# paper.pdf ## Metadata - PDFFormatVersion=1.6 - Language=en - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - CreationDate=D:20260706084453+05'30' - Creator=LaTeX with hyperref package - ModDate=D:20260706084517+05'30' - Producer=Acrobat Distiller 26.0 (Windows) - Title=Genetic architecture of the limbic white matter microstructure in aging and Alzheimer's Disease - Custom.WPS-ARTICLEDOI=10.1002/alz.71630 - Custom.WPS-JOURNALDOI=10.1002/(ISSN)1552-5279 - Custom.WPS-PROCLEVEL=3 - xmp:createdate=2015-08-18T08:05:36+05:30 - xmp:modifydate=2016-06-21T14:17:29-04:00 - xmp:metadatadate=2016-06-21T14:17:29-04:00 - pdf:producer=Acrobat Distiller 9.0 (Windows) - pdfx:wps-articledoi=10.1002/alz.71630 - pdfx:wps-journaldoi=10.1002/(ISSN)1552-5279 - pdfx:wps-proclevel=3 - jav:journal_article_version=VoR - prism:doi=10.1002/alz.71630 - dc:format=application/pdf - dc:title=Genetic architecture of the limbic white matter microstructure in aging and Alzheimer's Disease - xmpmm:documentid=uuid:2de97207-b596-3dea-97f1-55b895a626f1 - xmpmm:instanceid=uuid:0996b977-7fc7-3f60-9484-860868e0a6de ## Contents ### Page 1 Received: 29 January 2026 Revised: 7 May 2026 Accepted: 18 May 2026 DOI: 10.1002/alz.71630R E S E A R C H A R T I C L E Genetic architecture of the limbic white matter microstructure in aging and Alzheimer’s Disease Anna S Lorenz1,2 Aditi Sathe1 Yisu Yang1 Alaina Durant1 Yiyang Wu1 Michael E. Kim3 Chenyu Gao4 Nancy R. Newlin3 Karthik Ramadass3,4 Praitayini Kanakaraj3 Nazirah Mohd Khairi4 Zhiyuan Li4 Tianyuan Yao3 Yuankai Huo4 Logan Dumitrescu1,2,5,6 Niranjana Shashikumar1 Kimberly R. Pechman1 Shannon L. Risacher7,8 Lori L. Beason-Held9 Yang An9 Konstantinos Arfanakis10,11,12 Guray Erus13 Christos Davatzikos13 Mohamad Habes14 Di Wang14 Duygu Tosun15 Arthur W. Toga16 Paul M. Thompson17 Elizabeth C. Mormino18 Panpan Zhang1,19 Kurt Schilling20,21 Marilyn Albert22 Walter Kukull23 Sarah A. Biber23 Bennett A. Landman2,20,21,24 Sterling C. Johnson25,26 Barbara Bendlin25,26 Julie Schneider11 David A. Bennett11 Angela L. Jefferson1,5,6 Susan M. Resnick9 Andrew J. Saykin7,8 Jennifer E. Below1,2,27 Timothy J. Hohman1,2,5,6 Derek B. Archer1,2,5,6 Alzheimer’s Disease Neuroimaging Initiative (ADNI) The BIOCARD Study Team (BIOCARD) The Alzheimer’s Disease Sequencing Project (ADSP) CorrespondenceDerek B. Archer: Vanderbilt Memory and Alzheimer’s Center 3319 West End Ave, Floor 8, Room 848 Nashville, TN 37203. Email: derek.archer@vumc.orgFunding informationNational Institutes of Health, Grant/Award Numbers: K01-EB03289, R01-EB017230, R01-AG074339, P30-AG066515, K01-AG073584, U24-AG074855, 75N95D22P00141, R01-AG059716, UL1-TR000445, UL1-TR002243, R01-AG034962, R01-AG056534, U19-AG03655, NIAAG021155, NIAAG027161, NIAAG037639, NIAAG054047, U01AG024904; American Heart Association, Grant/Award Number: 26PRE1558066; Alzheimer’s Association, Grant/Award Number: IIRG-08-88733Abstract INTRODUCTION: Limbic white matter (WM) abnormalities are prevalent in aging and Alzheimer’s disease (AD), but genetic drivers are unclear.METHODS: In 2614 older adults (mean age ± SD: 73.7 ± 9.8 years; 26% cogni- tively impaired) from seven harmonized cohorts enriched for cognitive impairment, we quantified free-water–corrected diffusion MRI (dMRI) metrics in seven limbic tracts. We estimated single nucleotide polymorphism (SNP) heritability, performed cohort genome-wide association studies (GWASs) with meta-analysis, evaluated shared genetic architecture and enriched pathways, and assessed AD relevance using brain RNA-seq data.RESULTS: Limbic WM is heritable (h2 = 0.26–0.60; pFDR < 0.05). Meta-GWAS identified six loci (p < 5 × 10−8), including a signal implicating CDH19, an oligodendrocyte- enriched cell-adhesion gene. Additional loci were near the KC6, SENP5, RORA,FAM107B, and MIR548A1 genes. In brain tissue, RORA, FAM107B, and KC6 expression This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2026 The Author(s). Alzheimer’s & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer’s Association.Alzheimer’s Dement. 2026;22:e71630. wileyonlinelibrary.com/journal/alz 1 of 17https://doi.org/10.1002/alz.71630 ### Page 2 2 of 17 LORENZ ET AL.was associated with cognition and AD neuropathology. Results converged on insulin and immune biology and shared genetic architecture with lipid and cardiovascular traits.DISCUSSION: Limbic WM microstructure is genetically influenced and links oligoden- drocyte and vascular-inflammatory biology to AD-relevant outcomes.K E Y W O R D SAlzheimer’s disease, aging, diffusion MRI, GWAS, harmonization, white matter microstructureHighlights∙ Limbic free-water (FW)–corrected white matter (WM) integrity is heritable in late- life aging∙ Six genome-wide loci discovered for limbic WM traits in Alzheimer’s disease (AD)– enriched cohorts∙ Chr18 locus implicates CDH19, linking oligodendrocyte biology to WM microstruc- ture∙ RORA, FAM107B, and KC6 expression tracks cognitive decline and AD neuropathol- ogy1 BACKGROUNDThe integrity of nerve fibers and the myelination of axons, which characterize white matter (WM) microstructure, are essential for cog- nitive processing by enabling communication across brain regions.1,2 However, WM undergoes substantial changes during aging and in neu- rodegenerative diseases such as Alzheimer’s disease (AD).3 Although AD has been viewed as a disorder primarily affecting gray matter, emerging evidence highlights WM abnormalities, including axonal loss and demyelination.4–6 These changes, particularly within the limbic system, are prevalent across the AD diagnostic continuum,7 with accel- erated degeneration in individuals exhibiting abnormal aging8 and carriers of the apolipoprotein E (APOE) ε4 allele.9 They can manifest before the appearance of amyloid beta (Aβ) plaques and neurofibril- lary tangles,5 are detectable up to 22 years before symptom onset,10,11 and contribute to cognitive decline independently of gray matter atrophy.12 By estimating the directionality of water diffusion in the brain, diffusion magnetic resonance imaging (dMRI) enables in vivo quan- tification of WM microstructure through metrics including fractional anisotropy (FACONV), axial diffusivity (AxDCONV), mean diffusivity (MDCONV), and radial diffusivity (RDCONV).13–16 FACONV reflects the directional dependence of diffusion, with higher values typically indi- cating compact tracts. AxDCONV measures diffusion along the primary axis of fibers, where lower values can indicate axonal damage. MDCONV represents the average diffusion rate across tissue and extracellular compartments and is sensitive, but not specific, to multiple microstruc- tural properties. RDCONV measures diffusion perpendicular to the primary axis, with higher values often suggesting demyelination.13–16 Conventional metrics derived from single-tensor models can be con- founded by partial volume effects of extracellular free water (FW), such as cerebrospinal fluid. Advanced bi-tensor models address this by correcting these metrics for FW contributions, allowing differ- entiation between extracellular (i.e., FW) and intracellular contri- butions (FAFWcorr, AxDFWcorr, RDFWcorr, MDFWcorr) to diffusion.17 FW-corrected measures provide tissue-compartment analogs of con- ventional metrics, yielding more biologically specific indices of WM microstructure, and better capture WM neurodegenerative patterns in AD.7–9,12 However, they should not be interpreted as direct markers of any single histopathologic process. Measurements of WM microstructure in vivo have enabled the examination of underlying biological mechanisms and demonstrated that WM is influenced by genetics. Twin studies reported high heri- tability for dMRI metrics (up to 82%), indicating substantial genetic contributions to variations in WM.18,19 Genome-wide association studies (GWASs) in cognitively unimpaired populations, such as the UK Biobank, have identified more than 100 genomic regions influ- encing WM microstructure.20,21 However, these studies may not fully capture genetic pathways specific to neurodegeneration due to the underrepresentation of cognitively impaired individuals and those at risk for AD. Investigations focused on AD have implicated can- didate genes, such as APOE and BIN1, in WM alterations.22–25 In addition, a genetic predisposition for AD and several AD-risk vari- ants in TMEM106B, PTK2B, WNT3, and APOE are associated with microstructure.26,27 Nevertheless, a more comprehensive understanding of the genetic architecture influencing WM microstructural differences in older adults, particularly those at risk for or with AD, remains limited ### Page 3 LORENZ ET AL. 3 of 17due to the scarcity of combined diffusion and genetic data. There- fore, examining genetic influences on WM microstructure in cohorts enriched for cognitive impairment is critical for advancing our under- standing of neurodegenerative processes in aging and in populations at increased risk for AD. Large-scale harmonization efforts, such as the Alzheimer’s Disease Sequencing Project Phenotype Harmoniza- tion Consortium (ADSP-PHC), in combination with advanced FW- corrected dMRI methods, present an unprecedented opportunity to investigate these imaging genetic associations. This study assessed the role of genetic factors in WM microstruc- ture across seven tracts of the limbic system, including the cingulum, fornix, inferior longitudinal fasciculus (ILF), uncinate fasciculus (UF), and the transcallosal tracts of the inferior temporal gyrus (ITG), mid- dle temporal gyrus (MTG), and superior temporal gyrus (STG). These tracts are integral to memory function and substantially alter in aging, cognitive impairment, and AD-related neurodegeneration.7,12,24 By analyzing harmonized FW-corrected data from 2614 older