The effect of dual antiplatelet therapy in different patterns of watershed infarction: Subgroup analysis of the INSPIRES trial

Summary

Watershed infarction (WI) is heterogeneous. This study aims to explore the effect of clopidogrel-aspirin in patients with WI and which WI patterns could gain more benefits. Patients are classified into cortical WI (CWI) (n = 484), internal WI (IWI) (n = 372), CWI+IWI (n = 410), and non-WI (n = 4,033) according to diffusion-weighted magnetic resonance imaging. The results show that patients with WI treated with clopidogrel-aspirin have a lower risk of stroke recurrence compared to aspirin at 90 days (hazard ratio [HR], 0.67; 95% confidence interval [CI], 0.49–0.93). Specifically, patients receiving clopidogrel-aspirin show a lower rate of recurrent stroke than those receiving aspirin in IWI (HR, 0.54; 95% CI, 0.30–0.97), and with a similar trend in CWI+IWI, but not significant in CWI (p for interaction = 0.41). Clopidogrel-aspirin does not increase moderate-to-severe bleeding across WI patterns. This study reveals that the effect of clopidogrel-aspirin appears consistent across WI subgroups, but it might be more effective in patients with IWI. This study is registered at Clinicaltrials.gov (NCT03635749).

Graphical abstract

Highlights

• •Internal WI and mixed WI are associated with increased risks of stroke recurrence
• •Clopidogrel-aspirin may not be different between patients with and without WI
• •Clopidogrel-aspirin is consistent across WI patterns but may be more effective in IWI
• •Clopidogrel-aspirin does not increase moderate-to-severe bleeding across WI patterns

Liu et al. explore the effectiveness and safety of clopidogrel-aspirin in patients with different patterns of watershed infarction (WI). They find a higher risk of recurrent stroke in WI than in non-WI. The effect of clopidogrel-aspirin appears consistent across WI subgroups but might be more effective in patients with IWI.

Introduction

Watershed infarctions (WIs), also known as border zone infarcts, occur at the border between cerebral vascular territories where the tissue is the farthest from the arterial supply and thus most vulnerable to reductions in perfusion.¹ WIs account for 10% of all cerebral infarctions and are associated with a high risk of stroke recurrence and neurological deterioration at an early stage.^1,2,3

Possible etiologies of WI include large artery stenosis or occlusion, systemic hypotension, micro-embolism, decreased microemboli clearance, hemodynamic impairment, and hemorheological and anatomical variation.^4,5,6,7,8 Previous studies have shown that large artery stenosis or occlusion is the most frequent cause for WI in the Chinese population, which may be due to the high prevalence of intracranial atherosclerotic stenosis (up to 46%) in this population.^9,10 Classification of WI is based on location, with external border zone infarct, also known as cortical watershed infarction (CWI), occurring in the watershed territory between the anterior cerebral artery (ACA), middle cerebral artery (MCA), and posterior cerebral artery (PCA) territories. Internal border zone infarct, also known as an internal or subcortical watershed infarction (IWI), occurs in the watershed territory between ACA, MCA, and PCA and perforating medullary, lenticulostriate, recurrent artery of Heubner, and anterior choroidal artery territories.^5,6 Brain hypoperfusion and hemodynamic impairment caused by general circulatory impairment and severe large arterial stenosis or occlusion are the typical causes of WIs affecting the internal watershed territory.^6,11 Microembolism, caused by the tendency to form microemboli into distal arterioles or a decrease in microemboli clearance, contributes to infarcts in the external watershed territory where there is a lack of abundant collateral supply.^6,12

Current treatment of WIs includes antiplatelet agents and statins, correction of dehydration or hypotension, as well as endovascular or surgical re-vascularization.^2,13 Previous studies aimed at exploring effective therapeutic interventions for WIs have yielded contradictory results.^2,13,14 A possible explanation for the discrepancies among studies is that watershed infarcts are heterogeneous and composed of two distinct types: hypoperfusion and microemboli.^6,15 Few studies have provided evidence of the efficacy and safety of antiplatelet treatment for WIs according to their location and pathophysiology.

In the Intensive Statin and Antiplatelet Therapy for High-risk Intracranial or Extracranial Atherosclerosis (INSPIRES) trial, we compared combined clopidogrel-aspirin therapy initiated within 72 h after stroke onset with aspirin alone in patients with mild ischemic stroke or high-risk transient ischemic stroke (TIA) of presumed atherosclerotic cause and showed that dual antiplatelet therapy provided a non-significant 21% relative risk reduction in stroke recurrence but a higher risk of moderate-to-severe bleeding at 90 days.¹⁶ In this sub-analysis of INSPIRES trial, we aimed to determine whether clopidogrel-aspirin was more effective and safer compared to aspirin alone in patients with WI and which pattern of WI could gain more benefits from the dual antiplatelet treatment.

