DCE-MRI study on regional and Fazekas-stratified heterogeneity of blood-brain barrier leakage in white matter hyperintensities and related risk factors

Introduction

White matter hyperintensities (WMH) are core imaging markers of cerebral small vessel disease (CSVD). On T2-weighted fluid-attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI) sequences, WMH present as characteristic hyperintense lesions in periventricular or deep white matter regions, with pathological features including multidimensional alterations of demyelination, axonal degeneration, and gliosis (1). Mounting clinical evidence links WMH to neurological deficits (cognitive decline, gait disturbance), elevated risks of vascular dementia and acute ischemic stroke, and poor neurological prognosis post-stroke (2,3). As a key node in the biomarker network of cerebrovascular and neurodegenerative diseases, WMH have a complex clinical risk factor profile, and their pathological mechanisms have not yet been fully elucidated. Our team’s previous large-sample study based on 540 cases found that age and hypertension, as fundamental driving factors, exert a broad-spectrum effect on global WMH burden, whereas type 2 diabetes mellitus (T2DM) is specifically associated with deep white matter hyperintensities (DWMH) (4). The regional heterogeneity of risk factors suggests that there are clear anatomical subtype differences in the pathological mechanisms of WMH. Blood-brain barrier (BBB) damage is considered a potential key link in the occurrence and development of WMH, yet there are still many controversies regarding the characteristics of BBB leakage in WMH. In particular, whether there is heterogeneity in BBB leakage of WMH with different anatomical locations and severity grades, as well as the association pattern between this heterogeneity and clinical risk factors, remain unclear, which has become a core problem restricting the individualized accurate assessment of WMH and the effective implementation of clinical intervention programs (5).

Traditional consensus holds that moderate-to-severe WMH are accompanied by increased BBB leakage, which in turn elevates the risk of hemorrhagic transformation (HT) and malignant edema in stroke patients receiving intravenous thrombolysis (IVT). However, our team’s previous work found that WMH distribution and severity did not affect the incidence of intracranial HT or malignant edema after IVT in patients with acute anterior circulation ischemic stroke (6). This was further corroborated by Frey et al. (7), who confirmed no significant interaction between periventricular white matter hyperintensities (PWMH) and DWMH and post-IVT HT in the WAKE-UP trial. These findings collectively challenge the long-held assumption that “greater WMH burden equals more severe BBB leakage”, highlighting the need to re-evaluate this association using refined stratification. Recent basic research has advanced our understanding of the mechanisms underlying WMH-related BBB leakage. Downregulated expression or structural disruption of BBB tight junction proteins (occludin, claudin-5) is the core structural basis for leakage, mediated by oxidative stress, neuroinflammation, and aberrant cerebral perfusion (8,9). For clinical detection, dynamic contrast-enhanced (DCE)-MRI is the gold standard for in vivo quantitative BBB permeability assessment, with core parameters including volume transfer constant (K^trans) and permeability-surface area product (PS); dynamic susceptibility contrast (DSC)-MRI is more suitable for routine clinical use given its shorter acquisition time (10). Vikner et al. (11) highlighted via DCE-MRI that BBB leakage parameters reflect the product of vascular permeability and surface area; thus requiring correction for vascular density, the DCE-MRI-derived plasma volume fraction (Vp) is the core metric for local vascular density and perfusion, and abnormal perfusion directly biases BBB leakage quantification (12). Furthermore, recent studies have identified BBB hyperpermeability foci predominantly at the WMH-normal-appearing white matter (NAWM) junction, a key region for WMH progression (13-15).

Despite these advances, major controversies remain in the field. First, whether BBB leakage correlates linearly with WMH Fazekas grade is still debated (14,16-18): some studies report increasing BBB leakage with higher Fazekas grades, whereas others show decreased leakage metrics in severe WMH. Second, no consensus has been reached on the risk factors driving BBB leakage in PWMH versus DWMH, with conflicting findings regarding the roles of hypertension, diabetes, and other vascular risk factors (19-21). Third, although aberrant cerebral perfusion is known to bias BBB leakage quantification, no standardized protocol exists to quantify and correct this interference, preventing accurate characterization of WMH-related BBB leakage (5,14). These inconsistencies stem from the fact that most previous studies treated WMH as a single entity, failing to distinguish BBB disruption patterns across anatomical locations and lesion stages. This has not only led to conflicting findings, but also limited our mechanistic understanding of CSVD. Collectively, three critical research gaps remain unaddressed: 1) the lack of systematic, two-dimensional stratified analysis of BBB leakage characteristics in PWMH versus DWMH across Fazekas grades; 2) unclear mechanistic links between cerebral perfusion and BBB leakage across WMH subtypes, with no standardized correction for perfusion-related bias; and 3) unestablished independent risk factors for BBB leakage in anatomically distinct WMH subtypes, precluding targeted risk stratification. These gaps represent key bottlenecks to accurate pathological assessment and individualized intervention for WMH, highlighting the urgent need for targeted research.

Our team’s previous research based on a large-sample cohort has initially revealed the regional heterogeneity of WMH from the perspective of risk factors (4). However, it remains unclear whether WMH-related BBB leakage exhibits Fazekas grade-dependent severity specificity and regional heterogeneity, and how risk factors differ in their regulation of BBB leakage across WMH regions. Here, we used DCE-MRI with the Extended Tofts two-compartment model to quantify in vivo BBB permeability in PWMH and DWMH across Fazekas grades, systematically characterize BBB leakage heterogeneity by anatomical location and lesion severity, and identify risk factors for region-specific BBB permeability via multiple linear regression. Our findings provide novel imaging evidence for accurate WMH pathophysiological assessment, inform individualized treatment and clinical trial design for neurological diseases, and advance the development of a precision CSVD care framework. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0878/rc).

Methods

Study design and participants

This study was designed as a single-center retrospective observational study, with the participant screening workflow and grouping settings detailed in Figure 1. All participants were recruited from Jiangjin Central Hospital of Chongqing. A total of 437 participants who underwent DCE-MRI examinations between August 2022 and February 2026 were enrolled as the initial study population. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Jiangjin Central Hospital of Chongqing (Approval No. KY20240812-007). The requirement for written informed consent was waived due to the retrospective study design. We systematically collected baseline data of all enrolled cases, including demographic characteristics (age and gender), past medical history related to vascular risk factors (hypertension, T2DM, hyperlipidemia, and coronary heart disease), as well as lifestyle history (smoking history and alcohol drinking history). The exclusion criteria of this study were as follows: participants with a history of craniocerebral surgery or traumatic brain injury; patients with intracranial hemorrhagic lesions or large-area cerebral infarction; cases with intracranial neoplastic or non-neoplastic space-occupying lesions; participants with other systemic neoplastic diseases; those with MR images compromised by motion artifacts, susceptibility artifacts or other technical interference, which were of insufficient quality for quantitative analysis; and cases with incomplete clinical or imaging data. After strict screening in accordance with the above inclusion and exclusion criteria, a total of 75 eligible participants were finally included in this study.

