{"paper_id":"42da4243-8717-445a-9d37-ffcf4e6e75d0","body_text":"Glymphatic Dysfunction Coupled with Aberrant Distribution of Iron and Pathological proteins across Alzheimer’s Disease Continuum | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Glymphatic Dysfunction Coupled with Aberrant Distribution of Iron and Pathological proteins across Alzheimer’s Disease Continuum Chantat Leong, Aocai Yang, Ruisi Wang, Yu Sun, Jixin Luan, Manxi Xu, and 7 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9436391/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Introduction: Alzheimer’s disease (AD) involves impaired glymphatic clearance and abnormal iron-related tissue and pathological protein alterations, yet it remains unclear whether these processes represent independent abnormalities or a coordinated system-level architecture. Methods: We developed a multimodal MRI framework integrating multiple MRI indices capturing complementary aspects of glymphatic function with paramagnetic and diamagnetic susceptibility measures across AD spectrum. Cross-modal associations, stage-dependent deviations, and spatial progression patterns were examined to characterize clearance-tissue composition architecture linking fluid transport with susceptibility-derived alterations. Results: Glymphatic and susceptibility measures showed coordinated cross-modal associations at both ROI and network levels, with the strongest effects observed in limbic, default-mode, and frontoparietal networks. Across the AD continuum, coupled alterations were most prominent in deep nuclei and medial parietal regions, with deviations involving the thalamus, putamen, dentate nucleus, globus pallidus, precuneus, and posterior cingulate cortex. These multimodal patterns varied systematically with disease stage but were not significantly associated with cognitive performance. Conclusion: AD is characterized by progressive destabilization of a clearance-tissue composition system rather than isolated functional impairments. This cross-modal framework provides a systems-level perspective on neurodegenerative vulnerability and highlights integrated imaging markers sensitive to early physiological dysregulation along the AD continuum. Alzheimer's disease Cerebrospinal fluid flow Functional MRI Quantitative susceptibility mapping Glymphatic function Iron accumulation Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Background Alzheimer’s disease (AD) is characterized by progressive neurodegeneration arising from impaired clearance of metabolic waste, including amyloid-β (Aβ), tau, and other neurotoxic molecules. Inefficient removal of these proteins exacerbates their aggregation, amplifies oxidative and metabolic stress, and disrupts neuronal function, ultimately accelerating neurodegeneration. Converging evidence suggests that failure of brain fluid clearance systems, encompassing neurovascular coupling, blood-brain barrier (BBB) permeability, and glymphatic transport, plays a central mechanistic role in this process [ 1 – 4 ]. Dysregulated cerebrospinal fluid (CSF) transport is increasingly recognized as a critical component of impaired clearance in AD, and its disruption is expected not only to compromise solute removal but also to promote the accumulation of neurotoxic substances. The glymphatic system facilitates CSF-interstitial fluid exchange along perivascular pathways regulated by astrocytic aquaporin-4 (AQP4) and represents a key mechanism for metabolic waste clearance in the brain. AQP4 mislocalization, reduced CSF influx, and impaired perivascular solute transport are observed, leading to diminished clearance of Aβ and tau and promoting toxic protein accumulation [ 5 – 7 ]. Previous studies further show that AQP4 depolarization is associated with both amyloid burden and cognitive impairment, while genetic variations affecting AQP4 expression modulate AD risk and cognitive trajectories [ 8 , 9 ]. Advances in magnetic resonance imaging (MRI) enable in vivo quantification of multiple, complementary aspects of glymphatic and perivascular function. These associated MRI indicators include BOLD-CSF coupling (a proxy for neurovascular-driven CSF inflow), the DTI-ALPS index (for perivascular diffusivity), free-water (FW) content (for extracellular fluid mobility), and structural markers such as perivascular space (PVS) burden and choroid plexus (CP) morphology [ 10 – 16 ]. These metrics have consistently demonstrated glymphatic abnormalities across the AD spectrum, correlating with pathological protein burden and cognitive impairment. However, glymphatic indices are typically examined in isolation rather than within an integrated analytical framework. It remains unclear whether they reflect coordinated components of a unified fluid-transport system or converge onto shared vulnerability patterns in neurodegeneration. While glymphatic dysfunction is commonly considered a failure of Aβ and tau clearance, impaired CSF-interstitial fluid exchange is expected to influence the broader microstructural environment of brain tissue. Beyond the accumulation of soluble proteins, reduced clearance efficiency may promote iron retention and the accumulation of neurotoxic macromolecules, contributing to microstructural alterations in vulnerable regions [ 17 – 19 ]. Consistent with this view, iron homeostasis is profoundly altered in AD. Aging-related shifts in iron regulation, together with microglial activation and lysosomal dysfunction, promote iron retention in vulnerable cortical and subcortical regions. Excess iron catalyzes reactive oxygen species production, accelerates Aβ aggregation and tau phosphorylation, and contributes to mitochondrial damage and cellular death [ 20 – 23 ]. Previous studies also found that iron accumulation co-localizes with Aβ and tau aggregation, suggesting that altered tissue composition may emerge as a downstream consequence of impaired clearance-related processes [ 24 ]. Quantitative susceptibility mapping (QSM) provides a noninvasive approach to characterize microstructural tissue composition by capturing magnetic susceptibility signals sensitive to iron, myelin, and other macromolecular constituents. Conventional QSM reflects a composite voxel-wise signal in which paramagnetic and diamagnetic contributions may partially cancel each other, limiting biological specificity. Decomposition of QSM into paramagnetic (χ para ) and diamagnetic (χ dia ) components enables more specific separation of iron-dominant susceptibility from diamagnetic contributions related to neurotoxic and pathological proteins [ 25 – 28 ]. QSM studies consistently demonstrate regionally altered iron-related susceptibility across the AD spectrum, and vascular dysfunction, blood-brain barrier breakdown, and impaired interstitial drainage have been implicated in abnormal iron accumulation [ 29 , 30 ]. However, it remains unclear how these susceptibility alterations relate to clearance-related processes reflected by glymphatic function. Consequently, no integrative framework has yet unified glymphatic indices with sub-voxel QSM signals to determine whether these modalities share a coherent system-level organization, reflect coordinated processes, or exhibit structured deviations across neurodegenerative progression. Despite growing evidence for both glymphatic dysfunction and susceptibility alterations in AD, it remains unknown whether these represent independent pathological abnormalities or coordinated manifestations of a unified clearance-tissue composition architecture, encompassing iron and pathological protein alterations. Addressing this question requires an integrative framework capable of jointly characterizing fluid-transport dynamics and microstructural tissue properties across the whole brain. In this study, we introduced a multimodal MRI framework that integrates complementary glymphatic indices with paramagnetic and diamagnetic susceptibility components to examine whether glymphatic dysfunction and susceptibility alterations converge onto a shared physiological axis. By characterizing cross-regional associations between glymphatic indices and susceptibility-derived measures, this framework enables the capture of integrated clearance-tissue composition features across the brain. Such patterns may help delineate how disruption of the clearance-tissue composition axis unfolds along the AD continuum, providing insight into how physiological clearance processes interact with and shape the brain's microstructural environment (Fig. 1 ). Understanding these interactions is critical for clarifying how physiological clearance dysfunction contributes to the emergence of microstructural vulnerability in AD, thereby bridging molecular pathology with large-scale brain alterations. 2. Methods To examine the system-level relationships between glymphatic function and microstructural susceptibility, a multimodal MRI framework was used to integrate data acquisition, standardized preprocessing, glymphatic and susceptibility quantification, and cross-modal analytical strategies. A schematic overview of the framework is shown in Fig. 2 . 2.1 Participants This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and approved by the Ethics Committee of the China-Japan Friendship Hospital and Nanjing Drum Tower Hospital. Informed consent was obtained from all participants or their relatives who were fully informed about the study. For the multi-cohort and multimodal MRI study, a total of 838 participants (HC = 256, SCD = 154, MCI = 250, AD = 178) were recruited from the China-Japan Friendship Hospital and Nanjing Drum Tower Hospital. All participants underwent standardized neurological examinations, neuropsychological testing, and a comprehensive MRI protocol. Diagnoses followed the NIA–AA [ 31 ], and Petersen criteria [ 32 ], and participants were included if they were 50–85 years old, right-handed, and free of major neurological or psychiatric disorders (Fig. 2 A). Exclusion criteria comprised significant cerebrovascular lesions, substance misuse, MRI contraindications, or excessive head motion. Additional demographic information is provided in Supplementary Methods S1. 2.2 Image acquisition MRI data were acquired on two 3T MR platforms (GE MR750 with an 8-channel head coil at the China-Japan Friendship Hospital; Philips Ingenia CX with a 32-channel head coil at Nanjing Drum Tower Hospital) using site-matched multimodal protocols. 3D T1-weighted FSPGR (1 mm isotropic), 3D multi-echo GRE for QSM (1 mm isotropic), GRE-SS-EPI for resting-state fMRI, and diffusion-weighted spin-echo echo-planar (EPI) sequence for diffusion MRI. Full acquisition parameters for each modality are provided in Supplementary Methods S2. 2.3 MRI preprocessing All MRI data underwent standardized preprocessing pipelines (details in Supplementary Methods S3) to enable reliable extraction of glymphatic-related and susceptibility-based physiological metrics within a common anatomical framework. Structural T1-weighted images were corrected for bias field inhomogeneity, segmented using FreeSurfer, and registered to MNI152 space via boundary-based coregistration [ 33 ], providing anatomical features of glymphatic function. Diffusion data were denoised, distortion- and motion-corrected, modeled using DTI and bi-tensor free-water estimation, and coregistered to the T1 image [ 34 ]. These procedures enabled derivation of perivascular space (ALPS) and free water (FW) indices reflecting perivascular diffusivity and interstitial fluid mobility, which serve as proxies for glymphatic and extracellular transport function. Resting-state fMRI preprocessing included motion correction, nuisance regression, band-pass filtering, and spatial normalization [ 35 ], allowing extraction of BOLD–CSF coupling measures that capture neurovascular-driven CSF inflow dynamics. QSM was reconstructed using Laplacian unwrapping, LBV background removal and STAR-QSM inversion, and paramagnetic/diamagnetic susceptibility maps were computed through APART-QSM algorithms [ 36 ]. These steps enabled separation of χ para and χ dia components reflecting iron-related and macromolecular susceptibility sources. All susceptibility maps were subsequently normalized to MNI space via ANTs [ 37 ] to ensure spatial correspondence with glymphatic indices across subjects. 2.4 Computation of glymphatic- and iron-related measurements To jointly characterize clearance dynamics and microstructural tissue composition, we derived complementary glymphatic-related and susceptibility-based MRI metrics that capture distinct yet physiologically linked aspects of brain fluid transport and tissue microenvironment. Glymphatic-related indices were categorized into two major classes: coupling-related indices and perivascular indices. Coupling-related indices were designed to capture neurovascular-driven CSF inflow dynamics. Specifically, BOLD-CSF coupling was derived from resting-state fMRI by quantifying the temporal relationship between ventricular CSF signals and gray matter BOLD fluctuations, with the characteristic negative lag reflecting CSF inflow following neural and vascular activity [ 11 , 13 ]. This measure serves as an index of neurovascularly mediated glymphatic function. Perivascular indices were used to characterize solute transport and fluid mobility along perivascular and interstitial compartments. The DTI-ALPS index quantifies water diffusivity along perivascular directions relative to fiber orientations, providing a proxy for perivascular diffusivity and solute transport efficiency [ 34 , 38 ]. FW content was also estimated using DTI to isolate the extracellular water fraction, reflecting interstitial fluid mobility independent of tissue anisotropy [ 39 ]. In addition, volumetric measures of perivascular space (PVS), choroid plexus (CP), cerebrospinal fluid (CSF), and basal forebrain (BF) were extracted from structural MRI (Fig. 2 B). These measures represent structural correlates of CSF compartments and perivascular pathways that support glymphatic transport [ 8 , 40 , 41 ]. Iron-related and macromolecular tissue composition alterations were characterized using sub-voxel quantitative susceptibility mapping (QSM) metrics, namely χ dia and χ para . These metrics reflect paramagnetic susceptibility primarily associated with iron-related sources and diamagnetic susceptibility associated with pathological proteins and broader macromolecular tissue composition. To distinguish sources of susceptibility contrast, conventional QSM maps were decomposed into χ dia and χ para components using the APART-QSM framework [ 36 ], which separates paramagnetic contributions primarily associated with iron-related sources from diamagnetic contributions associated with pathological proteins and broader macromolecular tissue composition. Absolute diamagnetic susceptibility (|χ dia |) was used to avoid ambiguity in signal interpretation. Higher χ para and |χ dia | values indicate greater paramagnetic and diamagnetic tissue content, respectively (Fig. 2 B). To examine spatially specific susceptibility alterations relevant to AD pathology, region of interest (ROI) -level analyses were conducted across cortical and subcortical regions known to be early and frequently affected by amyloid and tau deposition. Twenty-eight ROIs were selected based on prior pathological evidence [ 28 , 42 ]. Regional χ para and |χ dia | values were computed by averaging voxel-wise values within each region after masking to exclude cerebrospinal fluid and non-brain voxels. Full parameters, regional definitions, and metric-specific computational procedures are provided in Supplementary Methods S3. 2.5 Statistical Analysis To characterize glymphatic-susceptibility relationships across the AD continuum from complementary perspectives, we employed a series of univariate, multivariate, and modeling approaches that together capture group differences, cross-modal associations, disease-relevant coupling patterns, and their cognitive and temporal implications. 2.5.1 Group-Level comparisons of imaging metrics Group differences in each glymphatic and susceptibility index were tested using ANCOVA, controlling for age, sex, and scanner. Post-hoc pairwise comparisons were FDR-corrected ( q < 0.05) [ 43 ]. Cohen’s d was computed to quantify effect sizes. 2.5.2 Glymphatic-Susceptibility association analysis Cross-modal associations between glymphatic indices and susceptibility measures were examined using complementary univariate and multivariate approaches. First, partial correlation analyses were performed between each glymphatic metric and each susceptibility measure (χ para and |χ dia |). Correlation matrices were computed separately for ROI-level (28 ROIs), which were selected based on previous studies identifying regions showing the most pronounced QSM abnormalities contributing to AD pathology [ 28 , 42 ], and network-level representations, including the Frontoparietal (FPN), Dorsal Attention (DAN), Ventral Attention (VAN), Limbic (LIM), Visual (VIS), Somatomotor (SOM), and Default Mode (DMN) networks. Statistical significance was assessed using two-tailed tests with FDR correction applied across all glymphatic-susceptibility pairs. Second, to identify coordinated cross-modal patterns not captured by pairwise correlations, partial least squares (PLS) analysis was conducted [ 44 , 45 ]. PLS was used to extract latent variables representing maximal shared variance between multiple glymphatic indices and susceptibility measures, capturing coordinated multivariate associations related to system-level clearance regulation (Fig. 2 C). Statistical significance of each latent component was assessed using permutation testing (3,000 permutations), and the stability of feature contributions was evaluated using bootstrap resampling (3,000 resamples). To ensure that multivariate associations were not driven by redundancy among glymphatic metrics, pairwise correlations among all glymphatic indices were examined, confirming the absence of near-collinearity (| r | > 0.80; Supplementary Fig. S1 ). 