key: cord-0330483-b38l1iut
authors: Ottoy, J.; Ozzoude, M.; Zukotynski, K.; Adamo, S.; Scott, C.; Gaudet, V.; Ramirez, J.; Swardfager, W.; Cogo-Moreira, H.; Lam, B.; Bhan, A.; Mojiri, P.; Kang, M. S.; Rabin, J. S.; Kiss, A.; Strother, S.; Bocti, C.; Borrie, M.; Chertkow, H.; Frayne, R.; Hsiung, R.; Laforce, R. J.; Noseworthy, M. D.; Prato, F. S.; Sahlas, D. J.; Smith, E. E.; Kuo, P. H.; Sossi, V.; Thiel, A.; Soucy, J.-P.; Tardif, J.-C.; Black, S. E.; Goubran, M.
title: Vascular burden and cognition: Mediating roles of neurodegeneration and amyloid-PET
date: 2021-12-27
journal: nan
DOI: 10.1101/2021.12.24.21267786
sha: 7470e10dd14bfc6901791712370d40cab1672a5d
doc_id: 330483
cord_uid: b38l1iut

INTRODUCTION: It remains unclear to which extent vascular burden promotes neurodegeneration and cognitive dysfunction in a cohort spanning low-to-severe small vessel disease (SVD) and amyloid-beta pathology. METHODS: In 120 subjects, we investigated 1) whether vascular burden, quantified as total or lobar white matter hyperintensity (WMH) volumes, is associated with different cognitive domains; and 2) whether the total WMH effect on cognition is mediated by amyloid (18F-AV45-PET), glucose metabolism (18F-FDG-PET), and/or cortical atrophy. RESULTS: Increased total WMH volume was associated with poorer performance in all cognitive domains tested, with the strongest effects observed for semantic fluency. These relationships were mediated mainly through cortical atrophy, particularly in the temporal lobe, and to a lesser extent through amyloid and metabolism. WMH volumes differentially impacted cognition depending on lobar location and amyloid status. DISCUSSION: Our study suggests mainly an amyloid-dependent pathway in which vascular burden affects cognitive impairment through temporal lobe atrophy.

White matter hyperintensities (WMH) detected on MRI are a common finding in stroke and dementia clinic patients [1, 2] . Yet, the contributions of WMH to cognitive decline remain poorly understood. While Alzheimer's disease (AD)-type pathologies such as amyloid-ß (Aß) plaques or tau neurofibrillary tangles are known to accelerate cognitive decline [3, 4] , the cognitive correlates of cerebral small vessel disease (SVD), specifically of WMH, have been both underestimated and understudied in AD [5] .

SVD is a group of diseases that affects small arteries, venules, and capillaries of the brain [6] . MRI-based markers of SVD include WMH of presumed vascular origin, cerebral microbleeds, lacunae, and enlarged perivascular spaces. Amongst these markers, WMH is the most studied proxy of SVD. Notwithstanding that a low burden of WMH is frequently observed in healthy elderly [7, 8] , prior research indicated that WMH often co-exist with AD pathology [9] . In fact, more than 80% of AD dementia cases at autopsy may display cerebrovascular disease [10] . WMH are associated with an increased risk of stroke and (AD) dementia and faster clinical progression [11] , highlighting that WMH should not just be interpreted as silent consequences of ageing [8] .

Network studies have shown that focal WMH may initiate a cascade of events affecting remote brain areas often involved in AD [12, 13] . In particular, WMH were found to be related to neurodegenerative processes (hypometabolism/atrophy) in AD-vulnerable regions [14] [15] [16] [17] [18] [19] [20] .

Specifically, these neurodegenerative processes were demonstrated to mediate the relationship between SVD burden and cognitive impairment [15, 18, 21] . Longitudinal studies, in turn, supported the potential causal relationship between SVD and cognitive impairment by showing that elevated WMH burden at baseline worsened neurodegeneration and cognition over time in AD cohorts [22, 23] . Similarly, studies have highlighted that baseline WMH predicted higher Aß . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ;  https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint deposition over time [24] [25] [26] , and even suggested that vascular abnormality precedes Aß deposition in the cascade of AD pathophysiology [27] .

However, hitherto, the majority of large-cohort studies in dementia that investigated the effects of WMH on neurodegeneration and cognition excluded subjects with considerable WMH burden, such as the Alzheimer's Disease Neuroimaging Initiative (ADNI). Others focused on cognitively normal elderly, with few enriched for high WMH burden [15] . Therefore, our knowledge is still limited regarding the effects of vascular pathology in patients of more extreme endophenotypes with evidence of significant WMH burden in addition to AD pathology [5] .

