Plaque phenotypes across remodeling patterns in vertebrobasilar atherosclerosis: a high-resolution magnetic resonance vessel wall imaging study

Introduction

A predominant etiology of ischemic stroke in Asian populations is intracranial atherosclerosis, accounting for approximately 30–50% of cases (1-3). Atherosclerotic progression is intrinsically linked to vascular remodeling processes (4), and in pathological transformations, blood vessels exhibit distinct remodeling phenotypes (5). Positive remodeling (PR) manifests as outward expansion of the vascular wall, preserving lumen patency while accommodating plaque growth. Conversely, negative remodeling (NR) involves a paradoxical concentric constriction at the lesion sites, exacerbating luminal compromise. Finally, intermediate remodeling (IR) is a transitional state between the positive and negative types (Figure 1). The characterization of vascular remodeling patterns can provide critical insights for stroke risk stratification and therapeutic decision-making (5).

Figure 1 Remodeling patterns. Categories of remodeling: negative (inward) remodeling is characterized by constriction of the outer wall (delineated by the green circle with a reduction in lumen size); expansion of the outer wall during plaque development with relative preservation of the lumen constitutes positive (outward) remodeling. Variables of interest include LA (area shown in black), WA (area shaded in gray), and VA (area circumscribed by the green outer wall). LA, lumen area; VA, vessel area; WA, wall area.

It has been recently established that positive vascular remodeling contributes to the accumulation of substantial plaque volume (6,7), which is strongly associated with the emergence of stroke (8) and may confer a higher risk of stroke compared to NR (7-9). Consequently, the accurate characterization of remodeling phenotypes—readily achievable through high-resolution magnetic resonance vessel wall imaging (HR-MRI) (5,7)—has become an essential stratification tool in clinical practice.

Beyond vascular remodeling assessment, HR-MRI demonstrates diagnostic equivalence to digital subtraction angiography (DSA) in quantifying stenosis severity (10). Crucially, HR-MRI enables the comprehensive characterization of intracranial arterial plaque features—including enhancement patterns (11-13), burden (5), surface morphology (9,13), longitudinal extent, and spatial distribution—unattainable through conventional angiographic techniques such as magnetic resonance angiography (MRA) or computed tomography angiography (CTA) (14). Critically, however, it is not clear whether plaque phenotypes exhibit differential expression across remodeling subtypes.

Intracranial vertebrobasilar atherosclerosis is the primary cause of posterior circulation stroke in East Asian populations. Crucially, incomplete characterization of the atheromatous plaque features may underlie the suboptimal therapeutic outcomes among patients with condition (14). Consequently, the comprehensive assessment of vertebrobasilar atherosclerotic phenotype may be a key clinical measure (15). We postulated that the stroke risk profiles differ significantly between positive and nonpositive remodeling (non-PR) patterns in patients with intracranial vertebrobasilar disease. To confirm this, we employed HR-MRI to systematically evaluate intracranial vertebrobasilar atheromatous lesions, characterize plaque pathomorphology across remodeling subtypes, and determine the associations between remodeling-specific plaque features and stroke occurrence. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1812/rc).

Methods

Study design and patient selection

This retrospective cohort study was approved by the Institutional Review Board of Zhongnan Hospital (approval No. 2025190K) and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The requirement for informed consent was waived due to the retrospective nature of the analysis.

We consecutively enrolled patients admitted to the Neurology Department of Zhongnan Hospital with suspected posterior circulation ischemia between January 2020 and July 2021. Patients exhibiting symptoms suggestive of posterior circulation ischemia (e.g., vertigo, diplopia, ataxia, dysarthria, or visual field defects) and/or with imaging evidence of infarction in the posterior circulation territory were selected for inclusion in the study.

MRI protocol

All participants underwent standardized MRI within 14 days of admission, which comprised the following sequences: (I) structural sequences, including T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and T2-fluid-attenuated inversion recovery (T2-FLAIR); (II) functional sequences, including diffusion-weighted imaging (DWI); and (III) angiographic sequences, including three-dimensional time-of-flight MRA (3D-TOF MRA) and contrast-enhanced T1W1 (CE-T1WI).

Inclusion criteria

The inclusion criteria were as follows: (I) radiologically confirmed intracranial vertebrobasilar atherosclerotic lesions; (II) ≥1 established atherosclerosis risk factor, including hypertension (≥140/90 mmHg or antihypertensive treatment), diabetes mellitus (fasting glucose ≥7.0 mmol/L or hypoglycemic agents), dyslipidemia [low-density lipoprotein cholesterol (LDL-C) ≥3.4 mmol/L or lipid-lowering therapy]; obesity [body mass index (BMI) ≥28 kg/m²] and smoking history; and (III) diagnostic-quality HR-MRI permitting plaque characterization.

Exclusion criteria

The exclusion criteria were as follows: (I) significant extracranial vertebral stenosis (>50% ipsilateral lumen reduction); (II) nonatherosclerotic vasculopathies, including intracranial aneurysms, moyamoya disease, vasculitis, arterial dissection, reversible cerebral vasoconstriction syndrome, and vascular malformations; (III) cardioembolic stroke risk factors, including atrial fibrillation, recent myocardial infarction (<6 months), patent foramen ovale with right-to-left shunt, significant valvular heart disease, and cardiomyopathy (ejection fraction <40%); and (IV) incomplete clinical/imaging data.

Patient stratification

Participants were classified into two groups according to DWI findings and clinical symptomatology. The stroke group consisted of patients with acute ischemic lesions in the posterior circulation territory on DWI and corresponding neurological deficits lasting >24 hours. Meanwhile, the nonstroke group consisted of patients lacking DWI abnormalities and having transient symptoms (<24 hours) or nonspecific findings.

