3) locus in our mALPS GWAS demonstrated an association with WM-PVS burden at a genome-wide significant p-value (P = 3.7 × 10−13) (Supplementary Data 23).
We employed LD-score regression to better understand the correlations between the ALPS-index and brain structure, white matter integrity, and brain function (Supplementary Data 24). All ventricular volume measures were significantly correlated with all three ALPS-indexes (Fig. 4A), with the most significant correlations found for lateral ventricular (in both sides: rg = −0.47 to −0.42, PBonferroni = 4.67 × 10−19 to 3.64 × 10−16) volume. Regarding the brain microstructures (Fig. 4B), the most significant genetic correlations with all three ALPS-indexes were observed in the fornix (rg = 0.42–0.43, PBonferroni = 1.42 × 10−14 to 2.92 × 10−12). However, no notable correlation was identified regarding the brain functional signatures after Bonferroni correction (Supplementary Data 24). We then examined the genetic correlations between ALPS-index and other neuropsychiatric phenotypes and vascular risk factors, including putative risk factors, cognition, neuropsychiatric and neurological disorders, CSF biomarkers of AD, and MRI-markers of cerebral small vessel disease (cSVD) (Supplementary Data 25). While no statistically significant associations were observed after applying the Bonferroni correction, nominally significant correlations were observed between WMH volume (rg = −0.27 to −0.26, P = 0.008–0.012) and all three ALPS-indexes. Greater left-hand grip strength, shorter reaction time, and higher pulse pressure were genetically correlated with at least one ALPS-index at the nominally significant level (Fig. 4C).
A Genetic correlation between the mean ALPS-index and brain volumes. All regions that are genetically correlated with the mean ALPS-index at nominal P < 0.05 are colored based on the magnitude of positive (red) or negative (blue) genetic correlation estimates. Regions with genetic correlation estimates that survived false discovery rate corrections are annotated with text. B Volcano plot displays the genetic correlation estimates (x-axis) and −log10 (p-value) (y-axis) between the mean ALPS-index and white matter microstructures. The top five significant white matter microstructures were labeled in the figure. C A bubble plot shows the genetic correlation between three ALPS-indexes and putative risk factors, cognitive measures, and other MRI markers of brain aging. The dot color represents the magnitude of positive (red) or negative (blue) genetic correlation, while the dot size denotes the −log10(P) of the genetic correlation. Correlation estimates with nominal P < 0.05 are annotated with dashed circles, and those that survived false discovery rate corrections are annotated with solid circles. The genetic correlations were estimated using LDSC. All p-values shown are two-tailed and are original p-values without multiple comparisons. AD Alzheimer’s disease, ADHD attention-deficit/hyperactivity disorder, ASD autism spectrum disorder, BG-PVS perivascular space in basal ganglia, BP bipolar disorder, cSVD cerebral small vessel disease, FTD frontotemporal dementia, BMI body mass index, CSF, cerebrospinal fluid, DBP diastolic blood pressure, FA fractional anisotropy, FDR false discovery rate, GC genetic correlation, HIP-PVS perivascular space in hippocampus, MRI magnetic resonance imaging, SBP systolic blood pressure, PP pulse pressure, WM-PVS perivascular space in white matter, WMH white matter hyperintensities.
We subsequently explored the polygenetic architecture of the ALPS-indexes and potential genetic overlap with neuropsychiatric phenotypes and vascular risk factors using conditional FDR (condFDR) and conjunctional FDR (conjFDR) analysis. By conditioning on the significance of associations with neuropsychiatric phenotypes and vascular risk factors at condFDR < 0.01, the power of loci discovery for the ALPS-indexes was markedly enhanced, with the number of identified genetic loci exceeding that of the discovery GWAS (each >15 in Fig. 5A). The conditional Q–Q plots indicated a clear genetic overlap between the ALPS-indexes and SCZ, HIP-PVS, WMH, CSF pTau, and CSF tau (Fig. 5B and Supplementary Fig. 11). At the conjFDR <0.05 threshold, several shared loci were identified between the ALPS-index and, primarily, psychiatric disorders, CSF (p)Tau, and cSVD biomarkers (Fig. 5A, Supplementary Data 27, and Supplementary Fig. 12).
A Enhanced discovery of genetic loci for the three ALPS-indexes conditioned on neuropsychiatric and neurological disorders and biomarkers by the condFDR analysis. ConjFDR analysis detected shared genetic loci across the three ALPS-indexes and neuropsychiatric disorders and biomarkers. B Conditional Q–Q plots for the mean ALPS-index conditioned on schizophrenia, CSF pTau, HIP-PVS, WMH, and vice versa, demonstrating genetic overlap. Conditional Q–Q plots of nominal versus empirical −log10 p-values (corrected for inflation) in one phenotype below the standard GWAS threshold of p < 5 × 10−8 as a function of significance of association with another phenotype, at the level of p ⩽ 0.1, p ⩽ 0.01, p ⩽ 0.001, respectively. The blue lines indicate all SNPs. The dashed lines indicate the null hypothesis. AD Alzheimer’s disease, ADHD attention-deficit/hyperactivity disorder, ASD autism spectrum disorder, BG-PVS perivascular space in basal ganglia, BP bipolar disorder, bvFTD behavioral variant frontotemporal dementia, CSF cerebrospinal fluid, HIP-PVS perivascular space in hippocampus, mALPS mean ALPS-index, SCZ schizophrenia, WM-PVS perivascular space in white matter, WMH white matter hyperintensities.
We then conducted bidirectional Mendelian randomization (MR) analyses to examine putative causal relationships. In forward MR, an increased level of left ALPS-index was associated with a reduced risk of ADHD after Bonferroni correction (OR [95% CI] = 0.647 [0.512–0.818], PIVW-Bonferroni = 0.009) (Supplementary Data 28). No significant results were found in reverse MR (Supplementary Data 29).
Gene-level analysis and functional enrichment
To gain insight into genes involved in the ALPS-index, gene-level analyses were conducted using genome-wide gene-based association studies (GWGAS, using MAGMA36; Fig. 6A) and physical position, expression quantitative trait locus (eQTL), and 3D chromatin interaction mapping strategies through FUMA31. Based on the GWAS results of three ALPS-indexes in the UKB discovery cohort, the three mapping strategies identified 151 unique genes (Supplementary Data 30). GWGAS identified 40 unique genes at a gene-wide significant level (p < 2.65 × 10−6), of which 30 overlapped with genes implicated by FUMA, resulting in a total number of 161 unique genes involved in ALPS-indexes (Fig. 6B and Supplementary Data 31). Of these, 111 (68.9%) overlapped between three ALPS-indexes (Fig. 6C), and 21 genes had the most robust evidence as they were simultaneously implicated by all four approaches (Supplementary Data 32).
A Manhattan plot of the GWGAS on the mean ALPS-index (n = 37,327 individuals). P values were computed by MAGMA gene-based test. The y-axis shows the −log10 p-value of each gene. The x-axis shows the chromosomal position (start position). The dashed line indicates the Bonferroni-corrected threshold for genome-wide significance of the gene-based test (P < 2.65 × 10−6; 0.05/18,878). B Venn diagram showing the number of genes associated with any of the three ALPS-indexes (mean, left, and right) implicated by positional mapping, eQTL mapping, chromatin interaction mapping, and GWGAS. C Upset plot displays the number of unique and shared genes among the three ALPS-indexes (mean, left, and right). eQTL expression quantitative trait loci, GWGAS genome-wide gene-based association analysis.
