Choroid plexus volume in pathology-confirmed Alzheimer's disease.
The journal of prevention of Alzheimer's disease · 2026
Abstract originale
INTRODUCTION: The choroid plexus (CP) increases in volume across the Alzheimer's disease (AD) continuum, suggesting its potential as a clearance-related biomarker. However, few studies have examined ante-mortem CP volume in relation to post-mortem AD pathology, the gold standard for diagnosis. METHODS: Participants who had structural magnetic resonance imaging and post-mortem pathology, with an interval of ≤ 5 years between imaging and death, were examined. Normalized CP volume (NCPV) was semi-automatically segmented from the lateral ventricles and analyzed using Bayesian linear regression to estimate associations with cognitive impairment (CI), AD pathology, and relevant clinical/demographic data. RESULTS: Intermediate and high levels of AD pathology and CI were associated with larger NCPV, whereas female sex was associated with lower NCPV. Subgroup analyses showed larger NCPV in individuals with greater CI despite comparable levels of AD pathology. DISCUSSION: These findings link CP enlargement to neuropathologically confirmed AD burden and CI, supporting further investigation of CP structure and function in AD.
Parole chiave
Leggi il testo integrale originale
Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.
Choroid plexus volume in pathology-confirmed Alzheimer's disease
Introduction
The choroid plexus (CP) increases in volume across the Alzheimer’s disease (AD) continuum, suggesting its potential as a clearance-related biomarker. However, few studies have examined ante-mortem CP volume in relation to post-mortem AD pathology, the gold standard for diagnosis.
Methods
Participants who had structural magnetic resonance imaging and post-mortem pathology, with an interval of ≤ 5 years between imaging and death, were examined. Normalized CP volume (NCPV) was semi-automatically segmented from the lateral ventricles and analyzed using Bayesian linear regression to estimate associations with cognitive impairment (CI), AD pathology, and relevant clinical/demographic data.
Results
Intermediate and high levels of AD pathology and CI were associated with larger NCPV, whereas female sex was associated with lower NCPV. Subgroup analyses showed larger NCPV in individuals with greater CI despite comparable levels of AD pathology.
Discussion
These findings link CP enlargement to neuropathologically confirmed AD burden and CI, supporting further investigation of CP structure and function in AD.
1Introduction
Alzheimer's disease (AD) is a progressive neurodegenerative disorder and the leading cause of dementia worldwide. The onset and progression of AD is pathologically characterized by the accumulation of amyloid-β (Aβ) plaques and neurofibrillary tangles [1]. Increasing evidence suggests that impaired clearance of these proteins, rather than their overproduction alone, plays a critical role in disease onset and progression [2]. Multiple clearance pathways have been implicated in AD, including transport across the blood–brain and blood–cerebrospinal fluid (CSF) barriers, intramural periarterial drainage, and the glymphatic system. Among these, CSF dynamics are thought to play an integral role, with disruptions in CSF production and/or flow contributing to the accumulation of AD pathology [3,4].
The choroid plexus (CP) is a highly vascularized epithelial structure located within the brain ventricles, which serves as the primary source of CSF production [[5], [6], [7], [8]]. Beyond CSF secretion, the CP is essential for formation of the blood-CSF barrier [9], immune surveillance [8], and active clearance of neurotoxic compounds from the CSF [10]. In AD, both the structure and function of the CP appear to be compromised, leading to alterations in secretory, barrier, and transport functions [6]. Such dysfunction has been linked to reduced Aβ clearance and impaired CSF turnover, processes that may contribute to downstream pathological accumulation. Preclinical studies have demonstrated a key role for the CP in Aβ removal and have further shown that this function is disrupted in AD models [11,12]. Consistent with these findings, post-mortem studies in individuals with AD have reported pronounced morphological alterations of the CP, including structural remodelling and accumulation of aberrant metabolites [6,13,14].
Neuroimaging studies have further examined CP volume across the cognitive continuum. Collectively, these studies show increased CP volume in individuals with dementia compared with cognitively normal controls [[15], [16], [17], [18], [19]]. Similarly, the onset and progression of CI in the AD spectrum has also been linked to increased CP volumes [6,15,16,18]. Elevated CP volume has also been linked to other established neuroimaging markers of AD pathology, including amyloid & tau positivity on positron emission tomography, hippocampal atrophy, and ventricular enlargement [[19], [20], [21]].
Despite these convergent lines of research, few studies have directly examined the relationship between in vivo CP volume and neuropathologically confirmed AD, the gold standard for diagnosis. The present study examined whether CP volume, measured during life, is associated with (1) the severity of post-mortem AD pathology and (2) the degree of CI. It was hypothesized that greater AD pathology burden and more severe CI are each associated with increased CP volume. To further clarify the relationship between CP volume and cognition, CP volume was also examined across individuals with comparable levels of AD pathology but differing levels of CI. It was hypothesized that, even within similar pathological strata, higher CP volume is associated with worse CI.
2Methods
2.1Source population
A retrospective analysis was conducted on patient data collected from the National Alzheimer's Coordinating Center (NACC), which included both clinical assessments and MRI data. Briefly, the NACC consists of multiple institute-funded AD research centers across the United States responsible for recruiting and collecting data on participants across the aging and AD spectrum. These participants make up the NACC Uniform Data Set (UDS), which contains data across multiple patient domains including standardized clinical, demographic and cognitive data [22,23]. A subset of these participants also underwent structural magnetic resonance imaging (MRI), as described below (see MRI Acquisition & Processing).
Neuropathological data were obtained from the NACC Neuropathology Data Set (NDS), which includes standardized post-mortem assessments of multiple neuropathological features, including Alzheimer’s disease-related neuropathological change (ADNC) [24]. Data from the UDS and NDS were merged and assessed for eligibility. The UDS, NDS and associated neuroimaging data is obtainable from the NACC via request.
Informed consent was obtained from participants and their informants at the contributing Alzheimer’s Disease Research Centers. All study protocols were approved by the University of Washington Institutional Review Board. Detailed descriptions of the data collection procedures, instruments, and variable definitions for the UDS and NDS have been published previously [25].