Results

Among the 6,100 patients recruited in the INSPIRES trial, 801 patients diagnosed with TIA were excluded. A total of 5,299 patients with ischemic stroke (age 63.5 ± 9.7 years, 3,450 male) were included in this analysis, including 1,266 with WI and 4,033 without WI. Among the 1,266 patients with WI (age 63.2 ± 9.9 years, 879 males) representing 23.9% of the ischemic strokes in the trial, 484 patients had only CWI, 372 had only IWI, and 410 had both CWI and IWI. Recruitment, randomization, and grouping are shown in Figure 1. The demographic and clinical characteristics of the patients according to different WI patterns are shown in Table 1. Patients with WI had a higher proportion of stenosis in culprit vessels, multiple infarction, and unilateral infarction, compared to patients without WI (p < 0.001, Table 1). Other characteristics including gender, previous ischemic stroke, hypertension, smoking status, and National Institutes of Health Stroke Scale (NIHSS) score differed between WI and non-WI groups and among the three WI groups (Table 1).

Figure 1.

Recruitment flow chart

CWI+IWI, combined cortical watershed infarction and internal watershed infarction.

Table 1.

Baseline clinical characteristics of patients with and without WI

	All patients with ischemic stroke (N = 5,299)	Patients with WI (n = 1,266)	Patients with CWI (n = 484)	Patients with IWI (n = 372)	Patients with CWI+IWI (n = 410)	p valuea	Patients with non-WI (n = 4,033)	p valueb
Age, mean (SD), years	63.5 (9.7)	63.2 (9.9)	63.65 (9.9)	63.5 (9.6)	62.5 (10.1)	0.19	63.6 (9.6)	0.25
Female	1,849 (34.9)	387 (30.6)	137 (28.3)	140 (37.6)	110 (26.8)	0.002	1,462 (36.3)	<0.001
BMI, mean (SD)	24.7 (3.3)	24.5 (3.2)	24.5 (3.1)	24.6 (3.3)	24.6 (3.2)	0.93	24.7 (3.3)	0.12
Medical history, no. (%)
Hypertension	3,484 (65.8)	798 (63.0)	296 (61.2)	247 (66.4)	255 (62.2)	0.26	2,686 (66.6)	0.02
Diabetes mellitus	1,424 (26.9)	317 (25.0)	113 (23.4)	90 (24.2)	114 (27.8)	0.28	1,107 (27.5)	0.09
Dyslipidemia	190 (3.6)	42 (3.3)	15 (3.1)	11 (3.0)	16 (3.9)	0.72	148 (3.7)	0.56
Previous ischemic stroke	1,480 (27.9)	351 (27.7)	164 (33.9)	96 (25.8)	91 (22.2)	<0.001	1,129 (28.0)	0.85
Previous TIA	56 (1.1)	12 (1.0)	6 (1.2)	2 (0.5)	4 (1.0)	0.57	44 (1.1)	0.66
Current smoker, no. (%)	1,602 (30.2)	460 (36.3)	174 (36.0)	106 (28.5)	180 (43.9)	<0.001	1,142 (28.3)	<0.001
Medication use, no. (%)
Aspirin	584 (11.0)	141 (11.1)	52 (10.7)	51 (13.7)	38 (9.3)	0.13	443 (11.0)	0.88
Clopidogrel	30 (0.6)	5 (0.4)	2 (0.4)	2 (0.5)	1 (0.2)	0.80	25 (0.6)	0.35
Lipid-lowering agents	406 (7.7)	97 (7.7)	45 (9.3)	28 (7.5)	24 (5.9)	0.15	309 (7.7)	0.99
Antihypertensives	2,542 (48.0)	566 (44.7)	214 (44.2)	182 (48.9)	170 (41.5)	0.11	1,976 (49.0)	0.01
Antidiabetics	1,203 (22.7)	259 (20.5)	88 (18.2)	75 (20.2)	96 (23.4)	0.15	944 (23.3)	0.03
NIHSS score, no. (%)
≤3	4,033 (76.1)	929 (73.4)	374 (77.3)	257 (69.1)	298 (72.7)	0.03	3,104 (77.0)	0.002
4 or 5	1,266 (23.9)	337 (26.6)	110 (22.7)	115 (30.0)	112 (27.3)		929 (23.0)
Time to randomization after onset of symptoms, no. (%)
≤24h	641 (12.1)	133 (10.5)	50 (10.3)	34 (9.1)	49 (12.0)	0.71	508 (12.6)	0.06
24h–48h	2,201 (41.5)	517 (40.8)	195 (40.3)	152 (40.9)	170 (41.5)		1,684 (41.8)
>48h	2,457 (46.4)	616 (48.7)	239 (49.4)	186 (50.0)	191 (46.6)		1,841 (45.7)
Statin therapy
Immediate use	2,621 (49.5)	624 (49.3)	237 (49.0)	187 (50.3)	200 (48.8)	0.90	1,997 (49.5)	0.88
Delayed use	2,678 (50.5)	642 (50.7)	247 (51.0)	185 (49.7)	210 (51.2)	–	2,036 (50.5)	–
Single/multiple infarction
Single infarction	1,174 (22.2)	3 (0.2)	3 (0.6)	0	0	0.08	1,171 (29.0)	<0.001
Multiple infarction	4,125 (77.8)	1,263 (99.7)	481 (99.4)	372 (0)	410 (0)		2,862 (71.0)
Location of artery stenosisc
None	1,012 (20.6)	139 (11.7)	60 (13.5)	47 (13.3)	32 (8.2)	0.09	873 (23.4)	<0.001
Intracranial	2,922 (59.4)	692 (58.3)	269 (60.3)	196 (55.5)	227 (58.4)		2,230 (59.8)
Extracranial	365 (7.4)	138 (11.6)	43 (9.7)	42 (11.9)	53 (13.6)		227 (6.1)
Both intra- and extra-cranial	619 (12.6)	219 (18.4)	74 (16.6)	68 (19.3)	77 (19.8)		400 (10.7)
Uni-/bilateral infarction
Unilateral	4,849 (92.7)	1,041 (95.2)	378 (96.2)	337 (95.2)	326 (94.0)	0.30	2,176 (91.5)	<0.001
Bilateral	255 (7.3)	53 (4.8)	15 (3.82)	17 (4.8)	21 (1.9)		202 (8.5)

