Deep medullar veins: predictors of clinical progression in Alzheimer’s disease

Introduction

Alzheimer’s disease (AD), the leading cause of dementia, has become one of the most expensive, deadly, and burdensome diseases worldwide (1). Its pathological hallmarks include tau hyperphosphorylation, amyloid-β (Aβ) plaques, neuroinflammation, and brain atrophy (2,3). However, despite extensive research, clinical drug trials have often failed, primarily because treatments are initiated too late in the disease course (4). This underscores the urgent need for early diagnosis and intervention to slow disease progression.

AD is characterized by a prolonged asymptomatic phase during which pathological changes are already underway, even in the absence of noticeable clinical symptoms (5). In addition to Aβ and tau pathology, other mechanisms—such as cerebral small vessel abnormalities (6,7) and perivenous glymphatic drainage impairment—are increasingly recognized as contributors to AD development (8,9). Notably, in 1995, Moody et al. (10) first described non-inflammatory degenerative changes in the subependymal and periventricular veins. These changes are now known as periventricular venous collagenosis, a condition classified as part of small vessel disease (11). Emerging evidence suggests that venous insufficiency and venous collagenosis may contribute to deep medullary veins (DMVs) degeneration, which are commonly observed in individuals with mild cognitive impairment (MCI) and AD (12-14).

Susceptibility-weighted imaging (SWI) is a highly sensitive technique that enables in vivo visualization of small intracranial veins by leveraging the susceptibility effects of deoxyhemoglobin. SWI-visible DMVs are integral to the brain’s glymphatic system and have been associated with aging, brain atrophy, and cognitive impairment (10,11,15-18). Despite these findings, direct evidence linking DMVs alterations to AD dementia and its clinical progression remains lacking.

Given these gaps, this longitudinal study aimed to investigate whether DMVs can serve as imaging biomarkers for AD progression. We hypothesize that DMVs damage occurs prior to the onset of AD dementia and may play a critical role in subsequent cognitive decline. Understanding the relationship between DMVs and AD progression could provide new insights into disease mechanisms and facilitate earlier intervention strategies. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-737/rc).

Methods

Participants and cognitive assessments

This retrospective longitudinal study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The data were obtained from the Open Access Series of Imaging Studies-3 (OASIS-3) project (https://www.oasis-brains.org), a publicly available longitudinal neuroimaging and clinical dataset hosted by the Washington University Knight Alzheimer’s Disease Research Center. OASIS-3 includes longitudinal magnetic resonance imaging (MRI), positron emission tomography (PET), clinical, and cognitive data from 1,098 participants ranging from cognitive normal (CN) to MCI and AD. Follow-up periods span up to 15 years, with assessments conducted at multiple timepoints.

People without subjective complaints or mood disturbances, with normal cognition typically score 24–30 on the Mini-Mental State Examination (MMSE) scores, and who have a Clinical Dementia Rating (CDR) of 0 are considered CN. (19). Those who present with preserved global cognition (MMSE =24–30) but show subtle decline (CDR =0.5), characterized by objective memory impairment indicated by their education-adjusted Wechsler Memory Scale Logical Memory II testing performance, and do not have impairment in other cognitive domains as well as do not meet the criteria for dementia, are diagnosed with MCI (19). Individuals who manifest with objective cognitive decline (MMSE <24, CDR ≥0.5), meeting the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association criteria through documented progressive memory dysfunction and impairment in multiple cognitive domains are diagnosed with AD dementia (19).

Figure 1 shows the flowchart of participant enrollment. A total of 845 participants who underwent brain SWI were initially included in the study. The interval between baseline cognitive assessments, MRI examinations, and PET scans was within three months. This study excluded 672 participants due to inferior SWI images quality that influenced the visualization of DMVs (e.g., motion artifacts or poor signal-to noise ratio), as well as 3 participants diagnosed with non-AD dementia. This left 170 participants for baseline analysis. Representative examples of excluded SWI images are provided in Figure 2.

Figure 1 Study flowchart showing participant recruitment. AD, Alzheimer’s disease; DMVs, deep medullary veins; OASIS-3, Open Access Series of Imaging Studies-3; PET, positron emission tomography; PIB, ¹¹C-labeled Pittsburgh compound B; SWI, susceptibility-weighted imaging.

