Predictors of early infarct recurrence in patients with symptomatic intracranial atherosclerotic disease

Abstract

Background and Purpose:

While prior studies identified risks factors for recurrent stroke in patients with symptomatic intracranial atherosclerotic disease (ICAD), few have assessed risk factors for early infarct recurrence.

Methods:

We performed a post-hoc analysis of the Mechanisms of earlY Recurrence in Intracranial Atherosclerotic Disease (MYRIAD) study of ICAD patients with recent (<21 days) stroke/TIA, 50–99% stenosis, and who underwent 6–8 week MRI per protocol. Infarct recurrence was defined as new infarcts in the territory of the symptomatic artery on brain MRI at 6–8 weeks compared to index brain MRI. Qualifying events and clinical and imaging outcomes were centrally ascertained by 2 independent reviewers. We assessed the association between baseline clinical and imaging variables and recurrent infarct in bivariate models and multivariable logistic regression to identify independent predictors of infarct recurrence.

Results:

Of 105 enrolled patients in MYRIAD, 89 (84.8%) were included in this analysis (mean age. 64±12 years, 54 (60.7%) were male, and 53 (59.6%) were white). The median time from qualifying event to MRI was 51+16 days, on which 22 (24.7%) patients had new or recurrent infarcts. Younger age (57.7 vs. 66.0 years; p<0.01), diabetes (32.6% vs. 14.6%, p=0.05), index stroke (31.3% vs. 4.6%, p=0.01), anterior circulation location of stenosis (29.7% vs. 12.0%, p=0.08), number of DWI lesions (>1: 40.0%, 1: 26.9% vs. 0: 4.4%, p<0.01), and borderzone infarct pattern (63.6% vs. 25.0%, p=0.01) on baseline MRI were associated with new or recurrent infarcts. Age (adj. OR 0.93, 95% CI 0.89–0.98, p<0.01) and number of DWI lesions (adj. OR 3.24, 95% CI 1.36–7.71, p<0.01) were independently associated with recurrent infarct adjusting for hypertension, diabetes, and stenosis location (anterior vs. posterior circulation).

Conclusions:

An index multi-infarct pattern is associated with early recurrent infarcts, a finding that might be explained by plaque instability and artery-to-artery embolism. Further investigation of plaque vulnerability in ICAD is needed.

Clinical Trial Registration Information:

https://clinicaltrials.gov/ct2/show/NCT02121028

INTRODUCTION

Clinical trials of intracranial atherosclerotic disease (ICAD) have found 12–20% of patients experience recurrent stroke at 1 year despite best medical management, with the highest risk of recurrence in the first weeks after randomization.^1–3 In the National Institutes of Health (NIH)-funded Mechanisms of earlY Recurrence in Intracranial Atherosclerotic Disease (MYRIAD) study, the risk of clinical stroke recurrence in the territory of the symptomatic artery at 1 year was 8.8%; however, nearly 25% of patients had recurrent infarcts on 6–8 week brain magnetic resonance imaging (MRI).⁴

Subclinical or imaging recurrence has been less studied, though a few Asian studies also suggest a high rate of infarct recurrence on brain magnetic resonance imaging MRI using diffusion-weighted imaging (DWI) and fluid attenuated inversion recovery (FLAIR) sequences.^{5, 6} The effects of subclinical or silent infarct accumulation on clinical outcomes in patients with ICAD are unknown. However, based on prior studies in non-ICAD cohorts, it is likely that subclinical infarcts increase the risk of developing dementia and accelerate cognitive decline.⁷ Furthermore, there have been limited data on predictors of early infarct recurrence in ICAD. Therefore, we sought to evaluate baseline clinical and imaging predictors of infarct recurrence at 6–8 weeks in the MYRIAD study.

