Perfusion deficit and vessel wall characteristics to predict recurrent ischemic events in medically treated patients with chronic symptomatic anterior circulation large vessel occlusion

Abstract

Background

To investigate the association between perfusion deficit, vessel wall characteristics, and risk of recurrent ischemic events in medically treated patients with chronic symptomatic anterior circulation large vessel occlusion.

Methods

We retrospectively reviewed chronic symptomatic patients due to anterior circulation large vessel occlusion in our center. All patients received multiparametric magnetic resonance imaging (including perfusion-weighted imaging and high-resolution vessel wall imaging) within 4 weeks to 3 months after symptom onset. The association between baseline clinical or imaging variables and recurrent ischemic events was assessed in bivariate models and multivariable logistic regression to identify independent predictors of recurrence.

Results

Among 71 enrolled patients, 21.1% (15/71) patients had recurrent ischemic events (nine ischemic strokes and six transient ischemic attacks) during a 2-year follow-up. In bivariate models, hypertension, occlusion with hyperintense signals, the presence of intraluminal thrombus, Tmax >4 s volume, Tmax >6 s volume, Tmax >8 s volume, and Tmax >10 s volume were associated with recurrence (all p < 0.05). In multivariate analysis, hypertension (p = 0.039, OR 10.057 (95% CI, 1.123–90.048)), higher deficit volume of Tmax >4 s (p = 0.011, OR 1.012 (95% CI, 1.003–1.021)) and occlusion with hyperintense signal (p = 0.030, OR 6.732 (95% CI, 1.200–37.772)) were still independent predictors of recurrent ischemic events.

Conclusions

Besides hypertension history, higher deficit volume of Tmax >4 s and occlusion with hyperintense signal determined using multiparametric MRI are strongly associated with risk for recurrent ischemic events in medically treated patients with chronic symptomatic anterior circulation large vessel occlusion. Future studies are needed to determine the utility of revascularization strategies in such high-risk patients.

Introduction

Even with optimal medical treatment, a considerable proportion of mild stroke or transient ischemic attack (TIA) patients combined with anterior circulation large vessel occlusion (ACLVO) continue to experience recurrent ischemic events in the chronic phase (beyond 4 weeks), which ranging from 7.3% to 22.7%.^1–3 The physical disability and cognitive function impairment resulting from recurrent events are significantly greater than those of the initial occurrence. ⁴ Identifying a subgroup at high risk for recurrence may facilitate risk stratification and influence treatment decision-making, such as aggressive endovascular intervention, to mitigate further morbidity and mortality in such patients over time.

Previous studies indicated that hypoperfusion and arterial-to-artery embolism often coexisted, or might work synergistically, as the primary cause of distal cerebral infarction in extracranial or intracranial atherosclerotic disease.^5,6 Recently, MR perfusion-weighted imaging (PWI) has played a pivotal role in triage for therapeutic intervention among ischemic stroke patients with emergent ACLVO, and time to maximum (Tmax) volume is a reliable used parameter of MR perfusion for such patients.^7,8 However, there are limited data on the utility of follow-up perfusion imaging in risk stratification for chronic medically treated patients with symptomatic ACLVO. Furthermore, the optimal Tmax threshold of perfusion deficit as a predictor of recurrent stroke/TIA in chronic patients remains to be established. Additionally, with the development of high-resolution vessel wall imaging (HR-VWI) for atherosclerotic diseases, we can obtain more detailed information regarding the signal intensity, enhancement characteristics, and intraluminal thrombus condition of the steno-occlusive segment.^9,10 Among chronic ACLVO, whether these vessel wall characteristics are related to recurrent ischemic events still needs to be explored.

In this retrospective study, we reviewed medically treated patients with chronic symptomatic ACLVO, and all of them received multiparametric magnetic resonance imaging (MRI) (including PWI and HR-VWI). We aimed to investigate the association between perfusion deficit, vessel wall characteristics, and risk of recurrent ischemic events in medically treated patients.

Methods

Patient selection

This study was approved by the ethics committee of our institution and conducted in accordance with the mandates of the Declaration of Helsinki (2008). The requirement for written informed consent was waived. We reviewed consecutive patients with chronic symptomatic ACLVO in our center between January 2019 and June 2021.