Figure 2 Representative SWI images excluded due to poor image quality. (A) Image degraded by severe motion artifacts, resulting in blurred venous structures. (B) Low signal-to-noise ratio compromised the visibility of DMVs. DMVs, deep medullary veins; SWI, susceptibility-weighted imaging.

To further assess the predictive value of DMVs score and longitudinal cognitive decline, we excluded individuals without follow-up data (n=2) and those diagnosed with AD (n=39). This resulted in a final cohort of 126 participants. Additionally, 62 participants were included in the mediation analysis.

Clinical progression

At each follow-up visit, participants were classified as either stable or progressive. The stable status was defined as the absence of cognitive conversion, whereas progressive status referred to conversion from CN to MCI/AD, or from MCI to AD (7). The final diagnosis was based on the participant’s status at the last visit. For survival analysis, progression time was calculated as the interval between the date of MRI scans and the date when conversion occurred.

SWI protocol and analysis

According to the OASIS-3 MRI protocol, SWI sequences were acquired from 3T MRI scanners. DMVs scoring was performed via dynamic assessment across five consecutive axial SWI slices (2 mm thickness each), enabling raters to evaluate venous continuity and morphology across adjacent layers. The selected slices spanned an anatomical range of approximately 10 mm within the periventricular region, from just above the basal ganglia to the level where the lateral ventricles were no longer visible (20). Only DMVs radiating perpendicularly from the lateral ventricles into the deep white matter were considered for scoring. According to medullar veins anatomy (21), DMVs were evaluated in six regions (bilateral), namely, the frontal, parietal, and occipital regions (Figure 3A). A 4-point score was used in each area: 0 points for all the veins exhibiting continuity and homogeneous signals; 1 point for veins with continuity but an inhomogeneous signal (one or more than one vein); 2 points for one or more veins exhibiting discontinuity and spot-like hypointensity; and 3 points for all veins having a lack of continuity (Figure 3B-3E). The total DMVs score which ranged from 0 to 18 was calculated by aggregating the scores of the six areas. Then, DMVs scores were dichotomized into a low-score group and a high-score group using a threshold of <8 vs. ≥8 (22). This cutoff was selected based on the median DMVs score within the study sample, which provided a data-driven approach to stratify participants. Median-based dichotomization is commonly used in exploratory analyses to achieve balanced group sizes when no established clinical threshold exists (23,24). SWI images were assessed by two trained radiologists who were blinded to the clinical data. Interrater reliability was good for the total DMVs score, with an intraclass correlation coefficient (ICC) of 0.79 [95% confidence interval (CI): 0.72–0.84]. In addition, Cohen’s κ was calculated for the dichotomized DMVs scores (κ=0.73, 95% CI: 0.63–0.83).

Figure 3 Anatomical reference and examples of SWI-visible DMVs. (A) Representative SWI images showing approximate boundaries of the frontal, parietal, and occipital regions. (B-E) Example of the four-point DMVs scoring system in the right parietal region, demonstrated across three consecutive slices. (B) 0 points: each vein is continuous with uniform spacing and signal (arrows indicate the same vein); (C) 1 point: each vein is continuous, but had inhomogeneous signal (arrow); (D) 2 points: one or more than one vein with discontinuity (arrow) and inhomogeneous signal; (E) 3 points: all the veins were not in continuity (arrows). DMVs scoring used dynamic consecutive slices; static images here may not fully capture all features. DMVs, deep medullary veins; SWI, susceptibility-weighted imaging.

MRI segmentation

To acquire total cortical volume and intracranial volume (ICV), T1-weighted images from the 3T scanners were segmented using FreeSurfer 5.3 (http://surfer.nmr.mgh.harvard.edu). The data processing steps followed those described in (25). Briefly, this processing includes motion correction, removal of non-brain tissue, automated Talairach transformation, segmentation, intensity normalization, tessellation, automated topology correction, and surface deformation. A trained lab member subsequently reviewed the automated segmentations generated by FreeSurfer 5.3 to ensure accuracy.

¹¹C-labeled Pittsburgh compound B (PIB) PET imaging and post-processing

Participants received 6–20 millicuries of carbon-11 labeled Pittsburgh Compound B (mCi ¹¹C-labeled PIB) intravenously with a 60-minute dynamic PET scan in a three-dimensional mode (24×5 s frames; 9×20 s frames; 10×1 min frames; 9×5 min frames).