METHODS

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study cohort:

With institutional review board approvals and informed consent, we enrolled consecutive patients at 10 recruiting sites in the United States.⁸ Eligibility criteria included recent (≤21 days) ischemic stroke or a transient ischemic attack (TIA) caused by ICAD of the intracranial carotid artery, middle cerebral artery M1 segment, basilar artery, or vertebral artery V4 segment, with 50–99% stenosis, in the absence of proximal cervical arterial stenosis >50% or a cardioembolic source. Ischemic stroke was defined as symptoms lasting >24 hours with confirmatory imaging of acute infarction while TIA was defined as symptoms lasting <24 hours and either diffusion weighted imaging (DWI) abnormality, or 2 or more stereotypical events with unequivocally ischemic symptoms. The degree of stenosis was calculated by established methods⁹using digital subtraction angiography, CT angiography, or MR angiography upon central review; flow gap on MRA was considered to represent >70% stenosis. We excluded those with contraindications to MRI, MR contrast agents, including allergy, creatinine >1.5 or GFR <30 mL/min/1.73 m², and those with planned endovascular treatment for ICAD. Enrolled patients underwent initial study-related biomarker imaging including quantitative MRA (QMRA), perfusion-weighted imaging (PWI) MRI, transcranial Doppler with vasomotor reactivity (TCD-VMR), and TCD embolic detection study (EDS) within 21 days of index stroke. All enrolled patients were treated with aggressive medical therapy based on the Stenting and Aggressive Medical Management for Preventing Recurrent Stroke in Intracranial Stenosis (SAMMPRIS) trial medical regimen.¹

Outcome of interest:

A study-paid brain MRI with FLAIR, DWI, and apparent diffusion coefficient (ADC) sequences was obtained at 6–8 weeks. A recurrent infarct at 6–8 weeks was defined as a lesion on DWI/ADC or FLAIR sequence that was localized to the territory of the stenotic artery and was not visualized on the baseline MRI; all cases of recurrent infarct included a baseline MRI for comparison. Recurrent infarcts were ascertained by 2 independent experienced vascular neurologists who reviewed clinical data and imaging studies, including that obtained at baseline and 6–8 weeks. In case of disagreement, they reviewed the case to reach consensus.

Baseline variables:

We recorded demographic (age, sex, race/ethnicity), medical history (hypertension, diabetes, hyperlipidemia, prior stroke, physical activity, body mass index, smoking, and alcohol), medications (anti-thrombotic, anti-lipid therapy), vital signs (blood pressure, body mass index), presenting symptom characterization (stroke vs. TIA, time and date of onset, symptoms), baseline National Institutes of Health Stroke Scale (NIHSS) score, baseline brain CT/MRI, baseline vascular imaging, baseline laboratory tests results (complete blood count with differential, complete metabolic profile, lipid profile, coagulation profile, glycohemoglobin in diabetic patients) at the time of the index event. Degree of stenosis was categorized as 50–69% vs. 70–99% stenosis and also by location (anterior vs. posterior circulation).

Index brain imaging was rated for presence of DWI lesion and, if present, number of discrete lesions. Furthermore, raters assessed for a borderzone infarct pattern defined as lesions in the corona radiata or centrum semiovale adhering to the internal borderzone and/or in the cortical borderzone regions between the middle and anterior or middle and posterior cerebral arteries.

Imaging biomarker variables:

For QMRA, abnormal “low flow” state was defined as volumetric flow rate (VFR) <25% of the lower end of normal age-, sex-, and artery-specific VFR at baseline. For PWI, abnormal perfusion was defined as Tmax >4 sec in a volume greater than 10 cc. For TCD-VMR, abnormal breath-holding index (BHI) was defined as <0.69. For TCD-EDS, microembolic signals (MES) distal to the stenotic artery were defined as present (at least 1 ES) or absent at baseline imaging. These definitions were used for the primary analysis. In this secondary analysis of MYRIAD, we assessed the following cutpoints: QMRA using the VERiTAS definition¹⁰ in posterior circulation patients; Tmax >4sec in volume greater than 5 cc; and BHI <0.90 and <0.50.

Statistical Analysis:

Analyses were performed using SAS software, version 9.4 (SAS Institute, Cary, NC). Categorical variables were presented as counts and percentages, and the difference between groups was tested using Pearson chi-square tests or Fisher’s exact tests, as appropriate. Continuous variables were presented as means and standard deviations (SD) or medians with 25^th and 75^th percentiles and the difference between groups was tested using the Mann-Whitney U test. We subsequently build step-wise models for the outcome of interest, new or recurrent infarct at 6–8 weeks, including initially baseline clinical and imaging variables associated with the outcome at p <0.10 in bivariate analysis. P <0.05 was considered significant in final models.