BMI, body mass index (calculated as weight in kilograms divided by height in meters squared).

All p values < 0.05 were marked as bold.

^ap value is calculated for the differences between three different WI patterns using chi-square test/Fisher’s exact test for categorical variables or Kruskal-Wallis test for continuous variables.

^bp value is calculated for the difference between WI and non-WI using chi-square test/Fisher’s exact test for categorical variables or Mann-Whitney U test for continuous variables.

^cThe location of the culprit vessel with at least 50% stenosis.

The overall rate of recurrent stroke was 11.9% (151 of 1,266) in patients with WI compared to 8.0% (323 of 4,033) in patients without WI at 90 days (hazard ratio [HR], 1.52, 95% confidence interval [CI], 1.26–1.85, p < 0.001, Table 2). When stratifying by WI patterns, the rate of stroke recurrence was 14.2% in CWI+IWI, 14.0% in IWI, and 8.5% in CWI. The risk of recurrent stroke was greater in patients with CWI+IWI (14.2% vs. 8.0%, HR, 1.82; 95% CI 1.38–2.41; p < 0.001) and IWI (14.0% vs. 8.0%, HR, 1.81; 95% CI 1.35–2.42; p < 0.001) compared to those without WI. Differences remained after adjusting for age, gender, body mass index, smoking status, NIHSS score, medical history of hypertension, stroke, hyperlipidemia, diabetes and cardiovascular disease, and early statin treatment (Table 2). Significant variables also include age, NIHSS score, and diabetes and ischemic stroke (p < 0.05). The timing of statin therapy did not have effect on stroke recurrence in WI population (now shown).

Table 2.

Cumulative probability of recurrent stroke in patients with different patterns of WI

Patterns of WI	No. of patients	Recurrent stroke, no. (%)	HR (95% CI)	p value	Adjusted HR (95% CI)a	p valueb
Non-WI	4,033	323 (8.0)	reference	–	reference	–
WI	1,266	151 (11.9)	1.52 (1.26–1.85)	<0.001	1.53 (1.26–1.86)	<0.001
CWI	484	41 (8.5)	1.07 (0.77–1.48)	0.70	1.08 (0.78–1.49)	0.65
IWI	372	52 (14.0)	1.81 (1.35–2.42)	<0.001	1.77 (1.32–2.37)	<0.001
CWI+IWI	410	58 (14.2)	1.82 (1.38–2.41)	<0.001	1.85 (1.39–2.45)	<0.001

All p values < 0.05 were marked as bold.

^aMultivariable Cox proportional hazards regression modeling adjusted for age, gender, BMI, current or previous smoking, National Institutes of Health Stroke Scale score, medical history of hypertension, hyperlipidemia, stroke, diabetes or cardiovascular disease, and whether receiving early statin treatment.

^bp value is for the multivariate model adjusted for the confounding factors.

Efficacy outcomes

The main efficacy outcomes for the clopidogrel-aspirin and aspirin groups are shown in Figure 2 and Table S1. In 1,266 patients with WI, patients treated with clopidogrel-aspirin had a lower risk of stroke recurrence at 90 days than those treated with aspirin alone (9.7% vs. 14.1%, adjusted HR, 0.67; 95% CI, 0.49–0.93; p = 0.02). In patients without WI, dual antiplatelet treatment was associated with a non-significant numerically lower rate of stroke recurrence than aspirin (7.3% vs., 8.8%, adjusted HR, 0.82; 95% CI, 0.66–1.02; p = 0.07). When stratified by WI patterns, patients administered clopidogrel plus aspirin had a significant lower rate of recurrent stroke than patients receiving aspirin in IWI (10.1% vs. 17.6%; adjusted HR, 0.54; 95% CI, 0.30–0.97; p = 0.04), with a similar but non-significant trend in CWI+IWI (11.3% vs. 17.0%; adjusted HR, 0.60; 95% CI, 0.35–1.02; p = 0.06), but not in CWI alone (8.20% vs. 8.80%; adjusted HR, 0.89; 95% CI, 0.48–1.64; p = 0.70). No interaction effect was found between WI status and treatment (p = 0.31), and between WI patterns and treatment (p = 0.41). The timing of statin treatment did not have effect on the stroke recurrence in any WI pattern or patients without WI. The Kaplan-Meier curves for the cumulative probability of recurrent stroke and the number of patients at risk are shown in Figure 3.

