2.1 Patient Selection and Clinical Assessment

This study was approved by the Ethics Committee of our hospital (Approval Number: KS2022023-1). Written informed consent was obtained from all participants or their legal guardians. From September 2020 to May 2023, patients with carotid artery stenosis admitted to the Department of Neurosurgery at our hospital were consecutively enrolled. All patients received endovascular treatment upon admission, and those who underwent carotid artery stenting (CAS) were included in this study. The inclusion criteria were as follows: (1) Eligibility for CAS versus carotid endarterectomy (CEA) was determined based on the North American Symptomatic Carotid Endarterectomy Trial (NASCET) criteria (i.e., asymptomatic carotid stenosis > 70% or symptomatic carotid stenosis > 50%) [20]. In addition, factors such as patient preference, medical comorbidities, lesion anatomy, and the relative risk of surgical versus endovascular intervention were considered. Patients who were deemed suitable candidates and opted for CAS were included in this study; (2) participants who failed maximal medical therapy, including dual antiplatelet therapy and intensive risk factor management; (3) those who underwent integrated PET/MR of the carotid artery within 1–3 days before CAS, with brain MRI on the same day as the PET/MR, and another brain MRI within 3 days post-CAS. The exclusion criteria were: (1) incomplete clinical data; (2) participants with non-atherosclerotic stenosis, such as dissection, arteritis, and Moyamoya disease; (3) participants with intracranial aneurysms in the target artery; (4) uncooperative participants or those with contraindications to MRI; (5) poor quality of PET/MR or MR images (score < 3). After screening, data from 47 eligible patients were collected (screening procedure detailed in Supporting Information).

Clinical and laboratory data were collected from patients upon admission. Baseline demographic data included age, sex, and BMI. Stroke risk factors included hypertension, hyperglycemia, hypercholesterolemia, hyperlipidemia, history of coronary heart disease, and smoking status. Laboratory data obtained within 24 h of admission included high-sensitivity C-reactive protein (CRP) and homocysteine. Additionally, essential baseline data, such as whether the patient was symptomatic and their previous medication use, were collected.

2.2 MRI Protocols

All enrolled participants underwent carotid artery integrated PET/MR and brain MR examinations 1–3 days before and after endovascular treatment. All patients were asked to fast for at least 6 h before ¹⁸F-FDG PET imaging. Imaging was performed only if fasting glucose was lower than 7.7 mmol/L before tracer injection. ¹⁸F-FDG was injected intravenously at a dose of 3–4 MBq/kg. The acquisition started 90–120 min after tracer injection using an integrated PET/MRI system (uPMR790, United Imaging Healthcare, Shanghai, China). MR vessel wall imaging was performed with an 8-channel carotid coil. The PET/MR image acquisition range was centered on the carotid bifurcation, with an 18 cm coverage.

High-resolution carotid vessel wall MRI sequences included 3D TOF-MRA, 3D T1, 3D T2, and post-contrast (gadolinium-DTPA, 0.1 mmol/kg, 2.5 mL/s) enhanced 3D T1 (T1C+). Specific parameters were as follows: (1) three-dimensional time of flight (3D-TOF), TR/TE(ms) = 17.5/4, FOV(mm) = 180*180, resolution (mm) = 0.75*0.60*1.5, slice number = 323;(2)3D T1 FSE, TR/TE(ms) = 800/14.52, FOV(mm) = 180 mm*180 mm, resolution(mm) = 0.60*0.60*0.60, slice number = 390;(3)3D T2 FSE, TR/TE(ms) = 2000/198.44, FOV(mm) = 180*180, resolution (mm) = 0.60*0.60*0.60, slice number = 390, (4)3D T1 FSE + C, TR/TE(ms) = 800/14.52, FOV(mm) = 180*180, resolution (mm) = 0.60*0.60*0.60, slice number = 390. PET images were reconstructed using an OSEM algorithm (matrix size = 256 × 256, thickness = 1.4 mm) with time-of-flight information following data corrections for attenuation, scatter, and random coincidences. Brain MRI Scanning was performed using GE Discovery 750 W. Routine brain MRI sequences included axial T1WI, T2WI, T2-FLAIR, and DWI, covering the range from the cranial vertex to the foramen magnum. DWI sequence parameters for the GE Discovery 750 W scanner: TR/TE = 3672/77.9 ms, FOV(mm) = 240 × 240 mm, Acquisition Matrix = 128 × 128, Slice thickness(mm) = 5, Flip angle(°) = 90, with b = 0, 1000s/mm².