RESULTS

Of 105 enrolled patients in MYRIAD, 89 (84.8%) met eligibility criteria to be included in this analysis. The median times from qualifying event to completion of biomarker imaging and 6–8 week MRI was 15±10 days and 51±16 days, respectively. The mean age was 64±12 years, 54 (60.7%) were male, and 53 (59.6%) were white. A total of 22 (24.7%) patients had new or recurrent infarcts on follow-up MRI. Of these, 14 were present on DWI, with corresponding ADC and FLAIR changes while 8 were present on FLAIR without DWI/ADC correlate.

On bivariate analysis, new infarcts on 6–8 week MRI was associated with younger age (57.7 vs. 66.0 years; p<0.01) and diabetes (32.6% vs. 14.6%, p=0.05). Those with index ischemic stroke were more likely to have recurrent infarcts than those with TIA (31.3% vs. 4.6%, p=0.01). There was no association with sex, race/ethnicity, hypertension, hyperlipidemia, prior stroke, body mass index, pre-stroke medications, current smoking, alcohol use, or physical activity (Table 1). Imaging predictors included anterior circulation (29.7% vs. 12.0%, p=0.08), presence of DWI abnormality (34.4% vs. 4.4%, p<0.01), number of DWI lesions (>1: 40.0%, 1: 26.9% vs. 0: 4.4%, p<0.01), and borderzone infarct pattern (63.6% vs. 25.0%, p=0.01). Degree of stenosis (70–99% or flow gap vs. 50–69%) was not associated with recurrent infarcts (28.0% vs. 7.1%, p=0.098).

Table 1.

Baseline clinical characteristics in those with and without infarct recurrence at 6–8 weeks

	New infarct (n=22)	No new infarct (n=67)	p-value
Age in years, mean (SD)	57.7 (12.2)	66.0 (11.6)	0.005
Male, n (%)	14 (63.6)	40 (59.7)	0.743
White, n (%)	13 (59.1)	40 (59.7)	0.960
Body mass index in kg/m2, mean (SD)	31.5 (6.9)	30.0 (8.0)	0.141
Hypertension, n (%)	21 (95.5)	54 (80.6)	0.067
Diabetes mellitus, n (%)^	15 (71.4)	31 (47.0)	0.051
Hyperlipidemia, n (%)*	13 (59.1)	47 (74.6)	0.169
Prior stroke, n (%)#	2 (10.5)	15 (23.8)	0.211
Current smoking, n (%)	6 (27.3)	9 (13.4)	0.132
No alcohol use, n (%)	17 (77.3)	53 (79.1)	0.856
No physical activity, n (%)	17 (77.3)	39 (58.2)	0.108
Index ischemic stroke, n (%)	21 (95.5)	46 (68.7)	0.011
NIHSS score, median (IQR)	2 (0–3)	1 (0–3)	0.643
SBP at enrollment, mean (SD)	139.2 (16.2)	146.0 (19.3)	0.141
DBP at enrollment, mean (SD)	81.7 (11.8)	80.2 (13.2)	0.636

^{^}n=2 missing,

^*n=4 missing,

^#n=7 missing

Using alternative cut-points for each imaging biomarker (Table 2), none reached statistical significance: BHI <0.50 (28.5% vs. 17.9%, p=0.321); BHI <0.90 (31.3 vs. 12.5%, p=0.141), and Tmax >4sec in volume >5 cc (26.1% vs. 11.5%, p=0.144). Having >1 study-specific biomarker abnormality trended towards increased risk of recurrent infarcts (33.3% vs. 17.0%, p=0.07). The presence of either Tmax >4sec in volume >5 cc or MES increased risk of recurrent infarcts (27.6% vs. 0%, p=0.06). Among posterior circulation patients (n=24), 25% had low flow on QMRA based on the VERiTAS definition. The risk of infarct recurrence was 33.0% vs. 5.6% in those with low compared to normal flow (p=0.14). When we assessed whether earlier biomarker imaging (≤14 days vs. >14 days) was predictive of recurrent infarcts, we found no association (data not shown).

