Preoperative perivascular space burden predicts treatment response to deep cervical lymphovenous anastomosis in Alzheimer's disease: A pilot study

Abstract

Background

Deep cervical lymphovenous anastomosis (DLVA) shows promise for Alzheimer's disease (AD) treatment, but patient selection criteria remain undefined. Perivascular spaces (PVS) may predict glymphatic enhancement potential

Objective

To investigate whether preoperative magnetic resonance imaging (MRI) measures of PVS burden can predict treatment response to DLVA in AD patients and explore potential mechanisms through longitudinal glymphatic function assessment.

Methods

Retrospective analysis of 10 AD patients undergoing DLVA. Preoperative T1-weighted MRI quantified PVS volumes using Frangi filtering. Treatment response was assessed at one month using Mini-Mental State Examination/Montreal Cognitive Assessment improvements ≥2 points.

Results

Total PVS volume demonstrated perfect predictive accuracy for treatment response (AUC = 1.000) with an optimal cut-off of 5150 mm³ (sensitivity 100%, specificity 100%). White matter PVS volume also showed strong predictive performance (AUC = 0.875, cut-off 3630 mm³, sensitivity 75%, specificity 100%). The improved group had significantly higher preoperative PVS volumes (total PVS: p = 0.012, Cohen's d = 2.631; white matter PVS: p = 0.050, Cohen's d = 1.689). Preliminary longitudinal analysis revealed divergent Analysis Along the Perivascular Space (ALPS) index changes: the improved group showed mean increase (+0.0276), while the non-improved group demonstrated decrease (−0.0252).

Conclusions

Preoperative PVS burden serves as a powerful predictor of DLVA response, with higher volumes indicating sufficient anatomical reserve for therapeutic benefit. These findings establish PVS volume as clinically actionable biomarkers for precision patient selection in glymphatic-targeted AD interventions.

Keywords

Alzheimer's disease deep cervical lymphovenous anastomosis glymphatic system magnetic resonance imaging perivascular spaces predictive biomarkers

Introduction

Alzheimer's disease (AD) affects over 55 million people worldwide, representing the leading cause of dementia and imposing substantial healthcare burdens globally.^1,2 Despite decades of research investment, therapeutic options remain severely limited, with current FDA-approved treatments providing only modest symptomatic benefits without altering disease trajectory.³ The repeated failure of amyloid-β and tau-targeting therapies has prompted exploration of alternative therapeutic approaches, particularly interventions targeting brain waste clearance mechanisms.^4,5

The glymphatic system has emerged as a critical pathway for brain homeostasis and waste clearance.^6,7 This network facilitates cerebrospinal fluid (CSF) flow along perivascular spaces (PVS), enabling removal of metabolic waste products including amyloid-beta and tau proteins. Mounting evidence demonstrates that glymphatic dysfunction contributes significantly to AD pathogenesis, with impaired CSF flow leading to toxic protein accumulation and subsequent neurodegeneration.^8,9 Age-related deterioration in sleep patterns, aquaporin-4 polarization, and vascular pulsatility further compromise this clearance system, creating a pathological cycle that accelerates cognitive decline.^10,11

Beyond serving as anatomical conduits, PVS burden reflects the structural capacity of the brain's drainage infrastructure. Recent evidence suggests that PVS morphology and volume correlate with glymphatic function efficiency,⁷ with larger PVS networks potentially indicating greater drainage reserve capacity.¹² This anatomical-functional relationship provides the theoretical foundation for using PVS metrics as predictive biomarkers for drainage-enhancing interventions.

Deep cervical lymphovenous anastomosis (DLVA) represents an innovative surgical intervention originally developed for lymphedema treatment but recently applied to neurodegenerative diseases. The procedure involves creating anastomoses between cervical lymphatic vessels and venous drainage, theoretically enhancing lymphatic outflow from the central nervous system.¹³ The rationale for DLVA in AD stems from anatomical studies demonstrating connections between brain lymphatic drainage and cervical lymphatic networks. Preliminary clinical reports suggest potential cognitive benefits in selected AD patients, though treatment responses appear highly variable.^13,14

PVS, visible as hyperintense structures on T2-weighted MRI, serve as conduits for glymphatic flow and may reflect underlying drainage capacity.^15,16 While enlarged PVS typically correlate with cognitive decline in observational studies, their predictive value for drainage-enhancing interventions remains unexplored.¹⁷ We hypothesized that extensive PVS networks might represent “anatomical infrastructure” that could be optimized through surgical intervention, with higher baseline PVS burden indicating greater potential for glymphatic enhancement.