Cell-type enrichment analysis was performed to assess whether genes are disproportionately expressed in specific brain cell types. For the mALPS-index, significant signals were enriched in oligodendrocyte progenitor cell (OPC) and oligodendrocyte (Supplementary Data 33). Oligodendrocytes are non-neuronal cells that provide support to axons and play a role in motor skill learning and memory formation37.
To investigate the biological properties underlying all identified genes involved in ALPS-indexes, we conducted enrichment analyses using the g:GOSt function in the g:Profiler webtool. Functional analyses revealed enrichment in biological processes of neurogenesis, anatomical structure development, and glial cell development, as well as cellular components of cell junction and cell projection, and in molecular functions of oxidoreductase activity and linoleoyl-CoA desaturase activity (Supplementary Data 34). The FUMA enrichment analysis revealed several enriched phenotypes related to neuropsychiatric phenotypes and metabolism, including Parkinson’s disease, cognitive ability, neuroticism, and lipid metabolism (Supplementary Data 35). Additionally, the functional analysis was conducted within GWGAS-defined genes, followed by MCL clustering. A cluster of genes (ARHGAP27, ARL17B, CRHR1, KANSL1, PLEKHM1, SPPL2C, STH) was found to be enriched in neurodegeneration, cognitive function, and neuropsychiatric disorders (Supplementary Data 36).
To investigate whether these identified genes could be potential targets for pharmacotherapy treatment, we subsequently searched for gene–drug interactions using the Drug Gene Interaction database (DGIdb)38. This search identified 25 unique genes with reported drug interactions involving 781 unique pharmaceuticals (Supplementary Data 37). The MAPT gene had the highest number of reported interactions (445 interactions), followed by the FEN1 gene (199 interactions). Several potential targets for pharmacotherapeutic intervention were observed (Supplementary Data 38), covering 82 unique genes. Of these, 60 (73.2%) had not been previously reported to interact with reported drugs.
We investigated the brain expression patterns of genes related to the ALPS-index from development to adulthood (see the “Methods” section). Several genes exhibited considerable variations in expression levels throughout the lifespan, with certain genes peaking during the prenatal period (e.g., NAV2, TRIOBP, FEN1, MAPT, DPYSL5) (Supplementary Fig. 13), suggesting that genetically determined influences on glymphatic activity may begin early in the prenatal period.
Discussion
The study investigated the genetic architecture of the ALPS-index, among up to 40,866 participants aged across 9–82 years old. The ALPS-indexes showed moderate heritability. We identified 17 unique genome-wide significant risk loci for the ALPS-indexes, predominantly for the mALPS-index, including two loci for the right ALPS-index. A total of 161 unique candidate genes were mapped through gene-level analysis. Consistent with our hypothesis, some of the identified genetic signals displayed changing effects during brain development and aging. Functional analyses revealed enrichment in biological processes of neurogenesis, anatomical structure development, and glial cell development, as well as phenotypes of neuropsychiatric diseases and metabolism. Genetic correlations were observed between higher ALPS-indexes and greater cognitive performance, better physical metrics, smaller brain ventricular volumes, and lower MRI-markers of cSVD. Genetic overlaps were identified between the ALPS-indexes and neuropsychiatric and neurological disorders, MRI markers of cSVD, and CSF biomarkers of AD, suggesting that genetic pleiotropy exists among these phenotypes that may share common underlying glymphatic pathological mechanisms. MR analysis further highlighted that a higher ALPS-index is associated with a reduced risk of ADHD. These results enhance the neurobiological understanding of glymphatic clearance activity and provide specific leads for functional experiments.
Strong evidence was found for the locus 12q23.3, tagged by rs12146713 (NUAK1), which had the highest posterior probability of causality (0.99) as detected by FINEMAP. The NUAK1 gene, a member of the AMP-activated protein kinase family of serine/threonine kinases39, was found to be associated with the fornix white matter microstructure40. The fornix functions in the formation of spatial memory, episodic memory, and executive functions41. It is aligned with our findings of the genetic correlation between the ALPS-indexes and fornix white matter microstructure and the previously observed association between the ALPS-index and cognition42. Additionally, the NUAK1 protein regulates tau levels by phosphorylating tau selectively at Ser35643, while the p-tau Ser356 has been implicated in the progression of AD pathology44. Inhibition of NUAK1 prevented the accumulation of tau and the formation of neurofibrillary tangles, and rescued memory deficits in a tauopathy mouse model43. Notably, the expression of the NUAK1 gene in the brain increases with age, peaking in the 20s (Supplementary Fig. 13). The study found that the rs12146713-C (NUAK1) alleles were associated with a lower mALPS-index in UKB and a slower increase of mALPS-index in ABCD. This suggests a genetic effect on decreased glymphatic activity in the aging brain and a slower development of glymphatic function in the young brain. Accordingly, the relatively low expression of the NUAK1 gene in early life is consistent with the glymphatic developmental process, and the elevated expression of this gene with age progression may be associated with subsequent age-dependent decline in glymphatic activity26. The NUAK1 gene is druggable45 (Supplementary Data 38), although there are currently no pharmacological treatments that target this gene for the management of neurological or psychiatric disorders. These collectively point to NUAK1 as a potential therapeutic entry for neuropsychiatric diseases43,44 with regulation of glymphatic clearance activity as a possible additional mechanism.
Using the longitudinal data from multiple cohorts, we observed that the mALPS-index increased over time in children’s brains but decreased over time in older brains. Previous work has highlighted age-related changes and the importance of age-dependent genetic effects on neuroimaging features of the human brain46,47. Given the age dependence nature of brain glymphatic activity quantified by the ALPS-index26, the genetic associations identified in the aging cohort mainly represent the degeneration process of glymphatic system, while that in younger cohorts may primarily implicate the maturation process. As we hypothesized that the genetic drivers of the ALPS-index may be distinct in young and older brains, most of the independent genetic signals identified in the UKB discovery cohort were not significant in young cohorts. The PRS derived from the summary data of the discovery GWAS did not show significance in young cohorts. This is unlikely due to the small sample size of these young cohorts, as the PRS constructed in another small cohort (ADNI) of older individuals yielded significant results. Despite these findings, several genetic signals, particularly in the 2p23.3 and 3q26.2 loci, consistently showed significant effects on glymphatic activity in both young and older individuals. No discordant effects were observed for these variants in meta-regression by age, further indicating their concordant role across the lifespan.
