Research

Clinical–MRI correlations in a multiethnic cohort with recent lacunar stroke: the

SPS3 trial
Oscar R. Benavente1,a*, Lesly A. Pearce2,a, Carlos Bazan3,a, Ana M. Roldan4,a,
Luciana Catanese5,a, Viveca M. Bhat Livezey6,a, Gabriela Vidal-Pergola7,a, Leslie A. McClure8,b,
and Robert G. Hart9,a for the SPS3 Investigators
Background Neuroimaging manifestations of small vessel hyperintensities were present in 50% and associated with
disease are heterogeneous, and correlation with patient fea- increased age (OR 4·3 per 20 years, CI 3·4, 5·4), hypertension
tures has not been adequately characterized. (OR 1·8, CI 1·4, 2·3), increased systolic blood pressure (OR 1·3
Aim Our goal was to correlate magnetic resonance imaging per 20 mmHg, CI 1·1, 1·5), increased diastolic blood pressure
findings with clinical features in a large multiethnic cohort (OR 1·2 per 10 mm, CI 1·0, 1·3), and prior stroke (OR 3·3, CI 2·3,
with recent lacunar stroke. 4·5). Infarct location varied significantly by race-ethnicity
Methods Patient characteristics were correlated with neuro- (P < 0·001), with Blacks and Hispanics having more infarcts in
imaging results in the Secondary Prevention of Small Subcor- the brainstem/cerebellum than non-Hispanic Whites, and by
tical Stroke study participants. gender with women more often having thalamic lacunes than
Results Among 3005 patients, mean age was 63 years; 62% men (P ≤ 0·001).
were men; and 51%, 30%, and 16% were non-Hispanic White, Conclusions In patients with recent lacunar stroke, infarct
Hispanic, and Black, respectively. Recent lacunar infarcts were location and number have distinctie associations with gender,
distributed between the subcortical hemisphere (31%), thala- vascular risk factors, and race-ethnicity, demonstrating the
mus (26%), brainstem/cerebellum (26%), and basal ganglia/ complex pathogenesis of lacunar stroke and cerebral small
internal capsule (16%). Multiple lacunar infarcts (i.e., acute and artery disease.
remote) were present in 40% and associated with increased Key words: lacunar stroke, MRI, African American, Hispanics, small vessel
age (OR 1·3 per 20 years, 95% CI 1·1, 1·5), male gender (OR 1·5, disease, white matter hyperintensity
CI 1·3, 1·7), hypertension (OR 1·5, CI 1·2, 1·8), increased systolic
blood pressure (OR 1·2 per 20 mmHg, CI 1·1, 1·3), and prior
Introduction
stroke (OR 3·8, CI 2·9, 5·0). Moderate-severe white matter