adults (26% cognitively impaired) across seven aging cohorts, we aim to identify novel biological pathways associated with limbic WM microstructure, providing insights into the mechanisms of WM degeneration in the context of aging, cognitive impairment, and AD risk.2 METHODS 2.1 ParticipantsdMRI and genetic data were leveraged from seven cohorts, including the Alzheimer’s Disease Neuroimaging Initiative (ADNI),* Biomarkers of Cognitive Decline Among Normal Individuals (BIOCARD), Baltimore Longitudinal Study of Aging (BLSA), National Alzheimer’s Coordinating Center (NACC), Religious Orders Study and Rush Memory and Aging Project (ROSMAP), Vanderbilt Memory and Aging Project (VMAP), and the Wisconsin Registry for Alzheimer’s Prevention (WRAP). Data used in the preparation of this article were obtained from the ADNI database (adni.loni.usc.edu). The ADNI was launched in 2003, as a public–private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial MRI, positron emission tomography (PET), other bio- logical markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impair- ment (MCI) and early AD.28 ADNI-GO, ADNI2, and ADNI3 phases with dMRI data were included in this study. BIOCARD began data collec- tion in 1995, at the National Institute of Mental Health (NIMH) and was later transferred to Johns Hopkins University with the goal of identifying preclinical biomarkers of cognitive decline and investigating variables that can predict future progression to AD among cognitively * Data used in preparation of this article were obtained from the Alzheimer’s Disease Neu- roimaging Initiative (ADNI) database (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 analysis or writing of this report. A complete listing of ADNI inves- tigators can be found at: http://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ADNI_ Acknowledgement_List.pdfRESEARCH IN CONTEXT1. Systematic review: We surveyed prior diffusion magnetic resonance imaging (MRI) studies of limbic white mat- ter (WM) degeneration in aging and Alzheimer’s disease (AD), including evidence that free-water (FW)–corrected metrics improve biological specificity over conventional tensor measures. We reviewed large-scale genome- wide association studies (GWASs) of WM microstructure (primarily in population-based, cognitively unimpaired cohorts) and studies linking AD risk genetics, vascular- metabolic factors, and immune dysregulation to WM injury and cognitive decline. 2. Interpretation: Using harmonized multi-cohort data enriched for cognitive impairment, we show that FW- corrected limbic WM is substantially heritable and identified six genome-wide loci underlying tract-specific microstructure. A strong chromosome 18 signal impli- cates CDH19, consistent with oligodendrocyte and myelin biology. Integration with human brain transcriptomics demonstrated that RORA, FAM107B, and KC6 expression relate to cognitive decline and AD neuropathology, strengthening relevance to AD mechanisms. Post-GWAS results converge on insulin signaling, immune processes, and neurotrophic pathways, and reveal shared genetics with lipid and cardiovascular traits. 3. Future directions: Further directions include replica- tion in multi-ancestry samples, incorporating longitudinal WM microstructural changes and validation of cell-type mechanisms in oligodendrocytes and neuroinflammation. normal middle-aged individuals. Participants undergo extensive longi- tudinal evaluations, including neuropsychological testing, MRI scans, and collection of blood and cerebrospinal fluid samples.29 The BLSA neuroimaging substudy began data collection in 1994, and it included dementia-free participants 55 to 85 years of age, with up to 10 years of prospective data collection at baseline. In 2009, the cohort was expanded to include BLSA participants 20 to 85 years of age, as well as 3T MRI-based acquisition of dMRI data. The NACC maintains a centralized data repository for the National Institute on Aging’s (NIA) Alzheimer’s Disease Research Centers (ADRC) Program, which cur- rently includes 33 centers and 4 exploratory centers across the United States.30–32 The ROS cohort, which began in 1994, is an ongoing lon- gitudinal study collecting clinical-pathological data on aging and AD. Participants in ROS are 65+ years of age Catholic nuns, priests, and brothers from various groups throughout the United States.33 MAP is a longitudinal study that started in 1997, recruiting cognitively unimpaired participants.33 The VMAP cohort began longitudinal data collection in 2012, with the goal of understanding the relationship between vascular health and brain aging in older adults with MCI.34 ### Page 4 4 of 17 LORENZ ET AL.TA B L E 1 Participant characteristics by cohort.ADNI BIOCARD BLSA NACC ROSMAP VMAP WRAP Total Number of participants 491 104 399 659 496 247 218 2614Number of female participants(%) 236 (48%) 66 (63%) 210 (53%) 371 (56%) 381 (77%) 92 (37%) 143 (66%) 1499 (57%)Age, years 74.7 (7.6) 73.4 (6.8) 73.2 (9.7) 71.3 (10.5) 81.1 (7.2) 73.7 (7.1) 62.4 (6.1) 73.7 (9.8)Education, years 16.4 (2.6) 17.5 (2.2) 17.1 (2.4) 15.6 (3.0) 15.8 (3.2) 15.9 (2.7) 16.8 (2.8) 16.2 (2.9)Number of APOE ε4 positive (%) 184 (37%) 34 (33%) 91 (23%) 281 (43%) 104 (21%) 88 (36%) 68 (31%) 850 (33%)Number of baseline diagnosisCU (%) / MCI (%) / AD (%) 282 (57%) / 171 (35%) / 38 (8%) 82 (79%) / 20 (19%) / 2 (2%) 395 (99%) / 3 (0.8%) / 1 (0.2%) 416 (63%) / 152 (23%) / 91 (14%) 403 (81%) / 88 (18%) / 5 (1%) 147 (60%)/ 99 (40%)/ 1 (0.4%) 214 (98%)/ 3 (1.4%)/ 1 (0.5%) 1939 (74%)/ 536 (21%)/ 139 (5%)Note: Values denoted as mean (SD) or frequency (percentage). Abbreviations: AD, Alzheimer’s disease; ADNI, Alzheimer’s Disease Neuroimaging Initiative; APOE, apolipoprotein E; BIOCARD, Predictors of Cognitive Decline Among Normal Individuals; BLSA, Baltimore Longitudinal Study of Aging; CU, cognitively unimpaired; MAP, Memory and Aging Project; MARS, Minor- ity Aging Research Study; MCI, mild cognitive impairment; NACC, National Alzheimer’s Coordinating Center; ROS, Religious Orders Study; VMAP, Vanderbilt Memory and Aging Project; WRAP, Wisconsin Registry for Alzheimer’s Prevention. The WRAP cohort began data collection in 2001, focusing on middle- aged adults with a parental history of AD. In 2004, the study expanded to include a control group of individuals without a parental history of AD. The primary objective of WRAP is to identify early biomarkers and risk factors for AD before clinical symptoms emerge.35,36 For all cohorts, participants provided written informed consent, and research was conducted in accordance with approved institu- tional review board protocols. Secondary analysis of these data was approved by the Vanderbilt University Medical School Institutional Review Board. The study included 2614 participants who were self- reported non-Hispanic White individuals between 50.1 and 100.9 years of age (mean age ± SD: 73.7 ± 9.8 years), and 57% were female. Table 1 provides an overview of data included for this study from the ADNI, BIOCARD, BLSA, NACC, ROSMAP, VMAP, and WRAP cohorts.2.2 dMRI acquisition and preprocessingAcross all cohorts, we had 78 different dMRI acquisition protocols; Table S1 provides relevant parameters (e.g., number of directions, b- values, resolution). dMRI data from all cohorts were preprocessed using the PreQual pipeline, which addresses motion, distortions, and eddy currents and performs slice-wise imputation.37,38 Imaging ses- sions were inspected manually by evaluating reports outputted by thePreQual pipeline. DTIFIT was used for the remaining data to compute conventional dMRI metrics, including FACONV, AxDCONV, MDCONV, and RDCONV. To account for FW content in each voxel, we calculated FW- corrected metrics, including FAFWcorr, AxDFWcorr, MDFWcorr, RDFWcorr, and FW.17 Symmetric normalization and linear interpolation were con- ducted using the Advanced Normalization Tools (ANTs) package to obtain a standard space representation of these maps by non-linearly registering the FACONV map to FMRIB58_FA atlas.39 The obtained warp from the registration was applied to all other microstructural maps. Individuals with significant age-regressed outliers (± 5 SD) in WM tract microstructural values were excluded.402.3 WM tractography templatesTractography templates for this study were leveraged from existing resources12,41,42 and can be retrieved from a publicly available Zenodo repository.43 The focus for this study was on seven WM tracts within the limbic system that are relevant to memory and are known to be vulnerable in aging and neurodegeneration, including AD, specifically the cingulum, fornix, ILF, UF, as well as ITG, MTG, and STG transcallosal tracts.442.4 Data harmonizationFor the dMRI data, a region of interest (ROI) approach was employed to calculate mean conventional and FW-corrected dMRI metrics for all tractography templates for each participant. These values were then harmonized using the Longitudinal ComBat package in R (version 4.1.0)45 applied to our entire in-house longitudinal dataset, including 5144 participants across 10,346 imaging timepoints. This harmoniza- tion process utilized a batch variable that had varying levels of speci- ficity depending on the cohort. Parameters used to create the batch variable can be found in Table S2.7–9 The batch variable accounted for various parameters across our cohorts, including scanner name, magnet strength, number of b-values/b-vectors, and resolution; this was optimized to reduce the number of batches while simultaneously accounting for parameters we most anticipated to explain between- batch heterogeneity. In total, we accounted for 34 unique batching levels. Additional covariates for the harmonization included mean- centered age, mean-centered