2.5.3 Multistage identification of AD-related coupled patterns To identify clearance-tissue composition patterns across the AD continuum, we integrated complementary statistical signatures, including system-level trend analysis, multivariate contributions, and group separation. First, a system-level trend analysis was performed to identify ROIs showing monotonic changes across the AD continuum using linear contrasts ( T ≥ 4, p < 0.001). Second, ROIs contributing significantly to disease-relevant PLS components were identified based on bootstrap ratios (|BR| ≥ 3), reflecting stable multivariate contributions. Third, for each candidate glymphatic–susceptibility pattern, Euclidean distances between group centroids were computed in the multivariate feature space to quantify the degree of separation across AD continuum. Glymphatic-susceptibility pairs satisfying all three criteria, including monotonic disease trend, significant multivariate contribution, and group separation, were classified as AD-related and retained for downstream analyses. 2.5.4 Structural equation modelling (SEM) Structural equation modelling (SEM) was employed to test whether multimodal glymphatic-susceptibility patterns mediated the relationship between disease status and cognitive performance [ 45 , 46 ]. Neurodegeneration latent factors (ND) were constructed from the selected coupled patterns, with highly collinear indicators removed to improve model stability. Models were estimated using robust maximum likelihood estimation, and standardized path coefficients and model-fit indices were extracted (Fig. 2 D). 2.5.5 Pseudotime-based disease progression modelling (DPM) To characterize the temporal organization of glymphatic-susceptibility alterations along disease progression, a pseudotime-based disease progression modeling (DPM) framework was applied [ 47 , 48 ]. Subject-level pseudotime scores were derived from the principal component structure of the ND feature matrix, rescaled to the [0, 1] interval, and oriented toward increasing disease severity. Pattern-wise trajectories were estimated using spline-smoothed resampling to capture continuous changes along pseudotime. Correlation-based clustering was used to group ND patterns into temporal classes, enabling characterization of heterogeneous progression profiles (Fig. 2 D). Full analytical details are provided in Supplementary Methods S4. 3. Results 3.1 Clinical characteristics and imaging features of glymphatic and susceptibility metrics Demographic variables (age, sex) and global cognition differed across groups along the AD continuum (Table 1). At the univariate imaging level, several glymphatic indices, including CSF volume, CP volume, FW, PVS, DTI-ALPS, BOLD-CSF coupling, and BF, showed progressive alterations across AD spectrum, consistent with reduced perivascular transport efficiency (Supplementary Table S1). Table 1. Demographic and clinical characteristics of AD, MCI, SCD and HC groups Characteristics AD MCI SCD HC F -value/ χ 2 p -value effect size Age 71.83±90.22 (N = 178) 66.74±8.02 (N = 250) 65.69±7.92 (N = 154) 64.21±8.71 (N = 256) 30.28 <0.0001 0.098 Sex(M/F) (51/95) (N = 178) (64/172) (N = 250) (28/106) (N = 154) (80/170) (N = 256) 8.45 <0.05 0.105 Education (years) 10.53±4.29 (N = 178) 10.73±3.42 (N = 250) 12.29±3.63 (N = 154) 11.23±4.47 (N = 254) 6.54 <0.05 0.023 MMSE 19.01±5.33 (N = 178) 27.18±2.41 (N = 250) 28.18±1.79 (N = 154) 28.33±1.72 (N = 256) 380.05 <0.001 0.59 MoCA 12.06±5.8 (N = 67) 21.37±3.97 (N = 83) 24.2±2.86 (N = 102) 25.09±2.93 (N = 89) 166.51 <0.001 0.6 HAMA 9.76±7.71 (N = 55) 5.33±5.85 (N = 81) 6.5±5.68 (N = 101) 3.5±6.37 (N = 66) 11.31 <0.001 0.1 HAMD 5.11±4.55 (N = 55) 3.89±4.76 (N = 81) 5.64±5.15 (N = 101) 2.68±3.28 (N = 66) 6.3 <0.05 0.059 RAVLT 6.28±12.13 (N = 71) 8.99±6.6 (N = 81) 12.13±7.16 (N = 101) 15.88±5.57 (N = 66) 18.03 <0.001 0.15 CDT 13.62±10.29 (N = 55) 23.32±6.51 (N = 81) 25.82±4.51 (N = 101) 27.5±2.3 (N = 66) 60.43 <0.001 0.38 BNT 16.04±6.16 (N = 71) 23.81±4.38 (N = 81) 26.16±3.47 (N = 101) 27.29±2.47 (N = 66) 101.83 <0.001 0.49 VFT 21.91±8.48 (N = 53) 29.23±6.31 (N = 30) 40.85±9.35 (N = 39) 48.33±7.91 (N = 12) 57.57 <0.001 0.57 Note: Data is presented as mean ± standard deviations (SD). MMSE: Mini-Mental State Examination, MoCA: Montreal Cognitive Assessment test, HAMD: Hamilton depression rating, HAMA: Hamilton anxiety rating, RAVLT: Rey Auditory Verbal Learning Test, CDT: Clock Drawing Test, BNT: Boston Naming Test, VFT: Verbal Fluency Test. The χ para and |χ dia | also showed widespread susceptibility alterations across temporal, parietal, and limbic cortices, with AD exhibiting lower diamagnetic and paramagnetic signals relative to SCD and MCI, alongside selective increases in the posterior superior temporal sulcus (Psts) and the dentate nucleus (DN). Network-level analyses further revealed reduced susceptibility signatures within limbic and default mode systems (Supplementary Table S2). Similar patterns of progressive alterations were observed across multiple glymphatic indices and susceptibility measures along the AD continuum. 3.2 Cross-modal glymphatic-susceptibility associations Univariate cross-modal analyses revealed structured, non-random coupling between glymphatic metrics and susceptibility-derived measures across multiple brain regions (Fig. 3A). gBOLD-CSF coupling and FW showed consistent negative correlations with both χ para and |χ dia |, whereas PVS, BF and DTI-ALPS exhibited positive correlations. These associations were most prominent in frontal, limbic, and subcortical regions. At the network level, parallel covariation patterns emerged: susceptibility within the LIM, DMN, and frontoparietal (FPN) networks showed strong cross-modal relationships with glymphatic indices, mirroring the anatomical organization observed at the ROI level. FW and PVS also demonstrate consistent significant correlations. 3.3 Multivariate glymphatic-susceptibility components identified by PLS PLS revealed a dominant latent component (LV1) that captured 55–57% of cross-modal covariance at the ROI level and 61–63% at the network level (|χ dia |: ROI 0.541, NET 0.626; χ para : ROI 0.568, NET 0.607). On the glymphatic side, gBOLD-CSF coupling contributed the dominant and most stable negative loadings across all models, particularly at the network level. FW also exhibited consistent negative contributions. In contrast, PVS, DTI-ALPS, and BF volume showed positive loadings. ALPS and PVS contributions were more prominent in |χ dia |, whereas χ para , showed comparatively weaker positive loadings. On the susceptibility side, for χ para , the strongest positive loadings were observed within LIM and DMN networks, accompanied by substantial contributions from visual (VIS) and FPN networks. |χ dia | showed prominent loadings within limbic and frontoparietal networks, together with prominent involvement of the ventral (VAN) and dorsal attention (DAN) networks and moderate contributions from somatomotor (SOM) regions. At the ROI level, |χ dia | additionally demonstrated pronounced loadings in deep nucleus and limbic regions, including the globus pallidus (GP), red nucleus (RN), putamen (Put), and thalamus (Tha). χ para loadings were primarily distributed across temporo-occipital and parahippocampal cortex, accompanied by moderate positive contributions in parietal and precuneus regions. Both susceptibility contrasts showed overlapping involvement of limbic, frontoparietal, and subcortical systems (Fig. 3B). 3.4 Spatially convergent glymphatic-susceptibility patterns across the AD continuum Multistage filtering identified a set of glymphatic-susceptibility pairs showing consistent alterations across the AD continuum based on trend analysis, PLS contributions, and group separation criteria (Supplementary Fig. S2). At the ROI level, both |χ dia | and χ para values demonstrated distinct glymphatic-susceptibility association patterns (Fig. 4). For |χ dia |, the strongest effects emerged in deep and posterior cortex, where DTI-ALPS and FW showed progressive alterations across disease stages. In contrast, χ para demonstrated stronger coupling with ALPS, FW, and PVS in medial and parietal regions, particularly in precuneus (Pcun). At the network level, a clearer and more integrated pattern emerged (Fig. 4). The strongest glymphatic-susceptibility associations were found within LIM, DMN, and FPN networks. This study also observed significant associations between the glymphatic function and susceptibility in the VIS and SOM networks. With regard to the glymphatic function, both FW and ALPS showed significant associations across susceptibility networks. 3.5 Associations between glymphatic-susceptibility patterns and cognition SEM models tested whether multimodal ND patterns mediated the relationship between diagnostic status and cognition (Fig. 5A). Diagnostic group significantly predicted both ND and MMSE across four models (ROI-level and network-level ND patterns derived from |χ dia | and χ para , p < 0.001), indicating that ND and cognitive performance captured graded differences across the AD continuum. In contrast, ND did not significantly predict MMSE in any model ( p > 0.05), and no indirect effects from diagnostic group to cognition through ND were observed. Although the magnitude and direction of the ND to MMSE paths varied slightly across glymphatic-susceptibility association domains, none reached statistical significance. 3.6 Sequential temporal progression of glymphatic-susceptibility alterations To characterize temporal organization, we examined DPM-derived pseudotime across ND indices from |χ dia | and χ para (Fig. 5B). ROI-based pseudotime showed significant increases from HC to AD, whereas network-based estimates showed no correspondence with diagnostic category. At the ROI level, both susceptibility modalities demonstrated substantial temporal heterogeneity. Deep and medial parietal regions deviated from healthy baselines at very early pseudotime positions, including Tha, Put, DN, GP, Pcun, and posterior cingulate cortex (PCC), while peak-change times varied widely across loci (Fig. 5B). The χ para indices often exhibited more prolonged or delayed trajectories relative to |χ dia |. At the network level, |χ dia | and χ para ND indices exhibited temporally staggered trajectories despite their weak correspondence with diagnostic groups. Early deviations were observed within limbic and default-mode networks, followed by mid-stage inflections in frontoparietal systems and later changes in somatomotor and dorsal-attention networks (Fig. 5B). 4. Discussion The present findings indicate that glymphatic function and susceptibility-related tissue composition are organized within a shared physiological system linking fluid clearance with iron-related and macromolecular properties across the brain. Rather than representing independent pathological processes, alterations in these domains appear to reflect the destabilization of a clearance-tissue composition architecture that is particularly prominent in metabolically demanding and vascularly complex regions. Across spatial scales, deviations from this coordinated organization become increasingly pronounced along the AD continuum, suggesting that disease progression is characterized by the progressive disruption of physiological mechanisms that regulate brain-environment homeostasis. This cross-modal framework therefore provides a systems-level perspective on how clearance inefficiency and tissue compositional alterations co-evolve, offering a unified lens through which regional vulnerability and disease progression can be understood. Glymphatic alterations in AD were broad yet patterned, reflecting coordinated disruption across multiple levels of the clearance pathway. Reductions in BOLD-CSF coupling and ALPS efficiency, together with elevated FW, PVS burden, and CSF and CP measures, point to convergent disruption of CSF inflow [ 11 , 13 ], perivascular transport, and interstitial fluid mobility [ 7 , 14 , 16 ]. Notably, several of these abnormalities were already present in SCD, indicating that glymphatic compromise emerges before the onset of overt cognitive decline. Parallel to these clearance-related disturbances, AD also exhibited distinct alterations in iron and neurotoxic proteins. The χ para values were reduced in the temporal, parietal, and occipital cortical regions, with focal elevations in the posterior superior temporal cortex and dentate nucleus. Increased χ para has been widely interpreted as reflecting iron accumulation. In contrast, cortical reductions in χ para are consistent with prior evidence of regionally selective iron redistribution in AD [ 49 ], a process that has been linked to mitochondrial stress and dysregulated cellular iron handling [ 50 , 51 ]. Diamagnetic susceptibility exhibited distinct patterns, with the strongest reductions in |χ dia | observed in frontal and parietal association cortices. In gray matter, |χ dia | has been shown to reflect macromolecular and structural composition beyond myelin content [ 25 – 27 , 51 ], and reductions in |χ dia | are consistent with alterations in protein density, cellular integrity, and extracellular organization accompanying neurodegeneration [ 28 ]. At the network level, both |χ dia | and χ para were prominently represented within LIM, DMN, and FPN, aligning their spatial expression with networks characterized by high metabolic demand, vascular burden, and early pathological vulnerability [ 52 ]. These susceptibility patterns collectively indicate that alterations in iron and macromolecular tissue composition undergo a transformation, occurring in parallel with the aforementioned clearance dysfunction and likely reflecting related pathological processes across these physiological domains. Beyond univariate alterations, the cross-modal analyses revealed that glymphatic and susceptibility features were not independently distributed but exhibited an organized pattern of coupling. The strongest positive associations emerged between PVS burden and ALPS efficiency, whereas negative associations were most pronounced between FW content and BOLD-CSF coupling. This organization reflects coordinated relationships between perivascular pathway integrity, interstitial fluid mobility, and large-scale solute exchange, linking microstructural tissue composition with fluid transport dynamics [ 17 , 19 ]. This directional pattern may reflect differences between structural capacity and effective exchange within the glymphatic system, whereby enlargement of perivascular pathways and increased transport demand coexist with reduced large-scale clearance efficiency. Notably, this coupling was preferentially expressed in regions such as Pcun and DN, areas characterized by high metabolic demand, dense vascular interfaces, and early AD vulnerability [ 52 , 53 ]. At the network level, LIM, DMN, and FPN played a dominant role in this cross-modal, indicating that coupling is organized according to large-scale functional hierarchies rather than being spatially widespread [ 11 , 54 , 55 ] and paramagnetic or diamagnetic changes observed in earlier analyses [ 51 , 56 ]. These alterations in tissue composition may represent downstream consequences of impaired solute exchange despite apparent increases in perivascular transport features. In particular, the preferential involvement of metabolically demanding and vascularly complex regions indicates that coupling between clearance-related and tissue compositional features is most prominent within circuits that rely heavily on efficient fluid transport and perivascular regulation. Taken together, these findings reflect that cross-modal physiological alterations are not randomly distributed but instead follow a structured spatial pattern aligned with known gradients of metabolic demand and vascular burden. When this clearance support is compromised, alterations in tissue composition are likely to arise within the same anatomical systems, providing a physiological explanation for the observed co-localization of glymphatic and susceptibility signatures in these metabolically and vascularly demanding circuits. AD progression was associated with region-specific changes in the relationships between glymphatic indices and susceptibility measures [ 2 ]. Rather than occurring uniformly across the brain, increases in ND and significant disease stage were preferentially clustered in deep nuclei and medial-parietal association cortices, including the DN, Tha, Pcun, and PCC. At the network level, they were also concentrated within LIM and DMN. The temporal ordering of these alterations revealed a heterogeneous but interpretable progression pattern in which these deep nuclei and limbic regions exhibited early deviations, followed by progressive engagement of frontoparietal and temporo-occipital association cortices [ 47 , 48 ]. Notably, the ordering of regional changes corresponded to areas where glymphatic dysfunction and susceptibility alterations were jointly expressed, indicating that disease progression is constrained by local clearance-tissue composition interactions rather than by conventional network organization. These findings suggest that AD does not affect glymphatic-susceptibility coupling features uniformly but progressively disrupts the organization of the clearance-tissue composition system according to regional dependence on fluid transport and metabolic support. Regions such as deep nuclei and medial parietal cortex, which rely heavily on efficient CSF-interstitial exchange to maintain local tissue homeostasis, appear particularly vulnerable to early disruption. As disease progresses, this disruption extends into higher-order association cortices, reflecting a hierarchical breakdown of the physiological coupling between clearance efficiency and tissue