The study protocol was approved by the Ethics Committee of The First Hospital of Harbin Medical University . All patients provided either written or verbal consent to participate in the study. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). Prospective inclusion of patients with atherosclerotic TIA in the MCA, treated at The First Hospital of Harbin Medical University, took place from October 2021 to July 2023 . The inclusion criteria were as follows: (1) age ≥ 18 years; (2) DWI images were negative; (3) patients had multiple (≥2) TIA symptoms in the past, with each attack lasting no more than 24 h; (4) contralateral limb numbness or weakness at the MCA stenosis in TIA patients; (5) MR examination performed within 7 days after onset; (6) presence of at least one atherosclerotic risk factor, such as hypertension, diabetes mellitus, dyslipidemia, alcohol consumption, or smoking. The exclusion criteria were as follows: (1) acute cerebral infarction confirmed by DWI; (2) contraindications to MRI enhancement; (3) previous history of cerebral infarction, cerebral hemorrhage, brain surgery, and cerebrovascular intervention; (4) combined ipsilateral carotid artery (ICA) stenosis ≥ 50%; (5) not due to atherosclerotic diseases, such as dissection, vasculitis, moyamoya disease, or cardiogenic embolism; (6) poor MRI image quality that prevented further analysis.

Clinical baseline characteristics and multimodal MR imaging of the patients were collected for the patients. According to the DSC-PWI images, the patients were divided into two groups: hypoperfusion and non-hypoperfusion. Furthermore, within the hypoperfusion group, segmented into low-(0–3) and high-(4–7) risk groups were performed based on the ABCD² score (1).

Demographic and clinical characteristics were collected from clinical records, including age, gender, BMI, history of hypertension, diabetes, total cholesterol, triglycerides, LDL, HDL, and hyperhomocysteinemia. Hypertension was defined as self-reported history of hypertension, mean blood pressure ≥ 140/90 mmHg, or taking anti-hypertensive medication (15); diabetes was defined as self-reported use of insulin or oral hypoglycemic drugs, or HbA1c ≥ 6.5% (16); total cholesterol ≥ 5.2 mmol/L, triglycerides ≥ 1.7 mmol/L, LDL ≥ 3.7 mmol/L, HDL ≤ 1.04 mmol/L, or self-reported history of dyslipidemia (17), and hyperhomocysteinemia defined as >15 μ mol/L (18). History of smoking: ≥1 cigarette per day for six consecutive months; history of alcohol consumption: >2 U for men and >1 U for women on average per day for the past year (19).

A 3.0 T MRI scanner (Achieva; Philips Healthcare, The Netherlands) with a 32-channel phased-array coil was used. Patients were trained to breathe before scanning to reduce the effect of motion artifacts on image quality. The scanning protocol and parameters included diffusion-weighted imaging (DWI), TR/TE 21/3.45 ms, thickness 0.6 mm, Flip angle 20°, FOV 194 mm × 194 mm; three-dimensional time-of-flight magnetic resonance angiography (3D-TOF-MRA), TR/TE 1,927/55 ms, thickness 4 mm, Flip angle 90°, FOV 230 mm × 230 mm; HR-VWI, and DSC-PWI. HR -VWI including T1-weighted three-dimensional volumetric isotropic turbo spin echo acquisition (T1-3D-VISTA), TR/TE 800/18 ms, thickness 0.6 mm, Flip angle 90°, FOV 180 mm × 144 mm; simultaneous non-contrast angiography and intraplaque hemorrhage (SNAP), TR/TE 10/5.8 ms, thickness 0.6 mm, Flip angle 11°, FOV 160 mm × 160 mm; T2-weighted three-dimension volumetric isotropic turbo spin echo acquisition (T2-3D-VISTA), TR/TE 2,400/90 ms, thickness 0.6 mm, Flip angle 90°, FOV 250 mm × 250 mm; and enhanced T1-3D-VISTA. The contrast agent was Gd-DTPA contrast, using 0.1 mL/kg. DSC-PWI, TR/TE 1,795/40 ms, thickness 0.6 mm, Flip angle 75°, FOV 224 mm × 224 mm, was performed by injecting Gd (0.1 mL/kg) via the anterior elbow vein at a rate of 5 mL/s with a high-pressure syringe, followed by flushing 20 mL of saline at the same rate and acquiring planar gradient echoes every 2 s (T2^*) images for the 60s.