Furthermore, to our knowledge, no studies have yet comprehensively studied the potential mediating roles of Aß, glucose metabolism, and atrophy in the vascular contributions to cognitive impairment [26] .

In this study, we investigated a unique cohort of clinically normal elderly and "real-world" patients capturing the spectrum of low to extensive WM disease and Aß pathology. The objectives were two-fold: 1) investigate whether vascular burden, quantified as total or lobar WMH volume, is associated with cognition; 2) investigate the potential mediating roles of Aß, glucose metabolism, and/or cortical atrophy in the WMH-cognition relationship.

The study included 120 subjects in total. Sixty subjects were recruited in a multicenter prospective observational study through seven participating sites as part of the C6 project in the . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 

A battery of cognitive tests was administered including: processing speed (Trail Making Test part-A, n=120); executive function (Trail Making Test part-B, n=119); semantic fluency (animal naming, n=120); and language (Boston Naming Test or BNT, n=120). Exploratory analyses included global cognition (Montreal Cognitive Assessment or MoCA, n=120) and global function (Functional Assessment Questionnaire or FAQ, n=110). Individual assessments are described in Supplementary Table 3. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 

Each subject underwent a 3T MRI scan, including 3D T1 and fluid-attenuated inversion recovery (FLAIR) structural sequences. Acquisition parameters are described in Supplementary   Table 4 and followed a common imaging protocol [30] . Each subject also underwent a 18 Fflorbetapir ( 18 F-AV45) PET scan to quantify brain Aß (n=120) and a 18 F-fluorodeoxyglucose ( 18 F-FDG) PET scan (n=117) to quantify brain glucose metabolism. Consistency of PET data between 

Atrophy was quantified using cortical thickness, extracted based on T1-weighted images using FreeSurfer v6.0. We employed a modified FreeSurfer workflow for patients with WMH burden [31] . An 'AD-signature' thickness meta-ROI was employed based on surface-weighted thickness averages of the entorhinal, fusiform, parahippocampal, mid/inferior temporal, and inferior parietal gyri [32] .

WMH were delineated based on our automated segmentation tool 'HyperMapper' that uses T1, FLAIR, and Bayesian 3D Convolutional Neuronal Networks (CNN) [33] . Subcortical lacunar . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint infarcts were masked before WMH segmentation [34] . ICV was extracted using our CNN-based 'iCVMapper' tool, which was shown to be more robust compared to other state-of-the-art skullstripping methods [35] . Finally, vascular burden was defined as WMH volume divided by ICV and log-transformed.

Lobar WMH volumes were determined by intersecting the total WMH mask and each individual lobar WM mask, as delineated in native T1 space based on the Desikan-Killiany-Tourville (DKT) atlas of FreeSurfer. Lobar WM masks included the frontal, parietal, temporal, occipital, insula, and cingulate lobes. Figure 1 shows examples of our structural MRI scans, WMH and lobar delineations, as well as a heatmap of WMH volumes across the MITNEC-C6 cohort.

PET images were processed using PetSurfer v6.0. This included motion-correction of the individual PET frames to the first frame and averaging to obtain one static frame, co-registration to T1-weighted image, smoothing to 8mm full-width-at-half-maximum to reach a common resolution across sites [36] , and generating standardized uptake value ratios (SUVR) maps. The 18 F-AV45 and 18 F-FDG SUVR maps were referenced to the cerebellum and the pons, respectively [37] . Partial volume correction (PVC) was applied using the geometric-transfer-matrix method with a point-spread-function of 8mm. Regional values were extracted based on the DKT atlas in native T1 space. AD-signature meta-ROIs were created for 18 F-AV45 and 18 F-FDG SUVR based on Jack's mask (volume-weighted average of the frontal, parietal, temporal, and cingulate regions) and the temporo-parietal lobe, respectively [37] [38] [39] . Based on our Aß-PET pipeline, 22% and 48% of ADNI and MITNEC-C6 subjects, respectively, were considered Aß-positive using a . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint quantitative 18 F-AV45 SUVR cut-off of 1.1. This cut-off was derived from Gaussian mixture modelling with two components using non-PVC SUVR in the AD-signature meta-ROI. Similar results were found when using PVC data.