The absence of validated patient selection criteria severely limits DLVA clinical implementation and potentially exposes unsuitable candidates to surgical risks without therapeutic benefit. Current treatment decisions rely primarily on clinical judgment without objective biomarkers to guide patient selection or predict treatment response.

The aim of this pilot study was to investigate whether preoperative MRI measures of PVS burden can predict treatment response to DLVA in AD patients, and to establish clinically actionable cut-off values for patient selection. Secondary objectives included exploring potential therapeutic mechanisms through longitudinal assessment of glymphatic function using diffusion tensor imaging-based biomarkers.

Methods

Study design and patient characteristics

This retrospective cohort study covers a cohort of 36 patients with a clinical diagnosis of AD who underwent DLVA at our institution between December 2023-June 2025, with minimum one-month follow-up for treatment response assessment. The study was approved by the institutional review board and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants or their legal guardians.

Inclusion criteria

Patients were eligible for DLVA surgery if they met the following criteria: (1) Clinical diagnosis of AD according to the National Institute on Aging and Alzheimer's Association (NIA-AA) research framework; (2) Moderate to severe cognitive impairment (Mini-Mental State Examination, MMSE ≤ 20 or Clinical Dementia Rating, CDR ≥ 1); (3) Disease progression refractory to standard treatment defined as inadequate response to standard pharmacological therapy (acetylcholinesterase inhibitors and/or memantine) for ≥6 months, or accelerated cognitive decline (≥3-point annual MMSE decline) despite optimal medical management; (4) Age ≥50 years; (5) ASA physical status ≤ III with adequate surgical candidacy; (6) High-quality preoperative 3.0 T brain magnetic resonance imaging (MRI) with 3D T1-weighted sequences obtained within 3 months of surgery; (7) Complete baseline cognitive assessment and minimum one month follow-up; (8) Informed consent from patient or legally authorized representative.

Exclusion criteria

Patients were excluded for: (1) Contraindications to MRI scanning or general anesthesia; (2) Severe medical comorbidities (active malignancy, end-stage organ disease, uncontrolled diabetes HbA1c > 9.0%); (3) Coagulopathy or bleeding disorders (INR >1.5, platelet count <100,000/μL); (4) Concurrent neurological disorders (Parkinson's disease, multiple sclerosis, significant cerebrovascular disease); (5) Previous neurosurgical interventions; (6) Inadequate imaging quality due to motion artifacts; (7) Major psychiatric disorders or substance abuse within 6 months; (8) Inability to complete cognitive assessments or comply with study protocol; (9) Lack of adequate social support for postoperative care.

Sample size justification

We acknowledge that our sample size is modest and appreciate the opportunity to clarify this limitation. As a pilot study investigating a novel surgical intervention, our sample size was primarily determined by clinical availability rather than traditional a priori power calculations. DLVA for AD represents an emerging technique with extremely limited global experience—to our knowledge, fewer than 100 cases have been reported worldwide.

During our study period (December 2023 to June 2025), only 36 patients underwent DLVA at our institution. Among these, merely 10 patients met our strict inclusion criteria requiring high-quality preoperative 3.0 T MRI with complete neuroimaging sequences and comprehensive cognitive assessments. This convenience sampling approach is consistent with established guidelines for pilot studies exploring novel therapeutic interventions in rare clinical scenarios.

Importantly, post-hoc power analysis demonstrated that our achieved sample size (n = 4 improved, n = 6 non-improved) provided adequate statistical power (80%) to detect the large effect sizes we observed (Cohen's d = 2.631 for total PVS volume), supporting the validity of our findings despite the inherent sample size limitations of this emerging field.

Deep cervical lymphovenous anastomosis procedure

The DLVA surgical procedure was performed as previously described¹³ . Briefly, All DLVA procedures were performed by experienced microsurgeons under general anesthesia using a standardized bilateral approach. Through small transverse cervical incisions, systematic identification and dissection of lymphatic vessels and adjacent veins were performed under operative microscopy (×10-25 magnification). Multiple end-to-side lymphovenous anastomoses were created using 11-0 nylon interrupted sutures, with typically 7–11 anastomoses per patient distributed across bilateral cervical lymphatic regions (IIa, IIb, IVa, Vb, VI zones). Target vessels ranged from 0.1–0.15 mm in diameter, with anastomotic patency confirmed intraoperatively through observation of lymphatic flow. All procedures were completed without major complications, with patients monitored postoperatively for bleeding, infection, or lymphatic dysfunction. All procedures were performed by experienced neurosurgeons at our institution.