composition. This spatial gradient therefore reflects not merely where abnormalities occur but also characterizes the spatiotemporal evolution of both clearance dysfunction and neurotoxic or tissue compositional burden. Although ND increased progressively across diagnostic stages, it showed little correspondence with individual differences in cognitive performance. This dissociation indicates that the cross-modal physiological alterations captured by ND reflect system-level vulnerability that is not directly mirrored by cognitive impairment. Rather than tracking cognitive decline, ND appears to capture early alterations in glymphatic function and susceptibility-related tissue properties that emerge before overt cognitive symptoms. This suggests that glymphatic-susceptibility disruption represents a physiological layer of disease expression that is partially decoupled from clinical symptom manifestation. The separation between ND and cognition, therefore, indicates that clearance-tissue composition alterations follow a trajectory distinct from, and potentially preceding, measurable cognitive deterioration across the AD continuum. These findings imply that multimodal physiological biomarkers may reflect upstream system vulnerability in AD that is not readily captured by conventional cognitive assessments. These findings suggest that iron-related and macromolecular tissue composition alterations in AD are not solely attributable to localized pathological burden or iron dysregulation in isolation but instead emerge within the context of a progressively weakened clearance-tissue composition system [ 6 , 52 , 57 ]. This study underscores the importance of viewing clearance dysfunction and tissue compositional alterations as interconnected dimensions of disease vulnerability. By capturing their coupled evolution across spatial and disease-stage gradients, multimodal imaging approaches offer a powerful means of characterizing early physiological dysregulation in AD and provide a systems-level lens through which the breakdown of brain environmental homeostasis can be understood. The cross-modal framework may also provide a useful basis for exploring physiological mechanisms underlying neurodegenerative processes and for developing imaging markers sensitive to system-level vulnerability before overt structural degeneration becomes apparent. Several limitations should be noted. First, the cross-sectional design precludes causal inference and prevents direct characterization of disease trajectories; longitudinal studies integrating MRI with fluid or PET biomarkers will be necessary to clarify the temporal ordering of clearance and susceptibility changes. Second, the glymphatic and susceptibility indices used here serve as indirect proxies, each capturing only part of the underlying biology and potentially influenced by vascular, inflammatory, or metabolic factors not assessed in this study. Finally, applying this framework to additional neurodegenerative or cerebrovascular conditions may help determine whether the cross-modal architecture is AD-specific or reflects a broader clearance–accumulation motif. Despite these constraints, the present multimodal approach provides a scalable foundation for integrating fluid-transport and susceptibility-based markers toward more comprehensive imaging indicators of neurodegenerative vulnerability. 5. Conclusion This study demonstrates that glymphatic dysfunction and susceptibility alterations in AD are not independent abnormalities but coordinated manifestations of a unified clearance-tissue composition architecture. As the disease progresses, this physiological architecture becomes progressively destabilized, with early deviations emerging in deep and limbic regions and extending to higher-order cortical systems. The dissociation between cross-modal physiological alterations and cognitive performance further indicates that glymphatic-susceptibility coupling reflects an upstream vulnerability distinct from overt clinical impairment. By integrating complementary indices of fluid transport and tissue composition, the present framework provides a systems-level perspective on neurodegenerative vulnerability and offers a conceptual basis for understanding how clearance dysfunction and microstructural alterations co-evolve across the AD continuum. These findings highlight the potential of cross-modal physiological markers as sensitive indicators of early disease processes and as tools for probing mechanisms underlying neurodegeneration. Declarations Ethics Statement This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and was approved by the Ethics Committee of the China-Japan Friendship Hospital and Nanjing Drum Tower Hospital. All the participants and/or their relatives were informed about this study and provided their written informed consent. Acknowledgements The authors thank all participants in the study. Funding source This study was supported by the STI2030-Major Projects (No. 2022ZD0213300), the National Natural Science Foundation of China (No. 82271953 and U25A20136), Key Industrial Core Technology Tackling Project of the Hengqin Guangdong-Macao Deep Cooperation Zone (No. 2430004045063), University of Macau (MYRG2022-00054-FHS, MYRG-GRG2023-00038-FHS-UMDF, and MYRG-GRG2024-00259-FHS), and the Macao Science and Technology Development Fund (FDCT 0014/2024/RIB1 and 0048/2021/AGJ). Author contribution CT.L.: conceptualization, methodology, validation, investigation, formal analysis, writing—original draft, writing—review and editing. AC.Y.: conceptualization, methodology, validation, investigation, formal analysis, writing—original draft, writing—review and editing. RS.W.: investigation, formal analysis. Y.S.: resources, data curation, validation. JX.L.: methodology, validation. MX.X.: methodology, validation, formal analysis. Y.G.: resources, data curation, validation. HW.Y.: resources, data curation, validation. YL.W.: resources, data curation, validation. N.S.: writing—review and editing. B. Z.: writing—review and editing. GL.M.: conceptualization resources, funding acquisition, project administration, writing—review and editing. Z.Y.: conceptualization resources, funding acquisition, project administration, writing—review and editing. Competing interests The authors declare no conflicts of interest Data and code availability Open-source software was used for all analyses. The data and codes analyzed in this study can be obtained from the corresponding authors upon reasonable request. References Bloom GS. Amyloid-β and tau: the trigger and bullet in Alzheimer disease pathogenesis. JAMA Neurol. 2014;71(4):505–8. Jagust W. Imaging the evolution and pathophysiology of Alzheimer disease. Nat Rev Neurosci. 2018;19(11):687–700. Weiner MW, et al. The Alzheimer's Disease Neuroimaging Initiative: a review of papers published since its inception. Volume 9. Alzheimer's & Dementia; 2013. pp. e111–94. 5. Apátiga-Pérez R, et al. Neurovascular dysfunction and vascular amyloid accumulation as early events in Alzheimer's disease. Metab Brain Dis. 2022;37(1):39–50. Jessen NA, et al. The glymphatic system: a beginner’s guide. Neurochem Res. 2015;40:2583–99. Tarasoff-Conway JM, et al. Clearance systems in the brain—implications for Alzheimer disease. Nat reviews Neurol. 2015;11(8):457–70. Choi JD, et al. Choroid plexus volume and permeability at brain MRI within the Alzheimer disease clinical spectrum. Radiology. 2022;304(3):635–45. Lin W, et al. Multimodal MRI reveals impaired glymphatic function with choroid plexus enlargement and cerebrospinal fluid expansion in alzheimer’s disease. Sci Rep. 2025;15(1):30409. Costa T, et al. Evaluating the robustness of DTI-ALPS in clinical context: a meta-analytic parallel on Alzheimer’s and Parkinson’s diseases. Sci Rep. 2024;14(1):26381. Kiviniemi V, et al. Ultra-fast magnetic resonance encephalography of physiological brain activity–glymphatic pulsation mechanisms? J Cereb Blood Flow Metabolism. 2016;36(6):1033–45. Fultz NE, et al. Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep. Science. 2019;366(6465):628–31. Chuang K-H, et al. Cholinergic basal forebrain neurons regulate vascular dynamics and cerebrospinal fluid flux. Nat Commun. 2025;16(1):5343. Han F, et al. Reduced coupling between cerebrospinal fluid flow and global brain activity is linked to Alzheimer disease–related pathology. PLoS Biol. 2021;19(6):e3001233. Steward CE, et al. Assessment of the DTI-ALPS parameter along the perivascular space in older adults at risk of dementia. J Neuroimaging. 2021;31(3):569–78. Yu P, Shen L, Tang L. Multimodal DTI-ALPS and Hippocampal Microstructural Signatures Unveil Stage-Specific Pathways in Alzheimer's Disease Progression. Front Aging Neurosci. 2025;17:1609793. Zhang Y et al. DTI-ALPS index-assessed glymphatic dysfunction mediates Alzheimer’s cognitive decline via amyloid-β-dependent pathways: multimodal PET/MRI study. Eur J Nucl Med Mol Imaging, 2025: pp. 1–13. Iliff JJ et al. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Science translational medicine, 2012. 4(147): p. 147ra111-147ra111. Murdock MH, et al. Multisensory gamma stimulation promotes glymphatic clearance of amyloid. Nature. 2024;627(8002):149–56. Xie L, et al. Sleep drives metabolite clearance from the adult brain. Science. 2013;342(6156):373–7. Conrad M, et al. Regulated necrosis: disease relevance and therapeutic opportunities. Nat Rev Drug Discovery. 2016;15(5):348–66. Everett J, et al. Evidence of redox-active iron formation following aggregation of ferrihydrite and the Alzheimer’s disease peptide β-amyloid. Inorg Chem. 2014;53(6):2803–9. Tao Y, et al. Perturbed iron distribution in Alzheimer's disease serum, cerebrospinal fluid, and selected brain regions: a systematic review and meta-analysis. J Alzheimer’s Disease. 2014;42(2):679–90. Zecca L, et al. Iron, brain ageing and neurodegenerative disorders. Nat Rev Neurosci. 2004;5(11):863–73. Becerril-Ortega J, et al. Iron overload accelerates neuronal amyloid-β production and cognitive impairment in transgenic mice model of Alzheimer's disease. Neurobiol Aging. 2014;35(10):2288–301. Schweser F, et al. Quantitative imaging of intrinsic magnetic tissue properties using MRI signal phase: an approach to in vivo brain iron metabolism? NeuroImage. 2011;54(4):2789–807. Liu C, et al. Quantitative susceptibility mapping: contrast mechanisms and clinical applications. Tomography. 2015;1(1):3. Lee J, Ji S, Oh S-H. So you want to image myelin using MRI: magnetic susceptibility source separation for myelin imaging. Magn Reson Med Sci. 2024;23(3):291–306. Ahmed M, et al. The diamagnetic component map from quantitative susceptibility mapping (QSM) source separation reveals pathological alteration in Alzheimer’s disease-driven neurodegeneration. NeuroImage. 2023;280:120357. Uchida Y, et al. APOE ɛ4 dose associates with increased brain iron and β-amyloid via blood–brain barrier dysfunction. J Neurol Neurosurg Psychiatry. 2022;93(7):772–8. Uchida Y, et al. Contributions of blood–brain barrier imaging to neurovascular unit pathophysiology of Alzheimer’s disease and related dementias. Front Aging Neurosci. 2023;15:1111448. Jack CR Jr, et al. NIA-AA research framework: toward a biological definition of Alzheimer's disease. Alzheimer's Dement. 2018;14(4):535–62. Petersen RC, et al. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol. 1999;56(3):303–8. Fischl B. FreeSurfer Neuroimage. 2012;62(2):774–81. Liu X, et al. Cross-vendor test-retest validation of diffusion tensor image analysis along the perivascular space (DTI-ALPS) for evaluating glymphatic system function. Aging disease. 2024;15(4):1885. Gu Y, et al. Transient arousal modulations contribute to resting-state functional connectivity changes associated with head motion parameters. Cereb Cortex. 2020;30(10):5242–56. Li Z, et al. APART-QSM: An improved sub-voxel quantitative susceptibility mapping for susceptibility source separation using an iterative data fitting method. NeuroImage. 2023;274:120148. Klein A, et al. Evaluation of volume-based and surface-based brain image registration methods. NeuroImage. 2010;51(1):214–20. Taoka T et al. Diffusion tensor image analysis ALong the perivascular space (DTI-ALPS): revisiting the meaning and significance of the method. Magnetic Resonance in Medical Sciences, 2024: p. rev. 2023 – 0175. Daducci A, et al. Accelerated microstructure imaging via convex optimization (AMICO) from diffusion MRI data. NeuroImage. 2015;105:32–44. Eisma JJ, et al. Deep learning segmentation of the choroid plexus from structural magnetic resonance imaging (MRI): validation and normative ranges across the adult lifespan. Volume 21. Fluids and Barriers of the CNS; 2024. p. 21. 1. Lin S, et al. Association of MRI indexes of the perivascular space network and cognitive impairment in patients with obstructive sleep apnea. Radiology. 2024;311(3):e232274. Kim H-G, et al. Quantitative susceptibility mapping to evaluate the early stage of Alzheimer's disease. Volume 16. NeuroImage: Clinical; 2017. pp. 429–38. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc: Ser B (Methodol). 1995;57(1):289–300. Krishnan A, et al. Partial Least Squares (PLS) methods for neuroimaging: a tutorial and review. NeuroImage. 2011;56(2):455–75. Cha J. Partial least squares. Adv Methods Mark Res. 1994;407:52–78. Rosseel Y. lavaan: An R package for structural equation modeling. J Stat Softw. 2012;48(1):1–36. Donohue MC, et al. Estimating long-term multivariate progression from short‐term data. Alzheimer's Dement. 2014;10:S400–10. Trapnell C, et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat Biotechnol. 2014;32(4):381–6. Yao J, et al. Distinct regional vulnerability to Aβ and iron accumulation in post mortem AD brains. Alzheimer's Dement. 2024;20(10):6984–97. Ayton S, et al. Brain iron is associated with accelerated cognitive decline in people with Alzheimer pathology. Mol Psychiatry. 2020;25(11):2932–41. van der Weijden CW, et al. Quantitative myelin imaging with MRI and PET: an overview of techniques and their validation status. Brain. 2023;146(4):1243–66. Buckner RL, et al. Molecular, structural, and functional characterization of Alzheimer's disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci. 2005;25(34):7709–17. Duyn JH, Schenck J. Contributions to magnetic susceptibility of brain tissue. NMR Biomed. 2017;30(4):e3546. Hablitz LM, et al. Increased glymphatic influx is correlated with high EEG delta power and low heart rate in mice under anesthesia. Sci Adv. 2019;5(2):eaav5447. Taoka T, Naganawa S. Glymphatic imaging using MRI. J Magn Reson Imaging. 2020;51(1):11–24. Madden DJ, Merenstein JL. Quantitative susceptibility Mapp brain iron healthy aging cognition NeuroImage. 2023;282:120401. Bouter Y, et al. 18F-FDG-PET and Multimodal Biomarker Integration: a Powerful Tool for Alzheimer’s Disease Diagnosis. Nuclear Medicine and Molecular Imaging; 2025. pp. 1–19. Additional Declarations No competing interests reported. Supplementary Files SupplementaryData20260416.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 02 May, 2026 Reviewers agreed at journal 29 Apr, 2026 Reviewers invited by journal 29 Apr, 2026 Editor assigned by journal 17 Apr, 2026 Submission checks completed at journal 17 Apr, 2026 First submitted to journal 16 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {\"props\":{\"pageProps\":{\"initialData\":{\"identity\":\"rs-9436391\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":false,\"archivedVersions\":[],\"articleType\":\"Research Article\",\"associatedPublications\":[],\"authors\":[{\"id\":634588607,\"identity\":\"7f23a12f-804b-42b4-a6a6-8dab6148a86a\",\"order_by\":0,\"name\":\"Chantat Leong\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"University of Macau\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Chantat\",\"middleName\":\"\",\"lastName\":\"Leong\",\"suffix\":\"\"},{\"id\":634588608,\"identity\":\"35e2d22d-5368-43c6-9a5e-fb97682e2943\",\"order_by\":1,\"name\":\"Aocai Yang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Aocai\",\"middleName\":\"\",\"lastName\":\"Yang\",\"suffix\":\"\"},{\"id\":634588609,\"identity\":\"106a65ac-bc18-4f91-8427-8759de920214\",\"order_by\":2,\"name\":\"Ruisi Wang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"University of Macau\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Ruisi\",\"middleName\":\"\",\"lastName\":\"Wang\",\"suffix\":\"\"},{\"id\":634588610,\"identity\":\"65554cfb-6155-41b5-a025-95cdf8e2bbef\",\"order_by\":3,\"name\":\"Yu Sun\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yu\",\"middleName\":\"\",\"lastName\":\"Sun\",\"suffix\":\"\"},{\"id\":634588611,\"identity\":\"e65f81b4-0ca2-43f8-8962-5eebf1fc6ce0\",\"order_by\":4,\"name\":\"Jixin Luan\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Jixin\",\"middleName\":\"\",\"lastName\":\"Luan\",\"suffix\":\"\"},{\"id\":634588612,\"identity\":\"8cddfd1f-f136-4f13-8dcb-e83acf985144\",\"order_by\":5,\"name\":\"Manxi Xu\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Manxi\",\"middleName\":\"\",\"lastName\":\"Xu\",\"suffix\":\"\"},{\"id\":634588615,\"identity\":\"fccb293a-ee71-426b-9011-3b95c596b996\",\"order_by\":6,\"name\":\"Yian Gao\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yian\",\"middleName\":\"\",\"lastName\":\"Gao\",\"suffix\":\"\"},{\"id\":634588616,\"identity\":\"5aabd49e-5070-4c6d-9c2f-3b499072ce3f\",\"order_by\":7,\"name\":\"Hongwei Yu\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Hongwei\",\"middleName\":\"\",\"lastName\":\"Yu\",\"suffix\":\"\"},{\"id\":634588621,\"identity\":\"48a6e445-fc61-4011-a4ca-ae892ca99f44\",\"order_by\":8,\"name\":\"Yuli Wang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Yuli\",\"middleName\":\"\",\"lastName\":\"Wang\",\"suffix\":\"\"},{\"id\":634588622,\"identity\":\"159de911-ed7e-4069-acb9-ffb48edcd976\",\"order_by\":9,\"name\":\"Ni Shu\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Beijing Normal University\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Ni\",\"middleName\":\"\",\"lastName\":\"Shu\",\"suffix\":\"\"},{\"id\":634588623,\"identity\":\"f42f47a4-63e4-41a3-9b21-c1dbff494afd\",\"order_by\":10,\"name\":\"Bing Zhang\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Nanjing Drum Tower Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Bing\",\"middleName\":\"\",\"lastName\":\"Zhang\",\"suffix\":\"\"},{\"id\":634588625,\"identity\":\"fe0596d3-8da5-4f16-92f7-de0d21a61153\",\"order_by\":11,\"name\":\"Guolin Ma\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"China-Japan Friendship Hospital\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Guolin\",\"middleName\":\"\",\"lastName\":\"Ma\",\"suffix\":\"\"},{\"id\":634588627,\"identity\":\"b0851e77-f686-496f-b216-806fb1f15bfb\",\"order_by\":12,\"name\":\"Zhen Yuan\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzUlEQVRIiWNgGAWjYDACCSjNDybZSNEi2UCyFoMDxGqRn9187DFvm12e8Y3sBIYPZYcZdNsb8GthnHMs3Zi3LbnY7EbuBsYZ5w4zmJ05gF8Ls0SOmTRvG3PiNqAWZt42oJYbCfi1sEnkfwNqqU/cPAOo5S8xWngkctiAWg4nbpAAamEkRouERJqZ5JxzxxNnnHm74WDPuXQegn6Rn5H8TOJNWXVif3vuxgc/yqzlzI434NcCAkw8UAbIeB48ChGA8QdRykbBKBgFo2DEAgC3VEOKf1tpcgAAAABJRU5ErkJggg==\",\"orcid\":\"\",\"institution\":\"University of Macau\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Zhen\",\"middleName\":\"\",\"lastName\":\"Yuan\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2026-04-16 09:53:24\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-9436391/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-9436391/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":108956659,\"identity\":\"cc613844-06d5-450f-9d62-6294cc8940ab\",\"added_by\":\"auto\",\"created_at\":\"2026-05-11 08:14:48\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":73949135,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eSchematic of conceptual model illustrating how glymphatic dysfunction may contribute to iron accumulation and macromolecular tissue composition through impaired CSF-ISF exchange along perivascular pathways in Alzheimer’s disease. This figure was generated with the assistance of AI-based image generation tools and subsequently edited by the authors for accuracy and clarity.