The study found significant age-dependent associations with the mALPS-index for several loci through the meta-regression with age and longitudinal GWAS. The G allele of the lead variant rs6012259 (SULF2) in 20q13.12 was associated with a lower mALPS-index in both young adults and older populations (i.e., the UKB and HCP cohorts), with a stronger negative effect at older ages. It was also associated with a faster decline of the mALPS-index in the UKB longitudinal cohort containing middle-aged to elderly individuals. Among children in the ABCD cohort, the significance of the 20q13.12 locus was attenuated. These findings suggest that this locus may primarily affect glymphatic dysfunction in older individuals. Furthermore, the mapped gene SULF2 encodes an enzyme involved in multiple heparan sulfate proteoglycan-dependent receptor tyrosine kinase signaling pathways48. It is highly expressed in the central nervous system and has roles in modulating neurite outgrowth and brain homeostasis48,49,50. Additionally, the A allele of the variant rs4757810 (NAV2) in 20q13.12 was associated with a lower mALPS-index in UKB and a slower increase of mALPS-index in the ABCD longitudinal cohort. It may play a detrimental role in the development of the glymphatic system in early life and in the exacerbation of glymphatic dysfunction in late life. The NAV2 gene plays a pivotal role in the development of the nervous system and the regulation of blood pressure51. It has previously been suggested as an AD risk gene, which was found to be associated with episodic memory performance in AD, as well as the age at onset and the risk of AD52,53. Furthermore, our study revealed discrepancies in the effect of identified genetic loci across children and older populations. For example, the 9q31.3 locus, tagged by (rs10817103) LPAR1, exhibited a robust association with the mALPS-index only in the middle-aged and older cohorts (i.e., ADNI and UKB), but not in any younger cohorts. One speculation is epigenetic regulation, where changes in epigenetic marks with age, such as DNA methylation and histone modifications54, may lead to altered expression of certain loci, which may affect the phenotype in later life. Epigenetic regulation of these loci may serve as glymphatic-targeting therapeutic targets for human diseases in the future.
The genetic architecture of ALPS-indexes on both sides and mALPS-index was highly similar. There was a significant genetic and phenotypic correlation between the three and highly consistent results for heritability and identified genetic loci. Nonetheless, significant genetic associations with better cognition, lower blood pressure, and BMI were observed for a higher ALPS-index on the left side but not on the right side. Previous cohort studies have demonstrated that patients with neurodegenerative diseases experience a more pronounced drop in the ALPS-index of the left hemisphere compared to the right hemisphere, but no significant difference was found in the bilateral ALPS-indexes29,55,56. Despite the plausibility of previous evidence for differences in the clinical relevance of the bilateral ALPS-index, it cannot be ruled out that measurement bias may be responsible for these differences. Handedness was considered a potential influencing factor15,29. However, there were no significant differences in bilateral ALPS-indexes between participants with varying handedness. From a genetic standpoint, given the subtle distinctions between the three genetic outcomes of bilateral versus mALPS index, it is still imperative to conduct a thorough investigation of these three traits separately. Further experimental studies are also needed to elucidate whether these differences have pathophysiological significance or are simply the consequence of measurement error.
This study explored the genetic determinants of glymphatic clearance activity using the non-invasive DTI-based ALPS-index15,16. Another non-invasive measure of the glymphatic system in humans is the PVS, a part of the glymphatic system visible on MRI57. There were multiple overlapping loci between ALPS-index and PVS, particularly the LPAR1 (9q31.3). This locus demonstrated genome-wide significant associations with both the mALPS-index and WM-PVS14 and is considered as a potential therapeutic target for central nervous system disorders58. For instance, the circular RNA encoded by LPAR1 may regulate neuronal cell death and neuroinflammation by sponging their target microRNAs, thereby increasing the risk of AD59. However, no significant genetic correlation was observed between ALPS-indexes and PVS burden, neither in white matter, hippocampus, nor basal ganglia. Although abnormal dilation of the PVS and decreased ALPS-index both have been hypothesized to be associated with glymphatic dysfunction22, the clinical relevance of ALPS-index is not comparable to that of the PVS burden10,60. The abnormality of the ALPS-index is closely related to pathological protein accumulation and cognitive decline in AD19,60 and even precedes and predicts accelerated Aβ deposition30. The PVS burden is more closely associated with conditions of cerebral small vessel disease14, but not with AD pathogenesis60,61,62,63. While it is beyond the scope of this paper, one potential explanation is that the alterations in ALPS-index and PVS loads may be underpinned by other discrete pathophysiological processes, apart from glymphatic activity.
Our results indicate a common underlying genetic mechanism for the ALPS-index, CSF tau pathology, and brain ventricular volume. The ALPS-index measures the fluid diffusivity along the PVS, which may represent the functional activity of the glymphatic system15,16. In alignment with its responding physiologic activity, the genetic correlation analysis revealed a strong association between a smaller ALPS-index and greater brain ventricular volume, which may indicate disturbed CSF flow and neurodegeneration64. Furthermore, a genetic overlap was observed between the ALPS-indexes and two critical biomarkers of neurodegenerative diseases65, namely CSF pTau and Tau. The glymphatic system plays a significant role in clearing tau protein species in the parenchyma2,17. The conjFDR analysis identified several overlapping loci, including two remarkable genomic locations, GMNC (3q28) and C16orf95 (16q24.2). The two genes were previously highlighted in association with both CSF pTau levels and brain ventricular volume66,67,68, implying that the same genetic risk factors contribute to both phenotypes as well as the ALPS-index, reinforcing the notion that glymphatic activity is correlated with tau pathology and neurodegeneration17,19,69. For both loci, the alleles that were associated with an increase in ALPS-indexes in the present study were also associated with smaller ventricular volumes but higher CSF (p)Tau levels. Counterintuitive effects of the two loci were previously discussed66,67. One possible explanation is the compensatory hypothesis68. Genetic factors associated with tau pathology lead to elevated CSF (p)Tau levels, which in turn induces a compensatory increase in fluid clearance processes and ventricular shrinkage for waste removal and brain reserve70.
The genetic correlations suggest a close involvement of the ALPS-index in brain structures and functional traits. Specifically, the ALPS-index is significantly genetically correlated with the fornix white matter and reaction time, a pair of significantly correlated phenotypes40. An intriguing finding is the significant association between a lower ALPS-index and higher blood pressure. It is plausible that blood pressure may affect the hemodynamics of the brain and, thus, the motion of water molecules along the PVS, as indicated by the ALPS-index. Despite the known association between the ALPS-index and vascular risk factors71, there is currently no previous evidence addressing the relationship between the ALPS-index and blood pressure. Further animal studies are needed to validate this finding and explore the underlying mechanisms. Another notable phenotype significantly genetically associated with ALPS is hand grip strength, which has been linked to cognitive ability and risk of dementia72,73. In addition, a polygenetic overlap between the ALPS-index and SCZ was identified. Studies have reported increased sulcal and subarachnoid CSF levels in patients with schizophrenia during adulthood74,75. Pending investigation of the role of brain fluid transporting activity in the SCZ, the polygenetic overlap between ALPS-index and SCZ provides insights into their shared basis.
A noteworthy discovery was the unidirectional causal relationship between ALPS-index and the risk of ADHD, indicating a potential protective effect of a higher ALPS-index against ADHD. Imaging evidence suggests that glymphatic dysfunction occurs in children with ADHD10,76. Our MR finding supports the protective role of the glymphatic system in the pathogenesis of ADHD. Possible pathophysiological mechanisms of the altered glymphatic activity in ADHD patients may be related to neuroinflammation10. The glymphatic network, in conjunction with the downstream meningeal lymphatic system77, serves as the immune cross-talk between the brain and periphery3. Impaired brain clearance pathways can lead to abnormalities in the processes of recruitment of immune cells from the blood and clearance of macromolecules from the brain77. Despite these speculations, further exploration is needed to fully comprehend the pathophysiology of the altered glymphatic system in neurodevelopmental diseases.