age squared, education, race/ethnicity, cognitive status (cognitively unimpaired or cognitively impaired), APOEε4 positivity, APOE ε2 positivity, and the interaction of mean-centered age and cognitive status. The harmonized values were then scaled by their SD. Next, the dataset was filtered to include only participants with genetic data and further narrowed to the baseline time point for cross-sectional ### Page 5 LORENZ ET AL. 5 of 17analysis. We ultimately used FW-corrected dMRI measures across seven limbic WM tracts (35 dMRI measures) for each participant.2.5 Genetic data quality control and imputationGenetic data were collected with various genotyping arrays across and within cohorts (ADNI: Illumina Human610-Quad BeadChip, Illumina HumanOmniExpress BeadChip, Illumina Omni 2.5 M, Illumnia Global Screening Array v2; BIOCARD: Illumina OmniExpress; BLSA: Illumina HumanOmni2.5 BeadChip, Illumina HumanOmniExpress BeadChip; NACC: several different arrays were used to collect genetic data—acquisition of all genetic data is outlined on the NACC website [https://naccdata.org/nacc-collaborations/partnerships]; ROSMAP: Global Screening Array-24 v3.0 BeadChip, Affymetrix GeneChip 6.0, Illumina HumanOmniExpress; VMAP: Illumina HumanOmniExpress; WRAP: Illumina Human610, Illumina OmniExpress).All genetic raw data underwent the same robust quality control and imputation pipelines.46 Variants that had a genotyping rate less than 95%, a minor allele frequency (MAF) less than 1%, or deviated from Hardy-Weinberg Equilibrium (p < 1×10−6) were removed. In addition, participants were excluded if genotyping efficiency was poor (missing > 1% of variants), if cryptic relatedness was present (PIHAT > 0.25), or if the reported and genotypic sex were not concordant. Imputation was performed on the University of Michigan Imputation Server using the TOPMed reference panel (hg38) with SHAPEIT phasing.47 Data were filtered to exclude variants with low imputation quality (R2 < 0.80), duplicated/multi- allelic variants, and MAF < 1%. Principal components analysis was conducted, and genetic ancestry outliers were excluded. Eight pairs of individuals were related across cohorts and were subsequently removed from the respective cohort with more individuals, namely BLSA, NACC, and ADNI (PIHAT > 0.25).2.6 Statistical analyses2.6.1 SNP-heritability tests Single nucleotide polymorphism (SNP) heritability was estimated for each FW-corrected dMRI metric using genome-wide complex trait analysis (GCTA). This method uses restricted maximum likelihood and genetic relatedness matrices to calculate heritability estimates.48 To account for multiple comparisons, results were adjusted across phenotypes using the false discovery rate (FDR) procedure.49 2.6.2 GWAS testing and meta-analysis GWAS in self-reported non-Hispanic White participants were con- ducted separately for each cohort (N = 7) for each dMRI metric (N =5) of each of the limbic WM tracts (N = 7) using PLINK software (Ver- sion 1.9, https://www.cog-genomics.org/plink/1.9). All models included covariates for age, sex, and the first three genetic ancestry princi- pal components. Next, we performed a meta-analysis across cohorts for each dMRI metric using genome-wide association meta-analysis (GWAMA).50 The significance threshold was set a priori to p < 5 ×10−8. We evaluated suggestive loci that had a p < 1 × 10−5. Reported genome-wide associations were filtered to include those present in at least six of seven cohorts. Variants were mapped to genes by proximity using the Open Targets platform. 2.6.3 eQTL analyses, replication, and databases Expression quantitative trait locus (eQTL) analyses were conducted on variants reaching genome-wide significance using the Genotype-Tissue Expression (GTEx) portal (https://gtexportal.org). Because expression regulation is broadly correlated across tissues and sample sizes for brain tissues in GTEx are small and therefore underpowered to detect modest effects, we considered all GTEx tissues in our post-GWAS eQTL analyses. To verify whether the identified variants have previously been associated with brain traits, genome-wide significant variants were validated using the Oxford Brain Imaging Genetics Server (BIG40) (https://open.win.ox.ac.uk/ukbiobank/big40/)51 and ENIGMA-VIS (https://enigma-brain.org/).52 The identified variants and their mapped genes were further explored using several databases, including GeneCards (https://www. genecards.org), Agora (https://agora.adknowledgeportal.org), and Open Targets (https://www.opentargets.org). A literature search was also conducted using PubMed and Web of Science to provide further biological context. 2.6.4 Association analysis with cognitive outcomes and AD pathologies using ROSMAP bulk RNA sequencing data Processed bulk RNAseq data in the ROSMAP cohort from three brain tissues (Table S3), including the dorsolateral prefrontal cortex (DLPFC), the posterior cingulate cortex (PCC), and the head of the caudate nucleus (CN),53 were used to fit linear regression models (for cross- sectional outcomes) and linear mixed-effects models (for longitudinal outcomes) to test associations between cognitive function and AD pathologies with expression of the genes closest to the identified genome-wide significant variants in the meta-analyzed GWAS. Overall cognitive function was represented using a global cog- nitive score, which was derived by converting raw scores from 19 cognitive tests to z-scores and subsequently averaged. For cross- sectional cognitive function, the global cognitive score at the last visit before death was used. For the longitudinal model, cognitive trajectory was quantified in a mixed-effects regression model. Cog- nitive trajectory was estimated as the individual-specific annual rate of change in global cognition. AD pathologies included Aβ load (immunohistochemistry staining), neurofibrillary tangles (silver stain- ing), tau tangle (immunohistochemistry staining), and neuritic plaque ### Page 6 6 of 17 LORENZ ET AL.pathology (silver staining) (https://www.radc.rush.edu/docs/var/ overview.htm?category=Pathology). For all cross-sectional outcomes, covariates included age at death, sex, postmortem interval (PMI), and interval between last visit and death. For the longitudinal models, in addition to the same set of covari- ates, time was modeled as the number of years between each visit and the final visit. To account for multiple comparisons, results were adjusted using the FDR correction.49 2.6.5 Gene- and pathway-level analysis The Multimarker Analysis of GenoMic Annotation (MAGMA v1.09) software was used to conduct gene- and pathway-level analyses.54 Results were adjusted using FDR correction.49 2.6.6 Genetic covariance analysis and local genetic covariance analysis The meta-analysis results for all dMRI metrics were used to perform genetic covariance analyses with the GWAS summary statistics of 65 complex traits (Table S4) as well as 3143 brain traits55 derived from the UK Biobank using the Genetic Covariance Analyzer (GNOVA) software.56 To further investigate local genetic covariance for signif- icant associations, we used the Super Genetic Covariance Analyzer (SUPERGNOVA) software.57 Results were adjusted across dMRI met- rics and traits using the FDR procedure.49 2.6.7 Transcriptome-wide association testing By leveraging predictive models that aggregate evidence across SNPs, transcriptome-wide association studies (TWASs) increase the power to detect associations with WM microstructure that might have been missed with the GWAS approach. S-PrediXcan was used to exam- ine associations between genetically predicted gene expression and dMRI metrics, leveraging precomputed predictive gene expression models from all GTEx tissues.58 To account for multiple compar- isons, results were adjusted across genes and tissues using the FDR procedure.493 RESULTS 3.1 Heritability of limbic WM microstructureFifteen of 35 FW-corrected dMRI metrics across all limbic tracts were heritable with SNP-heritability estimates ranging from 0.26 to 0.60 (pFDR < 0.05, Figure 1, Table S5). Specifically, all dMRI metrics of the cin- gulum, as well as four of five metrics of the fornix and the ILF showed significant heritability.3.2 Genome-wide significant signals in proximity to the genes CDH19, KC6, SENP5, RORA, FAM107B, and MIR548A1 were associated with limbic WM microstructureA cross-sectional GWAS for each dMRI metric of each tract was per- formed using linear regression models covarying for sex, age, and the first three ancestral principal components, followed by a meta- analysis across cohorts (Table S6 for meta-analysis summary statistics [p < 0.05]). Figure 2 presents an ideogram illustrating the 500 most significant genomic signals based on the smallest p-value associated with WM microstructure. In total, six genome-wide significant loci for different tract-by-microstructure combinations were discovered. We identified a locus with 38 genome-wide significant (p < 5 × 10−8) SNPs (Figure 3, lead SNP: rs12959877, effect allele frequency [EAF] = 0.44,β = 0.002 ± 3.18 × 10−4, p = 5.78 × 10−9, intronic CDH19) for MTG RDFWcorr, with additional suggestive associations (p < 1 × 10−5) at this site for RDFWcorr of STG and ITG. When evaluating eQTL (GTEx Portal) evidence, we found that all 38 SNPs were eQTLs for the gene CDH19in lung and spleen tissue (Table S7). This gene is highly expressed in oligodendrocytes and functions as a calcium-dependent cell adhesion glycoprotein.59 Additional genome-wide significant signals were found in proximity to the genes KC6, SENP5, RORA, FAM107B, and MIR548A1(Table 2). To contextualize the identified loci, we queried the Oxford Brain Imaging Genetics Server (BIG40)51 and ENIGMA-VIS.52 