2.3 Analysis of FDG PET/MR Images

PET/MR image analysis was conducted using a dedicated plaque analysis workstation (United Imaging Healthcare, Shanghai, China). The quality of the image was rated on a 4-point scale (1 = poor and 4 = excellent), determined by the clarity of the arterial wall, luminal margin, and plaque components [21]. Patients with an image rating < 3 were excluded. Three experienced neuroradiologists (SH Z, MM F, and FY; each with over 10 years of experience) analyzed the MR images independently and were blinded to all other information. Two (SH Z and MM F) independently reviewed the images, with discrepancies resolved by a third reviewer (FY). Plaque composition and surface status were classified based on AHA lesion types [22] (Table S1). The focus was on determining if a plaque was classified as AHA-LT VI and documenting these findings. For quantitative analysis, the degree of stenosis (NASCET), plaque area, and remodeling index were calculated at the most stenotic slice. By delineating the circular regions of interest for each slice, the maximum standardized uptake value (SUVmax) of the plaque was calculated; SUVmax ≥ 2.85 was considered indicative of high PET uptake [23]. The maximum target-to-background ratio (TBRmax) was calculated as the ratio of SUVmax to the venous blood pool SUVmean [24]. Inflammation levels were categorized based on TBR values: mild (TBR = 1.25–2), moderate (TBR = 2–3.2), and severe (TBR > 3.2) [25]. The symptomatic carotid atheroma inflammation lumen-stenosis(SCAIL) score was calculated for each patient, utilizing a 2-item scale that includes lumen stenosis and plaque inflammation (Table S2). Patients were categorized into low SCAIL score (< 3) and high SCAIL score (> 3) groups [26].

2.4 CAS Procedure Protocols

All CAS procedures were executed by the same surgical team under local anesthesia via the femoral artery approach. Patients were pre-treated with daily doses of clopidogrel and aspirin for antiplatelet therapy. Intraoperatively, intravenous heparin (100 U/kg) was administered for anticoagulation. Digital subtraction angiography (DSA) was used prior to stent placement to assess blood flow and stenosis. A 6-8F catheter was navigated to the ipsilateral common carotid artery proximal to the stenosis. After stent deployment, residual stenosis was confirmed to be less than 30%.

2.5 Evaluation of New Ischemic Lesions

Two radiologists (SH Z and MM F), blinded to clinical data and endovascular treatment details, independently reviewed all pre- and post-intervention DWI and apparent diffusion coefficient (ADC) maps. Acute periprocedural ischemic lesions were defined as new diffusion-restricted lesions (hyperintensity on DWI and hypointensity on ADC) appearing post-treatment but absent pre-treatment [5, 27]. New ischemic lesions were categorized into symptomatic ischemic strokes and asymptomatic new lesions. The frequency of any new ischemic lesions post-carotid stenting was recorded. We further analyzed the distribution, location, and number of new lesions post-carotid stenting. The distribution of new lesions was classified into ischemic lesions within the treatment vessel territory and those outside the treatment artery territory [27]. Lesion locations were categorized into peripheral brain regions supplied by the main branches of the carotid artery (including cortex and subcortical white matter) and deep brain regions supplied by perforating arteries [28]. The total number of new lesions per patient was also recorded. Ischemic lesions were considered independent if there was no continuity between lesions on the same slice and adjacent slices [29].

2.6 Statistical Analysis

Descriptive statistics are expressed as frequencies and percentages for categorical variables, while continuous variables are represented by mean (SD) or median (IQR), depending on suitability. The intraclass correlation coefficient (ICC) was calculated to assess the reproducibility of SUVmax, TBRmax, and the count of new ischemic lesions between raters. An ICC greater than 0.75 was considered to indicate good to excellent reliability. We first assessed the normality of all continuous variables using the Shapiro–Wilk test. For variables that followed a normal (Gaussian) distribution, results are reported as mean ± standard deviation, and differences were compared using the Student's t-test. For variables that did not pass the normality test, we report data as medians (interquartile range [IQR]) and employed nonparametric tests (e.g., Mann–Whitney U test, Kruskal-Wallis test) for comparisons. Categorical variables were expressed as numbers (percentages), and a P-value < 0.05 was considered statistically significant.

Bivariate and multivariate logistic regression analyses were conducted to determine how PET/MR evaluation metrics and other characteristics relate to the perioperative new ischemic lesions. Three multivariable logistic regression models were constructed, adjusting for clinically relevant covariates and potential confounders [30, 31]. In the first predictive scheme, we focused on two indicators—high PET uptake and high SCAIL score—and assessed their associations under each model: Model 1 represents the unadjusted (single-factor) analysis. Model 2 adjusts for demographic and cardiovascular risk factors (e.g., sex, age, hypertension, diabetes, hypercholesterolemia). Model 3 includes the covariates in Model 2 and imaging variables of plaque and vascular walls: plaque burden, positive remodeling, and type VI plaque. A p < 0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS (version 26.0; IBM Corporation).

3.1 Comparison of Patient Characteristics Between Groups

As shown in Figure S1, 47 CAS-treated patients met the inclusion criteria after screening. The mean age of the subjects was 65 ± 7 years, with 44 males (93.6%) and 28 out of 47 (59.6%) having symptomatic plaques. Postoperatively, 30 patients (63.8%) developed new ischemic lesions. The inter-observer and intra-observer consistency for radiological feature measurements was high (inter-group correlation coefficient, 0.95; intra-group correlation coefficient, 0.88).