Correspondence: Oscar R. Benavente*, Department of Medicine,

Cerebral small vessel disease (SVD) is highly prevalent, represent-
Division of Neurology, Brain Research Center, University of British ing more than a third of the total burden of symptomatic cere-
Columbia, S169-2211 Wesbrook Mall, Vancouver, BC V6T 2B5, Canada. brovascular disease and underlying most covert strokes (1–4).
E-mail: Oscar.benavente@ubc.ca Neuroimaging manifestations of SVD are heterogeneous and
1
Department of Medicine, Division of Neurology, Brain Research Center, include lacunar infarcts, intracerebral hemorrhages, white matter
University of British Columbia, Vancouver, Canada
2
Biostatistics Consultant, Minot, ND, USA
hyperintensities (WMH), enlarged Virchow–Robin spaces, and
3
Department of Radiology, University of Texas Health Sciences Center at cerebral microbleeds (5–8).
San Antonio, San Antonio, TX, USA Small subcortical strokes, commonly known as lacunar strokes,
4
Department of Neurology, University of Texas Health Science Center at are the most common clinical manifestation of SVD (8,9). WMHs
Houston, Houston, TX, USA appear as areas of hyperintense signal on FLAIR and T2-weighted
5
Department of Neurology, Boston University Medical Center (BUMC).
Boston, USA
magnetic resonance imaging (MRI) sequences and are common
6
South Puget Sound Neurology, Tacoma, WA, USA findings in patients with cerebrovascular disease, particularly in
7
Department of Neurology, University of Texas Health Sciences Center at those with lacunar strokes (7).
San Antonio, San Antonio, TX, USA The Secondary Prevention of Small Subcortical Strokes (SPS3)
8
Department of Biostatistics, University of Alabama at Birmingham, Bir- trial was an international trial that included multiethnic patients
mingham, AL, USA
9
Department of Medicine (Neurology), McMaster University, Hamilton,
with recent symptomatic lacunar infarcts confirmed by MRI.
ON, Canada Here, we describe the spectrum of MRI findings in this popula-
tion and explore differences in the location and multiplicity of
Received: 24 September 2013; Accepted: 3 March 2014; Published online
27 May 2014
lacunar infarcts and WMHs by risk factors, intracranial arterial
stenosis, ethnicity, and geographic region.
Funding: SPS3 was supported by a cooperative agreement
(U01NS038529) from the US National Institute of Health-National Insti-
tute of Neurological Disorders and Stroke (NIH-NINDS).
Methods
a
SPS3 Coordinating Center.
The rationale, design, and participant characteristics of SPS3 have
b
SPS3 Statistical Center. been published elsewhere (10–12). Briefly, SPS3 was a random-
‘Clinical Trial Registration’ www.clinicaltrials.gov NCT00059306.
ized, multicenter clinical trial conducted in 81 clinical centers in
North America, Latin America, and Spain. Patients with recent
Conflicts of interest: The authors declare that they have no conflicts of (within 180 days) lacunar stroke without surgically amenable
interest.
ipsilateral carotid artery disease or major-risk cardioembolic
DOI: 10.1111/ijs.12282 sources were randomized, in a 2-by-2 factorial design, to both an

© 2014 World Stroke Organization Vol 9, December 2014, 1057–1064 1057

 Research O. R. Benavente et al.