At a nominal threshold (p < 0.05), our genome-wide variants showed associations with diverse imaging-derived phenotypes, including dMRI measures, resting-state functional MRI measures, cortical thickness, and regional and tissue volume. However, none of the variants attained suggestive- level or genome-wide significance in these resources, indicating that our locus-WM trait associations likely represent novel findings pending replication (Table S8).3.3 Significant associations between gene expression of RORA, FAM107B, and KC6 with cognitive function and AD pathologyTo evaluate the clinical and neuropathologic relevance of the genes identified in the GWAS analysis, we investigated the associations of CDH19, KC6, SENP5, RORA, FAM107B, and MIR548A1 expression profiles in the PCC, DLPFC, and CN with cognitive measures (cross- sectional and longitudinal cognition) and AD pathologies (tau tangles, neuritic plaques, Aβ, neurofibrillary tangles). Significant associations were observed for RORA, FAM107B, andKC6 (Figure 4, Figure S1). Associations for RORA expression included cross-sectional cognition in CN (β = −0.654 ± 0.173; pFDR = 0.011) and DLPFC (β = −0.570 ± 0.145; pFDR = 0.001), and longitudinal cog- nition in CN (β = −0.082 ± 0.016; pFDR = 4.39 × 10−5) and DLPFC (β= −0.570 ± 0.014; pFDR = 1.31 × 10−3). Significant AD pathologies forRORA included tau tangles in CN (β = 0.751 ± 0.204; pFDR = 0.010) and ### Page 7 LORENZ ET AL. 7 of 17F I G U R E 1 SNP heritability estimates for limbic white matter (WM) microstructure. SNP heritability of 35 FW-corrected dMRI metrics from seven WM tracts in the limbic system. Abbreviations: AxD, axial diffusivity; dMRI, diffusion magnetic resonance imaging; FA, fractional anisotropy; FWcorr, free-water corrected; MD, mean diffusivity; RD, radial diffusivity; WM, white matter.TA B L E 2 Statistics for genome-wide significant SNPs (p < 5×10−8).Tract dMRI SNP Nearest gene EAF Beta SE p-valueMTG RDFWcorr rs12959877; for additional 37 SNPs see Table S4aCDH19 0.44 0.002 3.18 × 10−4 5.78 × 10−9 ITG FAFWcorr rs79952731 KC6 0.02 −0.013 0.002 7.37 × 10−9 ITG FAFWcorr rs142917632 KC6 0.02 −0.013 0.002 7.37 × 10−9 MTG FAFWcorr rs79952731 KC6 0.02 −0.012 0.002 2.58 × 10−9 MTG FAFWcorr rs142917632 KC6 0.02 −0.012 0.002 2.58 × 10−9 STG FAFWcorr rs79952731 KC6 0.02 −0.015 0.002 6.53 × 10−11 STG FAFWcorr rs142917632 KC6 0.02 −0.015 0.002 6.53 × 10−11 STG AxDFWcorr rs79952731 KC6 0.02 −0.020 0.004 1.95 × 10−8 STG AxDFWcorr rs142917632 KC6 0.02 −0.020 0.004 1.95 × 10−8 ILF FAFWcorr rs8026709 RORA 0.36 0.003 0.001 8.94 × 10−9 ILF AxDFWcorr rs8026709 RORA 0.36 0.004 0.001 5.31 × 10−9 STG RDFWcorr rs11542181 FAM107B 0.03 0.005 0.001 2.64 × 10−8 Cingulum RDFWcorr rs56017587 MIR548A1 0.13 0.002 4.25 × 10−4 2.98 × 10−8 Fornix AxDFWcorr rs78407651 SENP5 0.03 −0.012 0.002 4.77 × 10−8 Abbreviations: AxD, axial diffusivity; dMRI, diffusion MRI; EAF, effect allele frequency; FA, fractional anisotropy; FW, free water; ILF, inferior longitudinal fasciculus; ITG, inferior temporal gyrus transcallosal tract; MD, mean diffusivity; MTG, middle temporal gyrus transcallosal tract; RD, radial diffusivity; SE, standard error; SNP, single nucleotide polymorphism; STG, superior temporal gyrus transcallosal tract. DLPFC (β = 0.756 ± 0.165; pFDR = 7.95 × 10−5), neurite plaques in CN (β = 0.294 ± 0.081; pFDR = 0.010) and DLPFC (β = 0.165 ± 0.066; pFDR= 0.049), neurofibrillary tangles in DLPFC (β = 0.194 ± 0.051; pFDR =0.002), and Aβ in DLPFC (β = 0.515 ± 0.143; pFDR = 0.003). For FAM107B gene expression, there were significant associations with cross-sectional cognition in DLPFC (β = −0.242 ± 0.066; pFDR =0.003) and PCC (β = −0.259 ± 0.081; pFDR = 0.015), longitudinal cog- nition in DLPFC (β = −0.038 ± 0.006; pFDR = 2.21 × 10−8) and PCC (β = −0.025 ± 0.007; pFDR = 0.005). AD pathologies included tau tan- gles in DLPFC (β = 0.263 ± 0.075; pFDR = 0.003), neuritic plaques in DLPFC (β = 0.130 ± 0.030; pFDR = 3.43 × 10−4) and PCC (β = 0.119± 0.038; pFDR = 0.018), neurofibrillary tangles in DLPFC (β = 0.102 ±0.023; pFDR = 2.55 × 10−4), and Aβ in DLPFC (β = 0.325 ± 0.065; pFDR =3.98 × 10−5). Furthermore, KC6 expression in CN was associated with ### Page 8 8 of 17 LORENZ ET AL.F I G U R E 2 Ideogram of genomic signals associated with white matter (WM) microstructure. Ideogram of selected genomic regions associated with dMRI metrics. Shape codes for the p-value threshold (p < 5 × 10−8; p < 1 × 10−5), color codes for the WM tract. The 500 associations between genetic variants and dMRI metrics with the smallest p-value are represented in this plot. This figure was created using PhenoGram from Richie Lab Visualizations (https://visualization.ritchielab.org/phenograms/plot). Abbreviations: AxD, axial diffusivity; dMRI, diffusion-weigthed magnetic resonance imaging; FA, fractional anisotropy; FW, free water; ILF, inferior longitudinal fasciculus; ITG, inferior temporal gyrus transcallosal tract; MD, mean diffusivity; MTG, middle temporal gyrus transcallosal tract; RD, radial diffusivity; STG, superior temporal gyrus transcallosal tract; UF, uncinate fasciculus; WM, white matter. neurite plaques (β = −0.055 ± 0.017; pFDR = 0.019). All results for this analysis can be found in Table S9.3.4 Gene- and pathway-level analyses highlight biological mechanisms related to immune function, neurotrophic signaling, and cardiovascular traitsThe genetic architecture of WM microstructure was also investigated at the gene and pathway level. Gene-level analyses revealed a signifi- cant association between cingulum AxDFWcorr and the gene SERPINA12(z = 4.60, pFDR = 0.038, NSNPs = 132), which is known to influence obesity and atherosclerosis by modulating glycolipid metabolism.60–62 Table S10 depicts the statistics for all gene analyses (p < 0.05). Pathway analysis identified significant pathways (pFDR < 0.05) for ITG FAFWcorr, STG FAFWcorr and UF AxDFWcorr. Enriched pathways for ITG FAFWcorr were related to immune function, encompassing isotype switching of B cells, B cell–mediated immunity, and the production of immunoglobulins and antibodies. Beyond immune-related pathways, STG FAFWcorr was associated with several neurotrophic signaling path- ways involving neurotrophic receptor tyrosine kinease 3 (NTRK3) and NTRK2 via the RAS pathway, which are crucial for nervous system development and survival. This tract was also linked to insulin-like growth factor 1 receptor (IGF1R) signaling and insulin receptor bind- ing pathways. Furthermore, STG FAFWcorr showed associations with cancer mechanisms, including mutant forms and internalization of the epidermal growth factor receptor (EGFR). It also included pathways related to the overexpression of the human epidermal growth factor ### Page 9 LORENZ ET AL. 9 of 17F I G U R E 3 Genome-wide significant locus near CDH19 associated with medial temporal gyrus (MTG) free-water corrected radial diffusivity (RDFWcorr). (A) Location of the seven limbic white matter (WM) in the brain, with the MTG colored in light green. (B) Manhattan plot of GWAS results for MTG RDFWcorr. The red line indicates genome-wide significance (p < 5 × 10−8). The blue line indicates suggestive significance (p < 1 ×10−5). (C) Forest plot for the effect of variant rs12959877 for each cohort. (D) LocusZoom plot for variant rs12959877 on chromosome 18. Abbreviations: ADNI, Alzheimer’s Disease Neuroimaging Initiative; BIOCARD, Predictors of Cognitive Decline Among Normal Individuals; BLSA, Baltimore Longitudinal Study of Aging; chr, chromosome; GWAS, genome-wide association study; MAP, Memory and Aging Project; MTG, middle temporal gyrus transcallosal tract; NACC, National Alzheimer’s Coordinating Center; VMAP, Vanderbilt Memory and Aging Project; ROS, Religious Orders Study; WRAP, Wisconsin Registry for Alzheimer’s Prevention. receptor 2 (HER2) and signaling events mediated by the EphA2 recep- tor. For UF AxDFWcorr, enriched pathways included photoreceptor cell outer segment organization and the binding of the peptide hormone angiotensin to its receptor. Statistics for all pathway analyses can be found in Table S11 (p < 0.05).3.5 Genetic covariance between WM microstructure with cardiovascular, lipid, inflammatory, and neurological traitsTo investigate the extent of shared genetic factors between WM microstructure and complex traits (N = 65), we performed genetic covariance analyses using GNOVA. Traits that exhibited at least one FDR-corrected significant association with one dMRI metric are dis- played in Figure 5A. HDL cholesterol demonstrated multiple signif- icant genetic associations with WM microstructure. Specifically, we observed negative covariance with FAFWcorr and AxDFWcorr, alongside positive covariance with MDFWcorr, RDFWcorr, and FW. When examin- ing the local genetic covariance for HDL cholesterol with MDFWcorr and RDFWcorr, multiple distributed genomic regions exhibiting significant covariation were identified (Figure 5B, Table S12). The genetics of resting heart rate variability traits (SDNN, RMSSD, pvRSAHF) and triglyceride showed negative associations with those of MDFWcorr, RDFWcorr, and FW. For metabolic traits, type 2 dia- betes was positively associated with fornix FAFWcorr and AxDFWcorr, whereas hydroxyvitamin D demonstrated several positive associations with MDFWcorr and RDFWcorr. Moreover, multiple genetic associations between WM microstructure were identified with immune-related diseases, including ulcerative colitis, rheumatoid arthritis, primary ### Page 10 10 of 17 LORENZ ET AL.F I G U R E 4 Gene expression profiles in brain tissues of genes identified through GWAS. (A) Heatmap displaying t-statistics for associations between gene expression profiles in three brain tissues for GWAS-identified genes (RORA, SENP5, KCNK6, CDH19, FAM107B, and MIR548A1) with cognitive outcomes and AD pathologies. Only genes with at least one significant association (pFDR < 0.05) with a cognitive or AD outcome are included. (B,C) Exemplary volcano plots showing the association of FAM107B expression in PCC with (B) longitudinal cognitive performance and (C) neuritic plaque burden. Significance thresholds are indicated by color. All volcano plots can be found in Figure S1. Abbreviations: AD, Alzheimer’s disease; CN, caudate nucleus; DLPFC, dorsolateral prefrontal cortex; FDR, false discovery rate; GWAS, genome-wide association study; PCC, posterior cingulate cortex. sclerosing cholangitis, primary biliary cirrhosis, inflammatory bowel disease, eczema, celiac disease, and asthma. Neurological and psy- chiatric traits also showed significant associations, encompassing schizophrenia, neuroticism, ischemic stroke, frontotemporal dementia, epilepsy, autism spectrum disorder, anxiety disorder, and amyotrophic lateral sclerosis. Behavioral traits such as subjective well-being, smok- ing initiation, smoking cessation, cigarettes per day, cannabis depen- dence, sleep duration, risky behavior, loneliness, internalizing prob- lems, educational attainment, antisocial behavior, aggressive behavior, and age of initiation furthermore shared significant genetic archi- tecture with WM microstructural traits. Statistics for all computed tests can be found in Table S13. Most of the genetic covariance results remained significant and showed similar effect directions after removing the APOE region from the genome (Figure S2, Table S14). Genetic covariance analysis between FW-corrected dMRI metrics and 3143 neuroimaging-derived phenotypes from the UK Biobank identified 9218 significant associations after FDR correction of 110,005 tests performed (Table S15). Figure S3 highlights the neuroimaging-derived phenotypes with at least 10 significant genetic associations across the FW-corrected dMRI metrics. Notably, sub- stantial genetic overlap was observed with the UK Biobank dMRI measures (e.g., mode of the arcuate fasciculus, FACONV of the posterior corona radiata) and resting-state network metrics (e.g., NET100_0452, NET100_1425).3.6 TWAS identified an association between STG RDFWcorr and DNAJB14The summary statistics derived from the meta-analyzed GWAS of WM microstructure were used to compute TWAS using S-PrediXcan. A sig- nificant association between STG RDFWcorr and the gene DNAJB14 in colon transverse tissue (z = −5.47, pFDR = 0.016, NSNPs = 27) was identified. DNAJB14 is involved in chaperone cofactor-dependent pro- tein refolding and protein-containing complex assembly.63 Results (p <0.05) for all analyses can be found in Table S16.4 DISCUSSIONThis study investigated the genetic architecture of limbic WM microstructure in a large, harmonized sample of older adults enriched for cognitive impairment, utilizing advanced FW-corrected dMRI met- rics. We observed substantial heritability of limbic WM microstructure, especially within tracts such as the cingulum, fornix, and ILF. Our meta-analyzed GWAS identified several genetic loci associated with WM characteristics, implicating the genes CDH19, KC6, SENP5, RORA,FAM107B, and MIR548A1. Furthermore, we identified gene-level asso- ciation of WM microstructure with DNAJB14 using TWAS and withSERPINA12 using gene-level analysis. Pathway and genetic covariance ### Page 11 LORENZ ET AL. 11 of 17F I G U R E 5 Genetic covariance between complex traits and white matter (WM) microstructure. (A) Genetic covariance between diffusion MRI (dMRI) metrics and complex traits that have shown at least one association with a dMRI metric (pFDR < 0.05). “*” marks a pFDR < 0.05. (B) Local genetic covariance between HDL cholesterol with MDFWcorr and RDFWcorr. Only genomic regions with a p < 0.05 were included. Abbreviations: AxD, axial diffusivity; FA, fractional anisotropy; FDR, false discovery rate; FW, free water; ILF, inferior longitudinal fasciculus; ITG, inferior temporal gyrus transcallosal tract; MD, mean diffusivity; MTG, middle temporal gyrus transcallosal tract; RD, radial diffusivity; STG, superior temporal gyrus transcallosal tract; UF, uncinate fasciculus. ### Page 12 12 of 17 LORENZ ET AL.analyses revealed significant links between limbic WM microstructure and biological processes related to inflammation, vascular health, lipid metabolism, and various neurological conditions. The strongest GWAS signal identified was a locus on chromosome 18, with most SNPs acting as eQTLs for CDH19. This gene, also known as cadherin 19, encodes a calcium-dependent cell-adhesion glycopro- tein highly expressed in oligodendrocytes and Schwann cells, which are crucial for myelin formation and maintenance.59 This finding suggests a novel link between genetic variation influencing myelin-related cell adhesion processes and WM microstructural integrity in older adults. Another association was found near SENP5 on chromosome 3 with fornix AxDFWcorr. SENP5 regulates post-translational SUMOylation, critical for gene expression, DNA repair, and mitochondrial dynam- ics, and has been implicated previously in synaptogenesis and synaptic function in mature neurons.64,65 Based on our findings, SENP5 seems to be also involved in WM microstructural differences in later life. Several other GWAS-identified genes, including RORA, FAM107B, and KC6, showed significant associations of their brain tissue expres- sion levels with cognitive decline and neuropathologies (tau tangles, neurite plaques, neurofibrillary tangles, and Aβ plaques) in our bulk RNA-seq analyses. RORA encodes a protein that binds to hormone response elements upstream of multiple genes to enhance their expression. It regulates key genes involved in neurological functions and is implicated in autism spectrum disorders, highlighting its role in neuronal differentiation and synaptic plasticity.66 As a nuclear recep- tor transcription factor, RORA interacts with transcriptional regulators like insulin and brain-derived neurotrophic factors, both implicated in neurodegeneration and cognitive aging, including AD.67 In addi- tion, RORA supports neuronal survival and has a neuroprotective role in Parkinson’s disease by protecting neurons from oxidative stress.68FAM107B is widely expressed in the brain and associated with various brain structural features, such as cortical thickness, subcortical vol- ume, and overall brain morphology.69 KC6 is a less well-characterized RNA gene and has been associated with corneal biology and diseases.70 Although canonical AD-risk genes did not reach genome-wide sig- nificance, prior work in the same cohorts using a hypothesis-driven approach demonstrated associations between limbic FW-corrected WM microstructure and established AD-risk variants, including APOE,TMEM106B, PTK2B, and WNT3, as well as AD polygenic risk.27 Overall, these findings indicate that some of the identified genes are involved in general brain health, which is further supported by the observed genetic associations between WM microstructure and neurologi- cal and psychiatric conditions, such as multiple sclerosis, aggressive behavior, and internalizing problems. Our study strongly implicates inflammatory, lipid metabolism, and vascular mechanisms in limbic WM health. RORA is also involved in regulating cholesterol and glucose metabolism and blood vessel morphogenesis.67 SENP5, in addition to its importance for brain health, has a critical role in cardiac function by regulating mitochondrial balance. Excessive de-SUMOylation by SENP5, which is upregulated in human heart failure, has been linked to cardiomyopathies and cardiac dysfunction.71,72 Our post-GWAS analyses supported associa- tions between WM microstructure with lipid and vascular mechanisms, revealing gene-level associations for SERPINA12 and links to insulin and inflammation pathways. Notably, SERPINA12 is elevated in indi- viduals with obesity and is associated with obesity-linked metabolic traits, such as insulin resistance, impaired glycemic control, and cardio- vascular disease.73 The connection between WM and cardiovascular mechanisms was also demonstrated by genetic overlap between WM microstructure and HDL cholesterol, triglycerides, resting heart rate variability traits, ischemic stroke, type 2 diabetes, and several inflam- matory diseases. These findings suggest that cardiovascular risk and associated systemic inflammation may provide a pathophysiological basis for WM alterations observed in later life, which can in turn con- tribute to risk for neurodegenerative diseases and cognitive decline. Vascular conditions such as hypertension, atherosclerosis, and type 2 diabetes may compromise vascular integrity, leading to reduced blood flow or microvascular damage in the brain. Such vascular challenges adversely affect WM, potentially resulting in axonal and myelin dam- age. This highlights the importance of intact energy metabolism and the cardiovascular system in brain health. Although FW-corrected metrics provide improved tissue specificity relative to conventional dif- fusion measures, they remain sensitive to broader biological processes affecting WM. Consistent with our genetic covariance analyses, which demonstrated overlap with vascular, lipid, inflammatory, and neuro- logical traits, the identified loci likely reflect pathways influencing WM integrity through both microstructural and systemic mecha- nisms, including neurodegenerative and cerebrovascular processes. Further research is needed to disentangle signals related to vascular mechanisms from AD-specific pathologies. A key strength of this study is the investigation of genetic influence on WM microstructure within well-characterized cohorts of older