Demographic characteristics and clinical parameters were systematically collected from electronic medical records and comprised the following: basic demographics, including age, gender, and BMI; behavioral risk factors, including smoking history; comorbidities, including hypertension, diabetes mellitus, dyslipidemia, coronary artery disease (CAD); a history of previous stroke; and laboratory biomarkers, including fasting plasma glucose, a comprehensive lipid panel [total cholesterol, triglycerides, LDL-C, and high-density lipoprotein cholesterol (HDL-C)], and serum homocysteine concentration, among others. All parameters were extracted within 72 hours of hospital admission.

Protocol for HR-MRI

All scans were performed on a 3T whole-body MRI system (MAGNETOM Prisma; Siemens Healthineers, Erlangen, Germany) equipped with a 64-channel phased-array head/neck coil. 3D-TOF MRA was performed in the axial orientation. The imaging parameters were as follows: repetition time/echo time (TR/TE) =21.0/3.4 ms, field of view (FOV) =200 mm × 80 mm, slice thickness =0.6 mm, and acquisition matrix =320×250. The acquired high-resolution source images were subsequently processed with maximum intensity projection (MIP) reconstruction to generate three-dimensional angiographic views. The HR-MRI protocol included three sequences with one being sequence repeated postcontrast: precontrast T1-weighted fast spin-echo (FSE) [TR/TE =600 ms/13 ms, echo train length (ETL) =9; number of excitations (NEX) =2], T2-weighted FSE (TR/TE =1,300 ms/124 ms, slice thickness =0.9 mm, ETL =13, NEX =2), and postcontrast T1-weighted FSE (slice thickness =0.7 mm, in-plane resolution =0.7 mm × 0.7 mm, FOV =180 mm × 180 mm, matrix size =256×256). Proximal saturation bands were used for flow suppression, while clinical DWI and T2-FLAIR sequences were used for infarct identification.

Image analysis

All imaging datasets were independently evaluated on a picture archiving and communication system (PACS) workstation by two board-certified neuroradiologists blinded to the clinical data. Interrater discrepancies were adjudicated by a senior neuroradiologist (≥15 years’ experience) to establish a consensus. Intraobserver reproducibility was assessed through re-evaluation of 30 randomly selected cases by both readers at 4-week intervals, with the intraclass correlation coefficient (ICC) being calculated for agreement metrics.

Plaque characterization and remodeling metrics

Atherosclerotic plaques were radiologically defined as eccentric wall thickening visualized on both pre- and postcontrast MRI sequences, irrespective of luminal stenosis. Quantitative measurements were performed on cross-sectional T1-weighted images. Vessel area (VA) and lumen area (LA) were assessed on cross-sectional T1-weighted images at the maximum lumen narrowing (MLN) and reference sites. Reference sites were selected according to the Warfarin-Aspirin Symptomatic Intracranial Disease Study method (16). The plaque-free or minimal disease burden segment nearest to the stenosis site was selected as the reference site. If a proximal reference site was not available, the adjacent distal site was used. The remodeling index (RI) was calculated as follows: RI = VA_MLN/VA_Reference. RI ≤0.95 was defined as NR, RI >0.95 and <1.05 was defined as IR, and RI ≥1.05 was defined as PR (6). Non-PR was considered to be NR or IR (6). Measurement of plaque burden was conducted at the site of maximum stenosis, which was defined as follows: plaque burden = (1 – LA_MLN/VA_MLN) ×100% (5). The degree of stenosis was calculated as follows: degree of stenosis (1 − LA_MLN/LA_Reference) ×100% (16).

Plaque enhancement was quantified by comparing signal intensity (SI) ratios between pre- and postcontrast 3D T1-weighted sampling perfection with application-optimized contrasts using different flip-angle evolutions (SPACE) sequences, with the pituitary infundibulum serving as the internal reference. No enhancement was defined as an SI ratio less than or equal to that of the adjacent plaque-free vessel wall, mild enhancement as an SI ratio greater than that of the plaque-free wall but less than that of the pituitary infundibulum, and marked enhancement as an SI ratio greater or equal to that of the pituitary infundibulum (12). Nonmarked enhancement was considered to be no or mild enhancement (12). Plaque distribution was divided into focal (<50% axial vessel circumference involvement) and diffuse conditions (≥50% circumferential involvement). The plaque surface morphology was divided into regular and irregular types: regular, smooth luminal contour; irregular, and ulcerated/ruptured surface. Maximum plaque length was measured along vessel axis on postcontrast reconstructions at the site of maximal wall thickness, while maximum wall thickness was assessed perpendicular to lumen at the MLN site. Plaque geometry index was calculated as follows: plaque geometry index = (maximum length)/(maximum thickness) (Figure 2).

Figure 2 A 63-year-old male with a 2-day history of dizziness accompanied by right-sided limb weakness. DWI and ADC maps demonstrated restricted diffusion within the left brainstem (A,B), indicating an acute ischemic lesion. MRA revealed severe stenosis of the basilar artery (C). Vessel wall plaque characterization (3D-T1-SPACE): precontrast imaging perpendicular to the long axis of the basilar artery demonstrated an irregular plaque predominantly located in the ventral and left lateral vessel walls (D). Postcontrast imaging showed significant plaque enhancement (E,F). Vessel and lumen areas: VA and LA were measured at the site of MLN and at the proximal reference site (G,H). The RI was calculated as follows: RI = (VA at MLN/VA at reference) =0.1654 cm²/0.2164 cm² =0.76. The RI was <0.95, indicating negative remodeling. WA and plaque burden: WA =VA – LA. WA at MLN =0.1654 cm² – 0.01468 cm²=0.1507 cm². WA at reference =0.2164 cm² – 0.06376 cm² =0.1526 cm². Plaque burden at MLN = (WA at MLN/VA at MLN) =0.1507 cm²/0.1654 cm² =0.91 (91%). Luminal stenosis: percentage luminal area stenosis at MLN = [1 – (LA at MLN/LA at reference)] × 100% = [1 – (0.01468 cm²/0.06376 cm²)] × 100% =77.0%. The white arrows serve to highlight key pathological features: acute ischemic lesion (A,B), severe basilar artery stenosis (C), vulnerable plaque (irregular, enhancing) (D-F), quantitative measurement of plaque characteristics (G,H). 3D-T1-SPACE, three-dimensional T1-weighted sampling perfection with application-optimized contrasts using different flip-angle evolutions; ADC, apparent diffusion coefficient; DWI, diffusion-weighted image; LA, lumen area; MLN, maximal lumen narrowing; MRA, magnetic resonance angiography; RI, remodeling index; VA, vessel area; WA, wall area.