PET imaging analyses were performed using the PET unified pipeline (PUP, https://github.com/ysu001/PUP). PET images were first smoothed to a common spatial resolution of 8 mm. Inter-frame motion correction for dynamic images was applied using standard registration procedures. PET images were then registered to T1-weighted magnetic resonance (MR) images segmented using FreeSurfer 5.3 for brain parcellation. The regional spread function technique was used for the partial volume correction (26). The standardized uptake value ratio (SUVR) used 30–60 minutes post-injection as the time window with the cerebellar cortex as the reference region.

First, to assess global Aβ burden, the mean cortical SUVR was calculated as the arithmetic mean of the SUVRs from the bilateral precuneus, prefrontal cortex, gyrus rectus, and lateral temporal regions. Second, to achieve standardized quantification of Aβ analysis across centers, the Centiloid scale was employed to convert PiB PET data to a 0–100 scale (27). Finally, the cortical Centiloid values for PiB imaging were calibrated on the Centiloid scale using the equations provided in (28).

Statistical analysis

The statistical analysis was performed using the software SPSS 20.0 (IBM Corp., Armonk, NY, USA), R software (version 4.4.1; R Foundation for Statistical Computing, Vienna, Austria), and GraphPad Prism 10.3.0 for Windows (GraphPad Software, San Diego, CA, USA).

Sample characteristics

Missing data were assessed for all variables. Cases with missing variables were excluded from the relevant analyses. Analyses of variance (ANOVA), non-parametric Kruskal–Wallis tests, chi-squared tests, and Spearman correlations were used to compare baseline characteristics among diagnostic groups, or to test associations between DMVs score and demographic factors. Post hoc multiple comparisons were performed with Bonferroni correction. Logistic regression was used to assess the association between continuous DMVs scores and diagnosis (stable CN—those cognitively normal during long-term follow-up—vs. baseline-diagnosed AD). The odds ratio (OR) and P values were calculated to quantify this relationship. Continuous variables with normal distributions were presented as mean with standard deviation, whereas non-normal distributions were reported as median with interquartile range (IQR). Categorical variables were expressed as frequency and percentages.

Survival analysis

Participants who did not complete follow-up visits were excluded from survival analyses. DMVs scores were dichotomized based on the median value (<8 vs. ≥8) for Kaplan-Meier survival analysis. The log-rank test was used to assess group differences in progression-free survival. For Cox proportional hazards regression models, DMVs scores were treated as a continuous variable to preserve statistical power and provide precise estimates of risk. Covariates included age, sex, APOE ε4 status, MMSE score, cortical volume, and ICV. Additionally, we used receiver operating characteristic (ROC) curves to evaluate the diagnostic performance of the Cox regression models in predicting longitudinal cognitive decline. The Youden index was used to determine sensitivity and specificity.

Mediation analysis

We conducted a mediation analysis to investigate whether Aβ burden mediates the effect of DMVs damage on cognitive impairment. The mediation model employed a three-variable regression framework, with the DMVs score as the independent variable, the MMSE score as the dependent variable, and cortical Centiloid as the mediator. Age, sex, and cortical volume were included as covariates. The significance of the mediation effect was assessed through bootstrapping with 5,000 replications and a 95% CI; a CI that does not contain zero was considered significant. The partial mediation (PM) effect was used to quantify the contribution of the DMVs score to the total effect and was calculated as the ratio of the indirect effect to the total effect. This analysis was performed using the PROCESS macro.

Results

Participant characteristics

A total of 170 participants were enrolled in this study. The baseline characteristics of all participants are summarized in Table 1. Compared to individuals with MCI and AD, CN participants were younger and had a higher cortical volume. AD patients, in turn, had a higher proportion of APOE ɛ4 carriers, lower MMSE score, and higher cortical Centiloid compared to both CN and MCI groups. A significant group effect was also observed in ICV (ANOVA P=0.043), yet no pairwise differences reached significance after Bonferroni correction.

We then examined changes in DMVs scores across the AD continuum. The results showed a progressive increase in DMVs scores as the disease advanced, with significant differences among CN, MCI, and AD groups (P<0.001). We further compared DMVs scores between stable CN (n=72) and baseline AD (n=39). Logistic regression showed that higher DMVs scores were significantly associated with increased odds of baseline AD compared to stable CN (OR =2.59, 95% CI: 1.81–3.71, P<0.001). Furthermore, correlation analysis revealed that DMVs scores were significantly associated with age (r=0.371, P<0.001), MMSE scores (r=−0.494, P<0.001), cortical volume (r=−0.371, P<0.001), and cortical Centiloid values (r=0.498, P<0.001), but showed no significant correlation with ICV (P=0.051). The correlation results are illustrated in Figure 4.