Figure 1 Flowchart of participant enrollment. DCE-MRI, dynamic contrast-enhanced magnetic resonance imaging; DWMH, deep white matter hyperintensity; PWMH, periventricular white matter hyperintensity.

MRI protocol

All MRI scans were performed on a 3.0-tesla (3.0 T) MRI scanner (UMR780; United Imaging, Shanghai, China) with a 24-channel head-neck combined coil. All participants were placed in the supine position during scanning, with the head fixed by a sponge rubber pad to minimize motion interference. The parameters of conventional MRI sequences were set as follows: T2-weighted imaging (T2WI) with a matrix of 416×416, field of view (FOV) of 230×200 mm, repetition time/echo time (TR/TE) of 6,044/128 ms, 23 slices, slice thickness of 5 mm, and slice gap of 1 mm; T1WI-FLAIR with a matrix of 304×228, FOV of 230×200 mm, TR/TE of 2,000/6.3 ms, 23 slices, slice thickness of 5 mm, and slice gap of 1 mm; T2WI-FLAIR with a matrix of 288×230, FOV of 230×200 mm, TR/TE of 8,000/132 ms, 23 slices, slice thickness of 5 mm, and slice gap of 1 mm. The DCE-MRI sequence consisted of a multi-flip angle scan and a multi-phase DCE sequence. For the multi-flip angle scan, 5 single-phase flip angles (3°, 6°, 9°, 12°, and 15°) were adopted, with a matrix of 112×100, FOV of 230×200 mm, TR/TE of 4.11/1.84 ms, number of excitations of 1, and 20 slices (slice thickness of 6 mm, slice gap of 0 mm, slice direction interpolation of 2). For the DCE sequence, the flip angle was set to 10°, TR/TE was 2.51/0.92 ms, temporal resolution was 3 s, and the scan parameters (number of slices, slice thickness, gap, matrix, FOV) were consistent with those of the multi-flip angle sequence; a total of 90 phases (3,600 images) were collected, with a total scan time of 253 s. Gadobutrol, used as the contrast agent, was injected via the antecubital vein with a high-pressure syringe at a rate of 3 mL/s at the end of the 5th phase of the DCE sequence, with a total dose of 0.1 mmol/kg, followed by a 20 mL flush of 0.9% sodium chloride solution at the same injection rate.

Grouping

Fazekas grading (0 to 3) for PWMH and DWMH was performed on T2WI-FLAIR sequence images using the United Imaging artificial intelligence (AI)-based Cerebral Small Vessel Disease Analysis Module (UAI.OCR version R001; United Imaging, Shanghai, China). To ensure the accuracy and repeatability of the scoring results, all automatically segmented and scored results by the system were independently reviewed in a blinded manner by two senior neuroradiologists; in case of discrepancies in the scoring results between the two radiologists, a final consensus was reached through joint film reading and negotiation. A dual stratification strategy combining anatomical location and lesion severity was adopted for subgroup division in this study. First, all cases were divided into two major groups according to the anatomical location of the lesions, namely the PWMH group and the DWMH group. Each group was further divided into three subgroups based on Fazekas scoring results: the 0/1 group (no or mild lesions), the 2 group (moderate lesions), and the 3 group (severe lesions). The imaging characteristics and number of enrolled cases of each subgroup were as follows: the PWMH Fazekas 0/1 group (n=30) with lesions presenting as focal punctate, cap-like or pencil line-like mild changes without obvious fusion; the PWMH Fazekas 2 group (n=32) with lesions showing smooth halo-like confluent changes involving more than 25% of the circumference of the lateral ventricular body; the PWMH Fazekas 3 group (n=13) with extensive irregular confluent lesions, where hyperintensity extended to the deep white matter with further expanded involvement of the lateral ventricular body circumference; the DWMH Fazekas 0/1 group (n=50) with scattered punctate hyperintense lesions less than 5 mm in diameter without fusion; the DWMH Fazekas 2 group (n=16) with focal small-scale confluent lesions; and the DWMH Fazekas 3 group (n=9) with large confluent hyperintense lesions or bridging phenomenon between periventricular and deep white matter lesions.

BBB leakage assessment

The Extended Tofts two-compartment model was applied for quantitative analysis of DCE-MRI data to evaluate the level of BBB leakage in cerebral white matter. In the data preprocessing stage, motion correction of the dynamic sequence images was first performed using a rigid registration algorithm to eliminate artifacts caused by head micro-motion. The M1 segment of the middle cerebral artery was then selected to construct the arterial input function, with the superior sagittal sinus set as the venous output reference. Regions of interest (ROIs) were delineated blindly on T2WI-FLAIR sequences by a radiologist with more than 5 years of experience in neuroimaging analysis, with the area of a single ROI controlled between 50 and 100 mm². For the Fazekas 2 and 3 groups, ROIs were delineated in the core area of WMH lesions, whereas for the Fazekas 0/1 group, ROIs were delineated in the NAWM area of the corresponding anatomical site. All ROIs were placed to avoid non-target areas such as cerebral blood vessels and sulci. Finally, the arithmetic mean of the parameters from the bilateral homologous anatomical regions was calculated and included in the subsequent statistical analysis.

Statistical analysis

All statistical analyses and chart production were performed using GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA). Categorical data are presented as n (%), and intergroup comparisons were performed using the χ² test or Fisher’s exact test. For quantitative data, a normality test was performed first; normally distributed data were expressed as mean ± standard deviation, with intergroup comparisons using the two-sample t-test, whereas non-normally distributed data were described as median (interquartile range) [M (P25, P75)], with intergroup comparisons using the Mann-Whitney U test. Spearman rank correlation analysis was used to evaluate the correlation between BBB leakage indicators and local perfusion indicators, and stepwise multiple linear regression analysis was performed to explore the independent influencing factors of BBB leakage in WMH of different anatomical subtypes. All statistical tests were two-sided, with a significance level of α=0.05, and a P-value < 0.05 was considered statistically significant.