Two radiologists with more than 2 years of diagnostic experience conducted qualitative and quantitative analyses of the images while remaining blind to the clinical data of all recruited patients. In case of any disagreements between the observers, a consensus was reached through a joint reading. All parameter measurements were performed on the Philips Intellispace Portal workstation. The quality of MR images was classified as poor, marginal, good, and excellent according to the signal-to-noise, and images graded as poor were excluded from the study. On HR-VWI images, atherosclerotic plaques exhibited eccentric thickening of the vessel wall (17). According to the symptoms of TIA, the location of diseased blood vessels (left/right MCA) was defined. Plaque was defined as eccentric thickening of MCA wall. A culprit plaque was defined as the layer of the maximum lumen narrowing (MLN) of the MCA (20, 21). The reference lumen was the plaque-free segment proximal or distal to the largest luminal stenosis. The short axial T1-3D-VISTA images were magnified to 200% and the outer wall area (OWA) and inter wall area (IWA) were manually measured at the MLN site and the reference site. Intraplaque hemorrhage (IPH) was defined when the signal intensity (SI) of the culprit plaque was higher than 150% of gray matter (22). All cross-sections with plaque were categorized in accordance with whether the plaque was centered on the superior, inferior, dorsal, or ventral side of the vessel. If a plaque was distributed between two quadrants, the quadrant with the thickest plaque was selected (23). The plaque measurement is as follows: wall area (WA) = OWA-IWA; plaque area (PA) was calculated as WA_MLN-WA_reference (24); the degree of stenosis = (1 − IWA_MLN/IWA_reference) × 100% (12). The measurement of plaque burden is expressed by the normalized wall index (NWI), which is defined as WA_MLN/OWA_MLN × 100% (25); remodeling index (RI) = OWA_MLN/OWA_reference (RI > 1.05 was defined as positive remodeling and RI < 0.95 was defined as negative remodeling) (21). Additionally, SI was quantified with manually drawn regions of interest on HR-VWI images. The signal intensity of gray matter (SI_graymatter) was measured by manually drawing a round region of 8–10 mm² at the adjacent normal gray matter on matched pre and post-contrast T1-weighted images, respectively. The contrast enhancement ratio = [(SI_post-plaque/SI_{post-graymatter}) − (SI_pre-plaque/SI_{pre-graymatter})]/(SI_pre-plaque/SI_{pre-graymatter}) × 100% (26).

The DSC-PWI sequence was utilized to generate pseudo-color maps of region cerebral blood flow (rCBF), region cerebral blood volume (rCBV), mean transit time (MTT), and time to peak (TTP). It is challenging to compare hemodynamic parameters due to individual differences. Therefore, to avoid bias in the data, the relative rCBF (rrCBF), relative rCBV (rrCBV), relative MTT (rMTT), and relative TTP (rTTP) were calculated based on the affected/mirror contralateral side. Since there is no universally validated or consistent perfusion value for distinguishing risky from benign tissue. We based on previous research that, in the pseudo-color map, the hypoperfusion group was defined as reduced perfusion on the affected side compared with the contralateral side, with rrCBF < 75%. The non-hypoperfusion group was defined as no abnormal perfusion areas observed on the pseudo-color chart by visual observation (27, 28).

The statistical analysis of the data was performed using IBM SPSS 26.0 software. The Kolmogorov–Smirnov test was used to analyze the normality of continuous variables. Variables that followed a normal distribution were expressed as mean ± standard deviation (SD), and comparisons between groups were conducted using independent Student’s t-test. For variables with skewed distributions, median and interquartile range (IQR) were used, and Mann–Whitney U-test for comparison between groups. Categorical variables were represented as numbers and frequencies (%), and group comparisons were carried out using the chi-square test. In the chi-square test, if the expected count of any cell was less than 5, Pearson’s Chi-square test was applied. If any cell had an expected count was less than 5, a continuity correction test was used. If the minimum expected count was less than 1, Fisher’s exact test was employed . The interobserver reproducibility of continuous variables was assessed using intra-group correlation coefficients (ICC) and the agreement was classified as very good (r > 0.75), good (r = 0.4–0.75), or poor (r < 0.4). Cohen’s kappa was utilized to determine the interobserver agreement in identifying plaque distribution. The p < 0.05 difference was statistically significant.