Figure 4 Correlation between DMVs score and demographic factors. Correlation between DMVs score with age (A), MMSE score (B), cortical volume (C), and cortical Centiloid values for PIB (D). DMVs, deep medullar veins; MMSE, Mini-Mental State Examination; PIB, 11C-labeled Pittsburgh compound B.

DMVs predict clinical progression in AD

Among the baseline non-demented participants, 37 individuals (30.08%) experienced longitudinal cognitive decline. This included 9 individuals who progressed from CN to MCI, 12 who converted from CN to AD, and 16 who transitioned from MCI to AD. Kaplan-Meier analysis revealed that participants in the high-grade group (DMVs score ≥8) exhibited a significantly increased risk of progression to MCI/AD (log-rank P<0.001; Figure 5A). Furthermore, Cox regression analysis demonstrated that the DMVs score was a strong predictor of cognitive decline [hazard ratio (HR) =1.13, 95% CI: 1.02–1.24, P=0.014; Table 2], even after adjusting for age, sex, MMSE score, APOE ɛ4 status, cortical volume, and ICV. The Cox regression model showed a robust predictive value [area under the curve (AUC) =0.800, P<0.001; Figure 5B], with a sensitivity of 75.00% and a specificity of 77.91%.

Figure 5 Prognostic value of DMVs score for clinical progression to MCI/AD. Kaplan-Meier curves showing the association between DMVs grade and the risk of clinical progression in non-demented participants (A) and in CN participants (C). ROC curves illustrating the predictive performance of the Cox regression model for cognitive decline in non-demented participants (B) and in CN participants (D). AD, Alzheimer’s disease; AUC, area under the curve; CN, cognitive normal; DMVs, deep medullar veins; MCI, mild cognitive impairment; ROC, receiver operating characteristic.

These analyses were further conducted in the baseline CN subgroup, yielding similar results. The difference in diagnostic conversion risk across DMVs grades remained significant (log-rank P=0.014; Figure 5C). A higher baseline DMVs score was strongly associated with an increased risk of progression from CN to MCI/AD (HR =1.28, 95% CI: 1.00–1.63, P=0.049; Table 2). The predictive performance of the model in this subgroup was also excellent (AUC =0.845, sensitivity =76.19%, specificity =83.10%, P<0.001; Figure 5D).

Mediation among DMVs score, Aβ, and cognitive function

The cortical Centiloid of PIB showed a significant mediation effect between DMVs score and MMSE score after correcting for multiple covariates (PM =26.64%; total effect β=−0.578, P<0.001; direct effect β=−0.425, P=0.001; indirect effect β=−0.154, 95% CI: −0.337 to −0.017; Figure 6).

Figure 6 Medication effect of Aβ burden between DMVs score and MMSE score in AD spectrum. *, P<0.05; **, P<0.001. Aβ, amyloid-β; AD, Alzheimer’s disease; c, direct effect; c', indirect effect; CI, confidence interval; DMVs, deep medullar veins; MMSE, Mini-Mental State Examination.

Discussion

In this study, we investigated the potential of the DMVs score as a biomarker for clinical progression in AD. Survival analysis and ROC curves showed that an elevated DMVs score is a significant risk factor for cognitive decline, with higher scores indicating an increased likelihood of disease progression. Furthermore, we identified a strong association between increased DMVs scores and Aβ deposition. Mediation analysis revealed that Aβ burden partially mediates the relationship between DMVs alterations and cognitive function, highlighting the potential link between cerebrovascular pathology and AD pathogenesis.

AD follows a gradual trajectory, with prior studies estimating an annual conversion rate from MCI to AD ranging between 7% and 15% (29-32)—a rate nearly 10 times higher than that of CN individuals (33). Our findings align with these estimates, emphasizing the importance of early AD diagnosis. DMVs serve as the primary venous drainage pathway for the deep white matter (34), and their dysfunction may contribute to AD pathology. Recent research has established a connection between abnormalities in SWI-visible DMVs and cognitive impairment (12,18). Consistently, our study found that patients with MCI and AD exhibited higher DMVs scores. In addition to examining DMVs alterations across the cognitive spectrum, we compared individuals with stable cognition over long-term follow-up to those diagnosed with AD at baseline. This comparison revealed a markedly higher prevalence of elevated DMV scores in the AD group, underscoring a strong association between venous changes and established AD pathology. Using cognitively stable individuals as a reference enhances the clinical relevance of our findings, suggesting that DMVs alterations not only appear early but also become more pronounced in overt AD, supporting their potential as imaging biomarkers to distinguish AD from normal aging.