Statistical analysis

All analyses were conducted in R v.4.0.5 (The R Foundation for Statistical Computing, Vienna, Austria). Continuous variables are summarized as the mean ± standard deviation (SD) for data with a normal distribution (Shapiro-Wilk test) or as the median and interquartile range (IQR) for data with a nonnormal distribution. Categorical variables are reported as frequencies (percentages). Intergroup comparisons were conducted via the Kruskal-Wallis test for continuous variables and the Chi-squared test for categorical variables. Initial analysis compared the characteristics across the PR, NR, and IR groups. Due to the similarity in clinical profiles between the NR and IR groups, they were combined into a non-PR group for the primary comparative analysis to better distinguish the features of PR. All covariates were screened for their association with stroke occurrence within the PR and non-PR groups, respectively, via univariable logistic regression analysis. Covariates with P≤0.10 in the univariate models were included in stepwise backward elimination via multivariate analysis. The adjusted odds ratio (aOR) and 95% confidence intervals (CIs) were calculated, and multicollinearity was assessed via the variance inflation factor (VIF), with a VIF <5 indicating no concern for multicollinearity. Receiver operating characteristic (ROC) curves were generated for significant predictors, and area under the curve (AUC) values with the 95% CI were calculated via the DeLong method. The AUC served as a global measure of discriminatory power independent of the selected decision threshold and was interpreted as follows: 0.9–1.0, excellent discrimination; 0.8–0.9, good discrimination; 0.7–0.8, fair or acceptable discrimination; 0.6–0.7, poor discrimination; and 0.5–0.6, failure to discriminate (no better than chance).

Interobserver agreement for plaque features was determined according to the ICC (two-way mixed-effects model) as follows: ICC <0.40, poor consistency; ICC ≥0.40 and ≤0.75, moderate consistency; and ICC >0.75, excellent consistency. For all tests, a two-sided P<0.05 was considered statistically significant.

Results

Baseline characteristics of the remodeling subtypes

The study cohort comprised 208 patients with intracranial vertebrobasilar atherosclerosis. The median age was 61 years (IQR 54–68 years), the proportion of males was 73.1% (152/208), and the median BMI was 25.39 kg/m² (IQR 23.25–27.14 kg/m²). Patients were stratified by remodeling phenotype, with 133 cases of PR (stroke incidence: 27.8%, 37/133), 35 cases of IR (stroke incidence: 34.3%, 12/35), and 40 cases of NR (stroke incidence: 25.0%, 10/40) (Figure S1). Table 1 presents the comparison of the clinical characteristics and plaque morphology between the remodeling subtypes.

No statistically significant intergroup differences were observed in terms of baseline demographics or serological markers. This included age, sex distribution, and BMI; comorbidity profiles (hypertension, diabetes, and dyslipidemia); and metabolic parameters (fasting glucose, lipid profiles, and serum homocysteine). In terms of imaging characteristics, the NR group, as compared to the IR and PR groups, exhibited a significantly higher median plaque burden (0.87 vs. 0.80 and 0.87 vs. 0.77; P<0.001) and median stenosis rate (79% vs. 42% and 79% vs. 22%; P<0.001) (Figure 3). In addition, NR lesions, compared to PR lesions, had a higher prevalence of diffuse distribution (80.0% vs. 53.4%; P=0.008). Comparable results (P>0.05) were noted for plaque morphometrics (length, maximal thickness, and length-to-thickness ratio), enhancement patterns, surface morphology characteristics, and plaque location distribution patterns.

Figure 3 Vascular remodeling-dependent distribution of hemodynamic parameters. Violin plots with embedded boxplots demonstrating the distribution of (A) luminal stenosis severity rate and the (B) plaque burden rate across vascular remodeling phenotypes, including PR (n=133), IR (n=35), and NR (n=40). *, P<0.05; ***, P<0.001; NS, not significant. IR, intermediate remodeling; NR, negative remodeling; PR, positive remodeling.

Stroke risk factors in the non-PR group

Univariate logistic regression analysis of the non-PR group identified the risk factors for stroke to be BMI [odds ratio (OR) 1.23; 95% CI: 1.00–1.51; P=0.049], stenosis rate (per 10% increase; OR 1.27; 95% CI: 1.04–1.54; P=0.019), plaque burden (per 10% increase; OR 1.90; 95% CI: 1.13–3.18; P=0.015), plaque enhancement (OR 5.98; 95% CI: 2.03–18.9; P=0.001), and plaque location in the basilar artery (basilar artery vs. right vertebral artery; OR 3.38; 95% CI: 1.05–12.0; P=0.042). In addition, plaque surface irregularity was slightly associated with stroke occurrence (OR 3.37; 95% CI: 0.97–16.5; P=0.056) (Table 2). Multicollinearity assessment revealed no concerning correlations among the above-mentioned predictors (VIF <5). After covariate adjustment, multivariate analysis incorporating these variables identified two independent predictors: marked plaque enhancement (aOR 6.01; 95% CI: 1.48–24.38; P=0.012) and basilar artery plaque location (basilar artery vs. right vertebral artery; aOR 5.41; 95% CI: 1.21–24.19; P=0.027) (Figure 4).