Results

Baseline characteristics of the enrolled study population

A total of 75 eligible cases were finally included in this study after strict screening in accordance with the inclusion and exclusion criteria, and the demographic characteristics, baseline data of vascular risk factors, and the distribution of WMH Fazekas scores of the overall study population are detailed in Table 1. The age of the enrolled population was non-normally distributed, with a median of 67 years and an interquartile range of 58 to 75 years; 43 of the cases were male, accounting for 57.33% of the total study population. In terms of vascular risk factors, 39 cases (52.00%) had hypertension, 20 (26.67%) had T2DM, 27 (36.00%) had hyperlipidemia, and 17 (22.67%) had coronary heart disease among the enrolled population. For lifestyle history, 25 cases (33.33%) had a smoking history, and 18 (24.00%) had an alcohol drinking history. Regarding the distribution characteristics of WMH Fazekas scores: for PWMH, 30 cases (40.00%) were graded as Fazekas 0/1, 32 cases (42.67%) as Fazekas 2, and 13 cases (17.33%) as Fazekas 3; for DWMH, 50 cases (66.67%) were graded as Fazekas 0/1, 16 cases (21.33%) as Fazekas 2, and 9 cases (12.00%) as Fazekas 3.

Dual heterogeneity of BBB leakage in WMH across anatomical locations and Fazekas scores

The DCE-MRI-derived K^trans was used to quantify BBB permeability. WMH-related BBB leakage exhibited significant non-linear changes and dual heterogeneity across Fazekas grades and anatomical regions (Figure 2). Compared with Fazekas 0/1 WMH, K^trans was significantly higher in both PWMH and DWMH with Fazekas grade 2 (P=0.0014 and P=0.037, respectively), but significantly lower in Fazekas grade 3 PWMH and DWMH (P<0.0001 and P=0.0002, respectively), with no linear increase with lesion severity. Within the same Fazekas grade, K^trans was consistently higher in PWMH than in DWMH, with significant differences for grade 0/1 and 2 lesions (all P<0.05), but no significant difference for grade 3 lesions (P>0.05).

Figure 2 Differences in blood-brain barrier permeability of WMHs across different severity grades and anatomical locations. (A) Distribution of K^trans values in PWMHs with different severity grades (Fazekas 0/1, 2, and 3); (B) distribution of K^trans values in DWMHs with different severity grades (Fazekas 0/1, 2, and 3); (C-E) comparison of K^trans values between PWMH and DWMH with Fazekas 0/1 (C), 2 (D), and 3 (E) grades, respectively. DWMH, deep white matter hyperintensity; K^trans, volume transfer constant; PWMH, periventricular white matter hyperintensity; WMH, white matter hyperintensity.

Perfusion characteristics of WMH lesions and their correlation with BBB leakage

To clarify the potential pathophysiological mechanism underlying the differences in BBB leakage of WMH with different anatomical locations and severity grades, and simultaneously correct the interference of local perfusion level on the quantitative results of K^trans, the Vp parameter derived from DCE-MRI was used to evaluate the local microvascular perfusion characteristics of WMH regions in this study. The results showed that the local perfusion level of WMH also exhibited dual heterogeneity of Fazekas score and anatomical region (Figure 3). Taking WMH with Fazekas 0/1 as the reference, the Vp values of PWMH showed a continuous decreasing trend with the increase of Fazekas score, among which the Vp values of PWMH with Fazekas score 2 and 3 were significantly lower than those of the 0/1 group, with statistically significant differences (P=0.027 and P<0.0001, respectively); the Vp values of DWMH also showed a decreasing trend with the elevation of Fazekas score, and only the Vp value of DWMH with Fazekas score 3 was significantly lower than that of the 0/1 group (P=0.021). Under the same Fazekas score stratification, the Vp value of PWMH was significantly higher than that of DWMH in grade 0/1 WMH (P<0.0001); the Vp value of PWMH was slightly higher than that of DWMH in grade 2 WMH, with the difference at the critical level of statistical significance (P=0.054); there was no statistically significant difference in Vp values between PWMH and DWMH in grade 3 WMH (P=0.883). To further clarify the correlation between BBB leakage and local perfusion status in WMH regions, Spearman rank correlation analysis was performed between K^trans, the core marker of BBB leakage, and Vp, the evaluation index of local perfusion. The results are shown in Figure 4, K^trans values were positively correlated with Vp values in all subgroups except the PWMH group with Fazekas score 3. Among them, in the Fazekas 0/1 subgroup, K^trans was moderately positively correlated with Vp in the PWMH region (r=0.444, P=0.014), and weakly positively correlated with Vp in the DWMH region (r=0.288, P=0.043), both of which were statistically significant; in the lesion region of the Fazekas score 2 subgroup, K^trans still maintained a positive correlation trend with Vp, but the correlation did not reach a statistically significant level.

Figure 3 Differences in local perfusion levels of WMHs across different severity grades and anatomical locations. (A) Distribution of Vp values in PWMH with different severity grades (Fazekas 0/1, 2, and 3); (B) distribution of Vp values in DWMH with different severity grades (Fazekas 0/1, 2, and 3); (C-E) comparison of Vp values between periventricular and deep white matter in WMH with Fazekas 0/1 (C), 2 (D), and 3 (E) grades, respectively. DWMH, deep white matter hyperintensity; PWMH, periventricular white matter hyperintensity; Vp, plasma volume fraction; WMH, white matter hyperintensity.

Figure 4 Correlation analysis between K^trans and Vp values of WMHs with different severity grades. (A-C) The correlation between K^trans and Vp values in periventricular white matter with Fazekas 0/1 (A), 2 (B), and 3 (C) grades, respectively; (D-F) the correlation between K^trans and Vp values in deep white matter with Fazekas 0/1 (D), 2 (E), and 3 (F) grades, respectively. K^trans, volume transfer constant; Vp, plasma volume fraction; WMH, white matter hyperintensity.