Moreover, DMVs scores correlated with multiple pathological markers, including age, MMSE score, brain atrophy, and Aβ burden. Population-based studies have similarly reported an association between DMVs reduction and both aging and brain atrophy (11,17,35), corroborating our findings. Notably, Liu et al. (17) identified DMVs-related atrophy in the middle and inferior temporal lobes and hippocampus—regions particularly vulnerable in AD (36). These findings further support the role of DMVs alterations in AD pathophysiology.

Our longitudinal analysis underscores the predictive potential of the DMVs score for AD progression. Recent evidence suggests a growing association between impaired venous drainage and AD, with studies confirming a reduced venous drainage system in AD patients (37-39). One potential mechanism linking venous insufficiency to AD involves disruption of the glymphatic system, a perivascular network responsible for clearing metabolic waste from the central nervous system (40,41). Glymphatic dysfunction, exacerbated by venous insufficiency, impairs Aβ clearance, leading to its accumulation and subsequent cognitive decline (8). Glymphatic fluid stasis also contributes to venous stenosis and occlusion, further diminishing venule visibility on SWI sequences (42). Although glymphatic vessels are difficult to visualize using conventional neuroimaging techniques, they are known to surround cerebral small veins (43). These findings suggest that DMVs alterations may serve as an indirect marker of both venous and glymphatic dysfunction, providing valuable insight into AD pathophysiology.

Another proposed mechanism linking venous dysfunction to AD involves Aβ deposition within cerebral veins (37). Such deposits may contribute to venous collagenosis, leading to structural and functional impairments of cerebral veins (44). Cerebrovascular dysfunction, characterized by diminished reactivity and altered blood flow dynamics, may further exacerbate Aβ accumulation in small veins, ultimately accelerating AD pathogenesis (44). Furthermore, our mediation analysis confirmed that DMVs alterations may contribute to cognitive decline via Aβ deposition, reinforcing the multifactorial nature of AD beyond primary neurodegenerative pathology (45). These findings suggest that evaluating DMVs integrity could provide novel insights into brain microcirculation (15) and AD progression.

Despite these promising findings, our study has several limitations. First, the DMVs score provides a global assessment of small venous changes, but given the regional heterogeneity of AD pathology, future studies should focus on AD-specific regions to better characterize localized venous alterations. Second, cerebral perfusion data were lacking, limiting the ability to control for hemodynamic influences on DMVs visibility. Third, other common age-related comorbidities—such as chronic obstructive pulmonary disease, which can increase deoxyhemoglobin levels and exaggerate venous hypointensity, and lacunar infarcts, which may alter vein morphology—may confound DMVs scoring. Due to the retrospective nature of the study and limited clinical data, adjustment for these factors was not feasible. Fourth, given the relatively small sample size, caution is warranted in the interpretation of results and their generalization to broader populations. Larger, more diverse cohorts are needed to validate the longitudinal trajectory of DMVs changes in AD. Fifth, some participants were excluded due to poor SWI image quality, introducing selection bias. Lastly, although interrater reliability was excellent, the visual scoring method remains subjective and is limited by the relatively low field strength of the imaging modality. Future studies employing quantitative approaches using ultra-high-field (7T) MRI could provide more precise morphological assessments of DMVs (e.g., length, density, and tortuosity) (12) and reduce measurement bias.

Conclusions

Our findings underscore the association between DMVs alterations and cognitive dysfunction across the AD continuum. Furthermore, we highlight the potential of SWI-visible DMVs as imaging biomarkers for tracking disease progression in AD.