indi- viduals, including those with cognitive impairment, which enhances our ability to uncover biological mechanisms relevant to aging and neu- rodegenerative diseases. Unlike large-scale cohorts and studies such as UK Biobank and ENIGMA that have predominantly focused on WM in midlife, our study specifically addresses the genetic architecture of WM degeneration in the context of aging, cognitive impairment, and AD risk. In addition, the use of FW-corrected dMRI metrics provides a more accurate quantification of WM microstructure by mitigating confounds from extracellular FW. This study has several limitations. Our analyses were conducted on non-Hispanic White individuals, which, while reducing confound- ing from population stratification, limits the generalizability of our findings to more diverse populations. Future research should aim to replicate these findings in larger, multi-ethnic cohorts. The cross- sectional and correlational nature of our analyses precludes causal inferences. Although the use of single-shell dMRI data allowed for the inclusion of a large, harmonized dataset, the FW correction approach used here is based on a bi-tensor model whose parameter estimation is generally better constrained in multi-shell than in single-shell diffusion acquisitions.74 Accordingly, although FW-corrected metrics improve tissue specificity relative to conventional diffusion measures, indices derived from single-shell data remain model dependent and should be interpreted as nonspecific tissue-compartment diffusion measures rather than direct markers of a single biological process. Future studies ### Page 13 LORENZ ET AL. 13 of 17employing multi-shell dMRI approaches, such as Neurite Orientation Dispersion and Density Imaging (NODDI), could provide more detailed insights into distinct tissue compartments and potentially quantify aspects like neuroinflammation.74 Although FW correction reduces partial volume effects from extracellular FW, structural MRI measures such as regional brain volumes and WM hyperintensity (WMH) burden were not incorporated in the present study. As both macrostructural atrophy and WMH burden can influence diffusion-derived metrics, we cannot exclude residual confounding, and the identified associations may reflect a combination of microstructural, neurodegenerative, and cerebrovascular processes. Finally, only a small subset of our cohort (5%) had a formal clinical diagnosis of AD at the time of imaging. Enriching future cohorts with individuals across the AD spectrum may enhance the detection of neurodegeneration-specific genetic signals. Taken together, this imaging genetics study identified several novel genes linked to limbic WM microstructure in multiple, harmonized aging cohorts. Notably, bulk RNA-seq analyses demonstrated that some of these genes were associated with cognitive decline and Aβand tau burden, suggesting AD relevance. Furthermore, our findings revealed that the genetic architecture of limbic WM is strongly tied to vascular health and inflammation, highlighting these pathways as promising avenues for future therapeutic development.AFFILIATIONS1Vanderbilt Memory and Alzheimer’s Center, Vanderbilt University School of Medicine, Nashville, Tennessee, USA 2Vanderbilt Genetics Institute, Vanderbilt University Medical Center, Nashville, Tennessee, USA 3Department of Computer Science, Vanderbilt University, Nashville, Tennessee, USA 4Department of Electrical and Computer Engineering, Vanderbilt University, Nashville, Tennessee, USA 5Department of Neurology, Vanderbilt University Medical Center, Nashville, Tennessee, USA 6Vanderbilt Brain Institute, Vanderbilt University Medical Center, Nashville, Tennessee, USA 7Department of Radiology and Imaging Sciences, Indiana University School of Medicine, Indianapolis, Indiana, USA 8Indiana Alzheimer’s Disease Research Center, Indiana University School of Medicine, Indianapolis, Indiana, USA 9Laboratory for Behavioral Neuroscience, National Institute on Aging, National Institutes of Health, Baltimore, Maryland, USA 10Department of Biomedical Engineering, Illinois Institute of Technology, Chicago, Illinois, USA 11Rush Alzheimer’s Disease Center, Rush University Medical Center, Chicago, Illinois, USA 12Department of Diagnostic Radiology, Rush University Medical Center, Chicago, Illinois, USA 13Department of Radiology, University of Pennsylvania, Philadelphia, Pennsylva- nia, USA 14Neuroimage Analytics Laboratory and Biggs Institute Neuroimaging Core, Glenn Biggs Institute for Alzheimer’s and Neurodegenerative Diseases, Univer- sity of Texas Health Science Center, San Antonio, Texas, USA 15Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA 16Laboratory of Neuroimaging, USC Stevens Institute of Neuroimaging and Infor- matics, Keck School of Medicine, University of Southern California, Los Angeles, California, USA 17Imaging Genetics Center, Mark and Mary Stevens Institute for Neuroimag- ing and Informatics, Keck School of Medicine, University of Southern California, Marina del Rey, California, USA 18Department of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, California, USA 19Department of Biostatistics, Vanderbilt University Medical Center, Nashville, Tennessee, USA 20Department of Radiology & Radiological Sciences, Vanderbilt University Medi- cal Center, Nashville, Tennessee, USA 21Vanderbilt University Medical Center, Vanderbilt University Institute of Imag- ing Science, Nashville, Tennessee, USA 22Department of Neurology, Johns Hopkins School of Medicine, Baltimore, Maryland, USA 23National Alzheimer’s Coordinating Center, University of Washington, Seattle, Washington, USA 24Department of Biomedical Engineering, Vanderbilt University, Nashville, Ten- nessee, USA 25Wisconsin Alzheimer’s Disease Research Center, School of Medicine and Public Health, University of Wisconsin, Madison, Wisconsin, USA 26Wisconsin Alzheimer’s Institute, School of Medicine and Public Health, Univer- sity of Wisconsin, Madison, Wisconsin, USA 27Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee, USAACKNOWLEDGMENTSThis study was supported by several funding sources, including 26PRE1558066 (A.S.L.), K01-EB032898 (K.G.S.), R01-EB017230 (B.A.L.), R01-AG074339 (E.C.M.), P30-AG066515 (E.C.M.), K01- AG073584 (D.B.A.), U24-AG074855 (T.J.H.), 75N95D22P00141 (T.J.H.), R01-AG059716 (T.J.H.), UL1-TR000445 and UL1-TR002243 (Vanderbilt Clinical Translational Science Award), and S10-OD023680 (Vanderbilt’s High-Performance Computer Cluster for Biomedical Research). The research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH), National Institute on Aging (NIA). Study data were obtained from the Vanderbilt Memory and Aging Project (VMAP). VMAP data were collected by Vanderbilt Memory and Alzheimer’s Center Investigators at Van- derbilt University Medical Center. This work was supported by NIA grants R01-AG034962 (PI: Jefferson), R01-AG056534 (PI: Jefferson), U19-AG03655 (PI: Albert), and Alzheimer’s Association IIRG-08- 88733 (PI: Jefferson). The data contributed from the Wisconsin Registry for Alzheimer’s Prevention was supported by NIA AG021155, AG027161, AG037639, and AG054047. The BLSA is supported by the Intramural Research Program of the NIH, NIA. This research was supported in part by the Intramural Research Program of the NIH, NIA. Data collection and sharing for this project were funded (in part) by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (NIH Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). ADNI is funded by the NIA, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; ### Page 14 14 of 17 LORENZ ET AL.Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Com- pany; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharma- ceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. Data collection and sharing for the ADNI is funded by the NIA (NIH Grant U19 AG024904). The grantee organization is the North- ern California Institute for Research and Education. In the past, ADNI has also received funding from the National Institute of Biomedi- cal Imaging and Bioengineering, the Canadian Institutes of Health Research, and private sector contributions through the Foundation for the National Institutes of Health (FNIH), including generous contribu- tions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann- La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research &Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeu- tics. Data contributed from Memory and Aging Project (MAP) and the Religious Orders Study (ROS) was supported by NIA R01AG017917, P30AG10161, P30AG072975, R01AG056405, UH2NS100599, UH3NS100599, R01AG064233, R01AG015819, and R01AG067482, and the Illinois Department of Public Health (Alzheimer’s Disease Research Fund). Data can be accessed at www.radc.rush.edu. The National Alzheimer’s Coordinating Center (NACC) database is funded by NIA/NIH Grant U24 AG072122. NACC data are contributed by the NIA-funded Alzheimer’s Disease Research Centers (ADRCs): P50 AG005131 (PI James Brewer, MD, PhD), P50 AG005133 (PI Oscar Lopez, MD), P50 AG005134 (PI Bradley Hyman, MD, PhD), P50 AG005136 (PI Thomas Grabowski, MD), P50 AG005138 (PI Mary Sano, PhD), P50 AG005142 (PI Helena Chui, MD), P50 AG005146 (PI Marilyn Albert, PhD), P50 AG005681 (PI John Morris, MD), P30 AG008017 (PI Jeffrey Kaye, MD), P30 AG008051 (PI Thomas Wis- niewski, MD), P50 AG008702 (PI Scott Small, MD), P30 AG010124 (PI John Trojanowski, MD, PhD), P30 AG010129 (PI Charles DeCarli, MD), P30 AG010133 (PI Andrew Saykin, PsyD), P30 AG072975 (PI Julie Schneider, MD), P30 AG012300 (PI Roger Rosenberg, MD), P30 AG013846 (PI Neil Kowall, MD), P30 AG013854 (PI Robert Vassar, PhD), P50 AG016573 (PI Frank LaFerla, PhD), P50 AG016574 (PI Ronald