Intergroup equilibrium analysis of baseline risk factors among WMH subgroups

To exclude the interference of confounding variables such as baseline demographic characteristics and vascular risk factors on the aforementioned heterogeneity analysis results of BBB leakage and perfusion characteristics of WMH, univariate analysis was performed in this section to systematically evaluate the distribution equilibrium of baseline risk factors among WMH subgroups with different anatomical locations and lesion severity, and the results are detailed in Table 2. Among the PWMH subgroups with different Fazekas scores, only the age index showed a significant intergroup difference: compared with PWMH with Fazekas 0/1, the age of cases in the grade 2 and 3 subgroups was significantly higher, with statistically significant intergroup differences (all P<0.0001), whereas the distribution of all other baseline indicators including gender, hypertension, T2DM, hyperlipidemia, coronary heart disease, smoking history, and alcohol drinking history showed no statistically significant differences (all P>0.05). For the subgroup analysis of DWMH, with DWMH with Fazekas 0/1 as the fixed reference, the results showed that compared with the Fazekas 0/1 group, the DWMH Fazekas 2 group had statistically significant differences in the distribution of smoking history (P=0.04) and alcohol drinking history (P=0.006), yet there were no statistically significant differences in the distribution of other baseline risk factors between groups (all P>0.05). On this basis, we further compared the differences in baseline characteristics between PWMH and DWMH subgroups under the same Fazekas score stratification, and the results showed that only in the Fazekas 0/1 subgroup, the age of cases with DWMH was significantly higher than that of those with PWMH (P=0.007). Overall, the core vascular risk factors in this study cohort maintained a good equilibrium distribution among all WMH subgroups with different anatomical locations and severity grades, and only a few indicators such as age showed local intergroup differences, indicating that the Fazekas score heterogeneity and anatomical regional heterogeneity of WMH BBB leakage observed in this study were not caused by the intergroup distribution differences of baseline risk factors, which effectively excluded the interference of baseline confounding variables on the core research results, and provided reliable population baseline-level evidence for the subsequent in-depth exploration of the inherent pathophysiological mechanism of WMH heterogeneity and the identification of independent influencing factors of BBB leakage in different WMH subtypes.

Independent influencing factors of BBB leakage in WMH of different anatomical subtypes

Multiple linear regression analysis was performed to identify independent factors associated with BBB permeability (K^trans) in PWMH and DWMH (Fazekas 0–2 grades, Table 3). For PWMH, age (β=−0.526, 95% confidence interval (CI): −0.944 to −0.108, P=0.015) was an independent negative correlate of K^trans, whereas Vp (β=0.870, 95% CI: 0.031–1.709, P=0.043) and Fazekas grade 2 classification (β=20.970, 95% CI: 12.01–29.94, P<0.0001) were independent positive correlates. For DWMH, T2DM (β=9.770, 95% CI: 3.870–15.670, P=0.002) and Fazekas grade 2 classification (β=6.845, 95% CI: 0.977–12.710, P=0.023) were independent positive correlates of K^trans, with Vp showing a trend toward positive correlation (β=0.802, 95% CI: −0.034 to 1.638, P=0.060). No other factors were significantly associated with K^trans in either region (all P>0.05).

Discussion

This study addresses two long-standing core controversies in the field: the unclear linearity of the association between WMH-related BBB leakage severity and Fazekas grade, and the heterogeneous drivers of BBB disruption across anatomically distinct WMH subtypes, post-mortem histopathological validation and comprehensive imaging reviews have shown that Fazekas grading merely reflects lesion morphological burden rather than uniform pathological progression, as WMH encompass mixed pathological alterations including interstitial fluid accumulation, demyelination, and microvascular rarefaction at different disease stages (22,23). Using a dual-stratified design by anatomical location and lesion severity to systematically characterize BBB leakage features across WMH anatomical sites and Fazekas grades, as well as their correlations with clinical risk factors, our core findings are threefold: first, WMH-related BBB leakage does not increase linearly with lesion severity, but exhibits significant nonlinear dynamic changes, with the highest permeability in moderate (Fazekas 2) lesions and a marked decline in severe (Fazekas 3) lesions; second, PWMH show consistently higher BBB leakage than DWMH within the same Fazekas grade, confirming significant anatomical regional heterogeneity of WMH-related BBB damage; third, BBB leakage in PWMH and DWMH is governed by distinct regulatory mechanisms, with perfusion-dependent regulation in PWMH and T2DM-driven disruption in DWMH. These findings confirm the stage-dependent nonlinearity, anatomical regional specificity, and differential risk factor regulation of WMH-related BBB leakage, revealing fundamental pathophysiological differences between WMH subtypes and lesion stages, and providing novel imaging evidence to elucidate CSVD pathological heterogeneity, establish a precise WMH stratified evaluation system, and develop subtype-targeted intervention strategies.

Our findings confirm that WMH-related BBB leakage does not increase linearly with lesion burden, but exhibits significant nonlinear dynamics and anatomical regional heterogeneity. Building on our observation that Vp decreases in a lesion severity-dependent manner, and that K^trans is positively correlated with Vp in all subgroups except the Fazekas 3 PWMH group, we propose that WMH-related BBB disruption in sporadic CSVD has distinct stage-specific characteristics, with Fazekas 2 lesions potentially representing a pathologically active phase of BBB damage, although this interpretation requires confirmation in longitudinal studies. In this phase, disruption of endothelial tight junctions allows extensive contrast extravasation into the brain parenchyma, manifesting as elevated K^trans on imaging. As lesions progress to Fazekas 3, severe microvascular rarefaction, vascular fibrosis, and occlusion occur, leading to drastically reduced local perfusion and limited contrast delivery. Such perfusion decline is only one potential factor accounting for decreased K^trans; multiple coexisting pathological alterations including demyelination, axonal loss, and extracellular matrix remodeling jointly interfere with DCE quantitative readouts (24,25). The combined effect of these pathological changes may lead to apparent K^trans reduction, revealing a pathological shift from reversible functional BBB disruption to irreversible structural changes dominated by tissue necrosis and glial scarring, which partially explains inconsistent results across prior studies on Fazekas grade and BBB leakage correlation. This inherent limitation of K^trans quantification can be partially addressed by arterial spin labeling (ASL)-derived BBB water exchange parameter kw, which evaluates endothelial water permeability independent of gadolinium delivery and sensitively detects early subtle BBB dysfunction in CSVD and AD lesions (26). Notably, kw and DCE-derived K^trans reflect distinct BBB pathological dimensions: K^trans measures macromolecular gadolinium extravasation due to tight junction breakdown, whereas kw quantifies transendothelial water transport and glymphatic clearance efficiency, making the two markers mutually complementary for comprehensive microvascular assessment (27). Consistent multimodal imaging evidence from hereditary small vessel disease (SVD) also supports this interpretation: concurrent BBB hyperpermeability and microvascular depletion drive cerebral iron deposition detectable on quantitative susceptibility mapping (QSM), which coexists with attenuated DCE leakage metrics in advanced lesions (28). Furthermore, our multiple linear regression analysis identified distinct regulatory mechanisms underlying the regional heterogeneity of WMH BBB disruption: for PWMH, BBB damage is positively regulated by local cerebral perfusion (Vp) and independently negatively regulated by age, highlighting the key role of hemodynamic insufficiency and age-related endothelial ischemic injury in PWMH pathogenesis; for DWMH, BBB leakage is independently and positively driven by T2DM status, confirming the specific toxic effect of hyperglycemia on deep white matter microvascular endothelial integrity.