Figure 4 Multivariate predictors of stroke risk stratified by remodeling phenotype. Forest plot displaying the aORs from multivariate logistic regression models for the nonpositive remodeling group (n=75) and the positive remodeling group (n=133). *, per 10% increase odds ratio change. aORs, adjusted odds ratios; CI, confidence interval.

Stroke risk factors in the PR group

Univariate logistic regression analysis of PR group identified the risk factors for stroke to be stenosis rate (per 10% increase; OR 1.29, 95% CI: 1.13–1.49; P<0.001), plaque burden (per 10% increase; OR 1.95; 95% CI: 1.35–2.81; P<0.001), plaque enhancement (OR 7.16; 95% CI: 3.10–17.2; P<0.001), plaque surface irregularity (OR 5.69; 95% CI: 2.19–18.1; P<0.001), and diffuse distribution (OR 3.14; 95% CI: 1.40–7.54) (Table 3). Multicollinearity assessment indicated no concerning correlations among the above-mentioned predictors (VIF <5). These variables were included in multivariate logistic regression analysis, and after adjustment for covariates, the independent risk factors for stroke occurrence were marked plaque enhancement (aOR 6.93; 95% CI: 2.5–19.19; P<0.001), plaque surface irregularity (aOR 3.7; 95% CI: 1.17–11.7; P=0.026), and diffuse distribution (aOR 3.3; 95% CI: 1.12–9.77; P=0.031) (Figure 4).

Predictive performance of stroke risk factors

The variables included in the multivariate analysis were subsequently analyzed according to the ROC curve. In the non-PR group, the combined model (plaque enhancement + plaque location) had the highest AUC (AUC =0.802; 95% CI: 0.698–0.907), followed by plaque enhancement (AUC =0.701; 95% CI: 0.583–0.819), plaque burden (AUC =0.687; 95% CI: 0.558–0.816), stenosis rate (AUC =0.671; 95% CI: 0.543–0.799), BMI (AUC =0.662; 95% CI: 0.534–0.791), plaque location (AUC =0.651; 95% CI: 0.511–0.791), and plaque morphology (AUC =0.611; 95% CI: 0.513–0.709) (Figure 5A).

Figure 5 Discriminative performance of stroke predictors stratified by remodeling phenotype. (A,B) The ROC curves of the discriminative performance of individual plaque features and their combined models in predicting stroke occurrence in two distinct vascular remodeling subgroups: (A) non-positive remodeling and (B) positive remodeling. In both panels, the combined model outperformed any single feature, as evidenced by its higher AUC. (A) The combined model consisting of plaque enhancement and plaque location had an AUC of 0.802, while (B) the combined model consisting of plaque enhancement, surface morphology and distribution patterns had an AUC of 0.812, indicating slightly better predictive power in the positive remodeling group. Notably, plaque enhancement emerged as the strongest individual predictor across both groups [(A) AUC =0.701; (B) AUC =0.714], suggesting its consistent clinical relevance regardless of remodeling status. In contrast, features such as surface morphology and distribution patterns showed a greater contribution in the positive remodeling group when combined (B), suggesting that these features may be more informative when vascular remodeling is present. AUC, area under the curve; ROC, receiver operating characteristic curve.

In the PR group, the combined model (plaque enhancement + plaque morphology + plaque distribution) had the highest AUC (AUC =0.812, 95% CI: 0.729–0.896), followed by plaque enhancement (AUC =0.714; 95% CI: 0.626–0.803), plaque burden (AUC =0.710; 95% CI: 0.609–0.811), stenosis rate (AUC =0.710; 95% CI: 0.609–0.810), plaque surface morphology (AUC =0.672; 95% CI: 0.597–0.747), and plaque distribution (AUC =0.636; 95% CI: 0.548–0.724) (Figure 5B).

Relationship between plaque and stenosis in the remodeling phenotypes

Comparative analysis revealed patients with stroke, compared to nonstroke controls, had significantly elevated plaque burden (P<0.001) and stenosis severity (P<0.001), while remodeling indices were not significantly different between groups (Figure 6). Overall, a linear regression fit between plaque burden and stenosis rate indicated that the lumen began to narrow (i.e., remodeling could no longer preserve the lumen) when the plaque burden reached 59.4%. For the NR, IR, and PR patterns, the plaque burden corresponding to a stenosis of 0% was 43.3%, 59.0%, and 60.1%, respectively (Figure 7). These findings indicate that PR preserves luminal patency at significantly higher plaque volumes compared to NR (Δthreshold +16.8%; P<0.001).

Figure 6 Comparison of luminal stenosis degree, plaque burden, and vascular remodeling index between the stroke and nonstroke group. The violin plot compares three key vascular parameters between the stroke and nonstroke groups. The stroke group (red) showed significantly higher stenosis percentage (P<0.001) and plaque burden (P<0.001) compared to the nonstroke group (blue). No significant difference was observed in the remodeling index. Box plots within violins indicate the median, interquartile range, and data spread. ***, P<0.001; NS, not significant.

Figure 7 Relationship between plaque burden and stenosis across the remodeling phenotypes. Linear regression for stenosis degree as a function of plaque burden measurements. A stenosis of 0% corresponded to a plaque burden of 59.4% for all intracranial atherosclerotic lesions in the posterior circulation. A stenosis of 0% corresponded to plaque burdens of 43.3%, 59.0%, and 60.1% for the negative, intermediate, and positive remodeling patterns, respectively.

Interobserver agreement of MRI measurements

Interreader reliability, as assessed with the ICC, demonstrated excellent agreement across key plaque characteristics. Estimates for remodeling patterns, stenosis rate, plaque burden, plaque enhancement, plaque location, plaque surface morphology, and plaque distribution were 0.931, 0.808, 0.837, 0.928, 0.875, 0.865, and 0.8, respectively.