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Figure1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-9436391/v1/634d5a2f79f3018785a4c00b.png\"},{\"id\":108956682,\"identity\":\"9fc95dd2-12cf-43f7-a978-28bf76be1d26\",\"added_by\":\"auto\",\"created_at\":\"2026-05-11 08:14:59\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":37468505,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cstrong\\u003eOverview of study design and multimodal analysis framework. \\u003c/strong\\u003eA) Study cohort and multimodal MRI data acquisitions. Participants spanning the HC, SCD, MCI and, AD underwent multimodal MRI acquisition. B) Complementary glymphatic indices and sub-voxel susceptibility measures, were extracted at both ROI and network levels. MRI preprocessing and quantitative mapping steps are summarized for glymphatic function indices and for susceptibility imaging. B1) Glymphatic indices were derived from diffusion- and CSF-related measures, including, BOLD-CSF coupling, DTI-ALPS, FW, CP, CSF, and BF. B2) Susceptibility features were obtained via QSM reconstruction, enabling joint characterization of fluid-transport dynamics and microstructural tissue composition within the same anatomical framework. C) Cross-modal coupling analysis. Univariate correlation and multivariate PLS analyses were used to characterize glymphatic-susceptibility coupling patterns. Complementary criteria, including disease-stage trends, PLS contributions, and group-level separation, were integrated to identify robust cross-modal patterns. D) Derivation of latent neurodegeneration (ND) patterns and disease modeling. Identified cross-modal features were integrated to construct latent ND indices, which were further examined to characterize mediation effects and temporal trajectories along the AD continuum.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Figure2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-9436391/v1/8e4947d0b1296c22dc8f9525.png\"},{\"id\":108956746,\"identity\":\"cd6343f1-6727-4103-9888-17e084d268f6\",\"added_by\":\"auto\",\"created_at\":\"2026-05-11 08:15:09\",\"extension\":\"png\",\"order_by\":3,\"title\":\"Figure 3\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":45774848,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cstrong\\u003eCross-modal associations between glymphatic function and magnetic susceptibility. \\u003c/strong\\u003e(A) Univariate partial correlations between glymphatic indices (BOLD-CSF coupling, FW, CSF/CP/PVS volumes, DTI-ALPS, BF volume) and susceptibility metrics at network and ROI levels, shown separately for diamagnetic (|χdia|) and paramagnetic (χpara) components. Surface maps illustrate the spatial distribution of regional correlation patterns, and bar plots summarize the strongest cross-modal associations. Asterisks indicate FDR-corrected significance (\\u003cem\\u003ep\\u003c/em\\u003e \\u0026lt; 0.05). (B) Multivariate PLS results for the dominant latent variable (LV1) at both network and ROI levels for |χdia| and χpara. Surface maps show the spatial distribution of LV1 loadings, and bar plots indicate the strongest glymphatic and susceptibility contributors. Asterisks mark features with significant bootstrap ratios (|BR| ≥ 2.8).\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Figure3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-9436391/v1/cb2206e336e685bcb68d2237.png\"},{\"id\":108956662,\"identity\":\"135b13a7-dcbc-46fd-b666-63b429dbff58\",\"added_by\":\"auto\",\"created_at\":\"2026-05-11 08:14:49\",\"extension\":\"png\",\"order_by\":4,\"title\":\"Figure 4\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":17448236,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cstrong\\u003eSpatial patterns of glymphatic-susceptibility associations identified across the AD continuum.\\u003c/strong\\u003e (A) Consensus heatmaps summarizing glymphatic–susceptibility pairs that were consistently identified across complementary criteria, including trend analysis, PLS contribution, and group-level separation, with a consensus mask applied for visualization to highlight pairs jointly selected across methods. Colors represent consensus importance derived from the combined ranking of the three methods. Stars denote glymphatic-susceptibility pairs that were significant across all three criteria, highlighting robust and reproducible cross-modal associations. (B) Representative scatter plots illustrating individual-level associations between selected glymphatic indices and susceptibility measures across diagnostic groups (HC, SCD, MCI, AD). Regression lines indicate overall linear trends.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Figure4.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-9436391/v1/c3f9d687b85124509c8c8f8f.png\"},{\"id\":108956660,\"identity\":\"c847713b-c26d-4004-857a-1330693bcbed\",\"added_by\":\"auto\",\"created_at\":\"2026-05-11 08:14:48\",\"extension\":\"png\",\"order_by\":5,\"title\":\"Figure 5\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":50675006,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003e\\u003cstrong\\u003eStructural equation modeling and disease progression modeling of multimodal ND patterns. \\u003c/strong\\u003e(A)\\u003cstrong\\u003e \\u003c/strong\\u003eSEM testing whether multimodal neurodegeneration (ND) patterns mediate the association between diagnostic group and cognition. Four SEMs were estimated using ND constructs derived from network-level |χ\\u003csub\\u003edia\\u003c/sub\\u003e|, network-level χ\\u003csub\\u003epara\\u003c/sub\\u003e, ROI-level |χ\\u003csub\\u003edia\\u003c/sub\\u003e| and χ\\u003csub\\u003epara\\u003c/sub\\u003e. Rectangles denote observed variables (diagnostic group and glymphatic–susceptibility indicators), and the central ellipse represents the latent ND factor, which is linked to MMSE. Numbers indicate standardized path coefficients. (B) D DPM-derived pseudotime trajectories of ND indices across diagnostic groups. Left panels show individual pseudotime distributions for ROI- and network-level ND measures derived from ROI_DIA, ROI_PARA, NET_DIA, and NET_PARA, stratified by diagnostic group (HC, SCD, MCI, AD). Right panels depict DPM-derived ND trajectories resampled along the normalized pseudotime axis (0–1). For ROI-level and network-level ND indices derived from |χ\\u003csub\\u003edia\\u003c/sub\\u003e| and χ\\u003csub\\u003epara\\u003c/sub\\u003e, curves represent normalized trajectories clustered based on temporal similarity, with colors indicating trajectory clusters and vertical markers denoting estimated onset and peak-change times; clusters include trajectories showing increasing and decreasing ND progression across pseudotime.\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"Figure5.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-9436391/v1/0dfeeed7c795b0ab7193fc25.png\"},{\"id\":108956781,\"identity\":\"b0b46158-a6fe-4a42-bf80-ac7ead44b284\",\"added_by\":\"auto\",\"created_at\":\"2026-05-11 08:15:11\",\"extension\":\"docx\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":479367,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"SupplementaryData20260416.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-9436391/v1/9cf035f093bcb9acff1f7cb2.docx\"}],\"financialInterests\":\"No competing interests reported.\",\"formattedTitle\":\"Glymphatic Dysfunction Coupled with Aberrant Distribution of Iron and Pathological proteins across Alzheimer’s Disease Continuum\",\"fulltext\":[{\"header\":\"1. Background\",\"content\":\"\\u003cp\\u003eAlzheimer\\u0026rsquo;s disease (AD) is characterized by progressive neurodegeneration arising from impaired clearance of metabolic waste, including amyloid-β (Aβ), tau, and other neurotoxic molecules. Inefficient removal of these proteins exacerbates their aggregation, amplifies oxidative and metabolic stress, and disrupts neuronal function, ultimately accelerating neurodegeneration. Converging evidence suggests that failure of brain fluid clearance systems, encompassing neurovascular coupling, blood-brain barrier (BBB) permeability, and glymphatic transport, plays a central mechanistic role in this process [\\u003cspan additionalcitationids=\\\"CR2 CR3\\\" citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e1\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR4\\\" class=\\\"CitationRef\\\"\\u003e4\\u003c/span\\u003e]. Dysregulated cerebrospinal fluid (CSF) transport is increasingly recognized as a critical component of impaired clearance in AD, and its disruption is expected not only to compromise solute removal but also to promote the accumulation of neurotoxic substances.\\u003c/p\\u003e \\u003cp\\u003eThe glymphatic system facilitates CSF-interstitial fluid exchange along perivascular pathways regulated by astrocytic aquaporin-4 (AQP4) and represents a key mechanism for metabolic waste clearance in the brain. AQP4 mislocalization, reduced CSF influx, and impaired perivascular solute transport are observed, leading to diminished clearance of Aβ and tau and promoting toxic protein accumulation [\\u003cspan additionalcitationids=\\\"CR6\\\" citationid=\\\"CR5\\\" class=\\\"CitationRef\\\"\\u003e5\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e]. Previous studies further show that AQP4 depolarization is associated with both amyloid burden and cognitive impairment, while genetic variations affecting AQP4 expression modulate AD risk and cognitive trajectories [\\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e8\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR9\\\" class=\\\"CitationRef\\\"\\u003e9\\u003c/span\\u003e]. Advances in magnetic resonance imaging (MRI) enable \\u003cem\\u003ein vivo\\u003c/em\\u003e quantification of multiple, complementary aspects of glymphatic and perivascular function. These associated MRI indicators include BOLD-CSF coupling (a proxy for neurovascular-driven CSF inflow), the DTI-ALPS index (for perivascular diffusivity), free-water (FW) content (for extracellular fluid mobility), and structural markers such as perivascular space (PVS) burden and choroid plexus (CP) morphology [\\u003cspan additionalcitationids=\\\"CR11 CR12 CR13 CR14 CR15\\\" citationid=\\\"CR10\\\" class=\\\"CitationRef\\\"\\u003e10\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e16\\u003c/span\\u003e]. These metrics have consistently demonstrated glymphatic abnormalities across the AD spectrum, correlating with pathological protein burden and cognitive impairment. However, glymphatic indices are typically examined in isolation rather than within an integrated analytical framework. It remains unclear whether they reflect coordinated components of a unified fluid-transport system or converge onto shared vulnerability patterns in neurodegeneration.\\u003c/p\\u003e \\u003cp\\u003eWhile glymphatic dysfunction is commonly considered a failure of Aβ and tau clearance, impaired CSF-interstitial fluid exchange is expected to influence the broader microstructural environment of brain tissue. Beyond the accumulation of soluble proteins, reduced clearance efficiency may promote iron retention and the accumulation of neurotoxic macromolecules, contributing to microstructural alterations in vulnerable regions [\\u003cspan additionalcitationids=\\\"CR18\\\" citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e17\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e19\\u003c/span\\u003e]. Consistent with this view, iron homeostasis is profoundly altered in AD. Aging-related shifts in iron regulation, together with microglial activation and lysosomal dysfunction, promote iron retention in vulnerable cortical and subcortical regions. Excess iron catalyzes reactive oxygen species production, accelerates Aβ aggregation and tau phosphorylation, and contributes to mitochondrial damage and cellular death [\\u003cspan additionalcitationids=\\\"CR21 CR22\\\" citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e20\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR23\\\" class=\\\"CitationRef\\\"\\u003e23\\u003c/span\\u003e]. Previous studies also found that iron accumulation co-localizes with Aβ and tau aggregation, suggesting that altered tissue composition may emerge as a downstream consequence of impaired clearance-related processes [\\u003cspan citationid=\\\"CR24\\\" class=\\\"CitationRef\\\"\\u003e24\\u003c/span\\u003e].\\u003c/p\\u003e \\u003cp\\u003eQuantitative susceptibility mapping (QSM) provides a noninvasive approach to characterize microstructural tissue composition by capturing magnetic susceptibility signals sensitive to iron, myelin, and other macromolecular constituents. Conventional QSM reflects a composite voxel-wise signal in which paramagnetic and diamagnetic contributions may partially cancel each other, limiting biological specificity. Decomposition of QSM into paramagnetic (χ\\u003csub\\u003epara\\u003c/sub\\u003e) and diamagnetic (χ\\u003csub\\u003edia\\u003c/sub\\u003e) components enables more specific separation of iron-dominant susceptibility from diamagnetic contributions related to neurotoxic and pathological proteins [\\u003cspan additionalcitationids=\\\"CR26 CR27\\\" citationid=\\\"CR25\\\" class=\\\"CitationRef\\\"\\u003e25\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR28\\\" class=\\\"CitationRef\\\"\\u003e28\\u003c/span\\u003e]. QSM studies consistently demonstrate regionally altered iron-related susceptibility across the AD spectrum, and vascular dysfunction, blood-brain barrier breakdown, and impaired interstitial drainage have been implicated in abnormal iron accumulation [\\u003cspan citationid=\\\"CR29\\\" class=\\\"CitationRef\\\"\\u003e29\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR30\\\" class=\\\"CitationRef\\\"\\u003e30\\u003c/span\\u003e]. However, it remains unclear how these susceptibility alterations relate to clearance-related processes reflected by glymphatic function. Consequently, no integrative framework has yet unified glymphatic indices with sub-voxel QSM signals to determine whether these modalities share a coherent system-level organization, reflect coordinated processes, or exhibit structured deviations across neurodegenerative progression.\\u003c/p\\u003e \\u003cp\\u003eDespite growing evidence for both glymphatic dysfunction and susceptibility alterations in AD, it remains unknown whether these represent independent pathological abnormalities or coordinated manifestations of a unified clearance-tissue composition architecture, encompassing iron and pathological protein alterations. Addressing this question requires an integrative framework capable of jointly characterizing fluid-transport dynamics and microstructural tissue properties across the whole brain. In this study, we introduced a multimodal MRI framework that integrates complementary glymphatic indices with paramagnetic and diamagnetic susceptibility components to examine whether glymphatic dysfunction and susceptibility alterations converge onto a shared physiological axis. By characterizing cross-regional associations between glymphatic indices and susceptibility-derived measures, this framework enables the capture of integrated clearance-tissue composition features across the brain. Such patterns may help delineate how disruption of the clearance-tissue composition axis unfolds along the AD continuum, providing insight into how physiological clearance processes interact with and shape the brain's microstructural environment (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig1\\\" class=\\\"InternalRef\\\"\\u003e1\\u003c/span\\u003e). Understanding these interactions is critical for clarifying how physiological clearance dysfunction contributes to the emergence of microstructural vulnerability in AD, thereby bridging molecular pathology with large-scale brain alterations.