3Inclusion & exclusion criteria
UDS and NDS data available as of September 2024 were examined for eligibility. Inclusion criteria were: (1) availability of neuropathological data, including classification of ADNC; (2) availability of a 3 Tesla T1-weighted MRI; (3) an interval of ≤ 5 years between T1-weighted MRI acquisition and death/post-mortem examination; (4) age > 60 years at the time of MRI; and (5) availability of clinician-rated cognitive status at the study visit corresponding to the MRI.
To restrict analyses to late-onset AD, participants with primary neuropathological diagnoses other than AD were excluded. These included frontotemporal lobar degeneration and related subtypes (e.g., Pick’s disease, corticobasal degeneration, progressive supranuclear palsy, etc.), multiple system atrophy, prion disease, trinucleotide repeat disorders, neoplasms, primary white matter disease, and infectious processes (abscess, encephalitis, etc.). Definitions, identification criteria, and staging procedures for these pathologies are described in the NACC Neuropathology Data Set [24].
Given that AD pathology frequently co-occurs with other neurodegenerative and aging-related pathologies [26], participants with ADNC were retained in the sample if comorbid cerebrovascular pathology, Lewy body pathology, age-related tau astrogliopathy (ARTAG), and/or TAR DNA-binding protein 43 (TDP-43) pathology were present. Participants meeting these criteria were subsequently assessed for MRI quality.
4Pathological assessments
Neuropathological reporting of AD (obtained from the NDS) was reported using semi-quantitative ABC criteria consisting of Thal staging (criteria A) [30], Braak staging (criteria B) [27] and the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) (criterion C) [28].
Together, these criteria were combined to classify ADNC burden as “Not ADNC”, “Low”, “Intermediate” and “High”. Definitions for each pathological stage are described elsewhere [29].
For analyses in this study, the “Not” and “Low” ADNC categories were combined and referred to as “Slight ADNC.” This grouping was implemented to account for the relatively small number of participants with low pathological burden compared with the Intermediate and High ADNC groups.
5Clinical assessment
Cognitive status was assessed based on clinician-determined criteria described in the UDS [25]. In brief, normal cognition was defined as follows: global Clinical Dementia Rating scale = 0 [30]; neuropsychological testing within normal ranges and normal behaviour. Mild CI (MCI) was defined as: clinician-observed impairment in one or more cognitive domains (e.g., memory, language, executive function, and visuospatial skills); concern (by clinician or co-participant) of a change in cognition from previous levels; and currently preserved independence in functional abilities. Dementia was defined as: presence of cognitive and/or behavioural symptoms, decline in previous levels of functioning; interference with the ability to function at usual activities and/or work; CI detected/diagnosed through previous clinical history and not explained by delirium or another major psychiatric disorder. Across the sample, available instances of the completed Mini-Mental Status Examination (MMSE) [31] were also examined for exploratory sub-analysis.
Considering the vascular involvement of the CP [19,32,33], cardiovascular risk factors were also assessed. Body mass index (BMI), presence of hypertension and presence of hypercholesterolemia were extracted from the UDS. Hypertension and hypercholesterolemia were reported categorically as either recent/active, remote/inactive or absent. Additionally, impaired sleep, given its reported association with relevant clearance mechanisms in AD [34], [35] and CP-related dysfunction [[36], [37], [38]], was also examined. Sleep disturbance was operationalized as present or absent based on the Neuropsychiatric Inventory Questionnaire (NPI-Q), an informant-based instrument assessing the presence and severity of neuropsychiatric symptoms [39]. All clinical and demographic variables were obtained from the participant visit corresponding to MRI acquisition and therefore were collected within ≤ 5 years of death and post-mortem neuropathological assessment.
6MRI acquisition & processing
T1-weighted MRI scans were acquired using multiple imaging protocols and MRI systems across participating AD research centers. All images included in the present study underwent independent manual quality assessment by two raters to evaluate image suitability for analysis, including the presence of motion artifacts, focal brain lesions or injuries, and anatomical abnormalities that could compromise image processing (per criteria provided by Backhausen and colleagues [40]). Image inspection was performed using an MRI viewer (Insight Segmentation and Registration Toolkit, ITK-SNAP) [41]. See APPENDIX 1 for further information about visual inspection. Structural images that passed quality control were processed using FreeSurfer (version 6.0). The standard recon-all pipeline was applied to resample images to 1 mm isotropic voxel resolution and to perform automated atlas-based cortical and subcortical parcellation using the Desikan–Killiany atlas [42].
6.1CP segmentation
Estimated total intracranial volume (eTIV) and the CP volume from the lateral ventricles were obtained using the FreeSurfer recon-all pipeline. Although FreeSurfer provides automated CP segmentation, this approach has been shown to be inaccurate in some cases [20]. To improve segmentation accuracy, a lightweight Gaussian Mixture Model (GMM)–based algorithm was applied to the FreeSurfer-derived CP segmentation, an approach that has demonstrated improved performance relative to FreeSurfer alone [20] .All resulting CP segmentations were subsequently manually reviewed and corrected by a single rater, blinded to the clinical/pathological characteristics of each participant, to address cases of under- or over-segmentation using the Insight Segmentation and Registration Toolkit (ITK-SNAP) [41] (Fig. 1). Manual corrections involved the systematic inspection of voxels adjacent to the lateral ventricles that may have been incorrectly labeled as CP. This included neighboring white and gray matter structures such as the body and splenium of the corpus callosum, fornix, hippocampus, thalamus, vertical portion of the caudate nucleus, and surrounding cerebral white matter (see APPENDIX 1 for more information). Final CP volumes were calculated from the corrected segmentations and divided by eTIV to provide a normalized CP volume (NCPV) which accounts for interindividual differences in head size.