Regional statistics were performed in SPSS v24 (SPSS Inc., Chicago, IL) and vertex-wise statistics were performed in FreeSurfer v6.0. All metrics were z-scored to allow for direct comparison between models across predictors and outcome measures. Values were reported as mean ± standard error (SE), unless otherwise stated.

A two-tailed t-test for continuous variables and chi-square for categorical variables were used to detect significant group differences in the demographics. A two-way ANOVA (lobe*group), corrected for age, sex, education, and Sidak's multiple comparisons test, was used to detect group differences in lobar WMH volumes.

Linear regressions were employed to assess the associations between total/lobar WMH volume (independent variable) and each of the cognitive scores (dependent variable), adjusted for age, sex, and education. Bias-corrected bootstrapping with 5,000 replications was applied to . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. account for multiple comparisons and heteroscedasticity. Bias-corrected bootstrapping does not make any assumptions about normality in the sampling distribution and better controls type I errors.

For mediation analyses, the PROCESS macro v3.5 in SPSS was applied. Bias-corrected bootstrapping with 5,000 replications was performed for estimation of (in)direct and total effects.

We hypothesized that a serial mediation runs from 'total WMH àAß [24, 25] à atrophy [22, 23] à cognition'. Importantly, this mediation model allows not only to investigate indirect effects of WMH volumes on cognition through the hypothesized serial path, but also through the predictor and each of the mediators separately while adjusting for the remaining variables in the model. Due to the previously reported association of 18 F-FDG-PET with semantic fluency and executive function [40] , glucose metabolism was tested as an additional mediator within their path analyses: total WMH à Aß à metabolism [20, 41] à atrophy [42] à fluency or executive function; the cited references indicate prior research establishing this sequence of biomarkers becoming abnormal in AD. Imaging mediators were evaluated within AD-signature regions (see section 2.5).

Age, education, and sex were used as covariates regressed on the mediators and outcome simultaneously.

To test regional specificity of the mediation analyses, we performed additional path analysis whereby AD-signature meta-ROIs of the imaging markers were substituted by a more focal region. This region was selected through two separate whole-brain vertex-wise linear regressions of total WMH volume with (i) atrophy [cortical thickness] and (ii) Aß SUVR, to . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint identify one 'WMH-signature' region in relation to both atrophy and Aß. These vertex-wise regressions were adjusted for age, sex, education, and (i) additionally for global Aß SUVR. Vertexwise multiple comparisons correction was based on Monte-Carlo simulations with 5,000 iterations, which implemented a two-tailed cluster-forming P-value of 0.01. Table 1 represents the demographics within the low and high WMH groups. Greater total WMH volume was associated with poorer performance on the following assessments across all Volumes of lobar WMH were significantly higher in the high WMH group in the frontal, parietal, cingulate, and insula lobes ( Figure 2B ). Cognitive correlates of frontal WMH volumes were significant only in Aß-negative subjects, whereas cognitive correlates of temporal and parietal/cingulate WMH volumes were more prominent in the Aß-positive subjects (Supplementary Table 5 ). Only cingulate WMH volumes were significantly increased in the Aßpositive compared to the Aß-negative group (Figure 2B ).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 

In the path analysis for semantic fluency, we found significant indirect effects of total WMH volume through (i) 'WMHàatrophyàfluency', (ii) 'WMHàAßàatrophyàfluency', and 

Vertex-wise regression analyses showed that total WMH volume was positively associated with Aß load and inversely associated with atrophy, particularly in lateral temporal regions ( Figure 4A ). As such, we performed further path analysis on cognition using a temporal meta-ROI for all mediators. The temporal meta-ROI consisted of the inferior, superior, and middle . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

A post-hoc mediation model was investigated using Aß as the predictor and WMH as the mediator: 'AßàWMHàatrophyàcognition'. Rather than investigating the causal relationships between Aß and WMH, this model was applied to support our previous findings that the indirect effects of Aß and WMH through each other on atrophy and cognition have lower effect size compared to the effect of Aß or WMH alone. Figure 3) were found as described above for each of the cognitive domains (see section 3.2). Importantly, the indirect path 'Aßàatrophyàcognition' had larger effect size than its effect through WMH ('AßàWMHàatrophyàcognition'). This paralleled the results described above (see section 3.2), where the indirect path 'WMHàatrophyàcognition' had larger effect size than its effect through Aß ('WMHàAßàatrophyàcognition'). Taken together, this may suggest that WMH and Aß . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint exert more additive and less synergistic effects on atrophy (i.e., potentially driving further atrophy through different underlying mechanisms).