\\u003c/p\\u003e \\u003cp\\u003e \\u003c/p\\u003e\"},{\"header\":\"2. Methods\",\"content\":\"\\u003cp\\u003eTo examine the system-level relationships between glymphatic function and microstructural susceptibility, a multimodal MRI framework was used to integrate data acquisition, standardized preprocessing, glymphatic and susceptibility quantification, and cross-modal analytical strategies. A schematic overview of the framework is shown in Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003e.\\u003c/p\\u003e \\u003cp\\u003e \\u003c/p\\u003e \\u003cdiv id=\\\"Sec3\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003e2.1 Participants\\u003c/h2\\u003e \\u003cp\\u003e This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and approved by the Ethics Committee of the China-Japan Friendship Hospital and Nanjing Drum Tower Hospital. Informed consent was obtained from all participants or their relatives who were fully informed about the study.\\u003c/p\\u003e \\u003cp\\u003e For the multi-cohort and multimodal MRI study, a total of 838 participants (HC\\u0026thinsp;=\\u0026thinsp;256, SCD\\u0026thinsp;=\\u0026thinsp;154, MCI\\u0026thinsp;=\\u0026thinsp;250, AD\\u0026thinsp;=\\u0026thinsp;178) were recruited from the China-Japan Friendship Hospital and Nanjing Drum Tower Hospital. All participants underwent standardized neurological examinations, neuropsychological testing, and a comprehensive MRI protocol. Diagnoses followed the NIA\\u0026ndash;AA [\\u003cspan citationid=\\\"CR31\\\" class=\\\"CitationRef\\\"\\u003e31\\u003c/span\\u003e], and Petersen criteria [\\u003cspan citationid=\\\"CR32\\\" class=\\\"CitationRef\\\"\\u003e32\\u003c/span\\u003e], and participants were included if they were 50\\u0026ndash;85 years old, right-handed, and free of major neurological or psychiatric disorders (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003eA). Exclusion criteria comprised significant cerebrovascular lesions, substance misuse, MRI contraindications, or excessive head motion. Additional demographic information is provided in Supplementary Methods S1.\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec4\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003e2.2 Image acquisition\\u003c/h2\\u003e \\u003cp\\u003eMRI data were acquired on two 3T MR platforms (GE MR750 with an 8-channel head coil at the China-Japan Friendship Hospital; Philips Ingenia CX with a 32-channel head coil at Nanjing Drum Tower Hospital) using site-matched multimodal protocols. 3D T1-weighted FSPGR (1 mm isotropic), 3D multi-echo GRE for QSM (1 mm isotropic), GRE-SS-EPI for resting-state fMRI, and diffusion-weighted spin-echo echo-planar (EPI) sequence for diffusion MRI. Full acquisition parameters for each modality are provided in Supplementary Methods S2.\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec5\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003e2.3 MRI preprocessing\\u003c/h2\\u003e \\u003cp\\u003eAll MRI data underwent standardized preprocessing pipelines (details in Supplementary Methods S3) to enable reliable extraction of glymphatic-related and susceptibility-based physiological metrics within a common anatomical framework. Structural T1-weighted images were corrected for bias field inhomogeneity, segmented using FreeSurfer, and registered to MNI152 space via boundary-based coregistration [\\u003cspan citationid=\\\"CR33\\\" class=\\\"CitationRef\\\"\\u003e33\\u003c/span\\u003e], providing anatomical features of glymphatic function. Diffusion data were denoised, distortion- and motion-corrected, modeled using DTI and bi-tensor free-water estimation, and coregistered to the T1 image [\\u003cspan citationid=\\\"CR34\\\" class=\\\"CitationRef\\\"\\u003e34\\u003c/span\\u003e]. These procedures enabled derivation of perivascular space (ALPS) and free water (FW) indices reflecting perivascular diffusivity and interstitial fluid mobility, which serve as proxies for glymphatic and extracellular transport function. Resting-state fMRI preprocessing included motion correction, nuisance regression, band-pass filtering, and spatial normalization [\\u003cspan citationid=\\\"CR35\\\" class=\\\"CitationRef\\\"\\u003e35\\u003c/span\\u003e], allowing extraction of BOLD\\u0026ndash;CSF coupling measures that capture neurovascular-driven CSF inflow dynamics.\\u003c/p\\u003e \\u003cp\\u003eQSM was reconstructed using Laplacian unwrapping, LBV background removal and STAR-QSM inversion, and paramagnetic/diamagnetic susceptibility maps were computed through APART-QSM algorithms [\\u003cspan citationid=\\\"CR36\\\" class=\\\"CitationRef\\\"\\u003e36\\u003c/span\\u003e]. These steps enabled separation of χ\\u003csub\\u003epara\\u003c/sub\\u003e and χ\\u003csub\\u003edia\\u003c/sub\\u003e components reflecting iron-related and macromolecular susceptibility sources. All susceptibility maps were subsequently normalized to MNI space via ANTs [\\u003cspan citationid=\\\"CR37\\\" class=\\\"CitationRef\\\"\\u003e37\\u003c/span\\u003e] to ensure spatial correspondence with glymphatic indices across subjects.\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec6\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003e2.4 Computation of glymphatic- and iron-related measurements\\u003c/h2\\u003e \\u003cp\\u003eTo jointly characterize clearance dynamics and microstructural tissue composition, we derived complementary glymphatic-related and susceptibility-based MRI metrics that capture distinct yet physiologically linked aspects of brain fluid transport and tissue microenvironment.\\u003c/p\\u003e \\u003cp\\u003eGlymphatic-related indices were categorized into two major classes: coupling-related indices and perivascular indices. Coupling-related indices were designed to capture neurovascular-driven CSF inflow dynamics. Specifically, BOLD-CSF coupling was derived from resting-state fMRI by quantifying the temporal relationship between ventricular CSF signals and gray matter BOLD fluctuations, with the characteristic negative lag reflecting CSF inflow following neural and vascular activity [\\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e11\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e13\\u003c/span\\u003e]. This measure serves as an index of neurovascularly mediated glymphatic function.\\u003c/p\\u003e \\u003cp\\u003ePerivascular indices were used to characterize solute transport and fluid mobility along perivascular and interstitial compartments. The DTI-ALPS index quantifies water diffusivity along perivascular directions relative to fiber orientations, providing a proxy for perivascular diffusivity and solute transport efficiency [\\u003cspan citationid=\\\"CR34\\\" class=\\\"CitationRef\\\"\\u003e34\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR38\\\" class=\\\"CitationRef\\\"\\u003e38\\u003c/span\\u003e]. FW content was also estimated using DTI to isolate the extracellular water fraction, reflecting interstitial fluid mobility independent of tissue anisotropy [\\u003cspan citationid=\\\"CR39\\\" class=\\\"CitationRef\\\"\\u003e39\\u003c/span\\u003e]. In addition, volumetric measures of perivascular space (PVS), choroid plexus (CP), cerebrospinal fluid (CSF), and basal forebrain (BF) were extracted from structural MRI (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003eB). These measures represent structural correlates of CSF compartments and perivascular pathways that support glymphatic transport [\\u003cspan citationid=\\\"CR8\\\" class=\\\"CitationRef\\\"\\u003e8\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR40\\\" class=\\\"CitationRef\\\"\\u003e40\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR41\\\" class=\\\"CitationRef\\\"\\u003e41\\u003c/span\\u003e].\\u003c/p\\u003e \\u003cp\\u003eIron-related and macromolecular tissue composition alterations were characterized using sub-voxel quantitative susceptibility mapping (QSM) metrics, namely χ\\u003csub\\u003edia\\u003c/sub\\u003e and χ\\u003csub\\u003epara\\u003c/sub\\u003e. These metrics reflect paramagnetic susceptibility primarily associated with iron-related sources and diamagnetic susceptibility associated with pathological proteins and broader macromolecular tissue composition. To distinguish sources of susceptibility contrast, conventional QSM maps were decomposed into χ\\u003csub\\u003edia\\u003c/sub\\u003e and χ\\u003csub\\u003epara\\u003c/sub\\u003e components using the APART-QSM framework [\\u003cspan citationid=\\\"CR36\\\" class=\\\"CitationRef\\\"\\u003e36\\u003c/span\\u003e], which separates paramagnetic contributions primarily associated with iron-related sources from diamagnetic contributions associated with pathological proteins and broader macromolecular tissue composition. Absolute diamagnetic susceptibility (|χ\\u003csub\\u003edia\\u003c/sub\\u003e|) was used to avoid ambiguity in signal interpretation. Higher χ\\u003csub\\u003epara\\u003c/sub\\u003e and |χ\\u003csub\\u003edia\\u003c/sub\\u003e| values indicate greater paramagnetic and diamagnetic tissue content, respectively (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003eB).\\u003c/p\\u003e \\u003cp\\u003eTo examine spatially specific susceptibility alterations relevant to AD pathology, region of interest (ROI) -level analyses were conducted across cortical and subcortical regions known to be early and frequently affected by amyloid and tau deposition. Twenty-eight ROIs were selected based on prior pathological evidence [\\u003cspan citationid=\\\"CR28\\\" class=\\\"CitationRef\\\"\\u003e28\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR42\\\" class=\\\"CitationRef\\\"\\u003e42\\u003c/span\\u003e]. Regional χ\\u003csub\\u003epara\\u003c/sub\\u003e and |χ\\u003csub\\u003edia\\u003c/sub\\u003e| values were computed by averaging voxel-wise values within each region after masking to exclude cerebrospinal fluid and non-brain voxels. Full parameters, regional definitions, and metric-specific computational procedures are provided in Supplementary Methods S3.\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec7\\\" class=\\\"Section2\\\"\\u003e \\u003ch2\\u003e2.5 Statistical Analysis\\u003c/h2\\u003e \\u003cp\\u003eTo characterize glymphatic-susceptibility relationships across the AD continuum from complementary perspectives, we employed a series of univariate, multivariate, and modeling approaches that together capture group differences, cross-modal associations, disease-relevant coupling patterns, and their cognitive and temporal implications.\\u003c/p\\u003e \\u003cdiv id=\\\"Sec8\\\" class=\\\"Section3\\\"\\u003e \\u003ch2\\u003e2.5.1 Group-Level comparisons of imaging metrics\\u003c/h2\\u003e \\u003cp\\u003eGroup differences in each glymphatic and susceptibility index were tested using ANCOVA, controlling for age, sex, and scanner. Post-hoc pairwise comparisons were FDR-corrected (\\u003cem\\u003eq\\u003c/em\\u003e\\u0026thinsp;\\u0026lt;\\u0026thinsp;0.05) [\\u003cspan citationid=\\\"CR43\\\" class=\\\"CitationRef\\\"\\u003e43\\u003c/span\\u003e]. Cohen\\u0026rsquo;s \\u003cem\\u003ed\\u003c/em\\u003e was computed to quantify effect sizes.\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec9\\\" class=\\\"Section3\\\"\\u003e \\u003ch2\\u003e2.5.2 Glymphatic-Susceptibility association analysis\\u003c/h2\\u003e \\u003cp\\u003eCross-modal associations between glymphatic indices and susceptibility measures were examined using complementary univariate and multivariate approaches.\\u003c/p\\u003e \\u003cp\\u003eFirst, partial correlation analyses were performed between each glymphatic metric and each susceptibility measure (χ\\u003csub\\u003epara\\u003c/sub\\u003e and |χ\\u003csub\\u003edia\\u003c/sub\\u003e|). Correlation matrices were computed separately for ROI-level (28 ROIs), which were selected based on previous studies identifying regions showing the most pronounced QSM abnormalities contributing to AD pathology [\\u003cspan citationid=\\\"CR28\\\" class=\\\"CitationRef\\\"\\u003e28\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR42\\\" class=\\\"CitationRef\\\"\\u003e42\\u003c/span\\u003e], and network-level representations, including the Frontoparietal (FPN), Dorsal Attention (DAN), Ventral Attention (VAN), Limbic (LIM), Visual (VIS), Somatomotor (SOM), and Default Mode (DMN) networks. Statistical significance was assessed using two-tailed tests with FDR correction applied across all glymphatic-susceptibility pairs.\\u003c/p\\u003e \\u003cp\\u003eSecond, to identify coordinated cross-modal patterns not captured by pairwise correlations, partial least squares (PLS) analysis was conducted [\\u003cspan citationid=\\\"CR44\\\" class=\\\"CitationRef\\\"\\u003e44\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR45\\\" class=\\\"CitationRef\\\"\\u003e45\\u003c/span\\u003e]. PLS was used to extract latent variables representing maximal shared variance between multiple glymphatic indices and susceptibility measures, capturing coordinated multivariate associations related to system-level clearance regulation (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003eC). Statistical significance of each latent component was assessed using permutation testing (3,000 permutations), and the stability of feature contributions was evaluated using bootstrap resampling (3,000 resamples). To ensure that multivariate associations were not driven by redundancy among glymphatic metrics, pairwise correlations among all glymphatic indices were examined, confirming the absence of near-collinearity (|\\u003cem\\u003er\\u003c/em\\u003e| \\u0026gt; 0.80; Supplementary Fig. \\u003cspan refid=\\\"MOESM1\\\" class=\\\"InternalRef\\\"\\u003eS1\\u003c/span\\u003e).\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec10\\\" class=\\\"Section3\\\"\\u003e \\u003ch2\\u003e2.5.3 Multistage identification of AD-related coupled patterns\\u003c/h2\\u003e \\u003cp\\u003eTo identify clearance-tissue composition patterns across the AD continuum, we integrated complementary statistical signatures, including system-level trend analysis, multivariate contributions, and group separation. First, a system-level trend analysis was performed to identify ROIs showing monotonic changes across the AD continuum using linear contrasts (\\u003cem\\u003eT\\u003c/em\\u003e\\u0026thinsp;\\u0026ge;\\u0026thinsp;4, \\u003cem\\u003ep\\u003c/em\\u003e\\u0026thinsp;\\u0026lt;\\u0026thinsp;0.001). Second, ROIs contributing significantly to disease-relevant PLS components were identified based on bootstrap ratios (|BR| \\u0026ge; 3), reflecting stable multivariate contributions. Third, for each candidate glymphatic\\u0026ndash;susceptibility pattern, Euclidean distances between group centroids were computed in the multivariate feature space to quantify the degree of separation across AD continuum. Glymphatic-susceptibility pairs satisfying all three criteria, including monotonic disease trend, significant multivariate contribution, and group separation, were classified as AD-related and retained for downstream analyses.\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec11\\\" class=\\\"Section3\\\"\\u003e \\u003ch2\\u003e2.5.4 Structural equation modelling (SEM)\\u003c/h2\\u003e \\u003cp\\u003eStructural equation modelling (SEM) was employed to test whether multimodal glymphatic-susceptibility patterns mediated the relationship between disease status and cognitive performance [\\u003cspan citationid=\\\"CR45\\\" class=\\\"CitationRef\\\"\\u003e45\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR46\\\" class=\\\"CitationRef\\\"\\u003e46\\u003c/span\\u003e]. Neurodegeneration latent factors (ND) were constructed from the selected coupled patterns, with highly collinear indicators removed to improve model stability. Models were estimated using robust maximum likelihood estimation, and standardized path coefficients and model-fit indices were extracted (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003eD).\\u003c/p\\u003e \\u003c/div\\u003e \\u003cdiv id=\\\"Sec12\\\" class=\\\"Section3\\\"\\u003e \\u003ch2\\u003e2.5.5 Pseudotime-based disease progression modelling (DPM)\\u003c/h2\\u003e \\u003cp\\u003eTo characterize the temporal organization of glymphatic-susceptibility alterations along disease progression, a pseudotime-based disease progression modeling (DPM) framework was applied [\\u003cspan citationid=\\\"CR47\\\" class=\\\"CitationRef\\\"\\u003e47\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR48\\\" class=\\\"CitationRef\\\"\\u003e48\\u003c/span\\u003e]. Subject-level pseudotime scores were derived from the principal component structure of the ND feature matrix, rescaled to the [0, 1] interval, and oriented toward increasing disease severity. Pattern-wise trajectories were estimated using spline-smoothed resampling to capture continuous changes along pseudotime. Correlation-based clustering was used to group ND patterns into temporal classes, enabling characterization of heterogeneous progression profiles (Fig.\\u0026nbsp;\\u003cspan refid=\\\"Fig2\\\" class=\\\"InternalRef\\\"\\u003e2\\u003c/span\\u003eD).\\u003c/p\\u003e \\u003cp\\u003eFull analytical details are provided in Supplementary Methods S4.\\u003c/p\\u003e \\u003c/div\\u003e \\u003c/div\\u003e\"},{\"header\":\"3. Results\",\"content\":\"\\u003cdiv id=\\\"Sec14\\\" class=\\\"Section2\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e3.1 Clinical characteristics and imaging features of\\u0026nbsp;\\u003c/strong\\u003e\\u003cstrong\\u003eglymphatic\\u003c/strong\\u003e\\u003cstrong\\u003e\\u0026nbsp;and susceptibility metrics\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003cp\\u003eDemographic variables (age, sex) and global cognition differed across groups along the AD continuum (Table 1). At the univariate imaging level, several glymphatic indices, including CSF volume, CP volume, FW, PVS, DTI-ALPS, BOLD-CSF coupling, and BF, showed progressive alterations across AD spectrum, consistent with reduced perivascular transport efficiency (Supplementary Table S1).