6.2Statistical analysis
To examine the potential effects of cognition and AD path burden on NCPV, a Bayesian linear regression model was employed. Compared with more commonly used frequentist methods, the Bayesian framework allows for intuitive interpretation of results by reporting the probability of an effect for the proposed model, given the observed data. This framework integrates relevant priors and model likelihoods within a single flexible estimation process, without the strict assumptions (e.g., sample size, normality, homoscedasticity) that are often required of frequentist linear regression models. The use of weakly informative priors also helps to constrain effect sizes to plausible ranges and to stabilize estimates in smaller subgroups. Together, these features make the Bayesian framework well suited for examining subtle relationships among NCPV, AD pathology, and cognition.
Models were fit using Hamiltonian Monte Carlo sampling implemented in the brms package in R (version 4.3.1), with four chains of 4000 iterations each (including warm-up). Posterior convergence was assessed using standard diagnostic checks and visual inspection of sampling traces. Posterior draws were processed and visualized using tidybayes and ggplot2. Posterior distributions were summarized using posterior medians; 90% highest density intervals (HDIs); and the probability of direction (PD), defined as the posterior probability that an effect was greater (or less) than zero. Group-level marginal means and pairwise contrasts for categorical predictors were computed using emmeans extended to the Bayesian framework. Based on research aims and hypotheses, Bayesian linear regression models were specified with normally distributed likelihoods and weakly informative priors. Two models were implemented.
The first model evaluated the effects of ADNC, cognitive status, presence or absence of CVD risk factors, sleep disturbances, age and sex on NCPV:(1)NormalizedCPVolumei=β1(ADPathologyBurdeni)+β2(CognitiveStatusi)+β3(Agei)+β4(Sexi)+β5(CVDRiskFactorsi)+β6(SleepDisturbancesi),where i indexes the participants and coefficients β1- β6 represent the regression coefficients for each predictor in the model. Similarly, the second model was designed to assess the correlates of cognitive test performance (via the MMSE scores) in AD among participants with comparable levels of AD pathology:(2)NormalizedCPVolumei=β1(MMSEi)+β2(Agei)+β3(Sexi)+β4(CVDRiskFactorsi)+β5(SleepDisturbancesi).
Here, age, sex, the presence or absence of CVD risk factors and the presence or absence of sleep disturbances were modelled alongside MMSE scores to gauge the relationship to CI within MCI and dementia cohorts. Participants for this analysis were restricted to having high ADNC burden, controlling for additional variability in AD pathology while while retaining sufficient observations for exploratory subgroup analyses..
NCPV and non-categorical predictors were standardized (via Z-scoring - subtracting the sample mean and dividing by the sample standard deviation) before model fitting to place parameters on a common scale. Based on the distribution of these standardized values, weakly informative Normal priors were placed on regression coefficients (β∼N(0,2.5)). Default brms Student-t priors were used for the intercept (centered on the outcome mean in each analysis subset) and a half-Student-t prior for the residual standard deviation. For this study, the threshold for a credible effect was set at PD ≥ 85%. To ensure adequate sample representation, group comparisons were restricted to cognitive-pathological categories with ≥ 5 participants (Table 1). Posterior medians, HDIs and corresponding PD values for all variables are provided in the Supplementary Materials.Table 1Clinical, Pathological & Imaging Characteristics across the Sample.
Table 1 dummy alt textClinical & Imaging Characteristics AD Pathological Burden
Slight (n = 18) Intermediate (n = 23) High (n = 70)
Age at Death (mean ± SD) 78.6 ± 9 83.7 ± 5.7 77.2 ± 8.2
Females (n, % in pathology group) 4 (22%) 9 (39%) 26 (37%)
Active Hypertension (n, % in pathology group) 7 (39%) 9 (39%) 28 (40%)
Active Hypercholesterolemia (n, % in pathology group) 6 (33%) 13 (57%) 27 (39%)
Body Mass Index (mean ± SD) 28.6 ± 4.4 25.6 ± 3.8 26.1 ± 4.4
Presence of Sleep Disturbance (n, % in pathology group) 5 (28%) 5 (22%) 19 (27%)
Clinician-determined Cognitive Status
Cognitively Intact (n, % in pathology group) 12 (67%) 6 (26%) 5 (7%)
Mild Cognitive Impairment (n, % in pathology group) 2 (11%) 11 (48%) 17 (24%)
Dementia (n, % in pathology group) 4 (22%) 6 (26%) 48 (69%)
Imaging Findings Z Score
Normalized Choroid Plexus Volume (mean ± SD) −0.58 ± 1.00 0.1 ± 1.07 0.11 ± 0.96
Values are expressed as mean ± standard deviation or count (percentage) within AD pathological burden group (slight, intermediate, high).
7Results
The final sample consisted of 111 participants, for whom NCPV values were evaluated in relation to cognition, AD-related pathology and other demographic and clinical variables (Table 1). Participants were stratified into three AD pathological burden groups: slight (n = 18), intermediate (n = 23), and high (n = 70). Age at death did not differ markedly between groups, with mean values ranging from 77 to 84 years. Female representation was lower in the slight pathology group (22%) than in intermediate (39%) and high (37%) pathology groups. The prevalence of CI (i.e., the presence of MCI or dementia) increased with higher AD pathology burden, from 33% in slight ADNC to 93% in high ADNC. Additional demographic and clinical variables are summarized in Table 1. Results for the most salient findings of CP-related effects are summarized in Fig. 2, with additional distributions, effect sizes and comparisons listed in the Supplementary Materials.
7.1Main effects of AD pathology and cognitive impairment on CP volume
The NCPV was found to be increased in individuals with intermediate ADNC (median = 0.63, 90% HDI [0.009] – [1.26], PD = 90%) and high ADNC (median = 0.59, 90% HDI [−0.008] – [1.20], PD = 90%) compared to those with slight ADNC (Fig. 2). Pairwise comparisons showed no credible differences between high and intermediate ADNC (median = 0.04, 90% HDI [−0.30] – [0.38], PD = 56%) (Fig. 3A).