Figure 2.

Comparison of efficacy outcomes between dual antiplatelet treatment and aspirin alone in patients with different patterns of watershed infarction

NA, not applicable. p value is for the multivariate model adjusted for the confounding factors: age, gender, BMI, current or previous smoking, National Institutes of Health Stroke Scale score, medical history of hypertension, hyperlipidemia, stroke, diabetes, or cardiovascular diseases, and whether receiving early statin treatment. Data bars are represented as HRs (95% CI). All p < 0.05 were marked as bold.

Figure 3.

Kaplan-Meier curves for cumulative probability of recurrent stroke in different patterns of watershed infarction

(A–E) (A) Kaplan-Meier curves of the cumulative probability of recurrent stroke according to different infarction patterns. (B–E) Kaplan-Meier curves of the cumulative probability of new stroke between different therapeutic groups in the three watershed infarction patterns (with p values by log rank test). The number of patients at risk was shown under each figure.

Regarding secondary efficacy outcomes, the risk of composite vascular events and ischemic stroke were lower in patients who received dual antiplatelet treatment than those who received aspirin alone, in patients with IWI, and overall WI patients (p values <0.05, Figure 2; Table S1). The risk of hemorrhagic stroke was higher in patients receiving clopidogrel-aspirin compared to those receiving aspirin alone (0.6% vs. 0.2%; adjusted HR, 3.43; 95% CI, 1.12–10.57; p = 0.03) in patients with non-WI, but not in overall WI patients or patients with different WI patterns (Figure 2). The comparison of other efficacy outcomes including vascular death, TIA, and MI between two treatment groups is shown in Figure 2 and Table S1. The timing of statin treatment did not have effect on any efficacy outcome in any WI pattern or patients without WI (not shown).

Safety outcomes

A total of seven cases of moderate-to-severe bleeding (0.5%) occurred in patients with WI during 90-day follow-up, including two with CWI, one with IWI, and one with CWI+IWI in patients who received clopidogrel-aspirin and one with CWI and two with CWI+IWI in those who received aspirin alone (Table 3). The comparison of other secondary safety outcomes between two treatment groups is presented in Table 3. The risk of any bleeding was higher in patients receiving clopidogrel-aspirin compared to those receiving aspirin alone (3.6% vs. 2.2%; adjusted HR, 1.65; 95% CI, 1.14–2.41; p = 0.01) in patients with non-WI. No significant differences of any safety outcome were found between clopidogrel-aspirin and aspirin groups in overall WI patients or patients with different WI patterns (Table 3).

Table 3.

Comparison of safety outcomes between dual antiplatelet treatment and aspirin alone in patients with different patterns of WI

Outcomes	Patterns of WI	Clopidogrel-aspirin			Aspirin			HR/OR (95% CI)	p value	Adjusted HR/OR (95% CI)c	p value	p value for interaction
		No. of patients	No. of events	Incidence of events	No. of patients	No. of events	Incidence of events
Primary safety outcomes
Moderate-to-severe bleedinga	CWI	244	2	0.8%	240	1	0.4%	1.95 (0.18–21.45)	0.59	2.41 (0.20–29.70)	0.49	0.77
	IWI	179	1	0.5%	193	0	0	N/A	N/A	N/A	N/A
	CWI+IWI	204	1	0.5%	206	2	1.0%	0.51 (0.05–5.63)	0.58	0.27 (0.01–7.08)	0.43
	all WI	627	4	0.6%	639	3	0.5%	1.36 (0.30–6.06)	0.69	1.28 (0.28–5.79)	0.75	0.59
	non-WI	2,024	21	1.0%	2,009	10	0.5%	2.09 (0.99–4.44)	0.05	2.09 (0.98–4.45)	0.06
Secondary safety outcomes
All-cause death	CWI	244	4	1.6%	240	2	0.8%	1.95 (0.36–10.64)	0.44	1.98 (0.34–11.61)	0.45	0.99
	IWI	179	0	0	193	1	0.5%	N/A	N/A	N/A	N/A
	CWI+IWI	204	4	2.0%	206	1	0.5%	4.05 (0.45–36.21)	0.21	2.66 (0.19–38.16)	0.47
	all WI	627	8	1.3%	639	4	0.6%	2.03 (0.61–6.74)	0.25	1.75 (0.49–6.21)	0.39	0.57
	non-WI	2,024	28	1.4%	2,009	21	1.1%	1.33 (0.76–2.34)	0.33	1.29 (0.73–2.28)	0.38
Any bleedingb	CWI	244	3	1.2%	240	4	1.7%	0.73 (0.16–3.26)	0.68	0.62 (0.12–3.11)	0.56	0.28
	IWI	173	6	3.4%	193	3	1.6%	2.16 (0.54–8.64)	0.28	2.39 (0.58–9.87)	0.23
	CWI+IWI	204	4	2.0%	206	7	3.4%	0.58 (0.17–1.97)	0.38	0.38 (0.09–1.53)	0.17
	all WI	627	13	2.1%	639	14	2.2%	0.94 (0.44–2.00)	0.87	0.93 (0.43–1.99)	0.85	0.17
	non-WI	2,024	72	3.6%	2,009	44	2.2%	1.64 (1.12–2.38)	0.01	1.65 (1.14–2.41)	0.01
Mild bleedingb	CWI	244	1	0.4%	240	3	1.3%	0.33 (0.03–3.13)	0.33	N/A	N/A	0.38
	IWI	179	5	2.8%	193	3	1.6%	1.80 (0.43–7.53)	0.42	1.85 (0.42–8.05)	0.41
	CWI+IWI	204	4	2.0%	206	5	2.4%	0.80 (0.22–2.99)	0.75	0.53 (0.12–2.44)	0.42
	all WI	627	10	1.6%	639	11	1.7%	0.92 (0.39–2.17)	0.85	0.89 (0.37–2.12)	0.79	0.27
	non-WI	2,013	53	2.6%	2,009	35	1.7%	1.51 (0.99–2.31)	0.06	1.52 (0.99–2.33)	0.05
Intracranial hemorrhageb	CWI	244	1	0.4%	240	0	0	N/A	N/A	N/A	N/A	N/A
	IWI	179	0	0	193	0	0	N/A	N/A	N/A	N/A
	CWI+IWI	204	1	0.5%	206	1	0.5%	1.02 (0.06–16.31)	0.99	N/A	N/A
	all WI	627	2	0.3%	639	1	0.2%	2.04 (0.19–22.45)	0.56	1.44 (0.11–19.07)	0.78	0.97
	non-WI	2,024	14	0.7%	2,009	7	0.4%	1.99 (0.80–4.93)	0.14	2.05 (0.82–5.09)	0.12