\\u003c/p\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eTable 1. Demographic and clinical characteristics of AD, MCI, SCD and HC groups\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003ctable style=\\\"width: 100%;\\\"\\u003e\\n \\u003ctbody\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eCharacteristics\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eAD\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eMCI\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eSCD\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eHC\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e\\u003cem\\u003eF\\u003c/em\\u003e\\u003c/strong\\u003e\\u003cstrong\\u003e-value/ \\u0026chi;\\u003csup\\u003e2\\u003c/sup\\u003e\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e\\u003cem\\u003ep\\u003c/em\\u003e\\u003c/strong\\u003e\\u003cstrong\\u003e-value\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eeffect size\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eAge\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e71.83\\u0026plusmn;90.22\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;178)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e66.74\\u0026plusmn;8.02\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;250)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e65.69\\u0026plusmn;7.92\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;154)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e64.21\\u0026plusmn;8.71\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;256)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e30.28\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.0001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.098\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eSex(M/F)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e(51/95)\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;178)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e(64/172)\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;250)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e(28/106)\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;154)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e(80/170)\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;256)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e8.45\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.05\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.105\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eEducation\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003e(years)\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e10.53\\u0026plusmn;4.29\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;178)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e10.73\\u0026plusmn;3.42\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;250)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e12.29\\u0026plusmn;3.63\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;154)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e11.23\\u0026plusmn;4.47\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;254)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e6.54\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.05\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.023\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eMMSE\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e19.01\\u0026plusmn;5.33\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;178)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e27.18\\u0026plusmn;2.41\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;250)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e28.18\\u0026plusmn;1.79\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;154)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e28.33\\u0026plusmn;1.72\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;256)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e380.05\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.59\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eMoCA\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e12.06\\u0026plusmn;5.8\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;67)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e21.37\\u0026plusmn;3.97\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;83)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e24.2\\u0026plusmn;2.86\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;102)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e25.09\\u0026plusmn;2.93\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;89)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e166.51\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.6\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eHAMA\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e9.76\\u0026plusmn;7.71\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;55)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e5.33\\u0026plusmn;5.85\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;81)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e6.5\\u0026plusmn;5.68\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;101)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e3.5\\u0026plusmn;6.37\\u003c/p\\u003e\\n \\u003cp\\u003e(N = 66)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e11.31\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eHAMD\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e5.11\\u0026plusmn;4.55\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;55)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e3.89\\u0026plusmn;4.76\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;81)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e5.64\\u0026plusmn;5.15\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;101)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e2.68\\u0026plusmn;3.28\\u003c/p\\u003e\\n \\u003cp\\u003e(N = 66)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e6.3\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.05\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.059\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eRAVLT\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e6.28\\u0026plusmn;12.13\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;71)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e8.99\\u0026plusmn;6.6\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;81)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e12.13\\u0026plusmn;7.16\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;101)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e15.88\\u0026plusmn;5.57\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;66)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e18.03\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.15\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eCDT\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e13.62\\u0026plusmn;10.29\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;55)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e23.32\\u0026plusmn;6.51\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;81)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e25.82\\u0026plusmn;4.51\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;101)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e27.5\\u0026plusmn;2.3\\u003c/p\\u003e\\n \\u003cp\\u003e(N = 66)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e60.43\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.38\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eBNT\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e16.04\\u0026plusmn;6.16\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;71)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e23.81\\u0026plusmn;4.38\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;81)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e26.16\\u0026plusmn;3.47\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;101)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e27.29\\u0026plusmn;2.47\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;66)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e101.83\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.49\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eVFT\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e21.91\\u0026plusmn;8.48\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;53)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e29.23\\u0026plusmn;6.31\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;30)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e40.85\\u0026plusmn;9.35\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;39)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e48.33\\u0026plusmn;7.91\\u003c/p\\u003e\\n \\u003cp\\u003e(N =\\u0026nbsp;12)\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e57.57\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e\\u0026lt;0.001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd nowrap=\\\"\\\"\\u003e\\n \\u003cp\\u003e0.57\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/tbody\\u003e\\n \\u003c/table\\u003e\\n\\u003c/div\\u003e\\n\\u003cp\\u003eNote: Data is presented as mean \\u0026plusmn; standard deviations (SD). MMSE:\\u0026nbsp;Mini-Mental State Examination, MoCA:\\u0026nbsp;Montreal Cognitive Assessment\\u0026nbsp;test, HAMD:\\u0026nbsp;Hamilton depression rating, HAMA:\\u0026nbsp;Hamilton anxiety rating,\\u0026nbsp;RAVLT:\\u0026nbsp;Rey Auditory Verbal Learning Test, CDT:\\u0026nbsp;Clock Drawing Test, BNT:\\u0026nbsp;Boston Naming Test,\\u0026nbsp;VFT:\\u0026nbsp;Verbal Fluency Test.\\u003c/p\\u003e\\n\\u003cp\\u003eThe \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e and |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u0026nbsp;also showed widespread susceptibility alterations across temporal, parietal, and limbic cortices, with AD exhibiting lower diamagnetic and paramagnetic signals relative to SCD and MCI, alongside selective increases in the posterior superior temporal sulcus (Psts) and the dentate nucleus (DN). Network-level analyses further revealed reduced susceptibility signatures within limbic and default mode systems (Supplementary Table S2).\\u003c/p\\u003e\\n\\u003cp\\u003eSimilar patterns of progressive alterations were observed across multiple glymphatic indices and susceptibility measures along the AD continuum.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.2 Cross-modal glymphatic-susceptibility associations\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eUnivariate cross-modal analyses revealed structured, non-random coupling between glymphatic metrics and susceptibility-derived measures across multiple brain regions (Fig. 3A). gBOLD-CSF coupling and FW showed consistent negative correlations with both\\u0026nbsp;\\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e and |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|, whereas PVS, BF and DTI-ALPS exhibited positive correlations. These associations were most prominent in frontal, limbic, and subcortical regions.\\u003c/p\\u003e\\n\\u003cp\\u003eAt the network level, parallel covariation patterns emerged: susceptibility within the LIM, DMN, and frontoparietal (FPN) networks showed strong cross-modal relationships with glymphatic indices, mirroring the anatomical organization observed at the ROI level. FW and PVS also demonstrate consistent significant correlations.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.3 Multivariate glymphatic-susceptibility components identified by PLS\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003ePLS revealed a dominant latent component (LV1) that captured 55\\u0026ndash;57% of cross-modal covariance at the ROI level and 61\\u0026ndash;63% at the network level (|\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|: ROI 0.541, NET 0.626;\\u0026nbsp;\\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e: ROI 0.568, NET 0.607).\\u003c/p\\u003e\\n\\u003cp\\u003eOn the glymphatic side, gBOLD-CSF coupling contributed the dominant and most stable negative loadings across all models, particularly at the network level. FW also exhibited\\u0026nbsp;consistent negative contributions.\\u0026nbsp;In contrast, PVS, DTI-ALPS, and BF volume showed positive loadings. ALPS and PVS contributions were more prominent in |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|, whereas\\u0026nbsp;\\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e, showed comparatively weaker positive loadings.\\u003c/p\\u003e\\n\\u003cp\\u003eOn the susceptibility side, for \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e, the strongest positive loadings were observed within LIM and DMN networks, accompanied by substantial contributions from visual (VIS) and FPN networks. |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u0026nbsp;showed prominent loadings within limbic and frontoparietal networks, together with prominent involvement of the ventral (VAN) and dorsal attention (DAN) networks and moderate contributions from somatomotor\\u0026nbsp;(SOM) regions. At the ROI level, |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u0026nbsp;additionally demonstrated pronounced loadings in deep nucleus\\u0026nbsp;and limbic regions, including the globus pallidus (GP), red nucleus (RN), putamen (Put), and thalamus (Tha).\\u0026nbsp;\\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e loadings were primarily distributed across temporo-occipital and parahippocampal cortex, accompanied by moderate positive contributions in parietal and precuneus regions. Both susceptibility contrasts showed overlapping involvement of limbic, frontoparietal, and subcortical systems (Fig. 3B).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.4 Spatially convergent glymphatic-susceptibility patterns across the AD continuum\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eMultistage filtering identified a set of glymphatic-susceptibility pairs showing consistent alterations across the AD continuum\\u0026nbsp;based on trend analysis, PLS contributions, and group separation criteria (Supplementary Fig. S2).\\u0026nbsp;At the ROI level, both |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u003csub\\u003e\\u0026nbsp;\\u003c/sub\\u003eand \\u0026chi;\\u003csub\\u003epara\\u0026nbsp;\\u003c/sub\\u003evalues\\u0026nbsp;demonstrated\\u0026nbsp;distinct glymphatic-susceptibility association\\u0026nbsp;patterns (Fig. 4).\\u0026nbsp;For |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|, the strongest effects emerged in deep and posterior cortex, where DTI-ALPS and FW showed progressive alterations across disease stages. In contrast, \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e demonstrated stronger coupling with ALPS, FW, and PVS in medial and parietal regions, particularly in precuneus (Pcun).\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003eAt the network level, a clearer and more integrated pattern emerged (Fig. 4). The strongest glymphatic-susceptibility associations were found within LIM, DMN, and FPN networks. This study also observed significant associations between the glymphatic function and susceptibility in the VIS and SOM networks. With regard to the glymphatic function, both FW and ALPS showed significant associations across susceptibility networks.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.5 Associations between\\u003c/strong\\u003e \\u003cstrong\\u003eglymphatic-susceptibility patterns and cognition\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eSEM models tested whether multimodal ND patterns mediated the relationship between diagnostic status and cognition\\u0026nbsp;(Fig. 5A). Diagnostic group significantly predicted both ND and MMSE across four models (ROI-level and network-level ND patterns derived from |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u003csub\\u003e\\u0026nbsp;\\u003c/sub\\u003eand \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e, \\u003cem\\u003ep \\u0026lt;\\u003c/em\\u003e 0.001), indicating that ND and cognitive performance captured graded differences across the AD continuum.\\u003c/p\\u003e\\n\\u003cp\\u003eIn contrast, ND did not significantly predict MMSE in any model (\\u003cem\\u003ep\\u003c/em\\u003e \\u0026gt; 0.05), and no indirect effects from diagnostic group to cognition through ND were observed. Although the magnitude and direction of the ND to MMSE paths varied slightly across glymphatic-susceptibility association domains, none reached statistical significance.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003e3.6 Sequential temporal progression of glymphatic-susceptibility alterations\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eTo characterize temporal organization, we examined DPM-derived pseudotime across ND indices from |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u003csub\\u003e\\u0026nbsp;\\u003c/sub\\u003eand \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e (Fig. 5B). ROI-based pseudotime showed significant increases from HC to AD, whereas network-based estimates showed no correspondence with diagnostic category.\\u003c/p\\u003e\\n\\u003cp\\u003eAt the ROI level, both susceptibility modalities demonstrated substantial temporal heterogeneity. Deep and medial parietal regions deviated from healthy baselines at very early pseudotime positions, including Tha, Put, DN, GP, Pcun, and posterior cingulate cortex (PCC),\\u0026nbsp;while peak-change times varied widely across loci (Fig. 5B). The \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e indices often exhibited more prolonged or delayed trajectories relative to |\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|.