We acknowledge several limitations. First, it has not yet been fully validated whether the ALPS-index reflects the human glymphatic clearance activity78. While previous research has demonstrated the ALPS-index is highly reproducible and strongly correlated with glymphatic function as assessed through intrathecal contrast media administration16,79, it has not been thoroughly validated in pathophysiological studies. Caution is advised when interpreting the relationship between identified genetic loci linked to the ALPS-index and glymphatic function. A future direction of research will be the integration of multiple methods of the glymphatic system, as well as the transcriptome-wide association study, to further elucidate the underlying genetic basis of the glymphatic function. Second, our study cohorts did not encompass individuals in the prenatal and early postnatal stages, as well as those in young and middle adulthood, indicating a lack of comprehensive coverage of the entire lifespan. Third, there was little overlap in age among the various cohorts, and variations existed in sample collection, imaging, and genotyping methods. Therefore, the disparities observed in the meta-regression analysis may be influenced by residual confounding beyond age. Fourth, while we were able to validate PRS and a subset of genetic loci in two separate cohorts—the UKB non-British white and ADNI cohorts—certain genetic variants failed to demonstrate adequate replication. Although this may be attributed to the lack of robustness of these loci, the sample heterogeneity resulting from the limited statistical power due to the small sample size cannot be ignored. Large-scale samples will help to validate our findings and make new discoveries. Fifth, we used the default parameters in the FUMA pipeline to identify independent significant variants and candidate variants, which employed a relatively relaxed r2 threshold.
In summary, the current study identified 17 unique genetic risk loci and 161 genes for the ALPS-index, a promising marker of glymphatic clearance activity, examined the age-dependent and age-independent genetic effects across the lifespan, and demonstrated genetic associations with neuropsychiatric phenotypes. Our findings have improved the understanding of the genetic architecture of the ALPS-index and shed light on further research into the biological basis of glymphatic clearance activity across the lifespan and its correlation with brain health.
Methods
Study participants
This study utilized raw imaging and genetic data from participants in multiple cohorts: UKB (age range, 45–82 years), ADNI (age range, 55–75 years), ABCD (age range, 9–11 years), IMAGEN (age range, 13–18 years), and HCP (age range, 22–36 years). Prior to data collection, informed consent was obtained from all participants, and each study site received approval from their respective local ethics committees or institutional review boards.
Raw imaging data from approximately 40,000 middle-to-older aged UKB participants were acquired under application number 19542. UKB is a large, population-based prospective cohort study designed to investigate the genetic and non-genetic factors influencing diseases in middle-aged and older adults80. Multimodal neuroimaging data have been collected since 201481. The primary analysis focused on a subset of UKB individuals of white British ancestry (as defined in ‘ukb_sqc_v2.txt’), and an independent subset of UKB individuals identified as white but not British was used for internal replication80. The ADNI cohort, which also consists of middle-to-older aged participants, was utilized for external replication82. In order to generalize the primary results to younger groups, we also utilized data from children (the ABCD cohort)83, adolescents (the IMAGEN cohort)84, and young adults (the HCP cohort)85. To minimize population stratification, participants in all cohorts were restricted to white ancestry. Detailed characteristics of all participants for analysis are provided in Table 1.
Genotyping and quality control
The SNP array genotyping, phasing, imputation, and QC were carried out centrally by the UKB team86. Additional QC procedures were performed using PLINK v2.0, which involved removing individuals with a genotyping call rate lower than 0.95, discordant self-reported and genetic sexes, the presence of sex chromosome aneuploidy, at least ten third-degree relatives, or outliers in heterozygosity rates. At the variant level, we excluded variants with a call rate lower than 0.95, MAF < 0.01, Hardy–Weinberg equilibrium (HWE) P-value < 10−6, or imputation quality score lower than 0.5, resulting in 8.5 million autosomal variants. For ADNI, ABCD, IMAGEN, and HCP, we obtained raw genotype data and conducted phasing, imputation, and QC procedures similar to those in the UKB87,68. The rare variant association study (RVAS) was based on whole-exome sequencing (WES) data from the UKB. Details about the GWAS genotyping platforms, imputation panels, and genotype quality control for the different cohorts are presented in Supplementary Methods and Supplementary Data 39.
Diffusion tensor imaging (DTI) and ALPS-index calculation
In the UKB discovery cohort, imaging acquisition, pre-processing, and quality control are developed and run on behalf of UKB (https://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/brain_mri.pdf). Briefly, diffusion MRI was acquired with a standard (“monopolar”) Stejskal–Tanner pulse sequence (echo time = 92 ms) with 36 sections (resolution = 2.0 × 2.0 × 2.0 mm; field-of-view = 104 × 104 × 72 matrix) in 50 distinct diffusion-weighted directions (b-values = 1000 s/mm2). Pre-processing quality control included aligning modalities and removing artefacts. All scans were corrected for eddy currents and head motion. Gradient distortion correction was subsequently applied. Imaging details of the replication and generalization cohorts are provided in the Supplementary Methods. The MRI protocol was identical between the baseline and the follow-up in the UKB and ABCD cohorts.
A semi-automated and highly reliable pipeline developed and validated by Taoka et al. was used to calculate the ALPS-index from DTI in each cohort15. Diffusivity maps along the x-axis (right–left), y-axis (anterior–posterior), z-axis (inferior–superior), and color-coded fractional anisotropy (FA) maps were processed using DTIFIT (version 2.0; the FSL diffusion tensor fitting program, FMRIB) in UKB. We made ROI templates in each cohort separately to account for differences in image acquisition and participant characteristics across cohorts. To make the ROI templates, a group of age- and gender-matched participants in each cohort was randomly selected (n = 150 in UKB; n = 100 in ADNI, IMAGEN, ABCD, and HCP). Four 4-mm-diameter regions of interest (ROIs) were placed where the deep medullary veins were perpendicular to the ventricular body, on bilateral projection (superior and posterior corona radiata) and association (superior longitudinal fasciculus) fibers on the FA maps. Three neurologists (SYH, YRZ, and JD, with 4, 7, and 8 years of experience, respectively) blinded to clinical data independently placed ROIs to ensure reliability. These ROI maps were combined into a 4D map, which was subsequently averaged to generate a single 3D probability map. Manual visualization was adapted to confirm the accuracy of registration and the location of ROIs on the probability map. Various probability thresholds were utilized to select the probability map that optimally covered the projection and association fibers at the ventricular level. Selected ROI maps were warped to the MNI152 space using FNIRT and binarized as the templates, utilizing the estimated nonlinear deformation derived from the process of warping the individual FA to the 1 mm FMRIB58_FA template (spatial resolution, 1 × 1 × 1 mm3). Diffusivity maps along the x-axis, y-axis, z-axis, and color-coded FA maps were also warped to the MNI space. We recorded the diffusivity in the directions of the x-axis (Dx), y-axis (Dy), and z-axis (Dz) of four ROIs on the projection and association fibers as Dxproj, Dyproj, Dzproj, Dxassoc, Dyassoc, Dzassoc, respectively. The ALPS-indexes of the left and right hemispheres were calculated separately as \(\frac{{{ {Mean}}}\left({{ {Dxproj}}}+{{ {Dxassoc}}}\right)}{{{ {Mean}}}\left({{ {Dyproj}}}+{{ {Dzassoc}}}\right)}\). The mALPS-index was calculated as the average of the bilateral values. A higher ALPS-index represents greater diffusivity. We excluded individuals with prevalent dementia, stroke, hydrocephalus, and intracranial tumors before analysis (Supplementary Data 40) in the discovery cohort (prevalence: 1.5%) and prevalent dementia in the ADNI cohort (prevalence: 10.5%), as these diseases might influence the estimation of ALPS-index. We dropped ALPS outliers that fell outside three times the standard deviation from the mean. Prior to analyses, ALPS-indexes were normalized using a rank-based inverse normal transformation.