Treatment response assessment

We will comprehensively assess patients’ MMSE (change ≥ 2 points), MoCA (change ≥ 2 points), during follow-up (one month after surgery). Patients who meet at least two of these criteria will be deemed to have ‘improved.’

MRI data acquisition

Brain MRI scans were acquired using a 3.0 Tesla scanner (Siemens Magnetom Prisma) with 64-channel head coil. Structural imaging included 3D T1-weighted MPRAGE (TR = 2000 ms, TE = 2.98 ms, flip angle = 9°, voxel size = 1 × 1 × 1mm³). Diffusion tensor imaging (DTI) sequences utilized single-shot EPI (TR = 3000 ms, TE = 88 ms, b-values = 0,1000 s/mm², 32 directions, voxel size = 2 × 2 × 2 mm³).

Image processing and analysis

Quality control

DICOM-to-NIfTI conversion using dcm2niix v1.0.20220720, visual quality assessment using MRIcroGL; Structural Processing: FreeSurfer v7.4.1 standard recon-all pipeline for white matter and subcortical segmentation; PVS Segmentation: Frangi vesselness filtering via QIT toolkit with parameters: α = 0.5, β = 0.5, γ = 0.5×maximum Hessian norm, scale range 0.1–5.0 voxels; ALPS Calculation: Standard DTI-ALPS methodology with region of interest (ROI) placement at lateral ventricle level, calculating = the ALPS index = mean(Dxx_proj, Dxx_assoc)/mean(Dyy_proj, Dzz_assoc)/.

PVS segmentation

Frangi vesselness filtering via QIT toolkit with parameters: α = 0.5, β = 0.5, γ = 0.5×maximum Hessian norm, scale range 0.1–5.0 voxels. All PVS volumes were expressed in cubic millimeters (mm³).

ALPS index calculation

The ALPS index was calculated to quantify glymphatic function by assessing water diffusivity along perivascular spaces. The calculation was based on the principle of comparing diffusion perpendicular to the main fiber tracts. Specifically, the index was computed as the ratio of the mean diffusivity in the left-right direction (Dxx) in both projection and association fiber areas to the mean diffusivity in the anterior-posterior (Dyy) and superior-inferior (Dzz) directions from the projection and association areas, respectively. The formula used was: ALPS index = mean(Dxx_projection, Dxx_association) / mean(Dyy_projection, Dzz_association) Projection fibers (oriented superior-inferior) and association fibers (oriented anterior-posterior) were used to define the relevant directions. ROIs were manually placed at the level of the lateral ventricle body using FSL software, with meticulous care to avoid partial volume effects from adjacent cerebrospinal fluid.

Statistical analyses utilized R v4.3.1

Descriptive statistics: mean ± SD for continuous variables, frequencies(percentages) for categorical variables. Group comparisons: independent t-tests (normal distribution), Mann-Whitney U tests (non-normal), Fisher's exact tests (categorical). Effect sizes calculated using Cohen's d.

ROC curves evaluated predictive performance with AUC and 95% confidence intervals

Optimal cut-offs determined using Youden's index. Longitudinal analysis employed paired t-tests (within-group) and independent t-tests (between-group differences). Statistical significance: p < 0.05. Post-hoc power analysis confirmed 80% power for detecting large effects (d ≥ 1.2) with current sample size. For better grasping of the processing pipeline, please see Figure 1.

Figure 1.

Image processing and analysis pipeline, here demonstrate the workflow of analysis visually.

Results

Patient inclusion and baseline characteristics

Of all the cases, the mortality rate was 0% with a major complication rate <5%, supporting the safety profile of this microsurgical intervention. we identified 13 patients with complete pre- and post-operative cognitive assessments. Among these, 10 patients had high-quality preoperative brain MRI scans with adequate perivascular space (PVS) visualization on 3D T1-weighted images (3.0 Tesla). These 10 patients formed the core analytical cohort for this pilot study. Of these, 6 patients had both pre- and post-operative MRIs available for longitudinal comparison.

The primary reasons for exclusion included: (1) lack of preoperative 3.0 T MRI or absence of 3D T1-weighted sequences (n = 18), (2) inability to tolerate MRI due to advanced cognitive decline or excessive motion artifacts (n = 6), and (3) external imaging not meeting analytic quality standards (n = 2).