Table 2.

Baseline imaging characteristics in those with and without infarct recurrence at 6–8 weeks

	New infarct (n=22)	No new infarct (n=67)	p-value
70–99% stenosis or flow gap, n (%)	21 (95.5)	54 (80.6)	0.098
Anterior circulation, n (%)	19 (86.4)	45 (67.2)	0.082
Diffusion abnormality, n (%)^	21 (95.5)	40 (64.5)	0.005
Borderzone infarct pattern, n (%)@	7 (36.8)	4 (10.0)	0.013
Number of DWI lesions, n (%)^			0.007
0	1 (4.6)	22 (35.5)
1	7 (31.8)	19 (30.6)
2 or more	14 (63.6)	21 (33.9)
2 or more imaging biomarkers, n (%)	14 (63.6)	28 (41.8)	0.075
Tmax >4 sec in >5 cc	12 (54.5)	34 (50.7)	0.144
Breath holding index <0.5#	10 (45.5)	25 (37.3)	0.321
Breath holding index <0.9#	15 (68.2)	33 (49.5)	0.141
Low flow (VERiTAS definition)*	2 (66.7)	4 (19.0)	0.143
Tmax >4 sec in >5 cc or MES by TCD+	16 (100)	42 (80.8)	0.058

^{^}Available in 84 patients (22 with new infarcts and 62 without)

^@Assessed in 59 patients with anterior circulation stenosis (19 with new infarcts and 40 without)

^#Available in 63 patients (15 with new infarcts and 48 without)

^*Assessed in 24 patients with posterior circulation stenosis (3 with new infarcts and 21 without)

⁺Available in 68 patients (16 with new infarcts and 52 without)

In a multivariable model (Table 3) of baseline clinical and imaging factors, age (adj. OR 0.94, 95% CI 0.88–0.99, p=0.02) and number of DWI lesions (adj. OR 3.56, 95% CI 1.30–9.76, p=0.02) were independently associated with recurrent infarct adjusting for hypertension, diabetes, and stenosis location (anterior vs. posterior circulation). Other imaging biomarkers were not independently associated with infarct recurrence when adjusting for age and number of DWI lesions (data not shown).

Table 3.

Multivariable logistic regression model of infarct recurrence at 6–8 weeks

Variable	Adj. OR	95% CI	p-value
Age, per 1 year increase	0.93	0.88–0.98	0.012
Number of DWI lesions on baseline MRI	3.1	1.15–8.14	0.025

Model included: age, hypertension, diabetes, anterior (vs. posterior) circulation, severe stenosis, #DWI lesions, >1 imaging biomarker abnormality

DISCUSSION

In this secondary analysis of the observational multi-center MYRIAD study of biomarkers of recurrent stroke in patients with symptomatic ICAD, we found that a multiple DWI lesion pattern at baseline was independently associated with recurrent infarct at 6–8 weeks. This strong association implies that early hemodynamic factors (with borderzone infarct pattern) and artery-to-artery embolism may be important predictors of very early recurrent infarction in patients with symptomatic ICAD. Our findings of associations with borderzone infarct pattern and hypoperfusion using an alternate Tmax threshold are supportive of this hypothesis.

Borderzone infarcts have been shown to be associated with increased risk of recurrent stroke in patients in SAMMPRIS.¹¹ Furthermore, borderzone infarcts are markers of impaired perfusion.^{12, 13} Hypoperfusion detected by QMRA and PET imaging has been shown to predict stroke risk in patients with symptomatic vertebrobasilar¹⁰ and extracranial and intracranial atherosclerotic disease despite medical therapy, respectively.¹⁴ Recent studies using perfusion imaging in patients with symptomatic proximal anterior circulation ICAD using found that a Tmax >6sec on baseline (within 48 hours of stroke) CT or MR perfusion in the territory of stenosis was associated with higher rates of neurological deterioration compared to those without perfusion abnormality.^{15, 16}