Including criteria were as follows: (1) unilateral ACLVO as confirmed by MR angiography (MRA), CT angiography (CTA), or conventional angiography; (2) symptomatic occlusion with ischemic stroke, or TIA; (3) time of onset exceeds 4 weeks and is less than 3 months; (4) presence of one or more atherosclerotic risk factors including hypertension, diabetes mellitus, hyperlipidemia, coronary artery disease, and smoking history; (5) no contradiction to aggressive medical therapy; (6) having conducted multiparametric MRI; (7) completed follow-up.

Excluding criteria were as follows: (1) inadequate image quality for analysis; (2) patients with multiple large vessels severe stenosis or occlusion in intracranial arteries and extracranial arteries; (3) anterior cerebral artery occlusion; (4) nonatherosclerotic occlusion, such as cardioembolism or dissection; (5) massive cerebral infarction or mRS ≥ 3; (6) time of onset is less than 4 weeks or more than 3 months; (7) without long-term use of antiplatelet drugs; (8) subsequent endovascular or surgery treatment without symptoms recurrence; (9) lost to follow-up.

Imaging protocol and analysis

After admission, all patients commonly received multiparametric MRI on a 3.0 Tesla MRI system (Siemens Skyra; Erlangen, Germany), which was equipped with a 20-channel head/neck coil. The detailed MRI protocol ¹¹ included three-dimensional (3D) time-of-flight (TOF) MRA, diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC), dynamic susceptibility contrast-perfusion weighted imaging (DSC-PWI), and HR-VWI. Among that, the HR-VWI technique consisted of a 3D T1-weighted SPACE (sampling perfection with application-optimized contrast using different angle evolutions) sequence before and after contrast administration.

Regarding 3D T1-weighted HR-VWI, the imaging analysis was focused on identifying signal intensity, intraluminal thrombus, and plaque enhancement. Occlusion with hyperintense signals (HISs), which represented intraplaque hemorrhage or fresh thrombus,^12,13 was defined as ≥ 150% signal intensity of the normal-appearing vessel wall in the occluded segment. Intraluminal thrombus was defined as isointense or HISs, or a mixture of both, that were consistent with occlusion on MRA and signal loss in its source images. ¹⁰ The degree of plaque enhancement within the occluded segment was classified into two subgroups ¹⁴ : marked plaque enhancement indicating enhancement similar to or greater than that of the pituitary infundibulum; no marked plaque enhancement including enhancement similar to that of adjacent normal arterial walls without plaque or enhancement greater than that of the adjacent normal arterial walls but less than that of the pituitary infundibulum. Perfusion deficit volumes (Tmax > 4, 6, 8, and 10 s) were estimated by an automated software program (NeuBrainCARE, Neusoft Medical, Shenyang, China). ¹⁵ Two experienced neuroradiologists independently performed all image assessments, and discrepancies were resolved by consensus.

Aggressive medical treatment

Demographic data, vascular risk factors, neurologic deficits, time between stroke/TIA onset and multiparametric MRI, vessel wall characteristics, and perfusion deficit were documented for each patient. All patients were recommended with dual-antiplatelet therapy (aspirin [100 mg/d] and clopidogrel [75 mg/d]) for 3 months after admission, and life-long aspirin (100 mg/d) or clopidogrel (75 mg/d) monotherapy was continued thereafter. Patients were also treated with statin therapy and risk factor modification (blood pressure and diabetes control). In addition, lifestyle measures (such as smoking cessation and exercise) were recommended for secondary stroke prevention.

Follow-up

Patients were followed up every 3–6 months over a two-year period by telephone investigating or outpatient visit for signs of cerebrovascular events, risk factors, and current medical treatment. The last follow-up time was June 2023. A recurrent ischemic stroke or TIA in the same territory of the diseased ACLVO was determined according to newly developed neurological deficits and whether new infarcts were confirmed with brain MRI upon recurrence. ¹⁶ During follow-up, for patients who still had recurrent ischemic symptoms after aggressive medical treatment, endovascular recanalization was recommended.