Discussion

Our study established distinct stroke risk profiles across vascular remodeling phenotypes in patients with intracranial vertebrobasilar atherosclerosis. We found that intracranial atherosclerotic plaque characteristics exhibited remodeling-specific correlations with stroke risk, while vascular remodeling type did not demonstrate a significant association with stroke occurrence (P>0.05). Plaque enhancement emerged as a strong predictor of stroke across all remodeling subtypes and yielded a consistent multivariable-aOR >6.0 in both the PR and non-PR groups. In addition, vessel wall pathology features (enhancement and surface irregularity) outperformed conventional metrics, while stenosis severity and plaque burden showed limited predictive value in the adjusted models.

Atherosclerosis is a progressive pathophysiological cascade initiated by endothelial dysfunction, modified lipid infiltration, inflammatory activation, and maladaptive vascular remodeling (17). The mechanisms governing differential arterial remodeling involve a multitude of factors, including biological mediators [such as cytokine signaling and matrix metalloproteinase (MMP) activity] (18), neuroregulatory pathways (autonomic nervous system modulation) (5), and hemodynamic factors (low/oscillatory shear stress patterns) (19). These processes collectively drive arterial stiffening with elevated pulse wave velocity. Crucially, however, the specific triggers for the different remodeling phenotypes remain incompletely characterized. Inward remodeling—typically driven by low shear stress, vascular smooth muscle cell (VSMC) proliferation, and extracellular matrix (ECM) deposition—culminates in lumen narrowing. Shear stress, the tangential frictional force exerted by blood flow on the endothelium, is normally laminar; however, low or oscillatory shear stress activates the NF-κB pathway, upregulating endothelial adhesion molecules [e.g., VCAM-1 and intercellular adhesion molecule 1 (ICAM-1)] and pro-inflammatory cytokines, thereby promoting inward remodeling and atherogenesis (20,21). Conversely, outward remodeling, characterized by lumen expansion, is predominantly mediated by high shear stress, excessive MMP activation, and inflammation-induced disruption of the medial architecture. High shear stress stimulates endothelial mechanosensors (e.g., Piezo1 ion channels), upregulating nitric oxide and prostacyclin (PGI2) to maintain homeostasis through vasodilation and anti-inflammatory and antiproliferative effects (20,21). Overactivation of MMPs (e.g., MMP-9) degrades the basement membrane and ECM components, which disrupts the internal elastic lamina to facilitate VSMC migration into the intima and weakens the vascular wall, ultimately precipitating outward remodeling (22).

A complex and reciprocally causal relationship exists between vascular remodeling and atherosclerosis, and these two processes interact dynamically. Although vascular remodeling acts as a compensatory mechanism to preserve blood flow, it also provides the pathological substrate for plaque progression and rupture. During early atherosclerosis, positive vascular remodeling compensates for plaque accumulation by outward expansion, delaying luminal stenosis. Conversely, NR accelerates stenosis through concentric vessel constriction. In advanced disease, critical plaque burden (6) coupled with high-risk morphological features (e.g., thin fibrous cap and lipid-rich necrotic core) substantially elevates the risk of stroke. Notably, at equivalent stenosis severity, positively remodeled vessels harbor larger plaque volumes than do negatively remodeled segments (6,7). This volumetric advantage paradoxically increases vulnerability to fibrous cap rupture, intraplaque hemorrhage, thrombus formation. It has been suggested that an elevated RI independently predicts cryptogenic stroke (absence of significant stenosis in large vessels). In one study, PR was found to have more microembolic signal than NR in patients with symptomatic intracranial middle cerebral artery stenosis (23). Meta-analyses have associated PR with increased stroke risk (7-9,13,19,24). Notably, our study demonstrated comparable stroke risk between the PR and non-PR groups (P>0.05). Furthermore, when we excluded patients with arteries with <30% luminal stenosis, the median luminal stenosis of the entire cohort was 60.0%. We reanalyzed the association between PR and stroke. After exclusion, the association between PR and stroke remained nonsignificant (OR 1.42; 95% CI: 0.67–3.03; P=0.358) (Table S1). This suggests that the lack of association in our cohort may not be solely attributable to the inclusion of mild stenosis cases. Instead, it may reflect the heterogeneous nature of stroke mechanisms in our population. Specifically, in patients with mild-to-moderate stenosis, plaque instability (characterized by enhancement or irregular surface) might be a more direct trigger for ischemic events than the remodeling pattern itself. Stroke vulnerability appears to be driven by remodeling-specific plaque phenotypes rather than remodeling type. In the stratification of stroke risk, direct assessment of plaque vulnerability features (enhancement pattern, surface integrity, and distribution characteristics) should be prioritized over categorical remodeling classifications. PR does not inherently confer greater risk than NR when concomitant plaque pathology is controlled for. This underscores the necessity of HR-MRI in comprehensive risk assessment and the need to move beyond luminal stenosis evaluation to direct characterization of plaque.