N/A, not applicable; OR, odds ratio.

All p values < 0.05 were marked as bold.

^aModerate-to-severe bleeding events were defined according to the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries criteria (severe bleeding was defined as intracranial hemorrhage, or other substantial/fatal hemorrhage causing hemodynamic compromise requiring treatment; moderate bleeding was defined by the need for blood transfusion).

^bBleeding events are defined according to the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries criteria.

^cMultivariable Cox proportional hazards regression modeling or general linear modeling adjusted for age, gender, BMI, and current or previous smoking, National Institutes of Health Stroke Scale score, medical histories of hypertension, stroke, diabetes, hyperlipidemia and cardiovascular disease, and whether receiving early statin treatment.

Discussion

This WI substudy demonstrated a higher risk of recurrent stroke in patients with WI than those without WI. Clopidogrel-aspirin treatment may be consistently beneficial to patients with and without WI. While there was no interaction between the subtypes of WI and clopidogrel-aspirin treatment, the beneficial effect of dual antiplatelet treatment might be more pronounced in patients with IWI.

Our results showed that patients with WI had a higher risk of stroke recurrence than patients without WI. More importantly, patients with IWI or CWI+IWI had a higher risk of stroke recurrence than those with CWI alone. The findings demonstrate poor early clinical outcomes in patients with WI, especially infarctions in the internal border zone, which can be explained by the anatomy of this area.^13,17 The penetrating arteries from the internal watershed territory are the distal branches of the main vessels, which have the lowest perfusion pressure and minimal collateral compensatory vessels.^18,19 Therefore, these regions are vulnerable to decreased cerebral perfusion and difficult to establish collateral circulation.^4,20 For CWI+IWI pattern, both hypoperfusion and microembolism may contribute to the infarction. In the condition of chronic hypoperfusion, small emboli are more prone to enter distal arterioles. Simultaneously, reduced cerebral perfusion may lead to impaired embolus clearance, which in turn may cause thrombosis of vessels in the watershed area.^4,6,7,18 Therefore, combined infarction in both the internal and cortical zones is considered the most severe pattern with the poorest clinical outcomes, as shown in this trial. On the other hand, the cumulative incidence of events for those with CWI and those without any watershed infarcts is almost identical; it may be explained as CWI has a possible better collateral circulation through leptomeninges and dural anastomosis, and thus less hypoperfusion.²¹ Moreover, patients with WI were more likely to present with large artery stenosis compared to those without WI, which may also explain the higher rate of stroke recurrence in WI. Although the concomitant presence of non-WI lesions in WI patients might impact the overall recurrence risk, our results supported that WI, particularly the infarctions involving the internal border zone, may be an important imaging marker for predicting early stroke recurrence. These findings are supported by earlier studies.^2,22,23