Furthermore, the Aß-independent effect of WMH on atrophy was greater than the WMHindependent effect of Aß on atrophy in both the initial and post-hoc models. Taken together, this may suggest that the effect of WMH on atrophy exceeds the effect of Aß on atrophy.

This study investigated the relationship between WMH burden and standardized neuropsychological assessments in a unique cohort of clinically normal elderly and "real-world patients" capturing the spectrum of low to extensive WM disease and Aß pathology. We found that WMH volumes were associated with poorer semantic fluency, executive function, and global function. These relationships were mostly due to the association of WMH volumes with cortical atrophy, particularly in the temporal lobe, and to a lesser extent due to the association of WMH volumes with Aß or glucose metabolism.

The contribution of WMH to cognitive deficits has been thus far poorly understood.

Previous literature showed direct, indirect, and no effects of WMH on cognitive decline in various populations [26, 43] . Insofar, most large AD/dementia cohort studies excluded patients with considerate amount of vascular co-pathology as mixed disease. As such, potential contributions of WMH to cognitive impairment are often not well addressed in these "clean" cases of probable AD [7] . While a considerate portion of both cognitively normal and AD dementia cases display WMH [8, 10, 28] , the prevalence of WMH is substantially higher in stroke and dementia clinic patients [1, 2] . This group of patients thus provides an enriched source for recruitment of more . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint extreme vascular endophenotypes. In our WMH-enriched cohort (totaling n=120), we found an association between WMH volumes and cognitive impairment but not after controlling for atrophy, particularly of the temporal lobe. This suggests an indirect ('mediation') effect of WMH on cognition through neurodegenerative processes. Similarly, studies have shown that WMH contribute to atrophy beyond age-related effects, particularly in the frontal and temporal lobes [14, 16] , and that the effect of WMH volumes on cognition is mediated by atrophy [15, 18, 21] . Our findings may imply that imaging-visible WM lesions exert secondary widespread and neurodegenerative effects on the grey matter by damaging specific WM tracts that subserve these cortical regions. The (anterior) temporal lobe may be especially vulnerable to atrophy in WMHenriched dementia cohorts because of its susceptibility to both AD-and vascular-related damage [44] and the potential vulnerability of connecting tracts to WMH crossing, such as the uncinate fasciculus [45] .

Unlike previous studies, we employed PET-based biomarkers of Aß and glucose metabolism in addition to atrophy as potential mediators in the model. This allowed us to investigate whether effects of WMH volumes on cognition were also indirectly promoted by Aß or glucose metabolism. Indeed, additional indirect paths via Aß followed by neurodegeneration (metabolism or atrophy) mediated the WMH-cognition relationship. These indirect paths through Aß, however, had lower effect size, suggesting that WMH and AD pathophysiology exert mostly additive effects on cognitive impairment. The more prominent additive effects were also underscored by our alternative path model, where Aß affected cognition mostly through cortical atrophy rather than through WMH. Our findings are in line with prior studies reporting additive effects of WMH and AD pathology on cognitive decline at baseline and longitudinally [46, 47] , as well as a lower degree of AD pathology in the presence of vascular pathology for the same level . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint of cognitive impairment [48] and a substantial increase in dementia risk post-stroke [49] . Importantly, apart from additive effects, we also show that the indirect path through Aß (i.e., 'WMHàAßà(..)àcognition') was significant. As such, subjects with elevated WMH burden may represent an at-risk group not only for increased neurodegeneration but also for Aß deposition [24] . Thus, subjects with vascular co-pathology may represent an important target group for dementia prevention when vascular risk factors can be treated alongside AD hallmarks.

In relation to cognition, we showed that higher volumes of WMH correlated with slower processing (psychomotor) speed and impaired executive function [11] . While processing speed was directly impacted by WMH burden, we found that executive function was impacted only indirectly mainly through cortical atrophy. Another strong correlate of WMH in our study was semantic fluency, one of the earliest domains to become impaired in AD [40] and sensitive to discriminate AD/vascular dementia from normal aging [50] . Interestingly, we found that neurodegeneration of the temporal/AD-signature region was a stronger mediator in the WMHfluency relationship than in the WMH-executive relationship. Indeed, semantic fluency is sustained primarily by the left temporal lobe, with language processing being a critical component for this task [45] , while executive function is thought to be predominantly frontal-mediated (not being a region-of-focus in the current study).