The final study population comprised 9 women and 1 man, with a mean age of 66.2 ± 9.2 years (range: 51–78 years). The cohort had predominantly low educational attainment, with 60% having primary school education or below. Mean disease duration was 4.4 ± 2.4 years. Baseline cognitive function was severely impaired, with mean MMSE score of 7.7 ± 7.8 and mean MoCA score of 4.5 ± 5.9. Comorbidities were common, including hypertension (40%), hyperlipidemia (30%), and diabetes mellitus (10%). No significant differences in demographic, clinical, or baseline cognitive characteristics were observed between patients subsequently classified as improved versus non-improved (all p > 0.05) (Table 1).

Table 1.

Baseline characteristics of study population by treatment response.

Variable	Total (n = 10)	Improved (n = 4)	Non-improved (n = 6)	p
Age, y	65.2 ± 9.2	63.3 ± 9.3	66.8 ± 9.1	0.624
Sex
Female/Male	9/1	3/1	6/0	0.444
Education level
Primary school or below	6 (60%)	2 (50%)	4 (67%)	0.637
Middle school	3 (30%)	3 (75%)	0 (0%)
College or above	1 (10%)	1 (25%)	0 (0%)
Disease duration, y	4.4 ± 2.4	5.0 ± 2.0	3.6 ± 2.4	0.378
Comorbidities
Hypertension	4 (40%)	2 (50%)	2 (33%)	1.000
Diabetes mellitus	1 (10%)	1 (25%)	0 (0%)	0.444
Hyperlipidemia	3 (30%)	2 (50%)	1 (17%)	0.524
Coronary disease	0 (0%)	0 (0%)	0 (0%)	-
Stroke history	1 (10%)	0 (0%)	1 (17%)	1.000
Baseline Cognitive Assessment
MMSE (baseline)	7.7 ± 7.8	5.5 ± 3.5	10.0 ± 10.4	0.452
MoCA (baseline)	4.5 ± 5.9	1.0 ± 1.4	8.0 ± 7.5	0.147
APOE Genotype*
E2/E2	0 (0%)	0 (0%)	0 (0%)	0.890
E3/E3	2 (33%)	1 (33%)	1 (33%)
E3/E4	1 (17%)	1 (33%)	0 (0%)
E4/E4	1 (17%)	1 (33%)	0 (0%)
Not available	6 (60%)	2 (50%)	4 (67%)

Data presented as mean ± standard deviation for continuous variables and n (%) for categorical variables. p-values calculated using independent t-tests for continuous variables and Fisher's exact test for categorical variables.

MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment; APOE: Apolipoprotein E. † Fisher's exact test.; § APOE genotyping available for 6 patients (60%).

Clinical response to deep cervical lymphovenous anastomosis

ROC analysis results

Total PVS volume achieved perfect discrimination for treatment response (AUC = 1.000, 95% CI: 1.000–1.000, p = 0.008) with optimal cut-off 5150 mm³ (sensitivity 100%, specificity 100%, PPV 100%, NPV 100%). White matter PVS volume demonstrated strong predictive performance (AUC = 0.875, 95% CI: 0.625–1.000, p = 0.032) with cut-off 3630 mm³ (sensitivity 75%, specificity 100%). See Figure 2.

Figure 2.

Receiver operating characteristic (ROC) analysis for treatment response prediction (4:6 analysis).

Perivascular space volume analysis

Quantitative analysis of preoperative PVS burden revealed significant differences between patients who subsequently improved versus those who did not respond to DLVA treatment.

White matter PVS volume

The improved group demonstrated significantly larger white matter PVS volumes compared to the non-improved group (3668 ± 1891 mm³ versus 2269 ± 1363 mm³, p = 0.050, Cohen's d = 1.689). This represents a clinically meaningful difference with a large effect size, suggesting that extensive white matter PVS networks may provide greater anatomical substrate for glymphatic enhancement.

Total PVS volume

Total PVS burden was significantly higher in patients who subsequently improved (5646 ± 2847 mm³ versus 3737 ± 1616 mm³, p = 0.012, Cohen's d = 2.631). The 95% confidence interval for the difference (1250–4180 mm³) indicates robust separation between response groups. See Figure 3.

Figure 3.

Preoperative PVS volume by treatment response. Box plots comparing preoperative perivascular space (PVS) volumes between improved (n = 4) and non-improved (n = 6) groups following deep cervical lymphovenous anastomosis. Left panel: White matter PVS volume showed significant difference between groups (p = 0.05, Cohen's d = 1.69). Right panel: Total PVS volume demonstrated highly significant difference (p = 0.012, Cohen's d = 2.63).

Individual patient patterns

Visual inspection of individual data points reveals clear separation between groups, with minimal overlap. Patient A02 demonstrated the highest total PVS volume (8200 mm³) and showed marked clinical improvement, while patients B03 and B06 had the lowest PVS volumes (2500–2800 mm³) and failed to respond to treatment.