The main findings of our study showed that patients with WI receiving clopidogrel-aspirin exhibited a lower risk of recurrent stroke compared to those receiving aspirin alone. A non-significant trend was also found in patients without WI. No interaction was observed. The difference of statistical significance may be interpreted by the special anatomy and location of border zone, pathophysiology, and poor clinical outcomes of WI. The absence of interaction might be due to the concomitance of non-WI lesions in WI population as we classified patients according to whether they have WI (and the topography of WI) or not, regardless of other infarcts. More importantly, patients with IWI who received clopidogrel-aspirin showed a lower risk of stroke recurrence than those who received aspirin alone and a similar but not significant trend was observed in patients with CWI+IWI. However, no statistically significant benefit from dual antiplatelet treatment was observed in patients with isolated CWI (Figure 2). These results might be explained by the different pathophysiological mechanisms of the different WI patterns. It has been widely considered that large artery stenosis caused by atherosclerosis and the consequent hemodynamic impairment and hypoperfusion are primary causes of IWI, especially for those with infarctions in a “string-of-pearls” pattern.^6,7 Studies have also shown that in the area of hypoperfusion, blood flow slows facilitating platelet accumulation and aggregation, which are important elements in thrombus formation.²⁴ In addition, dual antiplatelet treatment might reduce the risk of atherosclerotic plaque rupture and thus decrease thrombogenesis. Therefore, it is possible that the benefits from dual antiplatelet treatment in patients with IWI may be attributed to the prevention of thromboembolism, alleviation of microcirculation disturbance, and correction of hemodynamics and reperfusion. Alternatively, the high prevalence of microembolic signals in CWI found by transcranial Doppler studies supports the microembolic etiology of CWI.^6,7,18 The compensatory mechanisms of the CWI territory are complex, which includes the anatomy of the circle of Willis, penetrating arteries, and collateral circulation.²¹ These special circumstances of CWI may explain the less impact of dual antiplatelet therapy. Another explanation for the lack of dual-antiplatelet benefit may be the relatively low risk of recurrent stroke inherent to patients with isolated CWI. Given by the benefit in IWI and neutral effect in CWI, the beneficial trend in CWI+IWI group was probably driven by the IWI component; therefore, the optimal treatment for CWI with IWI (CWI+IWI) requires further studies to be explored and verified. Our findings are supported by previous studies that showed the beneficial effect of dual antiplatelet and early endovascular therapy in patients with WI^2,14 and those with cerebral or carotid artery stenosis.^25,26 It is worth noting that although the INSPIRES trial excluded patients who would be scheduled for receiving intravenous thrombolysis or revascularization procedure, there are several patients who had extracranial vessel stenosis (6 of 356 overall and 3 of 138 with WI) received revascularization procedure during the follow-up period. The number of patients receiving revascularization procedure was small and balanced between two groups (overall: 4 vs. 2; WI: 2 vs. 1). Since intravenous thrombolysis was administered only to patients who had reached the study endpoint (developed recurrent stroke), it is not a confounding factor for the study results. We followed the clinical practice guidance and provided standard antiplatelet and statin therapy to patients with minor strokes or high-risk TIA, which ensures they receive good medical care.

For safety outcomes, we did not see significant difference of all bleeding events between two treatment groups in overall WI patients or within WI subtypes, while the risks of hemorrhagic stroke and any bleeding were statistically significantly higher in dual-antiplatelet group than those in aspirin group in patients with non-WI. There are several reasons that may explain the discrepancy. The number of bleeding events was quite small in WI patients, which makes several statistical analyses not applicable. Furthermore, cases of non-WI are caused by parent artery occluding penetrating artery infarction, artery-to-artery embolic strokes, and others,²⁷ and the embolic strokes have a higher risk of bleeding than other subtypes. Further analysis (not shown) revealed that the larger total number of bleeding events was primarily accounted for by the embolic strokes.

While hypothesis generating as a subgroup analysis, these results may have clinical relevance. Our study is to document the effect of dual antiplatelet therapy stratified by different patterns of WI. Patients with different infarction patterns have different pathophysiological mechanisms and clinical outcomes and thus may benefit differently from dual antiplatelet therapy. Our data provide evidence to further explore optimal therapies for different patterns and pathophysiologies of WI. WI may be considered an important imaging marker to predict early stroke recurrence. Therefore, effective interventions should be implemented as early as possible to reduce the risk of stroke recurrence and improve neurological function. Clopidogrel-aspirin treatment may be reasonable based on the positive results in stroke patients with WI and does not appear to increase moderate-to-severe bleeding. This is particularly relevant in the internal border zone and mixed patterns of WI. For the external border zone pattern of WI, future studies might test the effectiveness of anticoagulant therapy based on its pathogenesis of microembolism.

Overall, patients with WI showed a higher risk of recurrent stroke than those without WI. More specifically, patients with IWI, either alone or with additional CWI, had increased risks of stroke recurrence compared to those without WI. This work supports that WI may be considered an important imaging marker to predict early stroke recurrence. More importantly, the effect of clopidogrel-aspirin treatment appears consistent across WI subgroups, but it might be more effective than aspirin alone in patients with IWI. These results need to be further verified in large-scale studies.