\\u003c/p\\u003e\\n\\u003cp\\u003eAt the network level,\\u0026nbsp;|\\u0026chi;\\u003csub\\u003edia\\u003c/sub\\u003e|\\u003csub\\u003e\\u0026nbsp;\\u003c/sub\\u003eand \\u0026chi;\\u003csub\\u003epara\\u003c/sub\\u003e ND indices exhibited temporally staggered trajectories despite their weak correspondence with diagnostic groups. Early deviations were observed within limbic and default-mode networks, followed by mid-stage inflections in frontoparietal systems and later changes in somatomotor and dorsal-attention networks (Fig. 5B).\\u003c/p\\u003e\"},{\"header\":\"4. Discussion\",\"content\":\"\\u003cp\\u003eThe present findings indicate that glymphatic function and susceptibility-related tissue composition are organized within a shared physiological system linking fluid clearance with iron-related and macromolecular properties across the brain. Rather than representing independent pathological processes, alterations in these domains appear to reflect the destabilization of a clearance-tissue composition architecture that is particularly prominent in metabolically demanding and vascularly complex regions. Across spatial scales, deviations from this coordinated organization become increasingly pronounced along the AD continuum, suggesting that disease progression is characterized by the progressive disruption of physiological mechanisms that regulate brain-environment homeostasis. This cross-modal framework therefore provides a systems-level perspective on how clearance inefficiency and tissue compositional alterations co-evolve, offering a unified lens through which regional vulnerability and disease progression can be understood.\\u003c/p\\u003e \\u003cp\\u003eGlymphatic alterations in AD were broad yet patterned, reflecting coordinated disruption across multiple levels of the clearance pathway. Reductions in BOLD-CSF coupling and ALPS efficiency, together with elevated FW, PVS burden, and CSF and CP measures, point to convergent disruption of CSF inflow [\\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e11\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e13\\u003c/span\\u003e], perivascular transport, and interstitial fluid mobility [\\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e7\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR14\\\" class=\\\"CitationRef\\\"\\u003e14\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e16\\u003c/span\\u003e]. Notably, several of these abnormalities were already present in SCD, indicating that glymphatic compromise emerges before the onset of overt cognitive decline. Parallel to these clearance-related disturbances, AD also exhibited distinct alterations in iron and neurotoxic proteins. The χ\\u003csub\\u003epara\\u003c/sub\\u003e values were reduced in the temporal, parietal, and occipital cortical regions, with focal elevations in the posterior superior temporal cortex and dentate nucleus. Increased χ\\u003csub\\u003epara\\u003c/sub\\u003e has been widely interpreted as reflecting iron accumulation. In contrast, cortical reductions in χ\\u003csub\\u003epara\\u003c/sub\\u003e are consistent with prior evidence of regionally selective iron redistribution in AD [\\u003cspan citationid=\\\"CR49\\\" class=\\\"CitationRef\\\"\\u003e49\\u003c/span\\u003e], a process that has been linked to mitochondrial stress and dysregulated cellular iron handling [\\u003cspan citationid=\\\"CR50\\\" class=\\\"CitationRef\\\"\\u003e50\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR51\\\" class=\\\"CitationRef\\\"\\u003e51\\u003c/span\\u003e]. Diamagnetic susceptibility exhibited distinct patterns, with the strongest reductions in |χ\\u003csub\\u003edia\\u003c/sub\\u003e| observed in frontal and parietal association cortices. In gray matter, |χ\\u003csub\\u003edia\\u003c/sub\\u003e| has been shown to reflect macromolecular and structural composition beyond myelin content [\\u003cspan additionalcitationids=\\\"CR26\\\" citationid=\\\"CR25\\\" class=\\\"CitationRef\\\"\\u003e25\\u003c/span\\u003e\\u0026ndash;\\u003cspan citationid=\\\"CR27\\\" class=\\\"CitationRef\\\"\\u003e27\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR51\\\" class=\\\"CitationRef\\\"\\u003e51\\u003c/span\\u003e], and reductions in |χ\\u003csub\\u003edia\\u003c/sub\\u003e| are consistent with alterations in protein density, cellular integrity, and extracellular organization accompanying neurodegeneration [\\u003cspan citationid=\\\"CR28\\\" class=\\\"CitationRef\\\"\\u003e28\\u003c/span\\u003e]. At the network level, both |χ\\u003csub\\u003edia\\u003c/sub\\u003e| and χ\\u003csub\\u003epara\\u003c/sub\\u003e were prominently represented within LIM, DMN, and FPN, aligning their spatial expression with networks characterized by high metabolic demand, vascular burden, and early pathological vulnerability [\\u003cspan citationid=\\\"CR52\\\" class=\\\"CitationRef\\\"\\u003e52\\u003c/span\\u003e]. These susceptibility patterns collectively indicate that alterations in iron and macromolecular tissue composition undergo a transformation, occurring in parallel with the aforementioned clearance dysfunction and likely reflecting related pathological processes across these physiological domains.\\u003c/p\\u003e \\u003cp\\u003eBeyond univariate alterations, the cross-modal analyses revealed that glymphatic and susceptibility features were not independently distributed but exhibited an organized pattern of coupling. The strongest positive associations emerged between PVS burden and ALPS efficiency, whereas negative associations were most pronounced between FW content and BOLD-CSF coupling. This organization reflects coordinated relationships between perivascular pathway integrity, interstitial fluid mobility, and large-scale solute exchange, linking microstructural tissue composition with fluid transport dynamics [\\u003cspan citationid=\\\"CR17\\\" class=\\\"CitationRef\\\"\\u003e17\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR19\\\" class=\\\"CitationRef\\\"\\u003e19\\u003c/span\\u003e]. This directional pattern may reflect differences between structural capacity and effective exchange within the glymphatic system, whereby enlargement of perivascular pathways and increased transport demand coexist with reduced large-scale clearance efficiency. Notably, this coupling was preferentially expressed in regions such as Pcun and DN, areas characterized by high metabolic demand, dense vascular interfaces, and early AD vulnerability [\\u003cspan citationid=\\\"CR52\\\" class=\\\"CitationRef\\\"\\u003e52\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR53\\\" class=\\\"CitationRef\\\"\\u003e53\\u003c/span\\u003e]. At the network level, LIM, DMN, and FPN played a dominant role in this cross-modal, indicating that coupling is organized according to large-scale functional hierarchies rather than being spatially widespread [\\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e11\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR54\\\" class=\\\"CitationRef\\\"\\u003e54\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR55\\\" class=\\\"CitationRef\\\"\\u003e55\\u003c/span\\u003e] and paramagnetic or diamagnetic changes observed in earlier analyses [\\u003cspan citationid=\\\"CR51\\\" class=\\\"CitationRef\\\"\\u003e51\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR56\\\" class=\\\"CitationRef\\\"\\u003e56\\u003c/span\\u003e]. These alterations in tissue composition may represent downstream consequences of impaired solute exchange despite apparent increases in perivascular transport features. In particular, the preferential involvement of metabolically demanding and vascularly complex regions indicates that coupling between clearance-related and tissue compositional features is most prominent within circuits that rely heavily on efficient fluid transport and perivascular regulation. Taken together, these findings reflect that cross-modal physiological alterations are not randomly distributed but instead follow a structured spatial pattern aligned with known gradients of metabolic demand and vascular burden. When this clearance support is compromised, alterations in tissue composition are likely to arise within the same anatomical systems, providing a physiological explanation for the observed co-localization of glymphatic and susceptibility signatures in these metabolically and vascularly demanding circuits.\\u003c/p\\u003e \\u003cp\\u003eAD progression was associated with region-specific changes in the relationships between glymphatic indices and susceptibility measures [\\u003cspan citationid=\\\"CR2\\\" class=\\\"CitationRef\\\"\\u003e2\\u003c/span\\u003e]. Rather than occurring uniformly across the brain, increases in ND and significant disease stage were preferentially clustered in deep nuclei and medial-parietal association cortices, including the DN, Tha, Pcun, and PCC. At the network level, they were also concentrated within LIM and DMN. The temporal ordering of these alterations revealed a heterogeneous but interpretable progression pattern in which these deep nuclei and limbic regions exhibited early deviations, followed by progressive engagement of frontoparietal and temporo-occipital association cortices [\\u003cspan citationid=\\\"CR47\\\" class=\\\"CitationRef\\\"\\u003e47\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR48\\\" class=\\\"CitationRef\\\"\\u003e48\\u003c/span\\u003e]. Notably, the ordering of regional changes corresponded to areas where glymphatic dysfunction and susceptibility alterations were jointly expressed, indicating that disease progression is constrained by local clearance-tissue composition interactions rather than by conventional network organization. These findings suggest that AD does not affect glymphatic-susceptibility coupling features uniformly but progressively disrupts the organization of the clearance-tissue composition system according to regional dependence on fluid transport and metabolic support. Regions such as deep nuclei and medial parietal cortex, which rely heavily on efficient CSF-interstitial exchange to maintain local tissue homeostasis, appear particularly vulnerable to early disruption. As disease progresses, this disruption extends into higher-order association cortices, reflecting a hierarchical breakdown of the physiological coupling between clearance efficiency and tissue composition. This spatial gradient therefore reflects not merely where abnormalities occur but also characterizes the spatiotemporal evolution of both clearance dysfunction and neurotoxic or tissue compositional burden.\\u003c/p\\u003e \\u003cp\\u003eAlthough ND increased progressively across diagnostic stages, it showed little correspondence with individual differences in cognitive performance. This dissociation indicates that the cross-modal physiological alterations captured by ND reflect system-level vulnerability that is not directly mirrored by cognitive impairment. Rather than tracking cognitive decline, ND appears to capture early alterations in glymphatic function and susceptibility-related tissue properties that emerge before overt cognitive symptoms. This suggests that glymphatic-susceptibility disruption represents a physiological layer of disease expression that is partially decoupled from clinical symptom manifestation. The separation between ND and cognition, therefore, indicates that clearance-tissue composition alterations follow a trajectory distinct from, and potentially preceding, measurable cognitive deterioration across the AD continuum. These findings imply that multimodal physiological biomarkers may reflect upstream system vulnerability in AD that is not readily captured by conventional cognitive assessments.\\u003c/p\\u003e \\u003cp\\u003eThese findings suggest that iron-related and macromolecular tissue composition alterations in AD are not solely attributable to localized pathological burden or iron dysregulation in isolation but instead emerge within the context of a progressively weakened clearance-tissue composition system [\\u003cspan citationid=\\\"CR6\\\" class=\\\"CitationRef\\\"\\u003e6\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR52\\\" class=\\\"CitationRef\\\"\\u003e52\\u003c/span\\u003e, \\u003cspan citationid=\\\"CR57\\\" class=\\\"CitationRef\\\"\\u003e57\\u003c/span\\u003e]. This study underscores the importance of viewing clearance dysfunction and tissue compositional alterations as interconnected dimensions of disease vulnerability. By capturing their coupled evolution across spatial and disease-stage gradients, multimodal imaging approaches offer a powerful means of characterizing early physiological dysregulation in AD and provide a systems-level lens through which the breakdown of brain environmental homeostasis can be understood. The cross-modal framework may also provide a useful basis for exploring physiological mechanisms underlying neurodegenerative processes and for developing imaging markers sensitive to system-level vulnerability before overt structural degeneration becomes apparent.\\u003c/p\\u003e \\u003cp\\u003eSeveral limitations should be noted. First, the cross-sectional design precludes causal inference and prevents direct characterization of disease trajectories; longitudinal studies integrating MRI with fluid or PET biomarkers will be necessary to clarify the temporal ordering of clearance and susceptibility changes. Second, the glymphatic and susceptibility indices used here serve as indirect proxies, each capturing only part of the underlying biology and potentially influenced by vascular, inflammatory, or metabolic factors not assessed in this study. Finally, applying this framework to additional neurodegenerative or cerebrovascular conditions may help determine whether the cross-modal architecture is AD-specific or reflects a broader clearance\\u0026ndash;accumulation motif. Despite these constraints, the present multimodal approach provides a scalable foundation for integrating fluid-transport and susceptibility-based markers toward more comprehensive imaging indicators of neurodegenerative vulnerability.\\u003c/p\\u003e\"},{\"header\":\"5. Conclusion\",\"content\":\"\\u003cp\\u003eThis study demonstrates that glymphatic dysfunction and susceptibility alterations in AD are not independent abnormalities but coordinated manifestations of a unified clearance-tissue composition architecture. As the disease progresses, this physiological architecture becomes progressively destabilized, with early deviations emerging in deep and limbic regions and extending to higher-order cortical systems. The dissociation between cross-modal physiological alterations and cognitive performance further indicates that glymphatic-susceptibility coupling reflects an upstream vulnerability distinct from overt clinical impairment.\\u003c/p\\u003e \\u003cp\\u003eBy integrating complementary indices of fluid transport and tissue composition, the present framework provides a systems-level perspective on neurodegenerative vulnerability and offers a conceptual basis for understanding how clearance dysfunction and microstructural alterations co-evolve across the AD continuum. These findings highlight the potential of cross-modal physiological markers as sensitive indicators of early disease processes and as tools for probing mechanisms underlying neurodegeneration.\\u003c/p\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003eEthics\\u003c/strong\\u003e\\u003cstrong\\u003e\\u0026nbsp;Statement\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThis study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and was approved by the Ethics Committee of the China-Japan Friendship Hospital and Nanjing Drum Tower Hospital. All the participants and/or their relatives were informed about this study and provided their written informed consent.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eAcknowledgements\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe authors thank all participants in the study.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eFunding source\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThis study was supported by the STI2030-Major Projects (No. 2022ZD0213300), the National Natural Science Foundation of China (No. 82271953 and U25A20136),\\u0026nbsp;Key Industrial Core Technology Tackling Project of the Hengqin Guangdong-Macao Deep Cooperation Zone (No. 2430004045063), University of Macau (MYRG2022-00054-FHS, MYRG-GRG2023-00038-FHS-UMDF, and MYRG-GRG2024-00259-FHS), and the Macao Science and Technology Development Fund (FDCT 0014/2024/RIB1 and\\u0026nbsp;0048/2021/AGJ).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eAuthor contribution\\u0026nbsp;\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eCT.L.: conceptualization, methodology, validation, investigation, formal analysis, writing—original draft, writing—review and editing. AC.Y.: conceptualization, methodology, validation, investigation, formal analysis, writing—original draft, writing—review and editing. RS.W.: investigation, formal analysis. Y.S.: resources, data curation, validation. JX.L.: methodology, validation. MX.X.: methodology, validation, formal analysis. Y.G.: resources, data curation, validation. HW.Y.: resources, data curation, validation. YL.W.: resources, data curation, validation. N.S.: writing—review and editing. B. Z.: writing—review and editing. GL.M.: conceptualization resources, funding acquisition, project administration, writing—review and editing. Z.Y.: conceptualization resources, funding acquisition, project administration, writing—review and editing.\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eCompeting interests\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eThe authors declare no conflicts of interest\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eData and code availability\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003eOpen-source software was used for all analyses. The data and codes analyzed in this study can be obtained from the corresponding authors upon reasonable request.