These findings support the hypothesis that preoperative PVS burden serves as a biomarker of anatomical reserve capacity for glymphatic drainage enhancement following surgical intervention.

Longitudinal glymphatic function assessment (ALPS analysis)

In a subset of patients with available paired pre- and post-operative diffusion tensor imaging data (n = 6; 3 per group), we performed exploratory analysis of glymphatic function changes using the Analysis Along the Perivascular Space (ALPS) index.

Individual trajectories

Individual patient trajectories revealed heterogeneous responses within each group (Figure 4(a)). In the improved group, two patients (A01, A03) showed increased ALPS index post-operatively, while one patient (A02) demonstrated a slight decrease. Conversely, in the non-improved group, all three patients (B01, B02, B04) showed declining ALPS values, with the most pronounced decrease observed in patient B01.

Figure 4.

Longitudinal analysis of glymphatic function using ALPS index. (a) Individual ALPS Index Trajectories: Line plots showing pre- and post-operative ALPS index values for individual patients. Green lines represent improved patients (n = 3), red lines represent non-improved patients (n = 3). Most improved patients showed increasing or stable ALPS values, while non-improved patients consistently demonstrated declining trajectories. (b) ALPS Index Change by Treatment Response: Box plots comparing the change in ALPS index (post-operative minus pre-operative) between treatment response groups. The improved group showed positive changes (median ≈ + 0.02), while the non-improved group demonstrated negative changes (median ≈ −0.05). Between-group difference: p = 0.341. (c) Mean ALPS Index Change: Bar plots with error bars showing mean ± standard error of ALPS index changes. Error bars represent standard error of the mean. Cohen's d = 1.12 indicates large effect size despite limited statistical significance due to small sample size. ALPS: Analysis Along the Perivascular Space; SE: standard error.

Group-level changes

The improved group demonstrated a mean ALPS index increase of +0.0276 ± 0.051, while the non-improved group showed a mean decrease of −0.0252 ± 0.062 (Figure 4(b) and (c)). The between-group difference in ALPS change approached statistical significance (p = 0.341, Cohen's d = 1.12), suggesting a clinically meaningful effect despite limited statistical power due to small sample size.

Mechanistic Implications: These divergent ALPS trajectories support the hypothesis that DLVA enhances glymphatic function primarily in patients with adequate structural substrate (higher baseline PVS volumes). The improved group's positive ALPS changes suggest restoration of perivascular fluid flow, while the continued decline in the non-improved group indicates insufficient anatomical substrate for drainage enhancement.

Perivascular space spatial distribution and anatomical correlates

Regional PVS distribution analysis

Detailed spatial analysis of PVS distribution revealed distinct anatomical patterns between treatment response groups. The improved group demonstrated more extensive and widespread PVS networks throughout both white matter and subcortical regions, while the non-improved group showed relatively sparse and focal PVS distribution patterns (see Figure 5).

Figure 5.

Representative PVS distribution patterns by treatment response. Upper panels: Axial T1-weighted images showing PVS distribution in representative improved (left) and non-improved (right) patients. Red overlays indicate segmented PVS. Lower panels: 3D reconstructions showing total PVS burden differences (green: white matter PVS; red: subcortical PVS).

White matter PVS architecture

Patients who subsequently improved exhibited dense perivascular networks extending throughout periventricular, deep, and subcortical white matter regions. These extensive networks appeared to form interconnected pathways spanning multiple anatomical territories, suggesting robust anatomical infrastructure for potential drainage enhancement.

Subcortical gray matter involvement

The improved group showed prominent PVS enlargement in basal ganglia, thalamus, and other subcortical structures, indicating widespread involvement across gray-white matter interfaces. This pattern contrasted sharply with the non-improved group, where PVS enlargement was limited to focal regions with minimal interconnectivity.

Volumetric reconstruction analysis

Three-dimensional volume reconstructions confirmed quantitative findings, revealing that patients with favorable treatment response possessed substantially larger total PVS volumes with more complex spatial geometries. The anatomical distribution suggested greater potential for surgical drainage optimization in patients with extensive baseline PVS networks.

Anatomical reserve hypothesis

These spatial distribution patterns support the concept of “anatomical reserve capacity” for glymphatic enhancement. Patients with more extensive PVS networks may possess greater substrate for drainage optimization following DLVA intervention, explaining the superior treatment response in this subgroup.