Comparing the groups with and without new ischemic lesions (Table 1), there were no significant differences in cardiovascular risk factors, clinical and treatment profiles, or laboratory indicators (p > 0.05). In terms of angiographic profiles, there were no significant differences in lumen diameter and lumen area between the two groups. Similarly, there was no significant difference in plaque burden (0.95 [0.90, 0.97] vs. 0.93 [0.87, 0.96]), positive remodeling (46.7% vs. 47.1%), and severity of stenosis (63.3% vs. 58.8%). The individual features of vulnerable plaques (lipid-rich necrotic core, hemorrhage, calcification, and ulceration) did not show significant differences between the groups. However, the proportion of AHA-VI plaques was significantly higher in the new ischemic lesions group compared to the no new ischemic lesions group (50.0% vs. 17.6%, p = 0.028). To further investigate which specific features of these AHA VI plaques might drive this association, we analyzed the prevalence of IPH, ulceration, and thrombosis within this subgroup. Our data indicate that IPH was the most common feature (13/18, 72.2%) and appears to contribute most to postoperative new ischemic lesions (40% vs. 17.6%, p = 0.114). Among PET-related metrics, the new ischemic lesions group had a higher proportion of high PET uptake (43.3% vs. 11.8%, p = 0.026) and high SCAIL score (63.3% vs. 23.5%, p = 0.009) compared to the no new ischemic lesions group, though the inflammation grade did not show a significant difference(Figure 1).

3.2 New Ischemic Lesions After CAS Treatment

We analyzed 30 new ischemic lesions to determine the pathophysiological mechanisms of perioperative new ischemic lesions. Overall, 65% of participants developed new ischemic lesions, with 28 patients (59.6%) having asymptomatic new ischemic lesions and 2 patients (4.3%) experiencing TIA or ischemic stroke post-treatment. The first patient exhibited a standardized wall index (plaque burden) of 0.80, indicative of positive remodeling, along with a lipid-rich plaque featuring partial hemorrhage. The second patient presented with a standardized wall index (plaque burden) of 0.84, also demonstrating positive remodeling, and plaques characterized by both hemorrhage and ulceration. The most common pattern in the cohort was new ischemic lesions in both peripheral brain areas and deep brain areas (15 participant, 50%), followed by lesions in peripheral brain areas only (9 participant, 30%), and new lesions in Deep brain areas only (6 participant, 20%). Regarding the distribution of new lesions, 25.5% (12 participant) had lesions only in the territory of the treated artery, while 2% had lesions entirely beyond the treated artery's territory. A larger proportion, 36.2% (17 participant), had mixed territory ischemic lesions (Table 2). A total of 98 new ischemic lesions were identified among the 30 participants, with a median of 2 lesions per participant (IQR 2 to 4.75). Of these, 61 lesions (62.2%) were located in the territory of the treated artery (median of 2 lesions per participant, IQR 1 to 2.75), and 37 lesions (37.8%) were beyond the territory of the treated artery (median of 1 lesion per participant, IQR 0 to 2).

We then analyzed the distribution of new ischemic brain lesions between the high and low PET uptake groups, finding no significant differences (in the territory of the treated artery only: 30.8% vs. 41.2%; beyond the territory of the treated artery only: 15.4% vs. 0%; in the mixed area: 53.8% vs. 58.8%; p = 0.318). Other detailed distribution counts are presented in Table S3.

3.3 High PET Uptake/ High SCAIL Score and New Ischemic Lesions Within 3 Days

Previous comparisons between groups showed that the proportion of high PET uptake and high SCAIL scores was significantly higher in the new ischemic lesions group compared to the no new ischemic lesions group. Representative cases are shown in Figure 2. We evaluated each of these indicators separately, in multiple binary logistic regression analysis, adjusting for demographics and cardiovascular risk factors (sex, age, hypertension, diabetes, hypercholesterolemia) in model 2, high PET uptake was independently associated with new ischemic lesions (aOR = 6.92, 95% CI: 1.35, 57.62; p = 0.036). Further adjustment for imaging variables of vascular lesion severity (plaque burden, positive remodeling, AHA-VI plaque) in model 3 confirmed this association (aOR = 7.26, 95% CI: 1.22, 73.59; p = 0.049). Similarly, high SCAIL scores were independently associated with new ischemic lesions after adjustments in both models (model 2: aOR = 5.94, 95% CI: 1.51, 28.85; p = 0.016; model 3: aOR = 7.06, 95% CI: 1.50, 44.18; p = 0.020) (Table 3).

Given the close relationship between PET uptake and SCAIL scores with complex plaques [18], we created a new variable “type” by multiplying the two categorical variables: Type 1 = high SCAIL score(−) + AHA-VI plaque(−), Type 2 = high SCAIL score(+) + AHA-VI plaque(−), Type 3 = high SCAIL score(−) + AHA-VI plaque(+), and Type 4 = high SCAIL score(+) + AHA-VI plaque(+). Additional multivariable analysis showed that Type 4 was a strong predictor of new ischemic lesions after similar adjustments (aOR = 25.52, 95% CI: 2.79, 655.25; p = 0.013) (Table 4). A bar chart illustrated that the proportion of new ischemic lesions was highest in the Type 4 group (90% vs. 10%) and lowest in the Type 1 group (31% vs. 69%) (Figure 3).