Reproducibility of the ALPS-index is greatly affected by the ROI setting79, which should be placed in the projection fiber region and the association fiber region. In the extraction of ALPS-index, the assessment of segmentation accuracy prioritized the placement of ROIs22. To validate the accuracy of the segmentation in the semi-automated ALPS-index calculation using the group-average ROI template, we assessed the comparability of the ALPS-index derived from the group-average ROI templates and manually placed ROIs. The latter were visually inspected for accuracy, with particular avoidance of placement of ROIs in areas with mixing subcortical fibers. Specifically, we included 150 subjects from the UKB cohort and 100 subjects from the IMAGEN, both of which were originally randomly selected to generate the group-average ROI template. Those participants from each cohort were then equally divided into a training set and a test set based on the scanning time. The average ROI template was generated in the training set using the aforementioned method. Two sets of ALPS-index were generated in the test set based on this ROI template and manually placed ROIs, respectively. Pearson correlation analysis was performed on the two sets of ALPS-index.
Heritability estimation and genome-wide association analyses
We estimated the SNP-based heritability of the ALPS-indexes (mean, left, right) using the genome-based restricted maximum likelihood (GREML) approach as implemented in GCTA v1.93.2, adjusting for age, age2, sex, imaging center, and the first ten genetic principal components (PCs). GWAS-based heritability analysis was conducted using GWAS summary data in LDSC v1.0.188. Linkage equilibrium (LD) scores from the European population of the 1000 Genomes Project were used as the reference. The LDSC intercept was calculated to assess genomic inflation.
As the primary GWAS analysis, we conducted GWAS of the mALPS-index in UKB via PLINK v2.0. We applied a linear additive regression model and regressed the same covariates as in the GCTA analysis (see Supplementary Methods for details of genomic risk loci identification, functional annotation, and FINEMAP). Given the mALPS-index was averaged from the ALPS-index in the bilateral hemisphere, GWAS of the left and right ALPS-index were then performed separately as sensitivity analyses. To validate the robustness of independent significant SNPs and to exclude the potential influence of white matter integrity, we performed sensitivity analyses with brain mean FA and MD as additional covariates in the UKB discovery cohort (Supplementary Methods). The significance threshold was P < 5.0 × 10−8 for the primary and sensitivity genome-wide association analyses.
Replication and generalization
We conducted two replication analyses of ALPS-associated independent significant variants identified in the primary analysis. The internal replication was performed in participants with white but not British ancestry (age range = 45–81 years, sample for analysis = 3470). The external replication was carried out in the ADNI cohort (age range = 55–75 years, sample for analysis = 260). We regressed the same covariates and used the identical analytic model as in the primary analysis. Significant replication was defined as a consistent effect direction to the primary analysis and an FDR-corrected P value < 0.05.
To explore if the ALPS-associated independent significant variants have effects on ALPS at younger ages, we conducted generalization analyses in children (ABCD, N = 4445), adolescents (IMAGEN, N = 1888), and young adults (HCP, N = 228). The generalization analyses employed the same covariates and association model as the primary analysis. Significant generalization was defined as a consistent effect direction to the primary analysis and an FDR-corrected P value < 0.05.
Meta-regression
Meta-regression was employed to investigate the age-dependent genetic effects on the mALPS-index. We split the discovery sample into three tertiles according to age (45–61, 62–68, 69–82, Table 1) and re-ran genetic association analyses of independent significant variants of the mALPS-index within each age subgroup. The genetic associations of the three UKB age subgroups, along with ABCD, IMAGEN, HCP, and ADNI, were meta-analyzed using an inverse variance-weighted fixed effects method under linear and quadratic assumptions with the ‘metafor’ R package. Variants that passed the FDR-corrected P value below 0.05 were considered significant.
GWAS of longitudinal changes
To assess whether genetic variants associated with cross-sectional variation of the mALPS-index also affect its longitudinal changes, we conducted a genetic association analysis between mALPS-associated variants and the annual change rate of mALPS-index. The genetic association analysis was carried out in two cohorts, ABCD (mean age at baseline = 10 years, mean follow-up = 2.0 years, N = 2152) and UKB (mean age at baseline = 62 years, mean follow-up = 2.6 years, N = 3580, Table 1). Wilcoxon signed-rank test (paired samples Wilcoxon test) was used to compare the differences in ALPS-indexes between baseline and follow-up in the two longitudinal cohorts using R (version 4.2.0). Two-tailed P < 0.05 was considered statistically significant. The analytical model and covariates in the GWAS of longitudinal change were the same as those in the primary analysis.
Shared genetic variation with other phenotypes
We utilized GWAS summary statistics to estimate the overall genetic correlations between the ALPS-indexes and related phenotypes via LDSC v1.0.188, using the European population data from the 1000 Genomes Project as a reference. We conducted a brain-wide LDSC analysis to investigate the genetic correlations between ALPS-indexes and various brain phenotypes, including brain structure (represented by volumes89), white matter integrity (represented by FA90), and brain functional phenotypes (represented by node amplitude and edge metrics91). The imaging phenotypes were selected based on their biological significance and data availability. We examined the genetic correlations between the ALPS-indexes and putative risk factors (BMI92, sleep93, DBP, SBP and PP94, grip strength95), cognition (reaction time and general cognitive function96), clinical diagnoses (AD97, ADHD98, ASD99, bvFTD100, stroke and subtypes101, SCZ102, BP103), MRI-markers of cSVD (WMH104, PVS burden in basal ganglia, white matter, and hippocampus14), and CSF biomarkers of Alzheimer’s disease (CSF Aβ42, pTau, and tau105). Genetic correlations with Bonferroni corrected P-value < 0.05 were considered significant. Subsequently, the pleiotropy-informed conditional FDR analysis and bidirectional two-sample MR were used to identify genetic overlap and causal relationship between ALPS-indexes and neuropsychiatric and neurological disorders (AD97, ADHD98, ASD99, bvFTD100, stroke101, SCZ102, BP103) as well as disease-related biomarkers (WMH104, three PVS measures14, and three CSF Alzheimer’s biomarkers105) (see Supplementary Methods for details).