Our analysis identified two independent risk factors for stroke in non-PR lesions: plaque enhancement (aOR 6.01; P=0.012) and basilar artery location (aOR 5.41; P=0.027). The combined predictive model achieved an AUC of 0.802 (95% CI: 0.698–0.907), demonstrating good discriminative capacity. Plaque enhancement, detectable by T1-weighted MRI sequences, serves as a surrogate marker for plaque vulnerability (11). This imaging feature primarily reflects neovascularization and inflammatory cell infiltration within the vessel wall. These pathological processes induce endothelial dysfunction and vascular leakage, resulting in gadolinium accumulation in the perivascular space. Ultimately, these changes lead to intraplaque hemorrhage and plaque rupture and are thus direct causes for embolic ischemic events. Plaque enhancement have been consistently associated with the occurrence of stroke events in previous literatures (9,11-13). In addition, it also is a risk factor for stroke recurrence (25,26), and preoperative plaque enhancement and NR jointly predict post-stenting perforator stroke in patients with basilar artery stenosis (27,28). Notably, in our study, marked plaque enhancement demonstrated consistent predictive value across the remodeling phenotypes among patients with stroke, with a prevalence of 59.1% in the PR group and 59.5% in non-PR group; moreover, the multivariate-aOR was >6.0 in both the PR and non-PR groups (P<0.05), which is consistent with previous work (9,12,13). In our study, the degree of stenosis was higher in the negatively remodeled vessels. The two vertebral arteries converge to form the basilar artery. With the poorer compensatory capacity of the collateral circulation, the basilar artery is more prone to hypoperfusion and reduced microembolic clearance after severe stenosis and therefore has a higher risk of stroke. In summary, a plaque located in the basilar artery and strongly enhanced under NR conditions implies a high risk of stroke, thus warranting close attention from clinicians. In such cases, early intensive statin and loading antiplatelet therapy may alter the plaque characteristics, reduce the risk, and prevent catastrophic events.

Beyond plaque enhancement, multivariate analysis identified two additional independent predictors of stroke in positively remodeled lesions: irregular surface morphology (aOR 3.70; P=0.026) and a diffuse distribution pattern (aOR 3.30; P=0.031). The combined predictive model integrating enhancement, surface irregularity, and diffuse distribution achieved an AUC of 0.812 (95% CI: 0.729–0.896) and demonstrated excellent discriminative capacity. Atherosclerosis is essentially a continuous accumulation of inflammatory cells and dynamic necrotic core expansion under the intima (17). Plaque enhancement serves as a prognostic biomarker for accelerated stenosis progression (29) and is mediated through inflammatory-driven pathobiological processes (30). Although PR initially preserves luminal area, persistent inflammatory response exacerbates atherosclerosis, enables larger plaque volume accumulation, and contributes to the tendency of vessels to narrow (17). Rupture of the fibrous cap of the atheromatous plaque, manifesting as irregularity of the plaque surface on HR-MRI, has been histologically verified in carotid endarterectomy specimens (31). This association may also be applicable to intracranial arterial plaques. In our cohort, plaque surface irregularity significantly predicted stroke risk (OR 5.69; 95% CI: 2.19–18.1), aligning with landmark studies by Lee et al. (OR 3.94; 95% CI: 1.90–8.16) and Song et al. (9,13) (OR 4.50; 95% CI: 1.39–8.57). Previous studies have indicated that advanced atherosclerosis often exhibits diffuse distribution (6,28). In our study, while diffuse plaque distribution predominated in the NR group (32/40,80.0%), its stroke risk was higher in the PR group (aOR 3.3; 95% CI: 1.12–9.77). This suggests remodeling phenotype modulates the pathological consequences of diffuse disease. We suggest that diffuse distribution is a characteristic feature of advanced multisegmental atherosclerosis, in which PR may mask hemodynamic impairment through compensatory expansion and delay clinical presentation until the critical inflammation threshold. In conclusion, in PR condition, arterial plaques with marked enhancement, surface irregularity, and diffuse distribution, which are associated with a high risk of stroke, need to be treated with caution and diligence in the clinical setting.

Recent research (24,25) has underscored the insufficiency of relying solely on luminal stenosis for stroke risk stratification, as this fails to characterize vessel wall pathology. A body of evidence suggests that vessel wall characteristics offer incremental prognostic value over the degree of stenosis alone (25). The observed nonassociation between stenosis severity and ischemic stroke risk (14) could be explained by remodeling-specific plaque phenotypes. Across all remodeling subtypes examined in our study, stenosis severity positively correlated with plaque burden (β=2.12; P<0.001), exhibiting a threshold effect at >59.4% wall occupancy where compensatory mechanisms fail (Figure 6). Consistent with the compensatory function of PR, PR lesions, as compared to NR lesions, exhibited a significantly lower median plaque burden (0.77 vs. 0.87; P<0.001) and stenosis severity (22% vs. 79%; P<0.001). This confirms the ability of PR to delay luminal compromise through arterial expansion. The critical plaque burden threshold for luminal narrowing onset (59.4% in our cohort) aligns with a previous study on posterior circulation, which reported a threshold of 57.6% (5), suggesting the existence of a reproducible hemodynamic failure point across populations. Our analysis also revealed a critical methodological insight: while univariate modeling demonstrated associations between stenosis severity/plaque burden and stroke risk across both the PR and non-PR groups, these associations became nonsignificant following multivariate adjustment for other plaque characteristics. This strongly indicates that luminal stenosis represents an inadequate standalone indicator of stroke risk in atherosclerosis (14). The loss of independent predictive value after vessel wall pathology was controlled for highlights the necessity of redirecting clinical attention to features of underlying plaque vulnerability. Specifically, the use of HR-MRI should be directed to the assessment of remodeling pattern, plaque enhancement, and surface irregularities, as these characteristics demonstrated stronger independent associations with stroke outcomes in our multivariate models.

Several methodological constraints warrant consideration. First, the cohort reflects regional epidemiological patterns of intracranial atherosclerosis, which include significant ethnic and geographic heterogeneity. This may limit extrapolation of our findings to other populations. Second, the histopathological validation of plaque characteristics (particularly enhancement patterns) remains technically unfeasible in vivo. This represents a fundamental challenge in intracranial vessel wall MRI research (1). Third, the observational design precludes causal inferences. Furthermore, the medication history of the participants was not incorporated into the analysis. This represents a potential confounder, as pharmacological interventions, particularly statins, are known to modulate plaque characteristics. Finally, the absence of data on specific plaque components, such as intraplaque hemorrhage (32) and large lipid cores—which are established predictors of stroke—constitutes another limitation of this study. There is thus a need for future prospective studies that comprehensively characterize plaque morphology and document detailed medication information.