Petersen, MD, PhD), P30 AG019610 (PI Eric Reiman, MD), P50 AG023501 (PI Bruce Miller, MD), P50 AG025688 (PI Allan Levey, MD, PhD), P30-AG072946 (PI Linda Van Eldik, PhD), P50 AG033514 (PI Sanjay Asthana, MD, FRCP), P30 AG035982 (PI Russell Swerdlow, MD), P50 AG047266 (PI Todd Golde, MD, PhD), P50 AG047270 (PI Stephen Strittmatter, MD, PhD), P50 AG047366 (PI Victor Henderson, MD, MS), P30 AG049638 (PI Suzanne Craft, PhD), P30 AG053760 (PI Henry Paulson, MD, PhD), P30 AG066546 (PI Sudha Seshadri, MD), P20 AG068024 (PI Erik Roberson, MD, PhD), P20 AG068053 (PI Marwan Sabbagh, MD), P20 AG068077 (PI Gary Rosenberg, MD), P20 AG068082 (PI Angela Jefferson, PhD), P30 AG072958 (PI Heather Whitson, MD), and P30 AG072959 (PI James Leverenz, MD).CONFLICT OF INTEREST STATEMENTS.C.J. has received consulting fees from Enigma Biomedical and has served on an advisory board for ALZPath within the past 2 years. A.J.S. has received support from Avid Radiopharmaceuticals, a subsidiary of Eli Lilly, and has participated in scientific advisory boards for Bayer Oncology, Eisai, Novo Nordisk, and Siemens Medical Solutions. A.J.S. has also served on the Observational Study Monitoring Board for the Multi-Ethnic Study of Atherosclerosis (MESA) Medical Imaging with Deep Learning, an NIH NHLBI-funded study. E.C.M. has received con- sulting fees from Eli Lilly. B.B.B. has received consulting fees from New Amsterdam, Cognito Therapeutics, and Merry Life Biomedical. B.B.B. has also served on advisory boards for the Weston Advisory Board, Rush Alzheimer’s Disease Research Center External Advisory Board, Emory Alzheimer’s Disease Research Center External Advisory Board, and South Texas Alzheimer’s Disease Research Center Exter- nal Advisory Board. B.B.B. is the founder of Cognovance, LLC. T.J.H. has received consulting fees from Circular Genomics and has served on the scientific advisory board of Vivid Genomics. A.L.J. has served as Chair of the Observational Study Monitoring Board for the Diverse- VCID, and has also served as an advisory board member for Lantheus Diagnostic and Therapeutic Innovations. B.L. has received consulting fees from Silver Maple LLC, serves as a board member for Medical Imaging with Deep Learning (MIDL). S.R. has served on the external advisory board of the Adult Aging Brain Connectome and on inde- pendent scientific advisory boards for the Canadian Consortium on Neurodegeneration in Aging and Dementias Platform UK. A.W.T. serves on the RHU-SHIVA and ADNI steering committees. All other authors declare no competing interests. Author disclosures are available in the Supporting Information.CONSENT STATEMENTAll participants provided written informed consent in their respective cohort studies.REFERENCES1. Penke L, Maniega SM, Bastin ME, et al. Brain-wide white matter tract integrity is associated with information processing speed and general intelligence. Mol Psychiatry. 2012;17(10):955–955. https://doi.org/10. 1038/mp.2012.127 2. Ritchie SJ, Bastin ME, Tucker-Drob EM, et al. Coupled Changes in Brain White Matter Microstructure and Fluid Intelligence in Later Life. J Neurosci. 2015;35(22):8672–8682. https://doi.org/10.1523/ JNEUROSCI.0862-15.2015 ### Page 15 LORENZ ET AL. 15 of 173. Cox SR, Ritchie SJ, Tucker-Drob EM, et al. Ageing and brain white matter structure in 3,513 UK biobank participants. Nat Commun. 2016;7:13629. https://doi.org/10.1038/ncomms13629 4. Ihara M, Polvikoski TM, Hall R, et al. Quantification of myelin loss in frontal lobe white matter in vascular dementia, Alzheimer’s disease, and dementia with Lewy bodies. Acta Neuropathol. 2010;119(5):579– 589. https://doi.org/10.1007/s00401-009-0635-8 5. Sachdev PS, Zhuang L, Braidy N, Wen W. Is Alzheimer’s a disease of the white matter? Curr Opin Psychiatry. 2013;26(3):244. https://doi.org/ 10.1097/YCO.0b013e32835ed6e8 6. Simpson JE, Ince PG, Higham CE, et al. Microglial activation in white matter lesions and nonlesional white matter of ageing brains.Neuropathol Appl Neurobiol. 2007;33(6):670–683. https://doi.org/10. 1111/j.1365-2990.2007.00890.x 7. Yang Y, Schilling K, Shashikumar N, et al. White matter microstructural metrics are sensitively associated with clinical staging in Alzheimer’s disease. Alzheimers Dement (Amst). 2023;15(2):e12425. https://doi. org/10.1002/dad2.12425 8. Archer DB, Schilling K, Shashikumar N, et al. Leveraging longitudinal diffusion MRI data to quantify differences in white matter microstruc- tural decline in normal and abnormal aging. Alzheimers Dement (Amst). 2023;15(4):e12468. https://doi.org/10.1002/dad2.12468 9. Peterson A, Sathe A, Zaras D, et al. Sex and APOE ε4 allele differences in longitudinal white matter microstructure in multiple cohorts of aging and Alzheimer’s disease. Alzheimers Dement. 2025;21(1):e14343. https://doi.org/10.1002/alz.14343 10. Lee S, Viqar F, Zimmerman ME, et al. White matter hyperintensities are a core feature of Alzheimer’s disease: evidence from the domi- nantly inherited Alzheimer network. Ann Neurol. 2016;79(6):929–939. https://doi.org/10.1002/ana.24647 11. Nasrabady SE, Rizvi B, Goldman JE, Brickman AM. White matter changes in Alzheimer’s disease: a focus on myelin and oligodendro- cytes. Acta Neuropathol Commun. 2018;6:22. https://doi.org/10.1186/ s40478-018-0515-3 12. Archer DB, Moore EE, Shashikumar N, et al. Free-water metrics in medial temporal lobe white matter tract projections relate to longi- tudinal cognitive decline. Neurobiol Aging. 2020;94:15–23. https://doi. org/10.1016/j.neurobiolaging.2020.05.001 13. Basser PJ, Jones DK. Diffusion-tensor MRI: theory, experimental design and data analysis - a technical review. NMR Biomed. 2002;15(7– 8):456–467. https://doi.org/10.1002/nbm.783 14. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL. Neuroimage. 2012;62(2):782–790. https://doi.org/10.1016/j. neuroimage.2011.09.015 15. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval- Jeantet M. MR imaging of intravoxel incoherent motions: appli- cation to diffusion and perfusion in neurologic disorders. Radiol- ogy. 1986;161(2):401–407. https://doi.org/10.1148/radiology.161.2. 3763909 16. Le Bihan D, Breton E. Imagerie de diffusion in-vivo par résonance magnétique nucléaire. Comptes-Rendus de l’Académie des Sciences. 1985;93(5):27–34. Accessed February 25, 2024. https://hal.science/ hal-00350090 17. Pasternak O, Sochen N, Gur Y, Intrator N, Assaf Y. Free water elimination and mapping from diffusion MRI. Magn Reson Med. 2009;62(3):717–730. https://doi.org/10.1002/mrm.22055 18. Gustavson DE, Coleman PL, Wang Y, et al. Exploring the genetics of rhythmic perception and musical engagement in the vanderbilt online musicality study. Ann N Y Acad Sci. 2023;1521(1):140–154. https://doi. org/10.1111/nyas.14964 19. Vuoksimaa E, Panizzon MS, Hagler DJ Jr, et al. Heritability of white matter microstructure in late middle age: A twin study of tract-based fractional anisotropy and absolute diffusivity indices. Hum Brain Mapp. 2016;38(4):2026–2036. https://doi.org/10.1002/hbm.23502 20. Zhao B, Li T, Yang Y, et al. Common genetic variation influencing human white matter microstructure. Science. 2021;372(6548):eabf3736. https://doi.org/10.1126/science.abf3736 21. Zhao B, Zhang J, Ibrahim JG, et al. Large-scale GWAS reveals genetic architecture of brain white matter microstructure and genetic over- lap with cognitive and mental health traits (n = 17,706). Mol Psychiatry. 2021;26(8):3943–3955. https://doi.org/10.1038/s41380-019-0569- z 22. Bagepally BS, Halahalli HN, John JP, et al. Apolipoprotein E4 and brain white matter integrity in Alzheimer’s disease: tract-based spatial statistics study under 3-tesla MRI. Neurodegener Dis. 2012;10(1– 4):145–148. https://doi.org/10.1159/000334761 23. De Rossi P, Buggia-Prévot V, Clayton BLL, et al. Predominant expres- sion of Alzheimer’s disease-associated BIN1 in mature oligoden- drocytes and localization to white matter tracts. Mol Neurodegener. 2016;11:59. https://doi.org/10.1186/s13024-016-0124-1 24. Harrison JR, Bhatia S, Tan ZX, et al. Imaging Alzheimer’s genetic risk using diffusion MRI: A systematic review. Neuroimage Clin. 2020;27:102359. https://doi.org/10.1016/j.nicl.2020.102359 25. Heise V, Filippini N, Ebmeier KP, Mackay CE. The APOE ɛ4 allele mod- ulates brain white matter integrity in healthy adults. Mol Psychiatry. 2011;16(9):9. https://doi.org/10.1038/mp.2010.90 26. Foley SF, Tansey KE, Caseras X, et al. Multimodal brain imaging reveals structural differences in Alzheimer’s disease polygenic risk carriers: a study in healthy young adults. Biol Psychiatry. 2017;81(2):154–161. https://doi.org/10.1016/j.biopsych.2016.02.033 27. Lorenz A, Sathe A, Zaras D, et al. The effect of Alzheimer’s disease genetic factors on limbic white matter microstructure. Alzheimer’s Dement. 2025;21(4):e70130. https://doi.org/10.1002/alz.70130 28. Jack CR, Bernstein MA, Fox NC, et al. The Alzheimer’s disease neu- roimaging initiative (ADNI): MRI methods. J Magn Reson Imaging. 2008;27(4):685–691. https://doi.org/10.1002/jmri.21049 29. Albert M, Soldan A, Gottesman R, et al. Cognitive changes preceding clinical symptom onset of mild cognitive impairment and relation- ship to ApoE genotype. Curr Alzheimer Res. 2014;11(8):773–784. Accessed June 27, 2024. https://www.ncbi.nlm.nih.gov/pmc/articles/ PMC4163954/ 30. Besser LM, Kukull WA, Teylan MA, et al. The revised national Alzheimer’s coordinating center’s neuropathology form—available data and new analyses. J Neuropathol Exp Neurol. 2018;77(8):717–726. https://doi.org/10.1093/jnen/nly049 31. Weintraub S, Salmon D, Mercaldo N, et al. The Alzheimer’s disease centers’ uniform data set (UDS): the neuropsychologic test battery.Alzheimer Dis Assoc Disord. 