Statistical analysis

All statistical analysis was performed using SPSS version 25.0 (IBM, Armonk, NY). Categorical variables are presented as counts and percentages. Continuous variables are expressed as the means and standard deviations (SD) or as the medians and interquartile ranges (IQR). To identify significant differences in baseline parameters between patients with or without recurrence, univariate analysis was performed by the Fisher exact or x² tests for categorical variables and two-sample t tests or Mann-Whitney U-tests for continuous variables. To minimize the overfitting of the regression model, the multicollinearity of pretreatment variables in univariate analysis with p < 0.15 was detected using the variance inflation factor (VIF). Finally, variables with p < 0.15 and VIF <10 in the univariate analysis were enrolled in the multivariate binary logistic regression. A nomogram was constructed from the results of the multivariate analysis using an R package (version 3.4.4; R Package for Statistical Computing; www.r-project.org). Receiver operating characteristic (ROC) analysis with the area under the curve (AUC) calculation was used to assess and compare prognostic models. The Youden index was computed using sensitivity + specificity − 1 to identify the optimal cutoff values. Statistical significance was set at a two-tailed p-value of < 0.05.

Results

Baseline patient characteristics

Ultimately, a total of 71 patients were enrolled (Figure 1). The mean age was 56.0 years (SD: 10.6), and 80.3% of the patients were male. Based on the occurrence of symptoms, 20 (28.2%) and 51 (71.8%) patients were classified as TIA and stroke, respectively. The time interval between the symptom onset and multiparametric MRI was 37 days (IQR: 33–46). MRA confirmed that 21 had an ACLVO in the cervical internal carotid artery (ICA), seven in intracranial ICA, and 43 in the middle cerebral artery (MCA) trunk. DWI indicated that watershed infarction was presented in 57 (80.3%) patients. Regarding 3D T1-weighted HR-VWI, occlusion with HIS was detected in 20 (28.2%) patients on non-enhanced T1-weighted images, and marked plaque enhancement was identified in 62 (87.3%) patients. The presence of intraluminal thrombus was detected in 34 (47.9%) patients. The median perfusion deficit volume of Tmax > 4 s was 62.9 mL (IQR: 5.6–132.1). The detailed patient characteristics are presented in Table 1.

Figure 1.

Patient flow diagram. ACLVO: anterior circulation large vessel occlusion; mRS: modified Rankin Scale.

Table 1.

Patient characteristics.

Characteristics	Total (n = 71)	No recurrence (n = 56)	Recurrence (n = 15)	p-value
Mean age (years)	56.0 ± 10.6	55.9 ± 10.8	56.3 ± 10.2	0.879
Male sex, n (%)	57 (80.3%)	43 (76.8%)	14 (93.3%)	0.287
Traditional risk factors
Hypertension, n (%)	39 (54.9%)	26 (46.4%)	13 (86.7%)	0.005
Diabetes mellitus, n (%)	20 (28.2%)	14 (25.0%)	6 (40.0%)	0.410
Hyperlipidemia, n (%)	9 (12.7%)	7 (12.5%)	2 (13.3%)	1
Coronary artery disease, n (%)	6 (8.5%)	5 (8.9%)	1 (6.7%)	1
Smoking history, n (%)	38 (53.5%)	30 (53.6%)	8 (53.3%)	0.987
Drinking history, n (%)	28 (39.4%)	20 (35.7%)	8 (53.3%)	0.215
Clinical symptoms				0.078
TIA, n (%)	20 (28.2%)	19 (33.9%)	1 (6.7%)
Stroke, n (%)	51 (71.8%)	37 (66.1%)	14 (93.3%)
Baseline mRS score, median (IQR)	1 (0–1)	1 (0–1)	1 (1–1)	0.410
Multiparametric MRI characteristics
Onset to multiparametric MRI, median (IQR), days	37 (33–46)	38 (33–51)	36 (32–42)	0.468
Right lesion, n (%)	42 (59.2%)	34 (60.7%)	8 (53.3%)	0.606
Responsible artery				0.107
Cervical ICA, n (%)	21 (29.6%)	15 (26.8%)	6 (40.0%)
Intracranial ICA, n (%)	7 (9.9%)	4 (7.1%)	3 (20.0%)
MCA trunk, n (%)	43 (60.6%)	37 (66.1%)	6 (40.0%)
Infarct location, n (%)				0.073
Watershed infarction	57 (80.3%)	42 (75.0%)	15 (100.0%)
Non-watershed infarction	14 (19.7%)	14 (25.0%)	0 (0%)
Occlusion with HIS, n (%)	20 (28.2%)	11 (19.6%)	9 (60.0%)	0.006
Marked plaque enhancement, n (%)	62 (87.3%)	49 (87.5%)	13 (86.7%)	1
Presence of intraluminal thrombus, n (%)	34 (47.9%)	23 (41.1%)	11 (73.3%)	0.026
Perfusion deficit volume (mL)
Tmax >4 s volume, median (IQR)	62.9 (5.6–132.1)	41.9 (0–95.1)	171.4 (116.8–235.9)	<0.001
Tmax >6 s volume, median (IQR)	0 (0–26.1)	0 (0–11.8)	59.2 (10.5–126.9)	<0.001
Tmax >8 s volume, median (IQR)	0 (0–0)	0 (0–0)	22.1 (0–55.4)	<0.001
Tmax >10 s volume, median (IQR)	0 (0–0)	0 (0–0)	6.3 (0–19.5)	<0.001