Acknowledgments

We would like to thank the OASIS-3 Principal Investigators (T. Benzinger, D. Marcus, and J. Morris) for providing the data, and Avid Radiopharmaceuticals, a wholly owned subsidiary of Eli Lilly, for providing the AV-45 doses. We would also like to thank the reviewers for their helpful and constructive comments on this paper.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-737/rc

Funding: This study was supported by the National Natural Science Foundation of China (No. 82302157, to S.L.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-737/coif). S.L. declares that this study was supported by the National Natural Science Foundation of China (No. 82302157). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Scheltens P, De Strooper B, Kivipelto M, Holstege H, Chételat G, Teunissen CE, Cummings J, van der Flier WM. Alzheimer's disease. Lancet 2021;397:1577-90. [Crossref] [PubMed]
2. Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement 2018;14:535-62. [Crossref] [PubMed]
3. Blennow K, de Leon MJ, Zetterberg H. Alzheimer's disease. Lancet 2006;368:387-403. [Crossref] [PubMed]
4. Li Y, Wang L, Zhong J, Xu H, Han Y, Zuo C, Jiang J. Impaired glymphatic function as a biomarker for subjective cognitive decline: An exploratory dual cohort study. Alzheimers Dement 2024;20:6542-55. [Crossref] [PubMed]
5. Insel PS, Donohue MC, Sperling R, Hansson O, Mattsson-Carlgren N. The A4 study: β-amyloid and cognition in 4432 cognitively unimpaired adults. Ann Clin Transl Neurol 2020;7:776-85. [Crossref] [PubMed]
6. Chen W, Song X, Zhang YAlzheimer's Disease Neuroimaging Initiative. Assessment of the Virchow-Robin Spaces in Alzheimer disease, mild cognitive impairment, and normal aging, using high-field MR imaging. AJNR Am J Neuroradiol 2011;32:1490-5. [Crossref] [PubMed]
7. Wang ML, Zou QQ, Sun Z, Wei XE, Li PY, Wu X, Li YHAlzheimer’s Disease Neuroimaging Initiative. Associations of MRI-visible perivascular spaces with longitudinal cognitive decline across the Alzheimer's disease spectrum. Alzheimers Res Ther 2022;14:185. [Crossref] [PubMed]
8. Tarasoff-Conway JM, Carare RO, Osorio RS, Glodzik L, Butler T, Fieremans E, Axel L, Rusinek H, Nicholson C, Zlokovic BV, Frangione B, Blennow K, Ménard J, Zetterberg H, Wisniewski T, de Leon MJ. Clearance systems in the brain-implications for Alzheimer disease. Nat Rev Neurol 2015;11:457-70. [Crossref] [PubMed]
9. Nedergaard M. Neuroscience. Garbage truck of the brain. Science 2013;340:1529-30. [Crossref] [PubMed]
10. Moody DM, Brown WR, Challa VR, Anderson RL. Periventricular venous collagenosis: association with leukoaraiosis. Radiology 1995;194:469-76. [Crossref] [PubMed]
11. Ao DH, Zhang DD, Zhai FF, Zhang JT, Han F, Li ML, Ni J, Yao M, Zhang SY, Cui LY, Jin ZY, Zhou LX, Zhu YC. Brain deep medullary veins on 3-T MRI in a population-based cohort. J Cereb Blood Flow Metab 2021;41:561-8. [Crossref] [PubMed]
12. Bouvy WH, Kuijf HJ, Zwanenburg JJ, Koek HL, Kappelle LJ, Luijten PR, Ikram MK, Biessels GJUtrecht Vascular Cognitive Impairment (VCI) Study group. Abnormalities of Cerebral Deep Medullary Veins on 7 Tesla MRI in Amnestic Mild Cognitive Impairment and Early Alzheimer's Disease: A Pilot Study. J Alzheimers Dis 2017;57:705-10. [Crossref] [PubMed]
13. Keith J, Gao FQ, Noor R, Kiss A, Balasubramaniam G, Au K, Rogaeva E, Masellis M, Black SE. Collagenosis of the Deep Medullary Veins: An Underrecognized Pathologic Correlate of White Matter Hyperintensities and Periventricular Infarction? J Neuropathol Exp Neurol 2017;76:299-312. [Crossref] [PubMed]