Additionally, individuals with MCI showed higher NCPV (median = 0.461, 90% HDI [−0.03] – [0.96], PD = 88%) when compared to cognitively intact individuals (Fig. 2). Participants with dementia showed increased NCPV compared to cognitively intact individuals though the probability of direction did not meet credibility thresholds (median = 0.38, 90% HDI [−0.13] – [0.89], PD = 84%) Pairwise comparisons between MCI and dementia showed no credible differences in NCPV (median = 0.08, 90% HDI [−0.24] – [0.41], PD = 63%) (Fig. 3B). Biological sex exhibited the largest median effect on NCPV where females had lower NCPVs compared to males (median = −0.69, 90% HDI [−0.962] – [−0.42], PD > 99%). Age, CVD factors and sleep did not show any credible effects on NCPV across the overall sample (see Supplementary Materials for effect sizes and probability estimates; S2-S4). To further interrogate observations, sensitivity analyses were also performed (Supplementary Materials; S7-S12 & S15-S22, S24-S27). This included: exploring alternate model priors; comparing results to more simplified regression models; and modelling the effects of additional variables on CP volumes such as time interval between MRI and death, individual effects of co-pathologies, severity of co-pathologies (i.e., CAA) and lateral ventricular volume as a covariate.
Summarizing the results of these sensitivity analyses, the observed effects for intermediate ADNC, high ADNC, females and MCI were mostly retained with minor variations in effect sizes and PD ≥ 85% across sensitivity analyses involving altered priors and the addition of the interval between MRI and death as a covariate. Similar findings were observed when individually assessing the presence of CAA, TDP-43, ARTAG and Lewy bodies (See S15, S18-S20 in Supplementary Materials).
7.2Exploring CP volume with varying cognitive impairment in high AD pathology
A subset of participants (n = 10 with MCI, n = 23 with dementia) with available MMSE scores and high ADNC were further examined in terms of associations between NCPV and CI in a cohort with comparable burden of AD pathology. Among participants with MCI and high ADNC (Table 2), no credible associations were observed between NCPV and MMSE scores (median = −0.03, 90% HDI [−0.24] – [0.11], PD = 64%).Table 2Modelling the Association between NCPV and Varying CI within a Sub-group of MCI with High AD Pathology.
Table 2 dummy alt textNCPV & Effect of Varying CI in Participants with MCI with High AD Pathology (n = 10)
Parameter Age at Death MMSE Score Females Body Mass Index Active Hypertension Active Hypercholesterol-emia Active Sleep Disturbances
Descriptive Statistics 79.1 yrs ± 7.3 27 ± 2.1 3 (30%) 27.4 ± 3.3 5 (50%) 4 (40%) 1 (10%)
Effect Size (PD) −8.5 × 10–3(56%) -0.03 (64%) −3.76 (97%) 0.2 (76%) −2.2 (96%) −0.55 (90%) 1.27 (87%)
Descriptive statistics are expressed as mean ± standard deviations for continuous variables and counts (percentages) for categorical. Effect sizes are reported as posterior medians, with corresponding probabilities reported as PD. Bolded effect size values indicate credible findings (PD ≥ 85%).
Normalized Choroid Plexus Volume (NCPV), Cognitive Impairment (CI), Mini-Mental State Examination (MMSE), Mild Cognitive Impairment (MCI), Alzheimer's Disease (AD), Probability of Direction (PD).
Like the overall sample, biological females exhibited lower NCPV compared to biological males (median = −3.76, 90% HDI [−4.82] – [−1.49], PD = 97%). Participants with recent/active hypertension and hypercholesterolemia also exhibited lower NCPV compared to those with remote/inactive hypertension (median = −2.23, 90% HDI [−3.71] – [−0.94], PD = 96%) and remote/inactive hypercholesterolemia (median = −0.55, 90% HDI [−1.36] – [−0.005], PD = 90%), respectively. The presence of sleep disturbances was also associated with higher NCPV (median = 1.27, 90% HDI [−0.27] – [2.09], PD = 87%). Although these associations showed high probabilities of effect, they should be interpreted cautiously due to the limited sample size.
Twenty-three participants with dementia, high ADNC and reported MMSE scores were similarly modeled (Table 3). In this group, MMSE scores were negatively associated with NCPV such that lower MMSE scores, indicating greater CI, were associated with higher NCPV (median = −0.11, 90% HDI [−0.22] – [0.00], PD = 90%). Older age was also associated with a modest increase in NCPV (median = 0.07, 90% HDI [0.01] – [0.12], PD = 96%). Participants with active hypercholesterolemia demonstrated credible associations with lower NCPV (median = −0.71, 90% HDI [−1.36] – [0.02], PD = 91%). Lastly, sleep disturbances were credibly associated with increased NCPV (median = 0.71, 90% HDI [−0.05] – [1.50], PD = 88%). Although female sex was negatively associated with NCPV (median = −0.46, 90% HDI [−1.33, 0.45], PD = 75%), consistent with the overall results, the effect did not meet the established credibility threshold and was likely attributable to the small sample size.Table 3Modelling the Association between NCPV and Varying CI within a Sub-group of Dementia with High AD Pathology.
Table 3 dummy alt textNCPV & Effect of Varying CI in Participants with Dementia with High AD Pathology (n = 23)
Parameter Age at Death MMSE Score Females Body Mass Index Recent/Active Hypertension Recent/Active Hypercholesterol emia Recent Sleep Disturbances
Descriptive Statistics 74 yrs ± 8.1 21 ± 3.6 4 (17%) 27 ± 4.7 11 (48%) 9 (39%) 6 (26%)
Effect Size (PD) 6.7×10–2(96%) −0.11 (90%) −0.46 (75%) −0.02 (59%) −0.18 (63%) −0.71 (91%) 0.71 (88%)
Descriptive statistics are expressed as mean ± standard deviations for continuous variables and counts (percentages) for categorical. Effect sizes are reported as posterior medians, with corresponding probabilities reported as PD.
Normalized Choroid Plexus Volume (NCPV), Cognitive Impairment (CI), Mini-Mental State Examination (MMSE), Mild Cognitive Impairment (MCI), Alzheimer's Disease (AD), Probability of Direction (PD).