Our core conclusions regarding the nonlinear association between BBB leakage and WMH severity, as well as the regional heterogeneity and driver specificity of BBB disruption, are consistent with key studies in the field while offering critical novel insights, validating the multidimensional heterogeneity hypothesis of WMH pathogenesis and revealing the methodological root of prior controversies. First, our finding that WMH K^trans peaks at Fazekas 2 and declines significantly at Fazekas 3 revises the long-held linear assumption that greater WMH burden equals more severe BBB leakage. This aligns with studies by Chung et al. (29), Voorter et al. (16,30), and Verstappen et al. (31), which collectively confirm elevated BBB leakage in early-to-moderate WMH, and reduced K^trans in chronic late lesions due to tissue necrosis and microvascular rarefaction. This is further corroborated by our observation of a continuous, severity-dependent decline in Vp and its positive correlation with K^trans, highlighting the critical need for perfusion correction in BBB leakage assessment. In contrast, studies by Rudilosso et al. (13), Solé-Guardia et al. (8), and Wang et al. (32) have reported a linear increase in WMH BBB leakage with higher Fazekas grades, likely because their cohorts were predominantly enriched with mild-to-moderate WMH, failing to capture the microvascular depletion in severe lesions and fully correct for local perfusion interference. As emphasized by Verstappen et al. (31), BBB permeability parameters reflect the product of vascular permeability and surface area, and reduced leakage metrics in severe WMH may stem from two mechanisms: on the one hand, severe microvascular rarefaction leads to underestimation of total leakage per unit tissue; on the other hand, fibrosis and glial scar formation in late lesions induce vascular wall remodeling or brain tissue inactivation, reducing the actual BBB leakage function. Thus, reduced leakage parameters may represent either a measurement artifact or a true late pathological feature of severe WMH, requiring comprehensive interpretation alongside perfusion metrics. Second, our confirmation of higher BBB leakage in PWMH versus DWMH within the same Fazekas grade is consistent with findings from Chung et al. (29) and Ju et al. (33). Chung et al.’s review noted that PWMH is predominantly characterized by BBB dysfunction and interstitial fluid accumulation, whereas DWMH is more closely linked to impaired glymphatic clearance. Ju et al. further attributed this regional difference to the higher susceptibility of the periventricular region to inflammation-induced BBB damage, with pro-inflammatory factor LP-PLA2 activity only positively correlating with periventricular Fazekas grade. However, Voorter et al. (16) did not observe this regional difference, and Verstappen et al. (31) reported the opposite finding of higher leakage in DWMH. These discrepancies are likely explained by cohort differences: Voorter et al. enrolled a mixed cohort of sporadic and hereditary (CADASIL) SVD, where CADASIL-related BBB damage is driven by aquaporin dysfunction rather than the tight junction disruption seen in sporadic SVD, diluting regional leakage differences; Verstappen et al. used the leakage tissue volume ratio (vl), which reflects overall lesion leakage extent rather than the unit-area leakage intensity (K^trans) used in our study, with the more diffuse nature of DWMH lesions leading to naturally higher vl values despite lower unit microvascular leakage intensity. Finally, our finding that T2DM is an independent risk factor for DWMH BBB leakage is cross-validated by multiple studies. Yu et al. (34) linked soluble epoxide hydrolase-mediated linoleic acid metabolism abnormalities specifically to deep white matter BBB disruption, with pathway products damaging endothelial tight junctions and synergizing with hyperglycemia. Chen et al. (35) confirmed in vitro that age-related endothelial senescence only increases permeability in periventricular endothelial cells, whereas hyperglycemia specifically increases permeability in deep white matter endothelial cells, an effect reversible by metformin. Hashmat et al. (36) further demonstrated that hyperglycemia induces tight junction protein degradation and inflammatory factor upregulation specifically in deep white matter microvessels, independent of hypertension and hyperlipidemia. Another study (37) have failed to observe this driver specificity, largely due to cohort heterogeneity, limited sample size, and lack of subtype stratification; notably, Vipin et al. (38) linked elevated glycated hemoglobin to BBB dysfunction and WMH burden, but only attributed this to global perfusion impairment rather than the specific regulatory effect of diabetes on DWMH BBB leakage. In summary, discrepancies in prior studies stem primarily from the lack of WMH subtype stratification, incomplete perfusion correction, and differences in cohort disease stage. Our refined dual-stratified design with synchronous Vp perfusion correction more accurately characterizes the heterogeneous features of BBB leakage, providing targeted imaging evidence to resolve long-standing controversies in the field.