Gene-based analysis
We performed a GWGAS of 18,878 protein-coding genes using MAGMA v1.08 implemented in FUMA. The significance threshold was Bonferroni-corrected (P = 0.05/18,878 genes = 2.65 × 10−06). Furthermore, we utilized positional, eQTL, and 3D chromatin interaction data to map the ALPS-associated variants to genes. Protein–protein interaction analysis was performed using significant genes identified in GWGAS as input into the STRING database106. We also performed Markov clustering with default settings and examined functional enrichment by clusters in the STRING database.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
Our GWAS summary statistics for the ALPS-indexes can be found at https://figshare.com/articles/dataset/GWAS_summary_statistics_for_ALPS-index/25515331, or from the GWAS catalog with a GCP ID of GCP001081 (https://www.ebi.ac.uk/gwas/). The individual-level data used in the present study are available through UKB, ABCD, IMAGEN, HCP (http://www.humanconnectome.org/), and ADNI (adni.loni.usc.edu). Individual-level data from the UKB samples are available through UKB under application number 19542. The GWAS Catalog resource can be found at https://www.ebi.ac.uk/gwas/.The GWAS atlas can be found at https://atlas.ctglab.nl/PheWAS/. The DSigDB database can be found at https://dgidb.org/. The Human Protein Atlas can be found at https://www.proteinatlas.org/.
Code availability
This study used openly available software and codes, specifically R, PLINK, GCTA, IMPUTE, Michigan Imputation Server (https://imputationserver.sph.umich.edu/), FUMA, MAGMA (https://ctg.cncr.nl/software/magma/, also implemented in FUMA), Hail (https://hail.is), SAIGE-GENE+ (https://saigegit.github.io/SAIGE-doc/), condFDR/conjFDR (https://github.com/precimed/pleiofdr/), LDSC. The semi-automated pipeline for processing DTI-ALPS index is available on GitHub (https://github.com/h8264/ALPS-index.git). Custom scripts for the analyses in this paper are available through GitHub (https://github.com/h8264/ALPS_GWAS.git).
References
Iliff, J. J. et al. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci. Transl. Med. 4, 147ra111 (2012).
Iliff, J. J. et al. Impairment of glymphatic pathway function promotes tau pathology after traumatic brain injury. J. Neurosci. 34, 16180–16193 (2014).
Hablitz, L. M. & Nedergaard, M. The glymphatic system: a novel component of fundamental neurobiology. J. Neurosci. 41, 7698–7711 (2021).
Xie, L. et al. Sleep drives metabolite clearance from the adult brain. Science 342, 373–377 (2013).
Hablitz, L. M. et al. Circadian control of brain glymphatic and lymphatic fluid flow. Nat. Commun. 11, 4411 (2020).
Mason, G. M., Lokhandwala, S., Riggins, T. & Spencer, R. M. C. Sleep and human cognitive development. Sleep Med. Rev. 57, 101472 (2021).
Yang, F. N., Xie, W. & Wang, Z. Effects of sleep duration on neurocognitive development in early adolescents in the USA: a propensity score matched, longitudinal, observational study. Lancet Child Adolesc. Health 6, 705–712 (2022).
Jiang, F. Sleep and early brain development. Ann. Nutr. Metab. 75, 44–54 (2019).
Li, X. et al. Glymphatic imaging in pediatrics. J. Magn. Reson. Imaging 59, 1523–1541 (2023).
Chen, Y. et al. Assessment of the glymphatic function in children with attention-deficit/hyperactivity disorder. Eur. Radiol. 34, 1444–1452 (2023).
Shen, M. D. Cerebrospinal fluid and the early brain development of autism. J. Neurodev. Disord. 10, 39 (2018).
Nedergaard, M. & Goldman, S. A. Glymphatic failure as a final common pathway to dementia. Science 370, 50–56 (2020).
Peng, W. et al. Suppression of glymphatic fluid transport in a mouse model of Alzheimer’s disease. Neurobiol. Dis. 93, 215–225 (2016).
Duperron, M. G. et al. Genomics of perivascular space burden unravels early mechanisms of cerebral small vessel disease. Nat. Med. 29, 950–962 (2023).
Taoka, T. et al. Evaluation of glymphatic system activity with the diffusion MR technique: diffusion tensor image analysis along the perivascular space (DTI-ALPS) in Alzheimer’s disease cases. Jpn J. Radiol. 35, 172–178 (2017).
Zhang, W. et al. Glymphatic clearance function in patients with cerebral small vessel disease. Neuroimage 238, 118257 (2021).
Harrison, I. F. et al. Impaired glymphatic function and clearance of tau in an Alzheimer’s disease model. Brain 143, 2576–2593 (2020).
Zhang, Y. et al. Interaction between the glymphatic system and α-synuclein in Parkinson’s disease. Mol. Neurobiol. 60, 2209–2222 (2023).
Hsu, J. L. et al. Magnetic resonance images implicate that glymphatic alterations mediate cognitive dysfunction in Alzheimer disease. Ann. Neurol. 93, 164–174 (2023).
Si, X. et al. Neuroimaging evidence of glymphatic system dysfunction in possible REM sleep behavior disorder and Parkinson’s disease. NPJ Parkinsons Dis. 8, 54 (2022).
Kamagata, K. et al. Association of MRI indices of glymphatic system with amyloid deposition and cognition in mild cognitive impairment and Alzheimer disease. Neurology 99, e2648–e2660 (2022).
Taoka, T. et al. Diffusion Tensor Image Analysis ALong the Perivascular Space (DTI-ALPS): revisiting the meaning and significance of the method. Magn. Reson. Med. Sci. 23, 268–290 (2024).
Munk, A. S. et al. PDGF-B is required for development of the glymphatic system. Cell Rep. 26, 2955–2969.e2953 (2019).
Di Palma, C. et al. Cerebrospinal fluid flow increases from newborn to adult stages. Dev. Neurobiol. 78, 851–858 (2018).
Kress, B. T. et al. Impairment of paravascular clearance pathways in the aging brain. Ann. Neurol. 76, 845–861 (2014).
Taoka, T. et al. Diffusion-weighted image analysis along the perivascular space (DWI-ALPS) for evaluating interstitial fluid status: age dependence in normal subjects. Jpn. J. Radiol. 40, 894–902 (2022).
Zhou, Y. et al. Impairment of the glymphatic pathway and putative meningeal lymphatic vessels in the aging human. Ann. Neurol. 87, 357–369 (2020).
Zhong, J. et al. Unlocking the enigma: unraveling multiple cognitive dysfunction linked to glymphatic impairment in early Alzheimer’s disease. Front. Neurosci. 17, 1222857 (2023).
Shen, T. et al. Diffusion along perivascular spaces as marker for impairment of glymphatic system in Parkinson’s disease. NPJ Parkinsons Dis. 8, 174 (2022).
Huang, S. Y. et al. Glymphatic system dysfunction predicts amyloid deposition, neurodegeneration, and clinical progression in Alzheimer’s disease. Alzheimer’s Dement. 20, 3251–3269 (2024).
Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping and annotation of genetic associations with FUMA. Nat. Commun. 8, 1826 (2017).
Yang, J. et al. Conditional and joint multiple-SNP analysis of GWAS summary statistics identifies additional variants influencing complex traits. Nat. Genet. 44, 369–375 s361–363 (2012).
Benner, C. et al. FINEMAP: efficient variable selection using summary data from genome-wide association studies. Bioinformatics 32, 1493–1501 (2016).
Finucane, H. K. et al. Partitioning heritability by functional annotation using genome-wide association summary statistics. Nat. Genet. 47, 1228–1235 (2015).
MacArthur, J. et al. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res. 45, D896–D901 (2017).
de Leeuw, C. A., Mooij, J. M., Heskes, T. & Posthuma, D. MAGMA: generalized gene-set analysis of GWAS data. PLoS Comput. Biol. 11, e1004219 (2015).