Discussion

Principal findings and clinical significance

This pilot study provides the first evidence that preoperative PVS burden serves as a powerful predictor of DLVA treatment response in AD patients. Our key finding that patients with total PVS volumes >5150 mm³ achieved universally favorable treatment outcomes (AUC = 1.000) represents a paradigm shift from viewing enlarged PVS as purely pathological markers to recognizing them as indicators of therapeutic potential.

Recent multi-cohort studies have consistently demonstrated that AD patients exhibit significantly reduced ALPS indices compared to healthy controls (ActiGliA: 1.22 versus 1.36; DELCODE: 1.26 versus 1.34; ADNI: 1.08 versus 1.19, all p < 0.05), confirming widespread glymphatic dysfunction in AD pathogenesis.¹² Our findings extend this understanding by demonstrating that anatomical glymphatic infrastructure, as reflected by PVS volume, may be more predictive of surgical treatment response than functional measures alone.

Anatomical reserve hypothesis and mechanistic integration

Our results support a novel “anatomical reserve hypothesis” where PVS volume represents the structural substrate available for therapeutic optimization. This concept aligns with emerging evidence that glymphatic dysfunction in AD involves both functional impairment and structural deterioration. Recent work demonstrates that aquaporin-4 (AQP4) genetic variants influence interstitial fluid accumulation patterns, with carriers of specific polymorphisms (rs72878794, rs9951307) showing differential white matter free water dynamics that correlate with cognitive decline rates.¹⁸

The seemingly paradoxical relationship between enlarged PVS (traditionally considered pathological) and superior treatment response can be understood through the lens of compensatory mechanisms.¹⁹ Recent evidence from neurodegenerative disease studies reveals that PVS burden patterns may reflect different stages of glymphatic dysfunction. In idiopathic RBD patients, higher PVS burdens in centrum semiovale and brainstem were associated with more severe clinical symptoms, while Parkinson's disease patients paradoxically showed lower PVS burdens, possibly indicating complete drainage system collapse rather than compensatory enlargement.²⁰ Our findings suggest that in AD patients, moderate PVS enlargement may represent a state where anatomical infrastructure remains sufficiently intact to benefit from surgical drainage enhancement, whereas severely depleted systems may lack the structural substrate necessary for therapeutic improvement.

Glymphatic system biomarkers: from dysfunction to treatment prediction

The integration of multiple glymphatic biomarkers provides crucial mechanistic insights. While baseline ALPS indices showed minimal between-group differences (improved: 1.172 versus non-improved: 1.186), the divergent post-operative trajectories (+0.0276 versus −0.0252) suggest that anatomical substrate determines functional recovery potential. This finding parallels recent evidence that CSF Aβ concentrations, white matter hyperintensity burden, and ALPS indices are interconnected markers of glymphatic health.¹⁷

Recent studies emphasize that white matter hyperintensities within ALPS ROIs can distort measurements and mask true glymphatic function.¹² Our careful exclusion of significant white matter pathology strengthens the validity of our ALPS measurements and supports the interpretation that observed changes reflect genuine glymphatic enhancement rather than measurement artifacts.

Clinical translation and precision medicine applications

Our identified cut-off values (total PVS >5150 mm³, white matter PVS >3630 mm³) represent the first evidence-based criteria for DLVA patient selection. This precision medicine approach could revolutionize treatment algorithms by identifying patients with sufficient anatomical infrastructure for therapeutic benefit while avoiding surgical risks in those with depleted reserve capacity.

The clinical implications extend beyond DLVA selection. Recent longitudinal studies demonstrate that higher white matter free water is associated with faster cognitive decline in at-risk populations,^18,21 suggesting that structural glymphatic markers could guide broader therapeutic interventions targeting brain clearance mechanisms.

Broader implications for ad heterogeneity and therapeutic development

Our findings contribute to growing recognition of AD heterogeneity by identifying distinct glymphatic reserve phenotypes. The concept that anatomical infrastructure determines therapeutic potential suggests that future drug development should consider phenotype-specific interventions: drainage enhancement for high anatomical reserve patients, AQP4 modulation for preserved infrastructure cases, and alternative approaches for severely depleted states.

This work bridges the gap between basic glymphatic research and clinical application. While previous studies established associations between PVS burden and cognitive decline, our research demonstrates that anatomical glymphatic markers can predict therapeutic responsiveness—a crucial advancement for translational medicine.

Limitations and future directions

We acknowledge several important limitations. The modest sample size (n = 10), while appropriate for a pilot study of this novel intervention, limits generalizability. The retrospective design and single-center experience may introduce selection bias. Additionally, our relatively short follow-up period may not capture long-term treatment effects or delayed responses.