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\u003cli\\u003e\\u003cspan\\u003eBloom GS. Amyloid-β and tau: the trigger and bullet in Alzheimer disease pathogenesis. JAMA Neurol. 2014;71(4):505\\u0026ndash;8.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eJagust W. Imaging the evolution and pathophysiology of Alzheimer disease. Nat Rev Neurosci. 2018;19(11):687\\u0026ndash;700.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eWeiner MW, et al. The Alzheimer's Disease Neuroimaging Initiative: a review of papers published since its inception. Volume 9. Alzheimer's \\u0026amp; Dementia; 2013. pp. e111\\u0026ndash;94. 5.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eAp\\u0026aacute;tiga-P\\u0026eacute;rez R, et al. Neurovascular dysfunction and vascular amyloid accumulation as early events in Alzheimer's disease. Metab Brain Dis. 2022;37(1):39\\u0026ndash;50.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eJessen NA, et al. The glymphatic system: a beginner\\u0026rsquo;s guide. Neurochem Res. 2015;40:2583\\u0026ndash;99.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eTarasoff-Conway JM, et al. Clearance systems in the brain\\u0026mdash;implications for Alzheimer disease. Nat reviews Neurol. 2015;11(8):457\\u0026ndash;70.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eChoi JD, et al. Choroid plexus volume and permeability at brain MRI within the Alzheimer disease clinical spectrum. Radiology. 2022;304(3):635\\u0026ndash;45.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLin W, et al. Multimodal MRI reveals impaired glymphatic function with choroid plexus enlargement and cerebrospinal fluid expansion in alzheimer\\u0026rsquo;s disease. Sci Rep. 2025;15(1):30409.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eCosta T, et al. Evaluating the robustness of DTI-ALPS in clinical context: a meta-analytic parallel on Alzheimer\\u0026rsquo;s and Parkinson\\u0026rsquo;s diseases. Sci Rep. 2024;14(1):26381.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eKiviniemi V, et al. Ultra-fast magnetic resonance encephalography of physiological brain activity\\u0026ndash;glymphatic pulsation mechanisms? J Cereb Blood Flow Metabolism. 2016;36(6):1033\\u0026ndash;45.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eFultz NE, et al. Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep. Science. 2019;366(6465):628\\u0026ndash;31.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eChuang K-H, et al. Cholinergic basal forebrain neurons regulate vascular dynamics and cerebrospinal fluid flux. Nat Commun. 2025;16(1):5343.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eHan F, et al. Reduced coupling between cerebrospinal fluid flow and global brain activity is linked to Alzheimer disease\\u0026ndash;related pathology. PLoS Biol. 2021;19(6):e3001233.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eSteward CE, et al. Assessment of the DTI-ALPS parameter along the perivascular space in older adults at risk of dementia. J Neuroimaging. 2021;31(3):569\\u0026ndash;78.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eYu P, Shen L, Tang L. Multimodal DTI-ALPS and Hippocampal Microstructural Signatures Unveil Stage-Specific Pathways in Alzheimer's Disease Progression. Front Aging Neurosci. 2025;17:1609793.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eZhang Y et al. DTI-ALPS index-assessed glymphatic dysfunction mediates Alzheimer\\u0026rsquo;s cognitive decline via amyloid-β-dependent pathways: multimodal PET/MRI study. Eur J Nucl Med Mol Imaging, 2025: pp. 1\\u0026ndash;13.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eIliff JJ et al. \\u003cem\\u003eA paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β.\\u003c/em\\u003e Science translational medicine, 2012. 4(147): p. 147ra111-147ra111.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eMurdock MH, et al. Multisensory gamma stimulation promotes glymphatic clearance of amyloid. Nature. 2024;627(8002):149\\u0026ndash;56.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eXie L, et al. Sleep drives metabolite clearance from the adult brain. Science. 2013;342(6156):373\\u0026ndash;7.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eConrad M, et al. Regulated necrosis: disease relevance and therapeutic opportunities. Nat Rev Drug Discovery. 2016;15(5):348\\u0026ndash;66.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEverett J, et al. Evidence of redox-active iron formation following aggregation of ferrihydrite and the Alzheimer\\u0026rsquo;s disease peptide β-amyloid. Inorg Chem. 2014;53(6):2803\\u0026ndash;9.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eTao Y, et al. Perturbed iron distribution in Alzheimer's disease serum, cerebrospinal fluid, and selected brain regions: a systematic review and meta-analysis. J Alzheimer\\u0026rsquo;s Disease. 2014;42(2):679\\u0026ndash;90.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eZecca L, et al. Iron, brain ageing and neurodegenerative disorders. Nat Rev Neurosci. 2004;5(11):863\\u0026ndash;73.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBecerril-Ortega J, et al. Iron overload accelerates neuronal amyloid-β production and cognitive impairment in transgenic mice model of Alzheimer's disease. Neurobiol Aging. 2014;35(10):2288\\u0026ndash;301.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eSchweser F, et al. Quantitative imaging of intrinsic magnetic tissue properties using MRI signal phase: an approach to in vivo brain iron metabolism? NeuroImage. 2011;54(4):2789\\u0026ndash;807.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLiu C, et al. Quantitative susceptibility mapping: contrast mechanisms and clinical applications. Tomography. 2015;1(1):3.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLee J, Ji S, Oh S-H. So you want to image myelin using MRI: magnetic susceptibility source separation for myelin imaging. Magn Reson Med Sci. 2024;23(3):291\\u0026ndash;306.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eAhmed M, et al. The diamagnetic component map from quantitative susceptibility mapping (QSM) source separation reveals pathological alteration in Alzheimer\\u0026rsquo;s disease-driven neurodegeneration. NeuroImage. 2023;280:120357.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eUchida Y, et al. APOE ɛ4 dose associates with increased brain iron and β-amyloid via blood\\u0026ndash;brain barrier dysfunction. J Neurol Neurosurg Psychiatry. 2022;93(7):772\\u0026ndash;8.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eUchida Y, et al. Contributions of blood\\u0026ndash;brain barrier imaging to neurovascular unit pathophysiology of Alzheimer\\u0026rsquo;s disease and related dementias. Front Aging Neurosci. 2023;15:1111448.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eJack CR Jr, et al. NIA-AA research framework: toward a biological definition of Alzheimer's disease. Alzheimer's Dement. 2018;14(4):535\\u0026ndash;62.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003ePetersen RC, et al. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol. 1999;56(3):303\\u0026ndash;8.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eFischl B. FreeSurfer Neuroimage. 2012;62(2):774\\u0026ndash;81.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLiu X, et al. Cross-vendor test-retest validation of diffusion tensor image analysis along the perivascular space (DTI-ALPS) for evaluating glymphatic system function. Aging disease. 2024;15(4):1885.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eGu Y, et al. Transient arousal modulations contribute to resting-state functional connectivity changes associated with head motion parameters. Cereb Cortex. 2020;30(10):5242\\u0026ndash;56.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLi Z, et al. APART-QSM: An improved sub-voxel quantitative susceptibility mapping for susceptibility source separation using an iterative data fitting method. NeuroImage. 2023;274:120148.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eKlein A, et al. Evaluation of volume-based and surface-based brain image registration methods. NeuroImage. 2010;51(1):214\\u0026ndash;20.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eTaoka T et al. \\u003cem\\u003eDiffusion tensor image analysis ALong the perivascular space (DTI-ALPS): revisiting the meaning and significance of the method.\\u003c/em\\u003e Magnetic Resonance in Medical Sciences, 2024: p. rev. 2023\\u0026thinsp;\\u0026ndash;\\u0026thinsp;0175.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eDaducci A, et al. Accelerated microstructure imaging via convex optimization (AMICO) from diffusion MRI data. NeuroImage. 2015;105:32\\u0026ndash;44.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eEisma JJ, et al. Deep learning segmentation of the choroid plexus from structural magnetic resonance imaging (MRI): validation and normative ranges across the adult lifespan. Volume 21. Fluids and Barriers of the CNS; 2024. p. 21. 1.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eLin S, et al. Association of MRI indexes of the perivascular space network and cognitive impairment in patients with obstructive sleep apnea. Radiology. 2024;311(3):e232274.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eKim H-G, et al. Quantitative susceptibility mapping to evaluate the early stage of Alzheimer's disease. Volume 16. NeuroImage: Clinical; 2017. pp. 429\\u0026ndash;38.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBenjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc: Ser B (Methodol). 1995;57(1):289\\u0026ndash;300.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eKrishnan A, et al. Partial Least Squares (PLS) methods for neuroimaging: a tutorial and review. NeuroImage. 2011;56(2):455\\u0026ndash;75.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eCha J. Partial least squares. Adv Methods Mark Res. 1994;407:52\\u0026ndash;78.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eRosseel Y. lavaan: An R package for structural equation modeling. J Stat Softw. 2012;48(1):1\\u0026ndash;36.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eDonohue MC, et al. Estimating long-term multivariate progression from short‐term data. Alzheimer's Dement. 2014;10:S400\\u0026ndash;10.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eTrapnell C, et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat Biotechnol. 2014;32(4):381\\u0026ndash;6.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eYao J, et al. Distinct regional vulnerability to Aβ and iron accumulation in post mortem AD brains. Alzheimer's Dement. 2024;20(10):6984\\u0026ndash;97.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eAyton S, et al. Brain iron is associated with accelerated cognitive decline in people with Alzheimer pathology. Mol Psychiatry. 2020;25(11):2932\\u0026ndash;41.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003evan der Weijden CW, et al. Quantitative myelin imaging with MRI and PET: an overview of techniques and their validation status. Brain. 2023;146(4):1243\\u0026ndash;66.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBuckner RL, et al. Molecular, structural, and functional characterization of Alzheimer's disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci. 2005;25(34):7709\\u0026ndash;17.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eDuyn JH, Schenck J. Contributions to magnetic susceptibility of brain tissue. NMR Biomed. 2017;30(4):e3546.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eHablitz LM, et al. Increased glymphatic influx is correlated with high EEG delta power and low heart rate in mice under anesthesia. Sci Adv. 2019;5(2):eaav5447.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eTaoka T, Naganawa S. Glymphatic imaging using MRI. J Magn Reson Imaging. 2020;51(1):11\\u0026ndash;24.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eMadden DJ, Merenstein JL. Quantitative susceptibility Mapp brain iron healthy aging cognition NeuroImage. 2023;282:120401.\\u003c/span\\u003e\\u003c/li\\u003e \\u003cli\\u003e\\u003cspan\\u003eBouter Y, et al. 18F-FDG-PET and Multimodal Biomarker Integration: a Powerful Tool for Alzheimer\\u0026rsquo;s Disease Diagnosis. Nuclear Medicine and Molecular Imaging; 2025. pp. 1\\u0026ndash;19.\\u003c/span\\u003e\\u003c/li\\u003e\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":false,\"hideJournal\":false,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":false,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":false,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"bmc-medicine\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":false,\"externalIdentity\":\"bmed\",\"sideBox\":\"Learn more about [BMC Medicine](http://bmcmedicine.biomedcentral.com/)\",\"snPcode\":\"12916\",\"submissionUrl\":\"https://submission.nature.com/new-submission/12916/3\",\"title\":\"BMC Medicine\",\"twitterHandle\":\"\",\"acdcEnabled\":true,\"dfaEnabled\":true,\"editorialSystem\":\"stoa\",\"reportingPortfolio\":\"BMC/SO AJ\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":true},\"keywords\":\"Alzheimer's disease, Cerebrospinal fluid flow, Functional MRI, Quantitative susceptibility mapping, Glymphatic function, Iron accumulation\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-9436391/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-9436391/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003eIntroduction:\\u003c/p\\u003e \\u003cp\\u003eAlzheimer\\u0026rsquo;s disease (AD) involves impaired glymphatic clearance and abnormal iron-related tissue and pathological protein alterations, yet it remains unclear whether these processes represent independent abnormalities or a coordinated system-level architecture.\\u003c/p\\u003e \\u003cp\\u003eMethods:\\u003c/p\\u003e \\u003cp\\u003eWe developed a multimodal MRI framework integrating multiple MRI indices capturing complementary aspects of glymphatic function with paramagnetic and diamagnetic susceptibility measures across AD spectrum. Cross-modal associations, stage-dependent deviations, and spatial progression patterns were examined to characterize clearance-tissue composition architecture linking fluid transport with susceptibility-derived alterations.\\u003c/p\\u003e \\u003cp\\u003eResults:\\u003c/p\\u003e \\u003cp\\u003eGlymphatic and susceptibility measures showed coordinated cross-modal associations at both ROI and network levels, with the strongest effects observed in limbic, default-mode, and frontoparietal networks. Across the AD continuum, coupled alterations were most prominent in deep nuclei and medial parietal regions, with deviations involving the thalamus, putamen, dentate nucleus, globus pallidus, precuneus, and posterior cingulate cortex. These multimodal patterns varied systematically with disease stage but were not significantly associated with cognitive performance.\\u003c/p\\u003e \\u003cp\\u003eConclusion:\\u003c/p\\u003e \\u003cp\\u003eAD is characterized by progressive destabilization of a clearance-tissue composition system rather than isolated functional impairments. This cross-modal framework provides a systems-level perspective on neurodegenerative vulnerability and highlights integrated imaging markers sensitive to early physiological dysregulation along the AD continuum.\\u003c/p\\u003e\",\"manuscriptTitle\":\"Glymphatic Dysfunction Coupled with Aberrant Distribution of Iron and Pathological proteins across Alzheimer’s Disease Continuum\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2026-05-11 08:11:17\",\"doi\":\"10.21203/rs.3.rs-9436391/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0},{\"type\":\"reviewerAgreed\",\"content\":\"180655610008559104101803674639632411072\",\"date\":\"2026-05-02T06:00:26+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"176242419916246101615148428100610258710\",\"date\":\"2026-04-29T08:46:34+00:00\",\"index\":\"hide\",\"fulltext\":\"\"},{\"type\":\"reviewersInvited\",\"content\":\"\",\"date\":\"2026-04-29T08:24:54+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"editorAssigned\",\"content\":\"\",\"date\":\"2026-04-17T11:32:32+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"checksComplete\",\"content\":\"\",\"date\":\"2026-04-17T10:47:33+00:00\",\"index\":\"\",\"fulltext\":\"\"},{\"type\":\"submitted\",\"content\":\"BMC Medicine\",\"date\":\"2026-04-16T09:42:35+00:00\",\"index\":\"\",\"fulltext\":\"\"}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"bmc-medicine\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":false,\"externalIdentity\":\"bmed\",\"sideBox\":\"Learn more about [BMC Medicine](http://bmcmedicine.biomedcentral.com/)\",\"snPcode\":\"12916\",\"submissionUrl\":\"https://submission.nature.com/new-submission/12916/3\",\"title\":\"BMC Medicine\",\"twitterHandle\":\"\",\"acdcEnabled\":true,\"dfaEnabled\":true,\"editorialSystem\":\"stoa\",\"reportingPortfolio\":\"BMC/SO AJ\",\"inReviewEnabled\":true,\"inReviewRevisionsEnabled\":true}}],\"origin\":\"\",\"ownerIdentity\":\"d799fc0b-e84c-4a15-a1d8-c5ab75f033e8\",\"owner\":[],\"postedDate\":\"May 11th, 2026\",\"published\":true,\"recentEditorialEvents\":[{\"type\":\"reviewerAgreed\",\"content\":\"180655610008559104101803674639632411072\",\"date\":\"2026-05-02T06:00:26+00:00\",\"index\":37,\"fulltext\":\"\"},{\"type\":\"reviewerAgreed\",\"content\":\"176242419916246101615148428100610258710\",\"date\":\"2026-04-29T08:46:34+00:00\",\"index\":19,\"fulltext\":\"\"},{\"type\":\"reviewersInvited\",\"content\":\"21\",\"date\":\"2026-04-29T08:24:54+00:00\",\"index\":\"\",\"fulltext\":\"\"}],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"under-review\",\"subjectAreas\":[],\"tags\":[],\"updatedAt\":\"2026-05-11T08:11:18+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2026-05-11 08:11:17\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-9436391\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-9436391\",\"identity\":\"rs-9436391\",\"version\":[\"v1\"]},\"buildId\":\"XKTyCvWXoU3ODBz1xrDgd\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}