Our findings provide critical new insights into the clinical evaluation and intervention of CSVD, and identify key limitations of current imaging diagnostic paradigms. Our observation that BBB leakage metrics are lower in severe versus moderate WMH directly resolves a long-standing clinical paradox: patients with severe WMH have markedly elevated risks of stroke and cognitive decline, yet their BBB leakage indices do not rise synchronously and even decrease, a discrepancy that has long lacked definitive imaging-based clarification. The core mechanism underlying this deviation is severe microvascular rarefaction and drastically reduced local perfusion in advanced lesions, where limited contrast delivery masks the inherently high vascular wall permeability, rather than reflecting a true reversal of BBB damage. Representative cases are shown in Figure 5. Thus, clinical interpretation must integrate both perfusion and BBB leakage metrics to accurately characterize microvascular damage in severe lesions and avoid misjudging lesion activity; furthermore, incorporating leakage and perfusion assessments of the WMH-NAWM junction—the leading edge of lesion progression—enables early warning of WMH advancement and addresses the limitation of current diagnostics focused solely on established lesions. The confirmed regional heterogeneity of WMH necessitates a paradigm shift from a one-size-fits-all treatment model to subtype-specific precision therapy: we hypothesize that perfusion-improving strategies may benefit patients with predominant PWM, and tight glycemic control could alleviate endothelial injury in diabetic patients with DWMH; these subtype-targeted management conjectures remain unproven without longitudinal outcome and therapeutic response data, and need to be verified in subsequent clinical studies. Moderate lesions may represent a relatively active phase of BBB leakage and a potential intervention window; however, longitudinal studies assessing BBB leakage and lesion progression are needed to determine whether the observed Fazekas 2 pattern truly reflects a transitional phase of disease activity, and whether early targeted treatment can effectively halt lesion progression; in contrast, severe lesions with irreversible microvascular depletion exhibit lower apparent leakage values, which may be partially driven by marked local hypoperfusion alongside other late-stage tissue pathological changes, and such lesions offer minimal intervention benefit, requiring a focus on global brain function rather than BBB-targeted treatment alone. These findings also highlight the need for dual anatomical and etiological stratification in future clinical trials: enrollment should not be based solely on total lesion visual scores, but also on dominant lesion subtypes, as mixed pathogenesis may dilute apparent drug efficacy. From a pathophysiological perspective, our results delineate the multidimensional mechanisms of CSVD progression, extending beyond chronic hypoperfusion-induced ischemic demyelination to include specific metabolic toxicity-mediated endothelial injury of deep penetrating arteries, and late-stage reduced leakage due to microvascular bed collapse. These findings revise the long-held linear model of monotonically progressive BBB damage with disease course, clarifying the dynamic pathological evolution from functional hyperleakage to structural hypoperfusion, and explaining the poor vasodilator response observed in some patients with advanced CSVD. For elderly patients, the independent negative association of age with PWMH leakage likely reflects age-related impairment of cerebral microcirculatory compensatory capacity, meaning interventions must balance BBB protection and microcirculatory support to avoid exacerbating hypoperfusion via unregulated blood pressure reduction. In summary, this study underscores the critical value of integrating perfusion parameters into multimodal MRI assessment, provides a robust imaging and pathophysiological foundation for early screening, risk stratification, and individualized CSVD treatment, and advances the transition from generalized CSVD management to precision stratification, with important implications for improving patient prognosis and reducing the risks of stroke and cognitive impairment.

Figure 5 T2WI-FLAIR images and DCE-MRI quantitative maps of WMH with different Fazekas grades. (A) Both periventricular and deep WMH are Fazekas grade 1 (T2WI-FLAIR). (B) Corresponding K^trans map for (A), representing blood-brain barrier leakage. (C) Corresponding Vp map for (A), representing local microvascular perfusion. (D) Both periventricular and deep WMH are Fazekas grade 2 (T2WI-FLAIR). (E) Corresponding K^trans map for (D). (F) Corresponding Vp map for (D). (G) Both periventricular and deep WMH are Fazekas grade 3 (T2WI-FLAIR). (H) Corresponding K^trans map for (G). (I) Corresponding Vp map for (G). DCE-MRI, dynamic contrast-enhanced magnetic resonance imaging; T2WI-FLAIR, T2-weighted imaging fluid-attenuated inversion recovery; WMH, white matter hyperintensity.

This study has several limitations that warrant consideration when interpreting its findings and generalizability. First, this single-center retrospective cohort study enrolled 75 Han Chinese participants, with a limited overall sample size and a small severe WMH subgroup, which may introduce selection bias; our conclusions therefore require validation in large, multicenter cohorts. Second, we used the Fazekas visual semi-quantitative grading system for WMH severity stratification, a widely standardized tool in the CSVD field with inherent limitations: visual scoring is subject to observer dependence, and the 0–3 grading scale cannot accurately quantify absolute lesion volume and spatial distribution, which may confound intergroup comparisons of BBB leakage parameters. Future studies should incorporate three-dimensional quantitative lesion volume metrics for more refined stratification to mitigate bias from semi-quantitative visual grading. Third, DCE-MRI parameter quantification was performed using a United Imaging 3.0T MRI scanner and the Extended Tofts two-compartment model; MRI scanners with different field strengths and alternative post-processing algorithms may introduce systematic deviations in K^trans and Vp measurements, necessitating multi-device and multi-model consistency validation to standardize quantitative BBB leakage detection. Additionally, ROI delineation may be affected by partial volume effects, and future work could incorporate AI-based automatic segmentation to improve the accuracy and repeatability of lesion localization. Fourth, our cross-sectional design only characterizes BBB leakage features across WMH lesion stages, but cannot dynamically track temporal changes in BBB permeability during WMH progression, nor establish a direct association between BBB leakage metrics and clinical endpoints such as cognitive decline and ischemic stroke; the prognostic value of these imaging markers therefore requires validation in a long-term follow-up cohort. Furthermore, we did not include cerebrospinal fluid inflammatory factors, peripheral blood neurovascular injury biomarkers, or histopathological data for correlation analysis, and thus cannot fully distinguish whether reduced K^trans in severe WMH represents measurement bias from perfusion interference or a true late-stage pathological change; future studies combining multi-omics profiling and histopathological validation are needed to further elucidate the underlying molecular mechanisms. Finally, we excluded Fazekas 3 severe lesions when constructing our multiple linear regression model, and thus could not identify independent drivers of BBB leakage in severe lesions; this requires supplementary analysis after expanding the sample size of the severe WMH subgroup in future work.

Conclusions

In summary, through the refined dual-stratified design of anatomical location and lesion severity, this study systematically clarifies the dual heterogeneous characteristics of BBB leakage of WMH in sporadic CSVD population, revises the inherent belief in the field that BBB damage increases linearly with the increase of WMH burden, confirms that Fazekas 2 is the pathologically active intervention window of BBB leakage, and reveals that local hypoperfusion-induced contrast delivery limitation is a potential major factor leading to reduced apparent BBB leakage metrics in severe WMH, while concurrent vascular remodeling and microvascular loss also participate in modulating DCE quantitative parameters. This mechanistic inference needs to be validated in longitudinal histopathology or imaging cohorts. More importantly, this study confirms that BBB damage of PWMH and DWMH has differential pathological regulatory pathways, clarifies that age and perfusion level are the core driving factors of BBB damage in PWMH, whereas T2DM has a specific regulatory effect on BBB leakage of DWMH, providing direct imaging and pathophysiological basis for anatomical subtype classification of CSVD. This series of findings not only fills the research gap in the heterogeneous pathological mechanism of WMH, but also provides quantifiable imaging markers for the clinical realization of risk stratification, precise intervention of CSVD and individualized decision-making of stroke thrombolytic therapy.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0878/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0878/dss

Funding: This study was supported by the Scientific and Technological Research Program of Chongqing Municipal Education Commission (Nos. KJQN202400112 and KJQN202500102), Natural Science Foundation of Chongqing, China (No. CSTB2025NSCQ-GPX1064), Beijing Medical Award Foundation Project (No. YXJL-2023-0227-0089), and Cultivation Talents Project of Chongqing University Jiangjin Hospital (No. 2025YCXM001).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-0878/coif). The 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. This study was approved by the Ethics Committee of Jiangjin Central Hospital of Chongqing (Approval No. KY20240812-007). Written informed consent was waived due to the retrospective study design.