14. Hsiao WC, Chang HI, Hsu SW, Lee CC, Huang SH, Cheng CH, Huang CW, Chang CC. Association of Cognition and Brain Reserve in Aging and Glymphatic Function Using Diffusion Tensor Image-along the Perivascular Space (DTI-ALPS). Neuroscience 2023;524:11-20. [Crossref] [PubMed]
15. Wang X, Lyu J, Duan Q, Li C, Huang J, Meng Z, et al. Deep medullary vein damage correlates with small vessel disease in small vessel occlusion acute ischemic stroke. Eur Radiol 2024;34:6026-35. [Crossref] [PubMed]
16. Nyúl-Tóth Á, Negri S, Sanford M, Jiang R, Patai R, Budda M, et al. Novel intravital approaches to quantify deep vascular structure and perfusion in the aging mouse brain using ultrasound localization microscopy (ULM). J Cereb Blood Flow Metab 2024;44:1378-96. [Crossref] [PubMed]
17. Liu ZY, Zhai FF, Ao DH, Han F, Li ML, Zhou L, Ni J, Yao M, Zhang SY, Cui LY, Jin ZY, Zhu YC. Deep medullary veins are associated with widespread brain structural abnormalities. J Cereb Blood Flow Metab 2022;42:997-1006. [Crossref] [PubMed]
18. Xu Z, Li F, Xing D, Song H, Chen J, Duan Y, Yang B. A Novel Imaging Biomarker for Cerebral Small Vessel Disease Associated With Cognitive Impairment: The Deep-Medullary-Veins Score. Front Aging Neurosci 2021;13:720481. [Crossref] [PubMed]
19. Tsiknia AA, Edland SD, Sundermann EE, Reas ET, Brewer JB, Galasko D, Banks SJAlzheimer’s Disease Neuroimaging Initiative. Sex differences in plasma p-tau181 associations with Alzheimer's disease biomarkers, cognitive decline, and clinical progression. Mol Psychiatry 2022;27:4314-22. [Crossref] [PubMed]
20. Zhang R, Zhou Y, Yan S, Zhong G, Liu C, Jiaerken Y, Song R, Yu X, Zhang M, Lou M. A Brain Region-Based Deep Medullary Veins Visual Score on Susceptibility Weighted Imaging. Front Aging Neurosci 2017;9:269. [Crossref] [PubMed]
21. Lee C, Pennington MA, Kenney CM 3rd. MR evaluation of developmental venous anomalies: medullary venous anatomy of venous angiomas. AJNR Am J Neuroradiol 1996;17:61-70.
22. Yin X, Han Y, Cao X, Zeng Y, Tang Y, Ding D, Zhang J. Association of deep medullary veins with the neuroimaging burden of cerebral small vessel disease. Quant Imaging Med Surg 2023;13:27-36. [Crossref] [PubMed]
23. Okour M, Jacobson PA, Israni A, Brundage RC. Comparative Evaluation of Median Versus Youden Index Dichotomization Methods: Exposure-Response Analysis of Mycophenolic Acid and Acyl-Glucuronide Metabolite. Eur J Drug Metab Pharmacokinet 2019;44:629-38. [Crossref] [PubMed]
24. MacCallum RC, Zhang S, Preacher KJ, Rucker DD. On the practice of dichotomization of quantitative variables. Psychol Methods 2002;7:19-40. [Crossref] [PubMed]
25. Reuter M, Rosas HD, Fischl B. Highly accurate inverse consistent registration: a robust approach. Neuroimage 2010;53:1181-96. [Crossref] [PubMed]
26. Su Y, Blazey TM, Snyder AZ, Raichle ME, Marcus DS, Ances BM, et al. Partial volume correction in quantitative amyloid imaging. Neuroimage 2015;107:55-64. [Crossref] [PubMed]
27. Mintun MA, Larossa GN, Sheline YI, Dence CS, Lee SY, Mach RH, Klunk WE, Mathis CA, DeKosky ST, Morris JC. [11C]PIB in a nondemented population: potential antecedent marker of Alzheimer disease. Neurology 2006;67:446-52.
28. Su Y, Flores S, Hornbeck RC, Speidel B, Vlassenko AG, Gordon BA, Koeppe RA, Klunk WE, Xiong C, Morris JC, Benzinger TLS. Utilizing the Centiloid scale in cross-sectional and longitudinal PiB PET studies. Neuroimage Clin 2018;19:406-16. [Crossref] [PubMed]
29. 2022 Alzheimer's disease facts and figures. Alzheimers Dement 2022;18:700-89. [Crossref] [PubMed]
30. Petersen RC, Lopez O, Armstrong MJ, Getchius TSD, Ganguli M, Gloss D, Gronseth GS, Marson D, Pringsheim T, Day GS, Sager M, Stevens J, Rae-Grant A. Practice guideline update summary: Mild cognitive impairment [RETIRED]: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology 2018;90:126-35.