The patient selection process is summarized in Figure 1. During the trial period, 151 patients with TIA were prospectively recruited for MRI. Among the recruited patients, 95 patients were excluded due to the following reasons: poor image quality (n = 30); ipsilateral ICA stenosis ≥ 50% (n = 15), moyamoya disease (n = 12), vasculitis (n = 14), cerebral hemorrhage (n = 9), dissection (n = 9), and incomplete of clinical information (n = 6). Ultimately, 56 patients with a mean age of 53.45 ± 11.86 years were included in this study. The hypoperfusion group consisted of 41 patients (mean age 55 ± 11 years, 21 male), while the non-hypoperfusion group consisted of 15 patients (mean age 50 ± 14 years, 10 male).

Among the 56 patients with TIA, the prevalence of hypertension was 58.92%, diabetes mellitus was 32.14%, hyperhomocysteinemia was 57.14%, and dyslipidemia was 37.5%. In addition, 20 patients (35.71%) had a history of smoking, and 13 patients (23.21%) had a history of alcohol consumption. In the hypoperfusion group, the proportion of patients with hypertension (68.3% vs. 33.3%) and hyperhomocysteinemia (65.9% vs. 33.3%) was significantly higher than that in the non-hypoperfusion group, and the difference was statistically significant (p = 0.019, p = 0.029). There were no significant differences between the two groups in terms of gender, age, BMI, diabetes, cholesterol, triglycerides, LDL, HDL, smoking, and alcohol consumption (p > 0.05). The clinical characteristics of TIA patients in the hypoperfusion and non-hypoperfusion groups were compared in Table 1.

Good interreader reproducibilities were found: IWA_MLN (ICC 0.997, 95% CI 0.994–0.998); IWA_reference (ICC 0.91, 95% CI 0.985–0.995); OWA_MLN (ICC 0.994, 95% CI 0.990–0.997); OWA_reference (ICC 0.988, 95% CI 0.980–0.993); SI_pre-plaque (ICC 0.954, 95% CI 0.923–0.973); SI_{pre-graymatter} (ICC 0.878, 95% CI 0.801–0.927); SI_post-plaque (ICC 0.946, 95% CI 0.910–0.968); SI_{post-graymatter} (ICC 0.880, 95% CI 0.803–0.928); rrCBF (ICC 0.760, 95% CI 0.623–0.852); rrCBV (ICC 0.795, 95% CI 0.674–0.874); rMTT (ICC 0.802, 95% CI 0.684–0.879); rTTP (ICC 0.758, 95% CI 0.618–0.851); plaque distribution k = 0.818; IPH k = 0.707.

Measurement of vessel and plaque characteristics in hypoperfusion and non-hypoperfusion is shown in Figures 2, 3, respectively. In the hypoperfusion group, both luminal stenosis (41.79% vs. 17.62%) and NWI (57.1 % vs. 40.21) were significantly higher than in the non-hypoperfusion group, and the differences were statistically significant (p = 0.006,0.009). The remaining plaque characteristics did not significantly differ between the two groups (all p > 0.05). The comparison of plaque characteristics between the two groups is presented in Table 2. Figure 4 illustrates the differences in clinical baseline characteristics and plaque characteristics between the two groups.

Patients in the hypoperfusion group were divided into low-risk and high-risk groups according to the ABCD² score. The rMTT was higher in the high-risk group [108.36(100.67–119.92)] than the low-risk group [117.6(109.31–128.14)], and the difference was statistically significant (p = 0.037). The hemodynamic comparison between the low-risk and high-risk groups is shown in Table 3.