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. Osman A, Kanaan A, Azeroual S, Abubakr M, Jwayyed J, Bhutta R. Hypertension and Cerebral Small Vessel Disease: A Review of the Pathophysiology, Progression, and Prevention. Cureus 2025;17:e92760. [Crossref] [PubMed]
2. Okine DN, Kern KC, Luby M, Latour LL, Gottesman RF. Blood-brain barrier disruption mediates the association between cerebral small vessel disease and clinical outcome after stroke: a secondary analysis of the Lesion Evolution in Stroke and Ischemia on Neuroimaging study. Front Stroke 2024;3:1510359. [Crossref] [PubMed]
3. Chen Y, Kong Q, Wang Z, Li J, Luo Y, Zhang Y, Yu Z, Huang H, Luo X. Correlation between cerebral small vessel disease and late-life depression: Insights from neuroimaging studies. Brain Res 2025;1868:150001. [Crossref] [PubMed]
4. Li Y, Zhang X, Wang H, Jiang L, Gu Z, Zhou J, Liang R. Exploration of heterogeneity in risk factors associated with imaging subtypes of white matter hyperintensities on fluid-attenuated inversion recovery magnetic resonance imaging. Front Neurol 2025;16:1647065. [Crossref] [PubMed]
5. Kern KC, Zagzoug MS, Gottesman RF, Wright CB, Leigh R. Blood-brain barrier disruption and increased free water are associated with worse cognitive performance in patients with chronic cerebrovascular disease. Neuroimage Clin 2024;44:103706. [Crossref] [PubMed]
6. Xiufu Z, Ruipeng L, Jun Z, Yonglong L, Yulin W, Jian Z, Xianglin C, Lan S, Zuowen Z. Analysis of influencing factors of early neurological improvement after intravenous rt-PA thrombolysis in acute anterior circulation ischemic stroke. Front Neurol 2022;13:1037663. [Crossref] [PubMed]
7. Frey BM, Shenas F, Boutitie F, Cheng B, Cho TH, Ebinger M, et al. Intravenous Thrombolysis in Patients With White Matter Hyperintensities in the WAKE-UP Trial. Stroke 2023;54:1718-25. [Crossref] [PubMed]
8. Solé-Guardia G, Janssen A, Wolters R, Dohmen T, Küsters B, Claassen JA, Leeuw FE, Wiesmann M, Gutierrez J, Kiliaan AJ. Impact of hypertension on cerebral small vessel disease: A post-mortem study of microvascular pathology from normal-appearing white matter into white matter hyperintensities. J Cereb Blood Flow Metab 2025;45:1717-30. [Crossref] [PubMed]
9. Yılmaz H, Bayraktutan U. Cerebral Small Vessel Disease: Therapeutic Approaches Targeting Neuroinflammation, Oxidative Stress, and Endothelial Dysfunction. Curr Issues Mol Biol 2025;47:232. [Crossref] [PubMed]
10. Wang H, Ren X, Shen Y, Qiu Y, Li M, Luo S, Zhang Z, Wu L, Zhou M, Jiang Y, Zhou F, Hong D. Meningeal lymphatic dysfunction in sporadic cerebral small vessel diseases and cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy by DCE-MRI. Quant Imaging Med Surg 2025;15:6692-704. [Crossref] [PubMed]
11. Vikner T, Garpebring A, Björnfot C, Malm J, Eklund A, Wåhlin A. MRI contrast accumulation in features of cerebral small vessel disease: blood-brain barrier dysfunction or elevated vascular density? Fluids Barriers CNS 2025;22:74. [Crossref] [PubMed]
12. Ling C, Zhang J, Shao X, Bai L, Li Z, Sun Y, Li F, Wang Z, Xue R, Zhuo Y, Yang Q, Zhang Z, Wang DJJ, Yuan Y. Diffusion prepared pseudo-continuous arterial spin labeling reveals blood-brain barrier dysfunction in patients with CADASIL. Eur Radiol 2023;33:6959-69. [Crossref] [PubMed]
13. Rudilosso S, Stringer MS, Thrippleton M, Chappell F, Blair GW, Jaime Garcia D, Doubal F, Hamilton I, Janssen E, Kopczak A, Ingrisch M, Kerkhofs D, Backes WH, Staals J, Duering M, Dichgans M, Wardlaw JM. SVDs@target consortium. Blood-brain barrier leakage hotspots collocating with brain lesions due to sporadic and monogenic small vessel disease. J Cereb Blood Flow Metab 2023;43:1490-502. [Crossref] [PubMed]
14. Kerkhofs D, Wong SM, Zhang E, Staals J, Jansen JFA, van Oostenbrugge RJ, Backes WH. Baseline Blood-Brain Barrier Leakage and Longitudinal Microstructural Tissue Damage in the Periphery of White Matter Hyperintensities. Neurology 2021;96:e2192-200. [Crossref] [PubMed]
15. Maillard P, Fletcher E, Lockhart SN, Roach AE, Reed B, Mungas D, DeCarli C, Carmichael OT. White matter hyperintensities and their penumbra lie along a continuum of injury in the aging brain. Stroke 2014;45:1721-6. [Crossref] [PubMed]
16. Voorter PHM, Stringer MS, van Dinther M, Kerkhofs D, Dewenter A, Blair GW, et al. Heterogeneity and Penumbra of White Matter Hyperintensities in Small Vessel Diseases Determined by Quantitative MRI. Stroke 2025;56:128-37. [Crossref] [PubMed]
17. Dewey BE, Xu X, Knutsson L, Jog A, Prince JL, Barker PB, van Zijl PCM, Leigh R, Nyquist P. MTT and Blood-Brain Barrier Disruption within Asymptomatic Vascular WM Lesions. AJNR Am J Neuroradiol 2021;42:1396-402. [Crossref] [PubMed]
18. Lim C, Lee H, Moon Y, Han SH, Kim HJ, Chung HW, Moon WJ. Volume and Permeability of White Matter Hyperintensity on Cognition: A DCE Imaging Study of an Older Cohort With and Without Cognitive Impairment. J Magn Reson Imaging 2025;61:2260-70. [Crossref] [PubMed]