8Discussion
The present study explored relationships between CP volume, AD pathology (via ADNC determined post-mortem), and CI. Overall, intermediate and high AD pathology were associated with a high probability (PD = 90%) of larger CP volumes. Similarly, participants with MCI also showed a high probability (PD = 88%) of larger CP volumes compared to cognitively intact controls, along with near-credible effects (PD = 84%) of larger CP volumes in the dementia group. Additional exploratory analyses probed the relationships between CP volume and CI among participants with comparable levels of AD pathology. Among participants with dementia and high AD pathology, lower MMSE scores were related to larger CP volumes, indicating that CP enlargement may be more pronounced at more severe levels of CI. Across the overall sample, biological females had smaller CP volumes compared to biological males (PD = 99%), suggesting a sex-based effect despite normalization to eTIV.
With respect to AD pathology, CP volume has been examined in relation to blood biomarkers, CSF measures, and PET. Associations between CP volume and amyloid deposition via PET [15], tau deposition via PET 21, CSF levels of amyloid [43], and plasma biomarker presence of amyloid [44] have been reported. Collectively, these studies support the notion that CP volume is associated with increased AD-related pathology observed in vivo; however, these studies largely rely on ante-mortem biomarkers of pathology. The present study is the first to explore the relationship between in vivo CP volume and post-mortem AD pathological burden in the same participants. Both intermediate and high ADNC groups showed a high probability (PD = 90%) of larger CP volume compared to the slight ADNC group, though differences between intermediate and high ADNC were not detected (PD = 56%). A relationship between intermediate ADNC and CP volume suggests that CP volume enlargement may occur early in the development of AD.
In AD and other neurodegenerative diseases, an increase in CP volume is typically interpreted as a marker of CP dysfunction. However, the precise processes underlying the changes in CP volumes remain an active area of investigation. Recent histological analyses of the CP demonstrate aberrant accumulation of inflammatory signals and metabolites among the CP of AD patients compared to older adult controls [6]. The onset and progression of AD have also been observed to cause alterations in CP morphology via epithelial atrophy and thickening of the basement membrane leading to remodelling of the CP tissue [14,45]. The increased aggregation of amyloid and Tau also propagate additional injury-related cascades that further drive the changes in the CP basement layer, causing thickening of the CP basement membrane and irregularities in CP morphology, thereby altering CP function [7]. However, there has yet to be work done to directly relate mechanisms of CP alterations at the cellular level to broader in vivo changes in CP volume [7].
In parallel with relationships to AD pathology, several studies report larger CP volumes with greater CI (i.e., MCI and dementia) compared to cognitively normal controls [16,19,46]. Findings from the present study are consistent with these reports. Increased CP volumes were observed in the MCI group and neared credibility thresholds in the dementia group, suggesting a ceiling effect that arises during the onset and progression of CI. The absence of credible differences between the MCI and dementia groups further supports this interpretation.
The accumulation of AD pathology, and the myriad physiological consequences that follow are key factors in the onset and progression of CI [47,48]. However, individuals presenting with CI may harbor varying levels of AD pathology and individuals with similar pathological burden may follow different cognitive trajectories [49,50]. In the context of CP volume, heterogeneity in AD pathology and CI complicates the interpretation of whether CP enlargement primarily reflects pathological burden and/or additional physiological processes that contribute to cognitive decline. To further probe these relationships, additional analyses were conducted in a small subgroup of participants with comparable AD pathological burden but different levels of CI based on their MMSE scores. Among participants with dementia and high AD pathology, poorer cognition (indexed by lower MMSE scores) showed a high probability (PD = 90%) of increased CP volume. However, this relationship was not observed in the MCI group. These findings suggest that CP enlargement may reflect processes contributing to CI beyond AD pathology. However, this interpretation remains strictly speculative given the small subgroup sizes (n = 23 for dementia + high ADNC; n = 10 for MCI + high ADNC) and the limited sensitivity of the MMSE for detecting subtle cognitive deficits in MCI or early dementia [51,52].
Subgroup analyses revealed that sleep and cardiovascular variables demonstrated credible associations with CP volume not observed in the overall study sample. Specifically, the presence of sleep disturbances showed a positive association to CP volume among both the MCI (PD = 87%) and dementia (PD = 88%) sub-groups with high AD pathology. CP volume enlargement in the context of sleep impairments is consistent with emerging research examining the relationships between CP dysfunction, CI, and impaired clearance mechanisms related to sleep disruption [37,38,53]. The present findings, although limited in their generalizability because of the limited sample size, suggest a possible interplay between CP dysfunction, CI, and impaired clearance and supports future work explicitly examining these relationships in larger, more, representative samples.
Recently, CVD risk factors have also been increasingly examined alongside CP volume due to the vascular involvement of the CP [19,32,33]. Current analyses did not observe credible findings between CP volume and BMI, presence of hypertension or the presence of hypercholesterolemia across the overall sample. However, subgroup analyses showed that the presence of hypertension (PD = 96%) and hypercholesterolemia (PD = 90%) did associate with lower CP volumes in the MCI /high ADNC group whereas hypercholesterolemia alone (PD = 91%) was associated with lower CP volumes in the dementia /high ADNC group, calling for further investigation into the potential contributions of CVD risk factors on CP volume across the AD spectrum. While both sleep and CVD showed probable effects towards CP volume, the limited sample size and low prevalence of these variables within the subgroups limit interpretation compared to the findings from the sample at large.
Biological sex showed a highly negative effect on CP volume, despite normalization of CP volume to eTIV. Similar sex-related differences in CP volume have been reported in both cross-sectional and longitudinal studies across the AD continuum [16,54]. In the present study, the sex effect exceeded the median effects for cognitive status and AD pathology, highlighting its importance in evaluating CP physiology in normal aging and AD. However, this finding should be interpreted with caution as the sample was not prospectively matched for either age nor sex. Future studies should consider sex-stratified analyses and alternative normalization strategies beyond eTIV (e.g., lateral ventricular volume, CSF volume, etc.) to assess the robustness of these differences to different study methodologies [55].