Watson, A. E. S. et al. Fractalkine signaling regulates oligodendroglial cell genesis from SVZ precursor cells. Stem Cell Rep. 16, 1968–1984 (2021).
Cannon, M. et al. DGIdb 5.0: rebuilding the drug–gene interaction database for precision medicine and drug discovery platforms. Nucleic Acids Res. 52, D1227–d1235 (2024).
Sun, X., Gao, L., Chien, H. Y., Li, W. C. & Zhao, J. The regulation and function of the NUAK family. J. Mol. Endocrinol. 51, R15–R22 (2013).
Ou, Y. N. et al. The genetic architecture of fornix white matter microstructure and their involvement in neuropsychiatric disorders. Transl. Psychiatry 13, 180 (2023).
Senova, S., Fomenko, A., Gondard, E. & Lozano, A. M. Anatomy and function of the fornix in the context of its potential as a therapeutic target. J. Neurol. Neurosurg. Psychiatry 91, 547–559 (2020).
Wang, J. et al. Glymphatic function plays a protective role in ageing-related cognitive decline. Age Ageing 52, afad107 (2023).
Lasagna-Reeves, C. A. et al. Reduction of Nuak1 decreases tau and reverses phenotypes in a tauopathy mouse model. Neuron 92, 407–418 (2016).
Taylor, L. W. et al. p-tau Ser356 is associated with Alzheimer’s disease pathology and is lowered in brain slice cultures using the NUAK inhibitor WZ4003. Acta Neuropathol. 147, 7 (2024).
Faisal, M. et al. Development and therapeutic potential of NUAKs inhibitors. J. Med. Chem. 64, 2–25 (2021).
Bethlehem, R. A. I. et al. Brain charts for the human lifespan. Nature 604, 525–533 (2022).
Brouwer, R. M. et al. Genetic variants associated with longitudinal changes in brain structure across the lifespan. Nat. Neurosci. 25, 421–432 (2022).
Joy, M. T., Vrbova, G., Dhoot, G. K. & Anderson, P. N. Sulf1 and Sulf2 expression in the nervous system and its role in limiting neurite outgrowth in vitro. Exp. Neurol. 263, 150–160 (2015).
Kalus, I. et al. Sulf1 and Sulf2 differentially modulate heparan sulfate proteoglycan sulfation during postnatal cerebellum development: evidence for neuroprotective and neurite outgrowth promoting functions. PLoS ONE 10, e0139853 (2015).
Kalus, I. et al. Differential involvement of the extracellular 6-O-endosulfatases Sulf1 and Sulf2 in brain development and neuronal and behavioural plasticity. J. Cell. Mol. Med. 13, 4505–4521 (2009).
McNeill, E. M., Roos, K. P., Moechars, D. & Clagett-Dame, M. Nav2 is necessary for cranial nerve development and blood pressure regulation. Neural Dev. 5, 6 (2010).
Yan, J. et al. Hippocampal transcriptome-guided genetic analysis of correlated episodic memory phenotypes in Alzheimer’s disease. Front. Genet. 6, 117 (2015).
Wang, K. S., Liu, Y., Xu, C., Liu, X. & Luo, X. Family-based association analysis of NAV2 gene with the risk and age at onset of Alzheimer’s disease. J. Neuroimmunol. 310, 60–65 (2017).
Wu, Z., Zhang, W., Qu, J. & Liu, G. H. Emerging epigenetic insights into aging mechanisms and interventions. Trends Pharm. Sci. 45, 157–172 (2024).
Meng, J. C. et al. Correlation of glymphatic system abnormalities with Parkinson’s disease progression: a clinical study based on non-invasive fMRI. J. Neurol. 271, 457–471 (2024).
Zhang, X. et al. Glymphatic system impairment in Alzheimer’s disease: associations with perivascular space volume and cognitive function. Eur. Radiol. 32, 1314–1323 (2024).
Sepehrband, F. et al. Image processing approaches to enhance perivascular space visibility and quantification using MRI. Sci. Rep. 9, 12351 (2019).
Xiao, D., Su, X., Gao, H., Li, X. & Qu, Y. The roles of Lpar1 in central nervous system disorders and diseases. Front. Neurosci. 15, 710473 (2021).
Li, Y. L. et al. Circular RNA expression profile of Alzheimer’s disease and its clinical significance as biomarkers for the disease risk and progression. Int. J. Biochem. Cell Biol. 123, 105747 (2020).
Hong, H. et al. The relationship between amyloid pathology, cerebral small vessel disease, glymphatic dysfunction, and cognition: a study based on Alzheimer’s disease continuum participants. Alzheimers Res. Ther. 16, 43 (2024).
Gertje, E. C., van Westen, D., Panizo, C., Mattsson-Carlgren, N. & Hansson, O. Association of enlarged perivascular spaces and measures of small vessel and Alzheimer disease. Neurology 96, e193–e202 (2021).
Banerjee, G. et al. MRI-visible perivascular space location is associated with Alzheimer’s disease independently of amyloid burden. Brain 140, 1107–1116 (2017).
Jeong, S. H. et al. Association of enlarged perivascular spaces with amyloid burden and cognitive decline in Alzheimer disease continuum. Neurology 99, e1791–e1802 (2022).
Khasawneh, A. H., Garling, R. J. & Harris, C. A. Cerebrospinal fluid circulation: What do we know and how do we know it? Brain Circ. 4, 14–18 (2018).
Hansson, O. Biomarkers for neurodegenerative diseases. Nat. Med. 27, 954–963 (2021).
Jansen, I. E. et al. Genome-wide meta-analysis for Alzheimer’s disease cerebrospinal fluid biomarkers. Acta Neuropathol. 144, 821–842 (2022).
Vojinovic, D. et al. Genome-wide association study of 23,500 individuals identifies 7 loci associated with brain ventricular volume. Nat. Commun. 9, 3945 (2018).
Ge, Y. J. et al. Genetic architectures of cerebral ventricles and their overlap with neuropsychiatric traits. Nat. Hum. Behav. 8, 164–180 (2024).
Ishida, K. et al. Glymphatic system clears extracellular tau and protects from tau aggregation and neurodegeneration. J. Exp. Med. 219, e20211275 (2022).
Stern, Y. et al. Whitepaper: defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement. 16, 1305–1311 (2020).
Wei, Y. C. et al. Vascular risk factors and astrocytic marker for the glymphatic system activity. Radio. Med. 128, 1148–1161 (2023).
Sternäng, O. et al. Grip strength and cognitive abilities: associations in old age. J. Gerontol. B Psychol. Sci. Soc. Sci. 71, 841–848 (2016).
Kuo, K. et al. Associations of grip strength, walking pace, and the risk of incident dementia: a prospective cohort study of 340212 participants. Alzheimer’s Dement. 19, 1415–1427 (2022).
Narr, K. L. et al. Regional specificity of hippocampal volume reductions in first-episode schizophrenia. Neuroimage 21, 1563–1575 (2004).
Narr, K. L. et al. Increases in regional subarachnoid CSF without apparent cortical gray matter deficits in schizophrenia: modulating effects of sex and age. Am. J. Psychiatry 160, 2169–2180 (2003).
Vilor-Tejedor, N. et al. Assessment of susceptibility risk factors for ADHD in imaging genetic studies. J. Atten. Disord. 23, 671–681 (2019).