Multi-infarct patterns may also be attributed to artery-to-artery embolism. Unstable plaques are known to be friable and embolize atherothombotic material distally, especially in the first few days after index stroke or TIA. Prior studies have noted that MES are common when performed in the first week of index stroke presentation in patients with ICAD.^17–21 Though we did find a weak association between the combination of either Tmax abnormality or MES and infarct recurrence, TCD were obtained at median 15 days from index stroke in MYRIAD; furthermore dual antiplatelet therapy in the early period after an index stroke, which was standard in MYRIAD, would likely reduce MES on TCD monitoring^{22, 23} as well as clinical events,²⁴ and therefore decrease the likelihood of finding a stronger association.

The rate of recurrent infarcts we observed in MYRIAD is consistent with prior studies. In a study from South Korea that included 55 patients with moderate to severe intracranial stenosis, the rate of recurrent infarcts in the territory of the symptomatic artery was 50.9%.⁶ Another study from South Korea assessed 6-month recurrence on brain MRI and found 12.5% had recurrent infarcts in the territory of the stenotic artery.⁵ Thus, relative to clinical events, the rate of infarct recurrence using brain MRI may be 4–5 higher.

Our results also suggest that the very early period after an index stroke is a particularly vulnerable period for recurrent ischemia. Clinical trials have consistently demonstrated a high risk of recurrence in the first weeks after randomization,^1–3 despite intensive medical management with antithrombotics and high dose statins. In a real-life observational cohort, we previously showed that 30-day risk of recurrence was nearly 25%, 5 times higher than reported in prior clinical trials.²⁵ Nearly half of these were in the first week after index presentation, a period not often included in clinical trials.

Presence of prior infarcts in patients with ICAD is a risk factor for stroke recurrence,²⁶ and baseline infarct patterns such as mixed infarct, multiple infarct,⁵ and borderzone infarct patterns¹¹ have been associated with greater risk of stroke recurrence. While imaging patterns have been associated with clinical stroke recurrence, few have assessed the role of baseline imaging biomarkers on infarct recurrence. Asian cohort studies have identified several baseline infarct patterns that confer increased risk of infarct recurrence: mixed cortical-subcortical index infarct pattern, presence of large territorial infarct, multiple territorial infarcts.^{5, 6} Multi-infarct patterns have also been associated with increased stroke recurrence in patients with TIA or minor stroke.²⁷ In addition, those with severe stenosis or occlusion may be higher risk of infarct recurrence than those with moderate stenosis.⁶ We expand on these findings in a diverse, non-Asian cohort and have additionally shown that borderzone and multi-infarct patterns may be particularly useful biomarkers of heightened risk.

There are several limitations to this study. As it was not pre-specified and did not adjust for multiple comparisons, the results of this study are only hypothesis generating. Cutpoints for imaging variables were selected post-hoc and have not been previously validated. We also were likely underpowered to identify significant associations given relatively small sample sizes. Inter-rater reliability of assessing new infarcts on 6–8 week MRI was modest and raters reached consensus when there was disagreement; nevertheless, there is a potential for misclassification of infarct recurrence. Residual confounding could explain some of our findings. For example, we are unable to fully explain the paradoxical effect of age on infarct recurrence, suggesting the possibility of unmeasured or residual confounding.

SUMMARY

Early recurrent infarcts are common in patients with symptomatic ICAD, suggesting a high burden of subclinical disease. An index multi-infarct pattern is associated with early recurrent infarcts, a finding that might be explained by plaque instability and artery-to-artery embolism.

Further investigation of plaque vulnerability and other biomarkers, specifically in earlier time windows than have been done in clinical trials to date, is needed.

Non-standard Abbreviations and Acronyms

ICAD

intracranial atherosclerotic disease

MRI

magnetic resonance imaging

DWI

diffusion-weighted imaging

MYRIAD

mechanisms of early recurrence in intracranial atherosclerotic disease

FLAIR

fluid attenuated inversion recovery

TIA

transient ischemic attack

BHI

breath-holding index

MES

microembolic signals