While numerous studies reported brain-behaviour relationships using total WMH volumes, only a few focused on the role of regional (lobar) WMH. Our exploratory analyses support the hypothesis that regional WMH burden may differentially affect cognition [43] , and highlighted that this relationship may depend on Aß status. Specifically, frontal WMH volumes were associated with poorer processing speed, semantic fluency, and executive function, particularly in the normal (Aß-negative) subjects. Previous research likewise found an association of frontal . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint WMH volumes with the so-called dysexecutive syndrome in normal elderly subjects [43] .

Conversely, cingulate and temporal WMH volumes played a more prominent role in the Aßpositive subjects in our study. Indeed, emerging evidence has shown vulnerability of the cingulate and temporal areas in the relationship between AD and vascular pathology. For example, Van Westen and colleagues [9] reported an association between temporal WMH and elevated Aß using both fluid and PET-based biomarkers in AD dementia. Second, McAleese et al. [51] reported that parietal/cingulate WMH are related to AD rather than to SVD. While the latter authors concluded that WMH is primarily a consequence of Wallerian degeneration triggered by cortical AD-type pathology itself, other studies have supported the idea that SVD exacerbates AD-related pathology by inducing neuroinflammatory responses and by reducing the clearance of toxic proteins (including Aß) from the brain [26] . In fact, our group previously showed that collagenosis of the large deep medullary veins is strongly associated with periventricular WMH burden in patients with and without AD [52] . Longitudinal studies may help to establish potential causal relationships between WMH and AD pathology.

The present study had several limitations. First, our study had relatively small sample size (n=120), due in part to recruitment of a unique cohort of "real-world" patients who demonstrated moderate-to-severe WM disease. Therefore, we limited the number of mediation analyses by investigating 1) all subjects combined and 2) using total (rather than lobar) WMH volumes.

Second, we did not include a marker of hyperphosphorylated tau, which is expected to be associated with atrophy. The relationship between tau pathology and vascular burden remains debated in literature, with most of the in-vivo biomarker studies not reporting an association [53] .

Furthermore, we found a sustained association between WMH volumes and cognitive deficits in the Aß-negative subjects who are unlikely to have substantial AD-related tau pathology (see . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Table 2 ), indicating that the impact of WMH on cognition may not strongly depend on AD-related tauopathy.

The strengths of this study lie in the use of a broad dynamic range of WMH burden and cognitive scores by the inclusion of mixed SVD and AD pathologies, the addition of PET-based biomarkers of Aß and glucose metabolism, as well as the use of optimized segmentation tools to determine WMH volumes and cortical thickness. In conclusion, our study suggests that increased WMH volumes affect cognitive impairment mainly through cortical atrophy and to a lesser extent through alterations in Aß deposition and glucose metabolism.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint based on a two-way ANOVA that investigated whether lobar location and group influenced WMH burden [54] . Values were adjusted for multiple comparisons and age, sex, and education ( * P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001). Abbreviations: FAQ, functional assessment questionnaire;

MoCA, Montreal cognitive assessment; WMH, white matter hyperintensity volumes.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint [32, 37, 39] . Values are indicated as mean ± SE and 95% CI are . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint bootstrapped with 5,000 replications. Path c represents the total (direct + indirect) effect adjusted only for covariates, whereas c' represents the direct effect adjusted for covariates and indirect effects.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint as mean ± SE and 95% CI are bootstrapped with 5,000 replications. Path c represents the total (direct + indirect) effect adjusted only for covariates, whereas c' represents the direct effect adjusted for covariates and indirect effects.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint volume-weighted averages of AD-signature regions [37, 39] .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 27, 2021. ; https://doi.org/10.1101/2021.12.24.21267786 doi: medRxiv preprint

Vascular Risk Factors and Leuko-Araiosis

Cerebral white matter disease is associated with Alzheimer pathology in a prospective cohort

Periventricular white matter hyperintensities and the risk of dementia: a CREDOS study

Amyloid Burden, Neuroinflammation, and Links to Cognitive Decline After Ischemic Stroke

Associations between amyloid β and white matter hyperintensities: A systematic review

Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration

Alzheimer's Disease Neuroimaging Initiative. White matter hyperintensity burden in elderly cohort studies: The Sunnybrook Dementia Study, Alzheimer's Disease Neuroimaging Initiative, and Three-City Study

Clinical Correlates of White Matter Findings on Cranial Magnetic Resonance Imaging

Elderly People

Cerebral white matter lesions -associations with Aβ isoforms and amyloid PET

Contribution of cerebrovascular disease in autopsy confirmed neurodegenerative disease cases in the National Alzheimer's Coordinating Centre