Conclusions

Our study delineated distinct stroke risk stratification profiles across vascular remodeling phenotypes in patients intracranial vertebrobasilar atherosclerosis. PR preserves luminal patency through compensatory expansion yet is paradoxically associated with delayed stroke presentation due to prolonged subclinical progression; meanwhile, NR accelerates hemodynamic impairment via concentric constriction. Marked plaque enhancement independently predicted stroke occurrence across all phenotypes (aOR >6.0; P<0.01) and may thus function as a pathophysiological indicator of plaque vulnerability. Future investigations with HR-MRI should prioritize plaque enhancement patterns, remodeling phenotypes, and surface morphology over stenosis metrics for individualized risk stratification in patients with posterior circulation disease.

Acknowledgments

We would like to thank the R Development Core Team and contributors for the R packages used in our study. We also appreciate everyone involved in this study, including the patients, physicians, and medical workers.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1812/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1812/dss

Funding: This study was supported by the Science and Technology Innovation Project of Yichang Health Commission (No. WJ2025Y05), the Research Projects of Yiling People’s Hospital of Yichang City (No. YLRMYY-YNKY-202508), and the Key R & D Program of Hubei Province (No. 2020BCB028).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1812/coif). Z.L. reports the funding from the Science and Technology Innovation Project of Yichang Health Commission (No. WJ2025Y05). H.X. reports the funding from the Research Projects of Yiling People’s Hospital of Yichang City (No. YLRMYY-YNKY-202508). Y.L. reports the funding from the Key R & D Program of Hubei Province (No. 2020BCB028). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Zhongnan Hospital (No. 2025190K). Patient consent was waived due to the nature of the retrospective study.