A key strength of this study is the direct linkage of in vivo CP volume with post-mortem pathological burden of AD. Combining post-mortem evaluation with ante-mortem neuroimaging anchors interpretations in the gold-standard diagnosis of AD [56]. Neuropathological data provide precise information on the severity of AD pathology and co-pathologies that may present with AD-like symptoms but do not meet criteria for AD. Prior studies of CP volume have relied primarily on ante-mortem biomarkers, which cannot fully resolve these distinctions and may introduce bias. The use of semi-automated CP segmentation with manual correction also enhanced the accuracy of CP segmentation compared to fully automated pipelines, which have been shown to mistakenly over/under-segment regions of the CP [20]. Furthermore, the use of a conservative window of ≤ 5 years applied between MRI and death also increases confidence that post-mortem pathology reflected the disease state at the time of image acquisition [54].
The results of the present study should be interpreted considering several limitations. First, the current sample was unevenly distributed in terms of pathology burden, with a high prevalence of high ADNC (> 60% of sample; see Supplementary Materials for full break down of presence/severity of co-pathologies). This, combined with the convenience sampling used in the NACC dataset may impact the generalizability of the findings. Second, a representative assessment on the presence and severity of relevant co-pathologies (CAA, TDP-43, ARTAG, Lewy bodies, etc.) would provide more realistic observations on the influence of pathologies on CP volume, considering AD pathology rarely occurs in isolation [56] (See Supplementary Materials for sensitivity analyses in the presence of co-pathologies). Nevertheless, the present observations provide additional evidence for relationships between AD pathology and CP volume and support future efforts to further disentangle the effect of various neurodegenerative pathologies on CP volume enlargement.
Small sample sizes in various subgroups also limited analyses of CI, pathology and their influence on CP volume. Still, the Bayesian framework enable analyses of this unique dataset by reporting effect sizes and their associated probabilities for chosen parameters on CP volume, which may have gone undetected using strictly frequentist approaches. The current framework and observations still provide new evidence supporting relationships between CP volume, cognition and pathology, further informing future research in this space.
Lastly, the cross-sectional study design limits inference regarding CP volume changes across stages of increasing pathology. Repeated-measures analyses relating CP volume and CI to post-mortem pathology may be possible with future NACC releases. Future work incorporating longitudinal measures of CP volume in cognitively normal older adults and individuals at early stages of CI (e.g., subjective CI) would be valuable for characterizing CP volume changes across normal aging and early stages of pathological aging. Additionally, longitudinal studies examining CP volume alongside ante-mortem markers of AD pathology (blood-based and/or CSF biomarkers), biomarkers of immune activation, and other relevant neuroanatomical correlates (i.e., ventricular volume) would be helpful in parsing out whether CP-related changes are more sensitive to AD pathology or other physiological processes (e.g., neuroinflammation) implicated in the presence of AD pathology. Given the CP’s role in CSF production and homeostasis, future work examining CP volume and AD pathology alongside multiple measures of CSF dynamics would also provide evidence to interrogate CP volume as a measure of dysfunctional clearance in the aging brain.
9Conclusion
The CP has been observed to increase in volume across the AD continuum. The present findings demonstrate that CP enlargement measured using structural MRI is related to gold-standard AD pathology, clinician-determined cognitive status, and sex. By incorporating post-mortem AD burden, these findings contextualize prior observations of CP enlargement and provide additional evidence that changes in CP structure may be tied to underlying pathological accumulation. Future studies integrating longitudinal imaging and more direct measures of CSF dynamics will be critical for elucidating the mechanistic underpinnings of CP volume changes and their potential relevance to brain clearance mechanisms.
Data statement
Data from the National Alzheimer’s Coordinating Center is publicly available to investigators through a data request process at https://naccdata.org.
Funding sources
None.
Disclosures/conflict of interest
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Consent statement
Research using the NACC data was approved by the University of Washington Institutional Review Board. All contributing AD research centers obtained informed consent from their participants and received institutional review board approval before submitting data to NACC. Research was conducted in accordance with the Declaration of Helsinki.
Declaration of generative artificial intelligence use
N/A
CRediT authorship contribution statement
Francis Fernandes: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Avyarthana Dey: Writing – review & editing. Andy Ma: Software, Methodology. Nathan W. Churchill: Writing – review & editing, Methodology. Corinne E. Fischer: Writing – review & editing, Conceptualization. Simon J. Graham: Writing – review & editing, Methodology. David G. Munoz: Writing – review & editing, Conceptualization. Tom A. Schweizer: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