The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta-analysis

Structural network connectivity and cognition in cerebral small vessel disease: Structural Network and Cognition

Network Efficiency Mediates the Relationship Between Vascular Burden and Cognitive Impairment

White matter lesions and brain gray matter volume in cognitively normal elders

Cognitive impairments associated with periventricular white matter hyperintensities are mediated by cortical atrophy

White matter hyperintensities and imaging patterns of brain ageing in the general population

White matter hyperintensities are associated with disproportionate progressive hippocampal atrophy

The effect of white matter hyperintensities on cognition is mediated by cortical atrophy

The Effects of Longitudinal White Matter Hyperintensity Change on Cognitive Decline and Cortical Thinning over Three Years

Reduced glucose uptake and Aβ in brain regions with hyperintensities in connected white matter

The effect of white matter hyperintensities on verbal memory: Mediation by temporal lobe atrophy

Alzheimer's Disease Neuroimaging Initiative. The effect of white matter hyperintensities on neurodegeneration in mild cognitive impairment

The temporal relationships between white matter hyperintensities, neurodegeneration, amyloid beta, and cognition

White matter hyperintensities predict amyloid increase in Alzheimer's disease

White matter hyperintensities are associated with subthreshold amyloid accumulation

Cerebral Small Vessel Disease and Alzheimer's Disease: A Review

Early role of vascular dysregulation on late-onset Alzheimer's disease based on multifactorial data-driven analysis

The effect of white matter hyperintensity volume on brain structure, cognitive performance, and cerebral metabolism of glucose in 51 healthy adults

The Use of Random Forests to Classify Amyloid Brain PET

The Canadian Dementia Imaging Protocol: Harmonizing National Cohorts

Cortical Thickness Estimation in Individuals With Cerebral Small Vessel Disease, Focal Atrophy, and Chronic Stroke Lesions

A largescale comparison of cortical thickness and volume methods for measuring Alzheimer's disease severity

Deep Bayesian networks for uncertainty estimation and adversarial resistance of white matter hyperintensity segmentation. Neuroscience

Lesion Explorer: A Video-guided, Standardized Protocol for Accurate and Reliable MRI-derived Volumetrics in Alzheimer's Disease and Normal Elderly

Improved Segmentation of the Intracranial and Ventricular Volumes in Populations with Cerebrovascular Lesions and Atrophy Using 3D CNNs

Reducing between scanner differences in multi-center PET studies

11C PiB and structural MRI provide complementary information in imaging of Alzheimer's disease and amnestic mild cognitive impairment

Associations between cognitive, functional, and FDG-PET measures of decline in AD and MCI

Semantic loss marks early Alzheimer's disease-related neurodegeneration in older adults without dementia

Association between white matter lesions and cerebral glucose metabolism in patients with cognitive impairment

Tracking pathophysiological processes in Alzheimer's disease: an updated hypothetical model of dynamic biomarkers

Cerebral white matter hyperintensities in the prediction of cognitive decline and incident dementia

The Hippocampal Neuro-Glio-Vascular Network: Metabolic Vulnerability and Potential Neurogenic Regeneration in Disease

The tract terminations in the temporal lobe: Their location and associated functions

Vascular and amyloid pathologies are independent predictors of cognitive decline in normal elderly

Cognitive Profile of Amyloid Burden and White Matter Hyperintensities in Cognitively Normal Older Adults

The overlap between vascular disease and Alzheimer's diseaselessons from pathology

Interaction Between a Rat Model of Cerebral Ischemia and β-Amyloid Toxicity

Diagnostic utility of abbreviated fluency measures in Alzheimer disease and vascular dementia

Parietal white matter lesions in Alzheimer's disease are associated with cortical neurodegenerative pathology, but not with small vessel disease

Collagenosis of the Deep Medullary Veins: An Underrecognized Pathologic Correlate of White Matter Hyperintensities and Periventricular Infarction?

White matter hyperintensities: relationship to amyloid and tau burden

Cerebral amyloid burden is associated with white matter hyperintensity location in specific posterior white matter regions

Mediation analyses of WMH on processing speed, global function and global cognition, using temporal meta-ROIs

Alternative mediation models tested

Inclusion and exclusion criteria for MITNEC-C6

Demographics in the MITNEC-C6 and ADNI cohorts

Description of cognitive tests and functional activities

Description of MR imaging parameters

Table S5. Lobar WMH volumes vs cognitions

We would like to express our deepest gratitude towards all the participants