TIA: transient ischemic attack; mRS: modified Rankin scale; IQR: interquartile ranges; ICA: internal carotid artery; MCA: middle cerebral artery; HIS: hyperintense signals; Tmax: time to maximum.

Predictive factors of recurrence

Among the 71 patients who underwent multiparametric MRI, 21.1% (15/71) patients had recurrent ischemic events (nine ischemic strokes and 6 TIAs) during the two-year follow-up. The recurrent ischemic events occurred within the first year in 86.7% of patients (13/15) and within the second year in the remaining two patients. For those patients with recurrence, 10 underwent endovascular recanalization attempts, and successful recanalization was achieved in 7 (70.0%) patients. A representative recurrence case is presented in Figure 2.

Figure 2.

An elderly man presented with numbness and weakness in the right limb one month ago and a multiparametric MRI was performed after admission. DWI (A) revealed subacute infarction of the left watershed zone and MRA (B) was diagnosed with left cervical ICA occlusion. HR-VWI scans (C) confirmed the occlusion of the C1 segment (thin arrow) and the presence of HIS (thick arrows). PWI (D) showed large areas of hypoperfusion in the left hemisphere (Tmax >4 s volume was 171.4 mL). Although aggressive medical treatment, the patient had recurrent symptoms 2 months after onset. Follow-up DWI (E) showed a new infraction of the paraventricular area. Preoperative angiography (E) showed a tapered stump (thin arrow) and reversal of flow above the clinoid segment of the left ICA (thick arrow). After the microwire passed the occlusion site and balloon angioplasty, angiography (G) showed a long, rough vascular wall (arrowheads) and regional dissection (thin arrow). Successful recanalization with good antegrade perfusion was achieved followed by the insertion of one Neuroform EZ stent and two Wallstent stents (H). Follow-up CT perfusion (I) showed the hypoperfusion area of the left hemisphere disappeared. DWI: diffusion-weighted imaging; MRA: magnetic resonance angiography; ICA: internal carotid artery; HR-VWI: high-resolution vessel wall imaging; HIS: hyperintense signals; PWI: perfusion-weighted imaging; Tmax: time to maximum; MRI: magnetic resonance imaging.

In univariate analysis (Table 1), hypertension (p = 0.005), occlusion with HIS (p = 0.006), the presence of intraluminal thrombus (p = 0.026), Tmax > 4 s volume (p < 0.001), Tmax > 6 s volume (p < 0.001), Tmax > 8 s volume (p < 0.001), and Tmax >10 s volume (p < 0.001) were significantly associated with recurrent ischemic events.