31. Petersen RC, Caracciolo B, Brayne C, Gauthier S, Jelic V, Fratiglioni L. Mild cognitive impairment: a concept in evolution. J Intern Med 2014;275:214-28. [Crossref] [PubMed]
32. Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to dementia--meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 2009;119:252-65. [Crossref] [PubMed]
33. Polyakova M, Sonnabend N, Sander C, Mergl R, Schroeter ML, Schroeder J, Schönknecht P. Prevalence of minor depression in elderly persons with and without mild cognitive impairment: a systematic review. J Affect Disord 2014;152-154:28-38. [Crossref] [PubMed]
34. Zhang K, Zhou Y, Zhang W, Li Q, Sun J, Lou M. MRI-visible perivascular spaces in basal ganglia but not centrum semiovale or hippocampus were related to deep medullary veins changes. J Cereb Blood Flow Metab 2022;42:136-44. [Crossref] [PubMed]
35. Tian Y, Li S, Yang Y, Cai X, Jing J, Wang S, Meng X, Mei L, Jin A, Yao D, Wei T, Wang Y, Pan Y, Wang Y. Associations of deep medullary veins with vascular risk factors, laboratory indicators, and cerebral small vessel disease: A population-based study. Brain Behav 2023;13:e2974. [Crossref] [PubMed]
36. Bakkour A, Morris JC, Wolk DA, Dickerson BC. The effects of aging and Alzheimer's disease on cerebral cortical anatomy: specificity and differential relationships with cognition. Neuroimage 2013;76:332-44. [Crossref] [PubMed]
37. Pardo K, Khasminsky V, Keret O, Benninger F, Goldberg I, Shelef I, Auriel E, Glik A. Alzheimer's disease patients have smaller venous drainage system compared to cognitively healthy controls. Alzheimers Dement 2025;21:e14551. [Crossref] [PubMed]
38. Beggs C, Chung CP, Bergsland N, Wang PN, Shepherd S, Cheng CY, Dwyer MG, Hu HH, Zivadinov R. Jugular venous reflux and brain parenchyma volumes in elderly patients with mild cognitive impairment and Alzheimer's disease. BMC Neurol 2013;13:157. [Crossref] [PubMed]
39. Chung CP, Beggs C, Wang PN, Bergsland N, Shepherd S, Cheng CY, Ramasamy DP, Dwyer MG, Hu HH, Zivadinov R. Jugular venous reflux and white matter abnormalities in Alzheimer's disease: a pilot study. J Alzheimers Dis 2014;39:601-9. [Crossref] [PubMed]
40. Jessen NA, Munk AS, Lundgaard I, Nedergaard M. The Glymphatic System: A Beginner's Guide. Neurochem Res 2015;40:2583-99. [Crossref] [PubMed]
41. Simka M, Skuła M. Potential Involvement of Impaired Venous Outflow from the Brain in Neurodegeneration: Lessons Learned from the Research on Chronic Cerebrospinal Venous Insufficiency. Rev Recent Clin Trials 2019;14:235-6. [Crossref] [PubMed]
42. Mestre H, Kostrikov S, Mehta RI, Nedergaard M. Perivascular spaces, glymphatic dysfunction, and small vessel disease. Clin Sci (Lond) 2017;131:2257-74. [Crossref] [PubMed]
43. Iliff JJ, Wang M, Liao Y, Plogg BA, Peng W, Gundersen GA, Benveniste H, Vates GE, Deane R, Goldman SA, Nagelhus EA, Nedergaard M. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Sci Transl Med 2012;4:147ra111. [Crossref] [PubMed]
44. Morrone CD, Bishay J, McLaurin J. Potential Role of Venular Amyloid in Alzheimer's Disease Pathogenesis. Int J Mol Sci 2020;21:1985. [Crossref] [PubMed]
45. Hong H, Hong L, Luo X, Zeng Q, Li K, Wang S, Jiaerken Y, Zhang R, Yu X, Zhang Y, Lei C, Liu Z, Chen Y, Huang P, Zhang MAlzheimer’s Disease Neuroimaging Initiative (ADNI). The relationship between amyloid pathology, cerebral small vessel disease, glymphatic dysfunction, and cognition: a study based on Alzheimer's disease continuum participants. Alzheimers Res Ther 2024;16:43. [Crossref] [PubMed]