Limitations of the study

This study has limitations. As a subgroup analysis of the INSPIRES trial, the limited sample size of different patterns of WI and the number of events in the two treatment groups reduced power and statistical significance. Future studies should increase the sample size and include a larger population of WI to provide enough study power. In addition, we did not perform perfusion imaging and microemboli detection and did not record post-stroke blood pressure. Future studies with more detailed imaging and complete blood pressure recording to assess distinct pathophysiologies and treatments for WI are indicated. We could not classify the type of recurrent strokes as based on the definition; not all recurrent cases required MRI evidence. Future study may be designed to collect and provide more information. In the INSPIRES trial, the rationale for exclusively including patients with presumed large-artery atherosclerotic stenosis and an NIHSS score ≤5 within 72 h was the lack of robust evidence regarding the efficacy of clopidogrel-aspirin therapy in this specific population. Although the results should not be generalized to those with other pathogeneses and more severe neurological impairment, the study offers valuable evidence and, to a certain degree, addresses a research gap in this specific field. Moreover, while exclusion of patients who received thrombolysis and endovascular treatment may limit the generalizability of the findings, it minimized the confounding effect of recanalization therapies on the assessment of the study drugs. Last, INSPIRES was conducted in the Chinese population, which has a much higher prevalence of intracranial artery stenosis than Caucasians. The external generalizability to other races should be assessed in future studies.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Ling Guan (lguanm@gmail.com).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The datasets used in this study are derived from the INSPIRES trial, which is sponsored by The National Key R&D Program of China (2017YFC1307900 and 2017YFC1307905). The data from this study are not publicly available due to patient privacy considerations. To request access to the data, please contact the lead contact (Ling Guan: lguanm@gmail.com). The requester must clearly describe the objectives of the research project for which the data will be used. Data access will be considered for non-commercial, research-oriented purposes only.
• •To ensure participant privacy, access to personally identifiable information or sensitive clinical details will not be provided. All data access requests must adhere to the consent agreements established with study participants and comply with all applicable institutional and national regulations.
• •This study does not report original code.
• •Additional information required to reanalyze the results is available from the lead contact upon reasonable request.

Acknowledgments

We sincerely appreciate the contributions of all patients recruited in this study. We gratefully acknowledge all physicians and nurses in the 222 study sites in the INSPIRES trial. We also appreciate Dr. Jean-Paul Collet at The University of British Columbia, who provided insightful advice on study design and result interpretation. This study was supported by The National Natural Science Foundation of China (nos. 82271516, 81801187, and 82425101), National Key R&D Program of China (2024YFC3044800), Noncommunicable Chronic Diseases-National Science and Technology Major Project (2023ZD0504800), and Beijing Municipal Science & Technology Commission (no. Z231100004823036).

Author contributions

L.G. and Yilong Wang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. L.G. and Yilong Wang proposed the study conception and study design. C.L., Y.L., Z.L., and Y.G. conducted the study, collected the data, and processed the data analyses. C.L., L.G., and Y.L. interpreted the results and prepared the draft of the manuscript. Y.P., H.Y., and Q.Z. conducted statistical analyses and helped to interpret the statistical results. Y.G., W.C., Y.L., and Z.L. supported the study conduction. S.C.J., D.W., P.A., P.M.B., A.W., and Yongjun Wang provided detailed review and editing of the article. None of the authors received payment for writing the article. All authors approve the submission of this version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Software and algorithms
SAS version 9.4	SAS Institute	https://www.sas.com
R version 4.2.3	R-Project	https://www.r-project.org/
RadiAnt DICOM Viewer	DICOM viewer program	https://www.radiantviewer.com/

Experimental model and study participant details

Human participants

The WI substudy was a secondary analysis of the INSPIRES trial which included human participants from 222 hospitals in China. Briefly, participants were eligible for enrollment if they were aged 35 to 80 years old, had mild ischemic stroke (National Institutes of Health stroke scale score [NIHSS] score of 5 or less) or high-risk TIA (ABCD2 score of 4 or higher) with presumed atherosclerotic cause within 72 h, and had not undergone thrombolysis or thrombectomy. All patients diagnosed with ischemic stroke were recruited for this WI substudy. The sample size was 5299 and the eligible patients were randomly assigned in a 1:1:1:1 ratio into four groups (see the randomisation and masking section). The INSPIRES trial was approved by the ethics committees of Beijing Tiantan Hospital (ethical approval No. KY2017-065-02) and all other participating centers. Patients or their legal representatives provided written informed consent before enrollment in the study.

Method details

We followed good clinical practice guidelines and the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline. Data were collected in accordance with the Declaration of Helsinki.

Study design

The WI substudy was a secondary analysis of the INSPIRES trial. The INSPIRES trial was a multicenter, randomized, double-blind, placebo-controlled, two-by-two factorial trial running from September 2018 to October 2022. The detailed participant eligibility, consent, randomisation and masking, data collection, follow-up and monitoring of the study have been published previously,²⁸ and all versions of study protocol and statistical analysis plan are available with this article online (Data S1: Study Protocol and Statistical Analysis Plan).