To minimize the overfitting of the regression model, multicollinearity of pretreatment variables in univariate analysis with p < 0.15 was detected using VIF. Tmax > 6 s volume (VIF =24.5), Tmax > 8 s volume (VIF =33.3), and Tmax > 10 s volume (VIF = 10.3) were excluded due to multicollinearity with VIF > 10. In further multivariate binary logistic regression analysis (Table 2), we found that hypertension (p = 0.039, OR 10.057 (95% CI, 1.123–90.048)), higher deficit volume of Tmax >4 s (p = 0.011, OR 1.012 (95% CI, 1.003–1.021)) and occlusion with HIS (p = 0.030, OR 6.732 (95% CI, 1.200–37.772)) were still independent predictors of recurrent ischemic events. A nomogram was built for ease of use (Figure 3). As shown in Figure 4, in ROC analyses, the AUCs of hypertension, higher deficit volume of Tmax > 4 s, and occlusion with HIS were 0.701 (95% CI, 0.564–0.838), 0.851 (95% CI, 0.738–0.964) and 0.702 (95% CI, 0.542–0.862), respectively. The cutoff value with the highest Youden index for Tmax > 4 s deficit volume was 114.5 mL (sensitivity, 80.0%; specificity, 87.5%).

Table 2.

Multivariate analysis of predictors for recurrent ischemic events.

Independent variablesa	OR	95% CI for OR	p-value
Hypertension
Yes	10.057	1.123–90.048	0.039
No	1
Tmax >4 s deficit volume, mL	1.012	1.003–1.021	0.011
Occlusion with HIS
Yes	6.732	1.200–37.772	0.030
No	1

^a The following variables were removed from the model: clinical symptoms, responsible artery, infarct location, and presence of intraluminal thrombus. Tmax: time to maximum; HIS: hyperintense signals.

Figure 3.

The nomogram was built according to independent predictors of recurrent ischemic events. HIS: hyperintense signals; Tmax: time to maximum.

Figure 4.

Receiver operating characteristic curve analysis of the indicated parameters for recurrence prediction. AUC: area under the curve; HIS: hyperintense signals; Tmax: time to maximum.

Discussion

Although it is still controversial regarding the determinant factors of recurrent ischemic events in patients with chronic symptomatic ACLVO, our results identified hypertension as an independent risk factor for recurrent events. More notably this study suggested that higher perfusion deficit volume of Tmax > 4 s and occlusion with HIS on HR-VWI were significantly associated with recurrence.

Hypertension is widely recognized as a significant risk factor for recurrent strokes. Numerous studies have consistently identified hypertension as an independent and robust predictor of recurrent stroke.^17,18 The results of our study among chronic symptomatic ACLVO patients are in accordance with previous studies. On the one hand, hypertension is a group of chronic, progressive conditions that can promote large or small cerebral artery disease, which may increase stroke recurrence.^19,20 On the other hand, some results^21,22 suggest that among large artery steno-occlusive disease, patients with or without impaired cerebral perfusion should be demanded for different blood pressure control. Aggressive blood pressure lowering during follow-up may increase the risk of ischemic stroke in hypertension patients with impaired perfusion. Future large sample studies including hemodynamic measurement are needed to determine the optimal level to which blood pressure should be lowered to achieve maximal benefits in chronic ACLVO patients.