2009;23(2):91. https://doi.org/10.1097/ WAD.0b013e318191c7dd 32. Weintraub S, Besser L, Dodge HH, et al. Version 3 of the Alzheimer disease centers’ neuropsychological test battery in the uniform data set (UDS). Alzheimer Dis Assoc Disord. 2018;32(1):10. https://doi.org/ 10.1097/WAD.0000000000000223 33. Bennett DA, Buchman AS, Boyle PA, Barnes LL, Wilson RS, Schneider JA. Religious orders study and rush memory and aging project.J Alzheimers Dis. 2018;64(Suppl 1):S161–S189. https://doi.org/10. 3233/JAD-179939 34. Moore EE, Liu D, Pechman KR, et al. Increased left ventricular mass index is associated with compromised white matter microstructure among older adults. J Am Heart Assoc. 2018;7(13):e009041. https://doi. org/10.1161/JAHA.118.009041 35. Johnson SC, Koscik RL, Jonaitis EM, et al. The wisconsin registry for Alzheimer’s prevention: a review of findings and current directions.Alzheimers Dement (Amst). 2017;10:130–142. https://doi.org/10.1016/ j.dadm.2017.11.007 36. Sager MA, Hermann B, La Rue A. Middle-aged children of per- sons with Alzheimer’s disease: APOE genotypes and cognitive func- tion in the wisconsin registry for Alzheimer’s prevention. J Geri- ### Page 16 16 of 17 LORENZ ET AL.atr Psychiatry Neurol. 2005;18(4):245–249. https://doi.org/10.1177/ 0891988705281882 37. Cai LY, Yang Q, Hansen CB, et al. PreQual: An automated pipeline for integrated preprocessing and quality assurance of diffusion weighted MRI images. Magn Reson Med. 2021;86(1):456–470. https://doi.org/ 10.1002/mrm.28678 38. Schilling KG, Blaber J, Huo Y, et al. Synthesized b0 for diffusion distor- tion correction (Synb0-DisCo). Magn Reson Imaging. 2019;64:62–70. https://doi.org/10.1016/j.mri.2019.05.008 39. Avants BB, Epstein CL, Grossman M, Gee JC. Symmetric diffeomor- phic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Med Image Anal. 2008;12(1):26–41. https://doi.org/10.1016/j.media.2007.06.004 40. Kim ME, Ramadass K, Gao C, et al. Scalable, reproducible, and cost- effective processing of large-scale medical imaging datasets. Med- ical Imaging 2025: Imaging Informatics. 2025;13411:128–140. SPIE. https://doi.org/10.1117/12.3047273 41. Archer DB, Coombes SA, McFarland NR, DeKosky ST, Vaillancourt DE. Development of a transcallosal tractography template and its applica- tion to dementia. Neuroimage. 2019;200:302–312. https://doi.org/10. 1016/j.neuroimage.2019.06.065 42. Brown CA, Johnson NF, Anderson-Mooney AJ, et al. Development, val- idation and application of a new fornix template for studies of aging and preclinical Alzheimer’s disease. Neuroimage Clin. 2017;13:106– 115. https://doi.org/10.1016/j.nicl.2016.11.024 43. Archer D. Tractography Templates for White Matter Microstruc- ture Analysis in Aging and Alzheimer’s Disease. Published online September 30, 2024. doi:10.5281/zenodo.13863955 44. Peter C, Sathe A, Shashikumar N, et al. White matter abnormal- ities and cognition in aging and Alzheimer disease. JAMA Neu- rol. 2025;82(8):825–836. https://doi.org/10.1001/jamaneurol.2025. 1601 45. Beer JC, Tustison NJ, Cook PA, et al. Longitudinal ComBat: a method for harmonizing longitudinal multi-scanner imaging data. Neuroim- age. 2020;220:117129. https://doi.org/10.1016/j.neuroimage.2020. 117129 46. Eissman JM, Dumitrescu L, Mahoney ER, et al. Sex differences in the genetic architecture of cognitive resilience to Alzheimer’s dis- ease. Brain. 2022;145(7):2541–2554. https://doi.org/10.1093/brain/ awac177 47. Taliun D, Harris DN, Kessler MD, et al. Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program. Nature. 2021;590(7845):290–299. https://doi.org/10.1038/s41586-021- 03205-y 48. Yang J, Lee SH, Goddard ME, Visscher PM. GCTA: a tool for genome- wide complex trait analysis. Am J Hum Genet. 2011;88(1):76–82. https://doi.org/10.1016/j.ajhg.2010.11.011 49. Benjamini Y, Hochberg Y. Controlling the false discovery rate - a practi- cal and powerful approach to multiple testing. J Royal Statist Soc, Series B. 1995;57:289–300. https://doi.org/10.2307/2346101 50. Mägi R, Morris AP. GWAMA: software for genome-wide association meta-analysis. BMC Bioinformatics. 2010;11:288. https://doi.org/10. 1186/1471-2105-11-288 51. Smith SM, Douaud G, Chen W, et al. An expanded set of genome-wide association studies of brain imaging phenotypes in UK Biobank. Nat Neurosci. 2021;24(5):737–745. https://doi.org/10.1038/s41593-021- 00826-4 52. Shatokhina N, Grasby KL, Jahanshad N, Stein JL, Medland SE, Thompson PM. ENIGMA-Vis: a web portal to browse, navigate & visu- alize brain genome-wide association studies (GWAS). Biological Psy- chiatry. 2021;89(9):S136. https://doi.org/10.1016/j.biopsych.2021.02. 350 53. Seto M, Dumitrescu L, Mahoney ER, et al. Multi-omic characterization of brain changes in the vascular endothelial growth factor family dur- ing aging and Alzheimer’s disease. Neurobiol Aging. 2023;126:25–33. https://doi.org/10.1016/j.neurobiolaging.2023.01.010 54. Leeuw CA de, Mooij JM, Heskes T, Posthuma D. MAGMA: gen- eralized gene-set analysis of GWAS data. PLoS Comput Biol. 2015;11(4):e1004219. https://doi.org/10.1371/journal.pcbi.1004219 55. Elliott LT, Sharp K, Alfaro-Almagro F, et al. Genome-wide associ- ation studies of brain imaging phenotypes in UK Biobank. Nature. 2018;562(7726):210–216. https://doi.org/10.1038/s41586-018- 0571-7 56. Lu Q, Li B, Ou D, et al. A powerful approach to estimating annotation- stratified genetic covariance via GWAS summary statistics. Am J Hum Genet. 2017;101(6):939–964. https://doi.org/10.1016/j.ajhg.2017.11. 001 57. Zhang Y, Lu Q, Ye Y, et al. SUPERGNOVA: local genetic correlation analysis reveals heterogeneous etiologic sharing of complex traits.Genome Biol. 2021;22(1):262. https://doi.org/10.1186/s13059-021- 02478-w 58. Gamazon ER, Wheeler HE, Shah KP, et al. A gene-based association method for mapping traits using reference transcriptome data. Nat Genet. 2015;47(9):9. https://doi.org/10.1038/ng.3367 59. Zorniak M, Clark PA, Kuo JS. Myelin-forming cell-specific cadherin- 19 is a marker for minimally infiltrative glioblastoma stem-like cells. J Neurosurg. 2015;122(1):69–77. https://doi.org/10.3171/2014. 9.JNS132373 60. Breitfeld J, Horn K, Le Duc D, et al. Genetic dissection of serum vaspin highlights its causal role in lipid metabolism. Obesity (Silver Spring). 2023;31(11):2862–2874. https://doi.org/10.1002/oby.23882 61. Yang L, Chen SJ, Yuan GY, Wang D, Chen JJ. Changes and clinical significance of serum vaspin levels in patients with type 2 diabetes.Genet Mol Res. 2015;14(3):11356–11361. https://doi.org/10.4238/ 2015.September.25.2 62. Zhou B, Liu Y, Ren Y, et al. Serum vaspin levels in gestational diabetes mellitus: a meta-analysis. Metab Syndr Relat Disord. 2023;21(10):535– 544. https://doi.org/10.1089/met.2023.0089 63. Goodwin EC, Motamedi N, Lipovsky A, Fernández-Busnadiego R, DiMaio D. Expression of DNAJB12 or DNAJB14 causes coordinate invasion of the nucleus by membranes associated with a novel nuclear pore structure. PLoS One. 2014;9(4):e94322. https://doi.org/10.1371/ journal.pone.0094322 64. Akiyama H, Nakadate K, Sakakibara SI. Synaptic localization of the SUMOylation-regulating protease SENP5 in the adult mouse brain.J Comp Neurol. 2018;526(6):990–1005. https://doi.org/10.1002/cne. 24384 65. Yamada S, Sato A, Ishihara N, Akiyama H, Sakakibara SI. Drp1 SUMO/deSUMOylation by Senp5 isoforms influences ER tubulation and mitochondrial dynamics to regulate brain development. iScience. 2021;24(12):103484. https://doi.org/10.1016/j.isci.2021.103484 66. Sarachana T, Hu VW. Genome-wide identification of transcriptional targets of RORA reveals direct regulation of multiple genes associated with autism spectrum disorder. Mol Autism. 2013;4(1):14. https://doi. org/10.1186/2040-2392-4-14 67. Acquaah-Mensah GK, Agu N, Khan T, Gardner A. A regulatory role for the insulin- and BDNF-linked RORA in the hippocampus: implications for Alzheimer’s disease. J Alzheimers Dis. 2015;44(3):827–838. https:// doi.org/10.3233/JAD-141731 68. Al-Zaid FS, Hurley MJ, Dexter DT, Gillies GE. Neuroprotective role for RORA in Parkinson’s disease revealed by analysis of post-mortem brain and a dopaminergic cell line. NPJ Parkinsons Dis. 2023;9:119. https://doi.org/10.1038/s41531-023-00563-4 69. van der Meer D, Frei O, Kaufmann T, et al. Understanding the genetic determinants of the brain with MOSTest. Nat Commun. 2020;11(1):3512. https://doi.org/10.1038/s41467-020-17368-1 70. Rabinowitz YS, Dong L, Wistow G. Gene expression profile studies of human keratoconus cornea for NEIBank: a novel cornea-expressed ### Page 17 LORENZ ET AL. 17 of 17gene and the absence of transcripts for aquaporin 5. Invest Ophthalmol Vis Sci. 2005;46(4):1239–1246. https://doi.org/10.1167/iovs.04-1148 71. Liu YP, Zhang TN, Wen R, Liu CF, Yang N. Role of posttransla- tional modifications of proteins in cardiovascular disease. Oxid Med Cell Longev. 2022;2022(1):3137329. https://doi.org/10.1155/2022/ 3137329 72. Mendler L, Braun T, Müller S. The ubiquitin-like SUMO sys- tem and heart function: from development to disease. Circ Res. 2016;118(1):132–144. https://doi.org/10.1161/CIRCRESAHA.115. 307730 73. Kurowska P, Mlyczyńska E, Dawid M, et al. Review: vaspin (SERPINA12) expression and function in endocrine cells. Cells. 2021;10(7):1710. https://doi.org/10.3390/cells10071710 74. Zhang H, Schneider T, Wheeler-Kingshott CA, Alexander DC. NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage. 2012;61(4):1000–1016. https://doi.org/ 10.1016/j.neuroimage.2012.03.072SUPPORTING INFORMATIONAdditional supporting information can be found online in the Support- ing Information section at the end of this article.How to cite this article: Lorenz AS, Sathe A, Yang Y, et al. Genetic architecture of the limbic white matter microstructure in aging and Alzheimer’s Disease. Alzheimer’s Dement. 2026;22:e71630. https://doi.org/10.1002/alz.71630