1Jack C.R Bennett D.A Blennow K. A/T/N: an unbiased descriptive classification scheme for Alzheimer disease biomarkers Neurology 87 5 2016 539 547 10.1212/WNL.0000000000002923 27371494 PMC4970664
2Tarasoff-Conway J.M Carare R.O Osorio R.S Clearance systems in the brain-implications for Alzheimer disease Nat Rev Neurol 11 8 2015 457 470 10.1038/nrneurol.2015.119 26195256 PMC4694579
3Hong H. Chen Y. Tozer D.J MRI markers of cerebrospinal fluid dynamics predict dementia and mediate the impact of cardiovascular risk Alzheimer’s Dement 21 10 2025 e7069910.1002/alz.70699 PMC12547625 41128158
4Kant S. Stopa E.G Johanson C.E Choroid plexus genes for CSF production and brain homeostasis are altered in Alzheimer’s disease Fluids Barriers CNS 15 2018 34 10.1186/s12987-018-0120-7 30541599 PMC6291926
5Balusu S. Brkic M. Libert C. The choroid plexus-cerebrospinal fluid interface in Alzheimer’s disease: more than just a barrier Neural Regen Res 11 4 2016 534 10.4103/1673-5374.180372 27212900 PMC4870896
6Čarna M. Onyango I.G Katina S. Pathogenesis of Alzheimer’s disease: involvement of the choroid plexus Alzheimer’s Dement 19 8 2023 3537 3554 10.1002/alz.12970 36825691 PMC10634590
7Gião T. Teixeira T. Almeida M.R Choroid plexus in Alzheimer’s Disease—the current State of knowledge Biomedicines 10 2 2022 224 10.3390/biomedicines10020224 35203434 PMC8869376
8Liu R. Zhang Z. Chen Y. Choroid plexus epithelium and its role in neurological diseases Front Mol Neurosci 15 2022 15 10.3389/fnmol.2022.949231 PMC9633854 36340696
9Liddelow S.A. Development of the choroid plexus and blood-CSF barrier Front Neurosci 9 2015 10.3389/fnins.2015.00032 PMC4347429 25784848
10Kratzer I. Ek J. Stolp H. The molecular anatomy and functions of the choroid plexus in healthy and diseased brain Biochim Biophys Acta (BBA) - Biomembr 1862 11 2020 18343010.1016/j.bbamem.2020.183430 32750317
11González-Marrero I. Giménez-Llort L. Johanson C.E Choroid plexus dysfunction impairs beta-amyloid clearance in a triple transgenic mouse model of Alzheimer’s disease Front Cell Neurosci 9 2015 10.3389/fncel.2015.00017 PMC4319477 25705176
12Crossgrove J.S Li G.J Zheng W. The choroid plexus removes beta-amyloid from brain cerebrospinal fluid Exp Biol Med (Maywood) 230 10 2005 771 776 10.1177/153537020523001011 16246905 PMC3982214
13Philippe T.J Kearns N. Avey D.R Single-nuclei RNA-sequencing of postmortem choroid Plexus uncovers dysfunction Alzheimer’s disease Alzheimer’s Dement 20 S8 2024 e09499710.1002/alz.094997
14Serot J.M Béné M.C Foliguet B. Morphological alterations of the choroid plexus in late-onset Alzheimer’s disease Acta Neuropathol 99 2 2000 105 108 10.1007/pl00007412 10672315
15Jeong S.H Park C.J Cha J. Choroid plexus volume, amyloid burden, and cognition in the Alzheimer’s Disease continuum Aging Dis 16 1 2025 552 564 10.14336/AD.2024.0118 PMC11745423 38300638
16Novakova Martinkova J. Ferretti M.T Ferrari A. Longitudinal progression of choroid plexus enlargement is associated with female sex, cognitive decline and ApoE E4 homozygote status Front Psychiatry 14 2023 14 10.3389/fpsyt.2023.1039239 PMC10031049 36970283
17Xu Y. Wang M. Li X. Glymphatic dysfunction mediates the influence of choroid plexus enlargement on information processing speed in patients with white matter hyperintensities Cereb Cortex 34 6 2024 bhae26510.1093/cercor/bhae265 38912605
18Jiang J. Zhuo Z. Wang A. Choroid plexus volume as a novel candidate neuroimaging marker of the Alzheimer’s continuum Alz Res Ther 16 1 2024 149 10.1186/s13195-024-01520-w PMC11221040 38961406
19Choi J.D Moon Y. Kim H.-J. Choroid plexus volume and permeability at brain MRI within the Alzheimer Disease clinical spectrum Radiology 304 3 2022 635 645 10.1148/radiol.212400 35579521
20Tadayon E. Moret B. Sprugnoli G. Improving choroid plexus segmentation in the healthy and diseased brain: relevance for tau-PET imaging in dementia J Alzheimers Dis 74 4 2020 1057 1068 10.3233/JAD-190706 32144979 PMC9094634
21Ota M. Sato N. Nakaya M. Relationship between the tau protein and choroid plexus volume in Alzheimer’s disease Neuroreport 34 11 2023 546 550 10.1097/WNR.0000000000001923 37384934
22Beekly D.L Ramos E.M van Belle G. The National Alzheimer’s Coordinating Center (NACC) Database: an Alzheimer disease database Alzheimer Dis Assoc Disord 18 4 2004 270 277 15592144
23Chan K.CG. Xia F. Kukull W.A. NACC data: who is represented over time and across centers, and implications for generalizability Alzheimers Dement 21 9 2025 e7065710.1002/alz.70657 PMC12445991 40968249
24Neuropathology Data Set (NP) V10: Forms & Documentation | National Alzheimer’s Coordinating Center 2014. Available from: https://naccdata.org/data-collection/forms-documentation/np-10 [Last accessed: 12/17/2025].
25Besser L. Kukull W. Knopman D.S Version 3 of the National Alzheimer’s coordinating center’s uniform data set Alzheimer Dis Assoc Disord 32 4 2018 351 358 10.1097/WAD.0000000000000279 30376508 PMC6249084
26DeTure M.A Dickson D.W. The neuropathological diagnosis of Alzheimer’s disease Mol Neurodegener 14 1 2019 32 10.1186/s13024-019-0333-5 31375134 PMC6679484
27Braak H. Braak E. Neuropathological stageing of Alzheimer-related changes Acta Neuropathol 82 4 1991 239 259 10.1007/BF00308809 1759558
28Mirra S.S Hart M.N Terry R.D. Making the diagnosis of Alzheimer’s disease. A primer for practicing pathologists Arch Pathol Lab Med 117 2 1993 132 144 8427562
29Montine T.J Phelps C.H Beach T.G National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: a practical approach Acta Neuropathol 123 1 2012 1 11 10.1007/s00401-011-0910-3 22101365 PMC3268003
30Morris J.C. The Clinical Dementia Rating (CDR): current version and scoring rules Neurology 43 11 1993 2412 10.1212/wnl.43.11.2412-a 8232972
31Folstein M.F Folstein S.E McHugh P.R. Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician J Psychiatr Res 12 3 1975 189 198 10.1016/0022-3956(75)90026-6 1202204