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. Yang WJ, Wong KS, Chen XY. Intracranial Atherosclerosis: From Microscopy to High-Resolution Magnetic Resonance Imaging. J Stroke 2017;19:249-60. [Crossref] [PubMed]
2. Wang Y, Zhao X, Liu L, Soo YO, Pu Y, Pan Y, Wang Y, Zou X, Leung TW, Cai Y, Bai Q, Wu Y, Wang C, Pan X, Luo B, Wong KS. Prevalence and outcomes of symptomatic intracranial large artery stenoses and occlusions in China: the Chinese Intracranial Atherosclerosis (CICAS) Study. Stroke 2014;45:663-9. [Crossref] [PubMed]
3. Qureshi AI, Feldmann E, Gomez CR, Johnston SC, Kasner SE, Quick DC, Rasmussen PA, Suri MF, Taylor RA, Zaidat OO. Intracranial atherosclerotic disease: an update. Ann Neurol 2009;66:730-8. [Crossref] [PubMed]
4. Ward MR, Pasterkamp G, Yeung AC, Borst C. Arterial remodeling. Mechanisms and clinical implications. Circulation 2000;102:1186-91.
5. Qiao Y, Anwar Z, Intrapiromkul J, Liu L, Zeiler SR, Leigh R, Zhang Y, Guallar E, Wasserman BA. Patterns and Implications of Intracranial Arterial Remodeling in Stroke Patients. Stroke 2016;47:434-40. [Crossref] [PubMed]
6. Ma N, Jiang WJ, Lou X, Ma L, Du B, Cai JF, Zhao TQ. Arterial remodeling of advanced basilar atherosclerosis: a 3-tesla MRI study. Neurology 2010;75:253-8. [Crossref] [PubMed]
7. Zhang DF, Chen YC, Chen H, Zhang WD, Sun J, Mao CN, Su W, Wang P, Yin X. A High-Resolution MRI Study of Relationship between Remodeling Patterns and Ischemic Stroke in Patients with Atherosclerotic Middle Cerebral Artery Stenosis. Front Aging Neurosci 2017;9:140. [Crossref] [PubMed]
8. Zhao DL, Deng G, Xie B, Ju S, Yang M, Chen XH, Teng GJ. High-resolution MRI of the vessel wall in patients with symptomatic atherosclerotic stenosis of the middle cerebral artery. J Clin Neurosci 2015;22:700-4. [Crossref] [PubMed]
9. Lee HN, Ryu CW, Yun SJ. Vessel-Wall Magnetic Resonance Imaging of Intracranial Atherosclerotic Plaque and Ischemic Stroke: A Systematic Review and Meta-Analysis. Front Neurol 2018;9:1032. [Crossref] [PubMed]
10. Park JE, Jung SC, Lee SH, Jeon JY, Lee JY, Kim HS, Choi CG, Kim SJ, Lee DH, Kim SO, Kwon SU, Kang DW, Kim JS. Comparison of 3D magnetic resonance imaging and digital subtraction angiography for intracranial artery stenosis. Eur Radiol 2017;27:4737-46. [Crossref] [PubMed]
11. Gupta A, Baradaran H, Al-Dasuqi K, Knight-Greenfield A, Giambrone AE, Delgado D, Wright D, Teng Z, Min JK, Navi BB, Iadecola C, Kamel H. Gadolinium Enhancement in Intracranial Atherosclerotic Plaque and Ischemic Stroke: A Systematic Review and Meta-Analysis. J Am Heart Assoc 2016;5:e003816. [Crossref] [PubMed]
12. Wang E, Shao S, Li S, Yan P, Xiang Y, Wang X, Li J, Wang G, Sun Q, Du Y. A High-Resolution MRI Study of the Relationship Between Plaque Enhancement and Ischemic Stroke Events in Patients With Intracranial Atherosclerotic Stenosis. Front Neurol 2018;9:1154. [Crossref] [PubMed]
13. Song JW, Pavlou A, Xiao J, Kasner SE, Fan Z, Messé SR. Vessel Wall Magnetic Resonance Imaging Biomarkers of Symptomatic Intracranial Atherosclerosis: A Meta-Analysis. Stroke 2021;52:193-202. [Crossref] [PubMed]
14. Leng X, Wong KS, Liebeskind DS. Evaluating intracranial atherosclerosis rather than intracranial stenosis. Stroke 2014;45:645-51. [Crossref] [PubMed]
15. Liu Z, Zhong F, Xie Y, Lu X, Hou B, Ouyang K, Fang J, Liao M, Liu Y. A Predictive Model for the Risk of Posterior Circulation Stroke in Patients with Intracranial Atherosclerosis Based on High Resolution MRI. Diagnostics (Basel) 2022;12:812. [Crossref] [PubMed]
16. Chimowitz MI, Kokkinos J, Strong J, Brown MB, Levine SR, Silliman S, Pessin MS, Weichel E, Sila CA, Furlan AJ. The Warfarin-Aspirin Symptomatic Intracranial Disease Study. Neurology 1995;45:1488-93. [Crossref] [PubMed]
17. Ritz K, Denswil NP, Stam OC, van Lieshout JJ, Daemen MJ. Cause and mechanisms of intracranial atherosclerosis. Circulation 2014;130:1407-14. [Crossref] [PubMed]
18. Majesky MW. Vascular Development. Arterioscler Thromb Vasc Biol 2018;38:e17-e24. [Crossref] [PubMed]
19. Zhang D, Wu X, Tang J, Wang P, Chen GZ, Yin X. Hemodynamics is associated with vessel wall remodeling in patients with middle cerebral artery stenosis. Eur Radiol 2021;31:5234-42. [Crossref] [PubMed]
20. Silver AE, Vita JA. Shear-stress-mediated arterial remodeling in atherosclerosis: too much of a good thing? Circulation 2006;113:2787-9. [Crossref] [PubMed]
21. DeVerse JS, Sandhu AS, Mendoza N, Edwards CM, Sun C, Simon SI, Passerini AG. Shear stress modulates VCAM-1 expression in response to TNF-α and dietary lipids via interferon regulatory factor-1 in cultured endothelium. Am J Physiol Heart Circ Physiol 2013;305:H1149-57. [Crossref] [PubMed]
22. Ota R, Kurihara C, Tsou TL, Young WL, Yeghiazarians Y, Chang M, Mobashery S, Sakamoto A, Hashimoto T. Roles of matrix metalloproteinases in flow-induced outward vascular remodeling. J Cereb Blood Flow Metab 2009;29:1547-58. [Crossref] [PubMed]
23. Shi MC, Wang SC, Zhou HW, Xing YQ, Cheng YH, Feng JC, Wu J. Compensatory remodeling in symptomatic middle cerebral artery atherosclerotic stenosis: a high-resolution MRI and microemboli monitoring study. Neurol Res 2012;34:153-8. [Crossref] [PubMed]
24. Zhao JJ, Lu Y, Cui JY, Ma LQ, Zhang RP, Xu Z. Characteristics of symptomatic plaque on high-resolution magnetic resonance imaging and its relationship with the occurrence and recurrence of ischemic stroke. Neurol Sci 2021;42:3605-13. [Crossref] [PubMed]
25. Kim JM, Jung KH, Sohn CH, Moon J, Shin JH, Park J, Lee SH, Han MH, Roh JK. Intracranial plaque enhancement from high resolution vessel wall magnetic resonance imaging predicts stroke recurrence. Int J Stroke 2016;11:171-9. [Crossref] [PubMed]
26. Song X, Zhao X, Liebeskind DS, Wang L, Xu W, Xu Y, Hou D, Zheng Z, Wu J. Incremental value of plaque enhancement in predicting stroke recurrence in symptomatic intracranial atherosclerosis. Neuroradiology 2020;62:1123-31. [Crossref] [PubMed]
27. Lu X, Li C, Qu C, Huang J, Wang Q, Qiu B, Wang C, Li S, Xu H, Liu Y. A High Resolution MRI Study of the Relationship Between Plaque Enhancement and Perforator Stroke after Stenting for Symptomatic Vertebrobasilar Artery Stenosis. J Stroke Cerebrovasc Dis 2021;30:105558. [Crossref] [PubMed]
28. Ma N, Xu Z, Lyu J, Li M, Hou Z, Liu Y, Yang M, Mo D, Gao F, Song L, Sun X, Liu L, Liu X, Sui B, Shen M, Ma L, Wang Y, Wang Y, Miao Z, Lou X. Association of Perforator Stroke After Basilar Artery Stenting With Negative Remodeling. Stroke 2019;50:745-9. [Crossref] [PubMed]
29. Lee HS, Jung JM, Yang HB, Lee SH. Predicting Stenosis Aggravation in Follow-Up High-Resolution Magnetic Resonance Images of Patients with Intracranial Atherosclerosis. Cerebrovasc Dis 2022;51:608-14. [Crossref] [PubMed]
30. Millon A, Boussel L, Brevet M, Mathevet JL, Canet-Soulas E, Mory C, Scoazec JY, Douek P. Clinical and histological significance of gadolinium enhancement in carotid atherosclerotic plaque. Stroke 2012;43:3023-8. [Crossref] [PubMed]
31. Derksen WJ, Peeters W, van Lammeren GW, Tersteeg C, de Vries JP, de Kleijn DP, Moll FL, van der Wal AC, Pasterkamp G, Vink A. Different stages of intraplaque hemorrhage are associated with different plaque phenotypes: a large histopathological study in 794 carotid and 276 femoral endarterectomy specimens. Atherosclerosis 2011;218:369-77. [Crossref] [PubMed]
32. Zhao J, Zhao X, Peng Z, Liu J. A model based on high-resolution magnetic resonance vessel wall imaging for predicting stroke recurrence in patients with symptomatic intracranial atherosclerosis. Quant Imaging Med Surg 2025;15:10422-32. [Crossref] [PubMed]