Randomisation and masking

Randomisation allocation was generated by a separate statistical team using a centralized computer. Eligible patients were randomly assigned in a 1:1:1:1 ratio: a. intensive antiplatelet therapy (clopidogrel at a loading dose of 300 mg on day 1, then 75 mg daily for days 2–90, combined with aspirin at a dosage of 100–300mg on day 1, then 100 mg daily for days 2–21) plus immediate intensive statin therapy (atorvastatin at a dose of 80 mg daily for days 1–21, then 40 mg daily for days 22–90); b. intensive antiplatelet therapy plus delayed intensive statin therapy (atorvastatin placebo for days 1–3, followed by atorvastatin at a dose of 40 mg daily for days 4–90); c. standard antiplatelet therapy (clopidogrel placebo combined with aspirin at a dose of 100–300mg on day 1, then 100 mg daily for days 2–90) plus immediate intensive statin therapy; and d. standard antiplatelet therapy plus delayed intensive statin therapy. The study employed double-blind design. Both participants and researchers were blind to the randomized assignment. The outcome adjudication committee was also blinded to the group assessment.

Efficacy and safety outcomes

All study patients were followed for 90 ± 7 days during which the development of all symptoms relatable to study outcomes and the time of occurrence were recorded. The adjudication committee composed of experienced neurologists, reviewed and validated all outcomes. The patients’ overall health condition and any changes in their clinical status were also recorded.

The efficacy and safety outcomes in this analysis were identical to those of the INSPIRES trial.²⁸ The primary efficacy outcome was the development of secondary stroke (ischemic or hemorrhagic) within 90 days.²⁹ Ischemic stroke was defined as an acute focal infarction of the brain or retina with one of the following: sudden onset of a new focal neurological deficit lasting <24 h with clinical or imaging evidence of infarction, or rapid worsening of an existing focal neurological deficit lasting ≥24 h with imaging evidence of new ischemic changes clearly distinct from the index ischemic event. Hemorrhagic stroke was defined as the acute extravasation of blood into the brain parenchyma or subarachnoid space with associated neurological symptoms. The secondary efficacy outcomes included a composite of cardiovascular events (ischemic and hemorrhagic stroke, myocardial infarction, or death from cardiovascular causes), and each individual cardiovascular event occurred within 90 days.²⁹

The primary safety outcome was the total number of moderate or severe bleeding events (based on the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries [GUSTO] definition³⁰) at 90 days. Severe (or life-threatening) bleeding was defined as intracranial hemorrhage, or other substantial/fatal hemorrhage causing hemodynamic compromise requiring treatment. Moderate bleeding was defined by the need of blood transfusion. Secondary safety outcomes included death from any cause, mild bleeding, any bleeding, or intracranial hemorrhage.^30,31

Procedures of the watershed infarction substudy

All patients completed MRI examinations (3.0T or 1.5T) in INSPIRES²⁸ and diffusion-weighted imaging (DWI) sequences were reanalyzed for WI, including the infarct location and classification.

DWI data were collected from individual centers in digital format and reviewed centrally by two senior neurologists (ZXY. L and CH. L), who were blinded to the baseline information, treatment assignment and study outcomes. WI were identified as hyperintense lesions on DWI images in border zone location.¹ All patients with WI were classified into three groups: CWI, IWI, and CWI+IWI. CWI was identified if in the frontal cortex (ACA/MCA) infarctions extended from the anterior horn to the cortex; and in the parieto-occipital region (MCA/PCA) the infarctions extended from the posterior horn to the cortex^11,22 (graphic abstract). IWI was identified if there were more than three lesions with diameters >3 mm in a linear fashion parallel to the lateral ventricles in the centrum semiovale or corona radiata, which may be chain-like (“string of pearls”) or confluent (“cigar shape”)^11,22,32 (graphic abstract). Patients with combined CWI and IWI were grouped as CWI + IWI. Patients without WI were in non-WI group. Disagreements were resolved by a third reviewer (Y. G).

Quantification and statistical analysis

All analyses were performed following the intention-to-treat principle. Demographic and clinical characteristics were described as categorical or continuous variables. The chi-square test or Fisher’s exact test was used for categorical variables, and Kruskal-Wallis test or Mann-Whitney U test for continuous variables. Multivariable cox-proportional hazards regression and general linear modeling were used to 1) assess the rate of recurrent stroke, and 2) estimate the relative risk of efficacy and safety outcomes between two treatment groups, in patients with and without WI, and in different WI patterns. Interaction terms were added to test the potential effect modification by WI status (WI or WI-free) and WI patterns. Kaplan-Meier curves and log rank tests were also presented to compare the cumulative risks of the primary outcome (recurrent stroke) between two treatment groups in patients with different WI patterns at 90 days. Confounding factors were selected based on univariate analysis and previous literature including age, gender, BMI, current or previous smoking, NIHSS score, medical history of hypertension, stroke, hyperlipidemia, diabetes and cardiovascular disease, and whether receiving early statin treatment.

All tests were 2-sided, and p < 0.05 was considered significant. All statistical analyses were conducted using SAS statistical software, version 9.4 (SAS Institute, Cary, North Carolina, USA).

Additional resources

The trial was registered at clinicaltrials.gov (Unique identifier: NCT03635749).

Published: February 5, 2026