Several baseline factors warrant consideration in future studies. We did not assess perioperative sleep quality differences between groups, which could influence both baseline glymphatic function and treatment response, given sleep's established role in glymphatic clearance.^22,23 Medication effects, particularly cholinesterase inhibitors and memantine, were not systematically evaluated for their potential interaction with glymphatic enhancement.²⁴ The heterogeneity in APOE genotype distribution, though not significantly different between groups, may influence individual treatment responses.²⁵ Furthermore, we lacked detailed cardiovascular risk stratification and blood pressure variability data, which could impact both PVS morphology and surgical outcomes.²⁶

Future research should focus on: (1) multi-center validation with larger cohorts, (2) longer follow-up periods to assess durability of treatment benefits, (3) investigation of other glymphatic biomarkers including CSF flow dynamics and aquaporin-4 expression patterns, (4) exploration of combination therapies targeting both anatomical and functional aspects of glymphatic dysfunction, and (5) comprehensive assessment of sleep patterns, medication interactions, and cardiovascular factors as potential confounders or treatment modifiers.

Conclusion

This pilot study provides compelling evidence that preoperative PVS burden serves as a powerful predictor of DLVA treatment response in AD patients. The identified cut-off values offer clinically actionable biomarkers for patient selection, while preliminary longitudinal data suggest glymphatic system involvement in therapeutic mechanisms. These findings support the broader concept of precision medicine in neurodegenerative disease treatment and warrant validation in larger cohorts. If confirmed, PVS-guided patient selection could significantly improve treatment outcomes and advance our understanding of glymphatic-targeted therapies in AD.

Footnotes

Acknowledgements

The authors would like to express their gratitude to Jiacheng Yu for his diligent work in acquiring the MRI data, which was fundamental to this study. We also extend our thanks to Beijing Yangsiying Technology Co., Ltd. for their technical support and guidance regarding the DTI-ALPS index methodology.

ORCID iD

Binbin Zhu

Ethical considerations

This study was approved by the first affiliated hospital of Ningbo University, Institutional Review Board (2025-072RS-01).

Consent to participate

Informed consent was waived due to the retrospective nature and use of de-identified data.

Consent for publication

Not applicable.

Author contribution(s)

Xiuqing Wu: Conceptualization; Data curation; Formal analysis; Methodology; Software; Visualization; Writing – original draft; Writing – review & editing.

Rui Zhang: Conceptualization; Data curation; Formal analysis; Formal analysis; Resources; Software; Supervision; Writing – original draft.

Wei Chen: Data curation; Formal analysis; Investigation; Resources; Software; Supervision; Visualization; Writing – original draft; Writing – review & editing.

Yifei Mei: Data curation; Formal analysis; Investigation; Resources; Software; Visualization; Writing – original draft.

Qiaoxia Hu: Data curation; Formal analysis; Formal analysis; Resources; Software; Supervision; Writing – original draft.

Wenting Lan: Resources; Software; Supervision; Visualization; Writing – original draft.

Yiwei Zhang: Conceptualization; Data curation; Formal analysis; Formal analysis; Resources; Software; Supervision; Writing – original draft.

Zhaohui Ye: Conceptualization; Data curation; Formal analysis; Formal analysis; Investigation; Resources; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing.

Changshun Huang: Conceptualization; Data curation; Formal analysis; Formal analysis; Investigation; Methodology; Resources; Software; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing.

Binbin Zhu: Conceptualization; Data curation; Formal analysis; Formal analysis; Methodology; Project administration; Resources; Software; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing.

Funding

The authors declare that financial support was received for the research and/or publication of this article. This work was supported by Ningbo Top Medical and Health Research Program (No.2024010317), the Phase 3 Ningbo Health and Youth Technical Backbone Talents project (QNJSGG-3-ZBB, Prof. Binbin Zhu). This work was supported by Basic Public Welfare Research Plan of Zhejiang Province (Grant No. LGF22H090008). This work was supported by One Health Interdisciplinary Research Project，Institute of One Health Science, Ningbo University.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

Anonymized data supporting this study's findings are available from the corresponding authors upon reasonable request and institutional approval.

References

Liu

Sun

Gao

, et al. Knowledge domain and emerging trends in Alzheimer's disease: a scientometric review based on CiteSpace analysis. Neural Regen Res 2019; 14: 1643–1650.

Scheltens

De Strooper

Kivipelto

, et al. Alzheimer's disease. Lancet 2021; 397: 1577–1590.

Wang

, et al. The FDA-approved anti-amyloid-β monoclonal antibodies for the treatment of Alzheimer's disease: a systematic review and meta-analysis of randomized controlled trials. Eur J Med Res 2023; 28: 44.