Louveau, A. et al. Structural and functional features of central nervous system lymphatic vessels. Nature 523, 337–341 (2015).
Haller, S., Moy, L. & Anzai, Y. Evaluation of diffusion tensor imaging analysis along the perivascular space as a marker of the glymphatic system. Radiology 310, e232899 (2024).
Taoka, T. et al. Reproducibility of diffusion tensor image analysis along the perivascular space (DTI-ALPS) for evaluating interstitial fluid diffusivity and glymphatic function: CHanges in Alps index on Multiple conditiON acquIsition eXperiment (CHAMONIX) study. Jpn J. Radiol. 40, 147–158 (2022).
Sudlow, C. et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 12, e1001779 (2015).
Alfaro-Almagro, F. et al. Image processing and quality control for the first 10,000 brain imaging datasets from UK Biobank. Neuroimage 166, 400–424 (2018).
Hendrix, J. A. et al. The worldwide Alzheimer’s disease neuroimaging initiative: an update. Alzheimers Dement. 11, 850–859 (2015).
Casey, B. J. et al. The Adolescent Brain Cognitive Development (ABCD) study: imaging acquisition across 21 sites. Dev. Cogn. Neurosci. 32, 43–54 (2018).
Schumann, G. et al. The IMAGEN study: reinforcement-related behaviour in normal brain function and psychopathology. Mol. Psychiatry 15, 1128–1139 (2010).
Van Essen, D. C. et al. The Human Connectome Project: a data acquisition perspective. Neuroimage 62, 2222–2231 (2012).
Bycroft, C. et al. The UK Biobank resource with deep phenotyping and genomic data. Nature 562, 203–209 (2018).
Chen, S. D. et al. Genome-wide survival study identifies PARL as a novel locus for clinical progression and neurodegeneration in Alzheimer’s disease. Biol. Psychiatry 94, 732–742 (2023).
Bulik-Sullivan, B. et al. An atlas of genetic correlations across human diseases and traits. Nat. Genet. 47, 1236–1241 (2015).
Ge, Y. J., Fu, Y., Gong, W., Cheng, W. & Yu, J. T. Genetic architecture of brain morphology and overlap with neuropsychiatric traits. Trends Genet. 40, 706–717 (2024).
Zhao, B. et al. Common genetic variation influencing human white matter microstructure. Science 372, eabf3736 (2021).
Zhao, B. et al. Common variants contribute to intrinsic human brain functional networks. Nat. Genet. 54, 508–517 (2022).
Pulit, S. L. et al. Meta-analysis of genome-wide association studies for body fat distribution in 694 649 individuals of European ancestry. Hum. Mol. Genet. 28, 166–174 (2019).
Dashti, H. S. et al. Genome-wide association study identifies genetic loci for self-reported habitual sleep duration supported by accelerometer-derived estimates. Nat. Commun. 10, 1100 (2019).
Evangelou, E. et al. Genetic analysis of over 1 million people identifies 535 new loci associated with blood pressure traits. Nat. Genet. 50, 1412–1425 (2018).
Jiang, L., Zheng, Z., Fang, H. & Yang, J. A generalized linear mixed model association tool for biobank-scale data. Nat. Genet. 53, 1616–1621 (2021).
Davies, G. et al. Study of 300,486 individuals identifies 148 independent genetic loci influencing general cognitive function. Nat. Commun. 9, 2098 (2018).
Kunkle, B. W. et al. Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat. Genet. 51, 414–430 (2019).
Demontis, D. et al. Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder. Nat. Genet. 51, 63–75 (2019).
Grove, J. et al. Identification of common genetic risk variants for autism spectrum disorder. Nat. Genet. 51, 431–444 (2019).
Ferrari, R. et al. Frontotemporal dementia and its subtypes: a genome-wide association study. Lancet Neurol. 13, 686–699 (2014).
Mishra, A. et al. Stroke genetics informs drug discovery and risk prediction across ancestries. Nature 611, 115–123 (2022).
Pardiñas, A. F. et al. Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection. Nat. Genet. 50, 381–389 (2018).
Stahl, E. A. et al. Genome-wide association study identifies 30 loci associated with bipolar disorder. Nat. Genet. 51, 793–803 (2019).
Traylor, M. et al. Genetic variation in PLEKHG1 is associated with white matter hyperintensities (n = 11,226). Neurology 92, e749–e757 (2019).
Deming, Y. et al. Genome-wide association study identifies four novel loci associated with Alzheimer’s endophenotypes and disease modifiers. Acta Neuropathol. 133, 839–856 (2017).
Szklarczyk, D. et al. STRING v11: protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 47, D607–D613 (2019).
Acknowledgements
The authors are grateful to all participants and cooperating institutions. The UK Biobank analyses were conducted using the UKB Resource under application number 19542. Data used in the preparation of this article were also obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu/). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in the analysis or writing of this report. A complete listing of ADNI investigators can be found at: http://adni.loni.usc.edu/wpcontent/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf. This study was supported by grants from the Science and Technology Innovation 2030 Major Projects (J.-T.Y.: 2022ZD0211600), National Natural Science Foundation of China (J.-T.Y.: 81971032, 82071201, 92249305; W.C.: 82472055, 82071997), the National Key R&D Program of China (J.-F.F.: 2018YFC1312904 and 2019YFA0709502; W.C.: 2023YFC3605400), Shanghai Municipal Science and Technology Major Project (J.-T.Y.: No. 2018SHZDZX01), Research Start-up Fund of Huashan Hospital (J.-T.Y.: 2022QD002), Excellence 2025 Talent Cultivation Program at Fudan University (J.-T.Y.: 3030277001), Shanghai Talent Development Funding for The Project (J.-T.Y.: 2019074), Shanghai Rising-Star Program (W.C.: 21QA1408700), 111 Project (J.-F.F.: B18015), and ZHANGJIANG LAB, Tianqiao and Chrissy Chen Institute, and the State Key Laboratory of Neurobiology and Frontiers Center for Brain Science of Ministry of Education, Fudan University.
Author information
Authors and Affiliations
Consortia
Contributions
J.-T. Yu, W. Cheng, and Y.-R. Zhang designed the study. S.-Y. Huang, Y.-J. Ge, P. Ren, Y.-R. Zhang, B.-S. Wu, and J. Du organized data, carried out the statistical analysis, and participated in the first draft of the manuscript. S.-Y. Huang, and B.-S. Wu designed and drew the figures. S.-D. Chen, J.-J. Kang, and Q. Ma organized and analyzed data. S.-Y. Huang, Y.-J. Ge, P. Ren, W.-K. Gong, J.-F. Feng, W. Cheng, Y.-R. Zhang, and J.-T. Yu critically revised the manuscript. A.L.W. Bokde, S. Desrivières, H. Garavan, A. Grigis, H. Lemaitre, M.N. Smolka, and S. Hohmann are the principal investigators of IMAGEN Consortium; Imaging and genetic data in the IMAGEN project were acquired and provided by the IMAGEN Consortium. All authors read and approved the final manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks Marie-Gabrielle Duperron and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Huang, SY., Ge, YJ., Ren, P. et al. Genome-wide association study unravels mechanisms of brain glymphatic activity. Nat Commun 16, 626 (2025). https://doi.org/10.1038/s41467-024-55706-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-024-55706-9