32Ho A.J Raji C.A Becker J.T Obesity is linked with lower brain volume in 700 AD and MCI patients Neurobiol Aging 31 8 2010 1326 1339 10.1016/j.neurobiolaging.2010.04.006 20570405 PMC3197833
33Li C. Zhang H. Wang J. Choroid Plexus Volume in Rural Chinese Older Adults: Distribution and Association With Cardiovascular Risk Factors and Cerebral Small Vessel Disease JAHA 035941 2024 10.1161/JAHA.124.035941 PMC11935722 39424375
34Keil S.A Jansson D. Braun M. Glymphatic dysfunction in Alzheimer’s disease: a critical appraisal Science (1979) 389 6756 2025 10.1126/science.adv8269 40638744
35Bae Y.J Kim J.-M. Choi B.S Altered brain glymphatic flow at diffusion-tensor MRI in rapid eye movement sleep behavior disorder Radiology 307 5 2023 e22184810.1148/radiol.221848 37158722
36Shih N.-C. Barisano G. Lincoln K.D Effects of sleep on brain perivascular space in a cognitively healthy population Sleep Med 111 2023 170 179 10.1016/j.sleep.2023.09.024 37782994 PMC10591884
37Castillo P.R Patel V. Mera R.M The association between slow-wave sleep and choroid plexus calcifications in older adults. Results from the sleep disorders substudy of the Atahualpa Project cohort Clin Neurol Neurosurg 246 2024 10854110.1016/j.clineuro.2024.108541 39265485
38Voumvourakis K.I Sideri E. Papadimitropoulos G.N The dynamic relationship between the glymphatic system Aging Mem Sleep Biomed 11 8 2023 2092 10.3390/biomedicines11082092 PMC10452251 37626589
39Cummings J. The Neuropsychiatric Inventory: development and applications J Geriatr Psychiatry Neurol 33 2 2020 73 84 10.1177/0891988719882102 32013737 PMC8505128
40Backhausen L.L Herting M.M Buse J. Quality control of structural MRI images applied using FreeSurfer-A hands-on workflow to rate motion artifacts Front Neurosci 10 2016 558 10.3389/fnins.2016.00558 27999528 PMC5138230
41Yushkevich P.A Gao Y. Gerig G. ITK-SNAP: an interactive tool for semi-automatic segmentation of multi-modality biomedical images Annu Int Conf IEEE Eng Med Biol Soc 2016 2016 3342 3345 10.1109/EMBC.2016.7591443 28269019 PMC5493443
42Fischl B. Salat D.H Busa E. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain Neuron 33 3 2002 341 355 10.1016/s0896-6273(02)00569-x 11832223
43Tadayon E. Pascual-Leone A. Press D. Choroid plexus volume is associated with levels of CSF proteins: relevance for Alzheimer’s and Parkinson’s disease Neurobiol Aging 89 2020 108 117 10.1016/j.neurobiolaging.2020.01.005 32107064 PMC9094632
44Liang P. Li M. Chen Y. Associations of choroid plexus volume with white matter hyperintensity volume and susceptibility and plasma amyloid markers in cerebral small vessel disease Alzheimer’s Res Ther 17 1 2025 90 10.1186/s13195-025-01740-8 40270041 PMC12016351
45Delvenne A. Vandendriessche C. Gobom J. Involvement of the choroid plexus in Alzheimer’s disease pathophysiology: findings from mouse and human proteomic studies Fluids Barriers CNS 21 2024 58 10.1186/s12987-024-00555-3 39020361 PMC11256635
46Xia H. Feng Y. Zhu H. Association of choroid plexus volume with white matter microstructure, glymphatic function, and peripheral systemic inflammation in Alzheimer’s disease Transl Psych 15 1 2025 238 10.1038/s41398-025-03432-1 PMC12254307 40645956
47Jagust W.J Teunissen C.E DeCarli C. The complex pathway between amyloid β and cognition: implications for therapy Lancet Neurol 22 9 2023 847 857 10.1016/S1474-4422(23)00128-X 37454670
48Bejanin A. Schonhaut D.R La Joie R. Tau pathology and neurodegeneration contribute to cognitive impairment in Alzheimer’s disease Brain 140 12 2017 3286 3300 10.1093/brain/awx243 29053874 PMC5841139
49Clarke K.M. Etemadmoghadam S. Danner B. The Nun study: insights from 30 years of aging and dementia research Alzheimers Dement 21 2 2025 e1462610.1002/alz.14626 PMC11852352 39998266
50de Vries L.E. Huitinga I. Kessels H.W. The concept of resilience to Alzheimer’s disease: current definitions and cellular and molecular mechanisms Mol Neurodegener 19 2024 33 10.1186/s13024-024-00719-7 38589893 PMC11003087
51Nasreddine Z.S. Phillips N.A. Bédirian V. The montreal cognitive assessment, MoCA: a brief screening tool for mild cognitive impairment J Am Geriatr Soc 53 4 2005 695 699 10.1111/j.1532-5415.2005.53221.x 15817019
52Fasnacht J.S. Wueest A.S. Berres M. Conversion between the montreal cognitive assessment and the mini-mental status examination J Am Geriatr Soc 71 3 2023 869 879 10.1111/jgs.18124 36346002
53Becker J.L. MacDonald M.E. Vessey K.A. Choroid plexus volume and its association with cognitive performance across the lifespan: links to sleep quality and healthy brain aging Neurobiol Aging 156 2025 40 49 10.1016/j.neurobiolaging.2025.08.003 40839938
54Dallaire-Théroux C. Callahan B.L. Potvin O. Radiological-pathological correlation in Alzheimer’s disease: systematic review of antemortem magnetic resonance imaging findings J Alzheimer’s Dis 57 2 2017 575 601 10.3233/JAD-161028 28282807
55Pintzka C.W.S. Hansen T.I. Evensmoen H.R. Marked effects of intracranial volume correction methods on sex differences in neuroanatomical structures: a Hunt MRI study Front Neurosci 9 2015 10.3389/fnins.2015.00238 PMC4496575 26217172
56Gauthreaux K. Bonnett T.A. Besser L.M. Concordance of clinical Alzheimer diagnosis and neuropathological features at autopsy J Neuropathol Exp Neurol 79 5 2020 465 473 10.1093/jnen/nlaa014 32186726 PMC7160616
Come citare questo studio
Francis Fernandes, Avyarthana Dey, Andy Ma, Nathan W Churchill, Corinne E Fischer, Simon J Graham, David G Munoz, Tom A Schweizer. (2026). Choroid plexus volume in pathology-confirmed Alzheimer's disease.. The journal of prevention of Alzheimer's disease. DOI: 10.1016/j.tjpad.2026.100648
Consulta la pubblicazione originale ↗