19. Liakakis G, Vakrakou AG, Boufidou F, Constantinides V, Velonakis G, Paraskevas GP, Stefanis L, Kapaki E. Exploratory Analysis of Cerebrospinal Fluid IL-6 and IL-17A Levels in Subcortical Small-Vessel Disease Compared to Alzheimer’s Disease: A Pilot Study. Diagnostics (Basel) 2025;15:669. [Crossref] [PubMed]
20. Su C, Cui Z, Guo J. Insulin resistance in cerebral small vessel disease: a mini review. Front Neurosci 2026;20:1760558. [Crossref] [PubMed]
21. Wang LL, Williamson BJ, Zhang B, Sriwastawa A, Aziz YN, Antzoulatos E, Stephens CB, Vagal A, Khatri P, DiFrancesco MW. Microstructural and microvascular features of white matter hyperintensities and their association with small vessel disease markers. Sci Rep 2025;15:18567. [Crossref] [PubMed]
22. Haller S, Kövari E, Herrmann FR, Cuvinciuc V, Tomm AM, Zulian GB, Lovblad KO, Giannakopoulos P, Bouras C. Do brain T2/FLAIR white matter hyperintensities correspond to myelin loss in normal aging? A radiologic-neuropathologic correlation study. Acta Neuropathol Commun 2013;1:14.
23. Wardlaw JM, Valdés Hernández MC, Muñoz-Maniega S. What are white matter hyperintensities made of? Relevance to vascular cognitive impairment. J Am Heart Assoc 2015;4:001140.
24. Tofts PS, Brix G, Buckley DL, Evelhoch JL, Henderson E, Knopp MV, Larsson HB, Lee TY, Mayr NA, Parker GJ, Port RE, Taylor J, Weisskoff RM. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999;10:223-32. [Crossref] [PubMed]
25. Sourbron SP, Buckley DL. Classic models for dynamic contrast-enhanced MRI. NMR Biomed 2013;26:1004-27. [Crossref] [PubMed]
26. Uchida Y, Kano Y, Kan H, Sakurai K, Morita H, Akagawa Y, Matsukawa N, Oishi K. Blood-brain barrier water exchange and paramagnetic susceptibility alterations during anti-amyloid therapy: preliminary MRI findings. J Prev Alzheimers Dis 2025;12:100256. [Crossref] [PubMed]
27. Uchida Y, Kan H, Sakurai K, Horimoto Y, Hayashi E, Iida A, Okamura N, Oishi K, Matsukawa N. APOE ɛ4 dose associates with increased brain iron and β-amyloid via blood-brain barrier dysfunction. J Neurol Neurosurg Psychiatry 2022;jnnp-2021-328519.
28. Uchida Y, Kan H, Sakurai K, Arai N, Inui S, Kobayashi S, Kato D, Ueki Y, Matsukawa N. Iron leakage owing to blood-brain barrier disruption in small vessel disease CADASIL. Neurology 2020;95:e1188-98. [Crossref] [PubMed]
29. Chung J, Oh H, Gwak DS, Kim DE. Multimodal Magnetic Resonance Imaging Signatures of White Matter Hyperintensities: Mechanistic Insights Into Pathobiological Heterogeneity. J Stroke 2026;28:1-28. [Crossref] [PubMed]
30. Voorter PHM, van Dinther M, Jansen WJ, Postma AA, Staals J, Jansen JFA, van Oostenbrugge RJ, van der Thiel MM, Backes WH. Blood-Brain Barrier Disruption and Perivascular Spaces in Small Vessel Disease and Neurodegenerative Diseases: A Review on MRI Methods and Insights. J Magn Reson Imaging 2024;59:397-411. [Crossref] [PubMed]
31. Verstappen D, de Jong JJA, Voorter PHM, van Dinther M, van Oostenbrugge RJ, Staals J, Jansen JFA, Backes WH. DCE-MRI reveals spatial pattern in heterogeneous blood-brain barrier leakage within white matter in cerebral small vessel disease. J Cereb Blood Flow Metab 2025;45:2104-14. [Crossref] [PubMed]
32. Wang X, Shi Y, Chen Y, Gao Y, Wang T, Li Z, Wang Y. Blood-Brain Barrier Breakdown is a Sensitive Biomarker of Cognitive and Language Impairment in Patients with White Matter Hyperintensities. Neurol Ther 2023;12:1745-58. [Crossref] [PubMed]
33. Chen J, He L, Mu X, Fang L. The association of lipoprotein-associated phospholipase A2 with cerebral white matter hyperintensity and dizziness in patients with cerebral small vessel disease. J Stroke Cerebrovasc Dis 2026;35:108537. [Crossref] [PubMed]
34. Yu D, Liang N, Zebarth J, Shen Q, Ozzoude M, Goubran M, et al. Soluble Epoxide Hydrolase Derived Linoleic Acid Oxylipins, Small Vessel Disease Markers, and Neurodegeneration in Stroke. J Am Heart Assoc 2023;12:e026901. [Crossref] [PubMed]
35. Chen JXY, Vipin A, Sandhu GK, Leow YJ, Zailan FZ, Tanoto P, Lee ES, Lee KL, Cheung C, Kandiah N. Blood-brain barrier integrity disruption is associated with both chronic vascular risk factors and white matter hyperintensities. J Prev Alzheimers Dis 2025;12:100029. [Crossref] [PubMed]
36. Hashmat A, Ya J, Kadir R, Alwjwaj M, Bayraktutan U. Hyperglycaemia perturbs blood-brain barrier integrity through its effects on endothelial cell characteristics and function. Tissue Barriers 2025;13:2350821. [Crossref] [PubMed]
37. Leigh R, Kern KC, Wang NY, Gottesman RF, Wright CB. Disruption of the Blood-Brain Barrier Predicts Progression of Cerebral Small Vessel Disease White Matter Hyperintensities. Ann Neurol 2026;99:437-48. [Crossref] [PubMed]
38. Vipin A, Chen JXY, Tee M, Hilal S, Leow YJ, Konstandin S, Eickel K, Günther M, Petr J, Mutsaerts HJ, Kandiah N. Cerebral haemodynamics and white matter hyperintensities: findings using non-invasive brain imaging. J Prev Alzheimers Dis 2026;13:100479. [Crossref] [PubMed]