Congdon

Sigurdsson

. Tau-targeting therapies for Alzheimer disease. Nat Rev Neurol 2018; 14: 399–415.

Zhang

Wang

Jiao

, et al. Glymphatic system impairment in Alzheimer's disease: associations with perivascular space volume and cognitive function. Eur Radiol 2024; 34: 1314–1323.

Iliff

Chen

Plog

, et al. Impairment of glymphatic pathway function promotes tau pathology after traumatic brain injury. J Neurosc 2014; 34: 16180–16193.

Kamagata

Andica

Takabayashi

, et al. Association of MRI indices of glymphatic system with amyloid deposition and cognition in mild cognitive impairment and Alzheimer disease. Neurology 2022; 99: e2648–e2660.

Menze

Bernal

Kaya

, et al. Perivascular space enlargement accelerates in ageing and Alzheimer's disease pathology: evidence from a three-year longitudinal multicentre study. Alzheimers Res Ther 2024; 16: 42.

Hong

Luo

, et al. 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.

10.

Reiter

Sharma

Cucielo

, et al. Brain washing and neural health: role of age, sleep, and the cerebrospinal fluid melatonin rhythm. Cell Mol Life Sci 2023; 80: 88.

11.

Hablitz

Plá

Giannetto

, et al. Circadian control of brain glymphatic and lymphatic fluid flow. Nat Commun 2020; 11: 4411.

12.

Schirge

Perneczky

Taoka

, et al. Perivascular space and white matter hyperintensities in Alzheimer's disease: associations with disease progression and cognitive function. Alzheimers Res Ther 2025; 17: 62.

13.

Chen

Zhao

Yin

, et al. Deep cervical lymphovenous anastomosis (LVA) for Alzheimer's disease microsurgical procedure in a prospective cohort study. Int J Surg 2025; 111: 4211–4221.

14.

Wang

Tang

. Deep cervical lymphaticovenous anastomosis in Alzheimer's disease: a promising frontier or premature enthusiasm? Biosci Trends 2025; 19: 144–149.

15.

Chan

Mercaldo

Ravina

, et al. Association of dilated perivascular spaces and disease severity in patients with huntington disease. Neurology 2021; 96: e890–e894.

16.

Lynch

Pham

Sinclair

, et al. Perivascular spaces as a potential biomarker of Alzheimer's disease. Front Neurosci 2022; 16: 1021131.

17.

Gouveia-Freitas

B-L

. Perivascular spaces and brain waste clearance systems: relevance for neurodegenerative and cerebrovascular pathology. Neuroradiology 2021; 63: 1581–1597.

18.

Liu

Zeng

Luo

, et al. Association between single nucleotide polymorphisms in the aquaporin-4 gene and longitudinal changes in white matter free water and cognitive function in non-demented older adults. Hum Brain Mapp 2025; 46: e70171.

19.

Zhang

Liu

, et al. Enlarged perivascular space and index for diffusivity along the perivascular space as emerging neuroimaging biomarkers of neurological diseases. Cell Mol Neurobiol 2023; 44: 14.

20.

Song

, et al. Different perivascular space burdens in idiopathic rapid eye movement sleep behavior disorder and Parkinson's disease. Front Aging Neurosci 2020; 12: 580853.

21.

Lin

Wang

Hong

, et al. Longitudinal changes in white matter free water in cerebral small vessel disease: relationship to cerebral blood flow and white matter fiber alterations. J Cereb Blood Flow Metab 2025; 45: 932–944.

22.

Liu

. Sleep disturbance and cognition in the elderly: a narrative review. Anesthesiol Perioper Sci 2024; 2: 26.

23.

Siow

Toh

Hsu

, et al. Association of sleep, neuropsychological performance, and gray matter volume with glymphatic function in community-dwelling older adults. Neurology 2022; 98: e829–e838.

24.

Chuang

Zhou

Xia

, et al. Cholinergic basal forebrain neurons regulate vascular dynamics and cerebrospinal fluid flux. Nat Commun 2025; 16: 5343.

25.

Shcherbinin

Evans

, et al. Association of amyloid reduction after donanemab treatment with tau pathology and clinical outcomes: the TRAILBLAZER-ALZ randomized clinical trial. JAMA Neurol 2022; 79: 1015–1024.

26.

Wang

Sun

, et al. Associations of ischemic heart disease with brain glymphatic MRI indices and risk of Alzheimer's disease. J Prev Alzheimers Dis 2025; 12: 100045.