Among chronic ACLVO patients, the presence of collaterals and variability of vascular anatomy strongly influence perfusion downstream. ²³ However, these image parameters are determined by merely visual judgment. Recently, watershed infarction has been reported as a common sign of impaired perfusion downstream in steno-occlusive disease.^24,25 In a SAMMPRIS post-hoc analysis of anterior circulation infarctions, ²⁴ patients with qualifying events attributable to internal and cortical borderzones were at higher rates of recurrent infarction compared to non-borderzone infarctions at a median follow-up of 31 months (26.4% vs. 10.4%, p = 0.054). However, the evaluation of watershed infarction is empirical and subjective, and some patients have combinations of several infarct patterns at the same time, which have led to inconsistency in the conclusions of previous studies. ²⁶ In this study, we employed Tmax parameter maps to perform absolute quantification of perfusion deficit volumes. These parameter maps, based on automated perfusion post-processing software, can be obtained quickly and exhibit high repeatability in both CT and MRI modalities. ²⁷ Sacchetti et al. ²⁸ explored quantitative measures of reduced MR perfusion among intracranial atherosclerotic disease and found that a mismatch volume > 15 mL, based on Tmax > 6 s quantified low perfusion, helps predict the recurrence of stroke. In acute stroke patients with ACLVO managed medically, Wang et al. ²⁹ also showed that patients with CT perfusion lesions (measured by Tmax > 6 s) ≥ 65 mL, compared with patients with perfusion lesions < 65 mL, showed a much higher rate of poor functional outcome (32% vs. 14%, p = 0.011). Consistent with these studies, our results showed that for chronic patients with symptomatic ACLVO, the perfusion deficit volume (measured by Tmax > 4, 6, 8, and 10 s) in the recurrence group was significantly higher than that in the non-recurrence group (all p < 0.001). In further multivariate analysis, instead of the traditional qualitative index (watershed infarction), the higher perfusion deficit volume of Tmax > 4 s was the independent risk factor associated with the recurrence of ischemic events (p = 0.011). Notably, a perfusion deficit volume of 114.5 mL (Tmax > 4 s) was the optimal cut-point to identify recurrent ischemic events (sensitivity, 80.0%; specificity, 87.5%), which is higher than the optimal Tmax threshold of previous studies. We thought that this could be ascribed to several factors. Firstly, previous studies focused on acute ACLVO always selected perfusion deficit volume of Tmax > 6 s and deficit volume of Tmax > 4 s should be greater than that of Tmax > 6 s for the same patient. Secondly, for non-acute patients with ACLVO, especially after the onset of more than 4 weeks and the disease has reached relative stability, the patient's tolerance to brain hypoperfusion might be better than that in the acute phase.

HR-VWI has been used as a promising approach to detect atherosclerotic plaque vulnerability.^14,30 Recently, it has been reported that plaque enhancement could be a useful imaging marker associated with the intraplaque inflammatory reaction that may increase the risk of stroke recurrence. ³¹ In this chronic symptomatic ACLVO cohort, plaque enhancement was prevalent and observed in 62/71 (87.3%) of patients. However, no significant difference was detected between patients with or without recurrent ischemic events. Indeed, since plaque enhancement usually changes followed by aggressive medical treatment, only those culprit lesions with persistent enhancement may contribute to the occurrence of a new cerebral infarction.^32,33 Remarkably, the univariate and multivariate analysis in this study showed occlusion with HIS on HR-VWI was independently associated with recurrent ischemic events. On the one hand, HIS in the occluded segment represents intraplaque hemorrhage or fresh thrombus, which is a subtype of vulnerable plaque or unstable thrombus, ³⁴ and has a tendency to further increase and progress over time. On the other hand, the presence of microembolic signals (MES) on transcranial Doppler has drawn significant attention in steno-occlusive diseases due to its potential as a predictive marker for future stroke risk in such patients. ³⁵ Occlusion with HIS detected on HR-VWI may increase the risk of MES in the downstream, which contributes to subsequent embolism infarction in chronic ACLVO. For patients with intraluminal fresh thrombus, there is still insufficient clinical evidence on whether early antiplatelet combined with anticoagulant therapy is effective. Further studies are needed to clarify this speculation.

This study has several limitations. First, it is retrospectively analyzed based on single-center data and selection bias cannot be avoided. Second, the recurrence rate of ischemic events may be underestimated because some patients may be asymptomatic, and subsequent diagnostic brain imaging is less likely to be performed without new symptoms. Third, due to the lack of follow-up MRA or CTA data, we cannot know whether spontaneous recanalization of occluded vessels occurs. Finally, we did not set a validation group for our nomogram model due to the limited cases enrolled. We look forward to future prospective studies with large samples for further validation.

Conclusions

Besides hypertension history, higher deficit volume of Tmax > 4 s and occlusion with HIS determined using multiparametric MRI are strongly associated with risk for recurrent ischemic events in medically treated patients with chronic symptomatic ACLVO. Future studies are needed to determine the utility of revascularization strategies in such high-risk patients.