antiplatelet intervention and one of two target levels of systolic The dimensions of the lesions were measured using an elec-
blood pressure (BP). Participants with a lacunar stroke syndrome tronic cursor or a caliper when films were used. The maximal
were required to meet MRI criteria to be eligible: no evidence of right-to-left or anterior-to-posterior dimensions of each lesion
recent or remote cortical infarct, old large (>15 mm) subcortical were recorded. The superior-to-inferior dimension was computed
infarct, or prior intracerebral hemorrhage; the presence of by multiplying the number of axial sections (minus 1⁄2 section) on
microbleeds was not exclusion. The MRI had to demonstrate at which the lesion appeared by the sum of the slice and skip thick-
least one of the following four specific criteria: (1) acute infarct on nesses. Maximum dimension of the lesion was then the maximum
diffusion-weighted imaging (DWI) with lesion ≤20 mm in size at of the right-to-left, anterior-to-posterior, and superior-to-
largest dimension (including rostro-caudal extent) corresponding inferior dimensions.
to the clinical syndrome; (2) acute or subacute infarct well- Intrarater reliability of the above findings/measurements was
delineated focal hyperintensity ≤20 mm in size at largest dimen- assessed (C. B.) by having the neuroradiologist re-evaluate a set of
sion (including rostro-caudal extent) on FLAIR or T2 and clearly 75 MRIs at the end of the study, unaware of his earlier interpre-
corresponding to the clinical syndrome; (3) multiple hypointense tations. Kappa (κ) intraclass correlation coefficients and percent
lesions (consistent with infarcts) of size 3–15 mm at largest agreement were computed for categorical variables, and 95% con-
dimension (including rostro-caudal extent) only in the cerebral fidence intervals (CI) for the slope and intercept were computed
hemispheres on FLAIR or T1 in patients whose qualifying event is using Passing–Bablok regression for continuous variables. Find-
clinically hemispheric (if qualifying event was clinically brainstem ings for presence/absence of acute DWI lesion (κ = 0·79, 94%
or cerebellar, this criterion alone was not sufficient for study agreement), dimension of lesion (95% CI for intercept −0·1, 0·09;
entry); (4) well-delineated hypointense lesion (consistent with 95% CI for slope 0·91, 1·08), presence/absence of hyperintensity
infarct) ≤15 mm in size at largest dimension (including rostro- on Flair/T2 compatible with acute infarct (κ = 0·78, 89% agree-
caudal extent) on FLAIR or T1 corresponding to the clinical syn- ment), presence/absence of lacunes on Flair/T1 (κ = 0·64, 88%
drome. The four specific criteria are in hierarchical order, i.e. the agreement), new or old cortical stroke (κ = 0·79, 97% agreement),
lowest numbered criterion present is reported. Lacunar ‘TIAs’ or presence/absence of intracerbral hemorrhage (κ = 0·82, 97%
required positive (bright) MRI DWI lesion to be eligible. agreement) showed little within-rater variability.
Patients underwent MRI before study entry for clinical man- Vascular imaging of the intracranial arteries proximate to study
agement: MRIs were not obtained with a prespecified data acqui- entry was required, and de-identified images [magnetic resonance
sition protocol. De-identified neuroimaging data were sent to the angiogram (MRA), computed tomographic angiography (CTA),
SPS3 Coordinating Center, and each case was interpreted by an and/or cerebral angiography] were to be submitted to the SPS3
experienced neuroradiologist (C. B.) unaware of patient charac- Coordinating Center. Stenosis of the intracranial arteries on
teristics. Old subcortical infarcts were defined as hypointense MRA, CTA, or angiogram was graded by the neuroradiologist (C.
areas on FLAIR and/or T1 measuring ≥3 mm, but no more than B.) using the NASCET criteria (15). When a participant had more
15 mm in maximum dimension. Hypointense lesions at the level than one vascular image, the measurement of stenosis was made
of the anterior commissure, convexity, or midbrain, were consid- on the modality offering the clearest visualization of stenosis.
ered enlarged peri-vascular spaces (EPVS) and not classified as WMHs were evaluated visually on FLAIR images using the age-
infarcts, unless the lesion was surrounded by a hyperintense halo related white matter change (AWRMC) (range 0–16) (16) by four
on FLAIR. Large subcortical infarcts were hypointense/ readers unaware of clinical information (L. C., O. R. B., V. M. B. L.,
hyperintense lesions on FLAIR and/or T2 measuring >15 mm. A. M. R.). A priori, scores of 0–4 on the ARWMC scale were
For anatomic localization, lesions were assigned to one or more of defined as none-mild disease, 5–8 moderate, and 9+ severe. Inter-
the following anatomical regions defined by gross anatomic and rater agreement was good-excellent on a sample of 40 MRIs
vascular characteristics: (1) basal ganglia (caudate, pallidum, (range: κ = 0·64, 77% agreement to κ = 0·89, 95% agreement).
putamen), (2) internal capsule, (3) subcortical white matter Patient demographic, clinical characteristics, and MRI findings
(centrum semiovale and corona radiate), (4) thalamus, (5) brain- were compared between groups using a chi-square test, Student’s
stem (midbrain, pons, medulla), and (6) cerebellum (13). Lesions t-test, or analysis of variance as appropriate. Multinomial and
in basal ganglia were not differentiated into individual nuclei. binomial logistic regressions were used to identify patient features
A cortical infarct was considered acute when an abnormal independently associated with lesion location and multiple vs.
bright signal on DWI (and dark on ADC if available), FLAIR, single lesions, respectively. All statistical tests were two sided, and
proton density or T2 was present on cerebral or cerebellar cortex significance was accepted at the 0·05 level. All analyses were done
in determined vascular territory (14). Subacute and chronic cor- using spss Statistics for Windows, Version 20 (Armonk, NY,
tical infarcts were hypointense/hyperintense dark lesions on USA).
FLAIR or proton density measuring >10 mm in maximal
dimension. Results
Old intracerebral hemorrhages were defined as hypointense
lesions on T2 gradient echo or hypointense on T2 (excluding Of 3020 participants, 3015 underwent an MRI after the qualifying
calcified lesions) measuring >10 mm in maximal dimension and stroke, and MR images for 3005 patients were available for central
localized in any part of the brain. Most of these were asymptom- review. The median time from qualifying lacunar stroke to MRI
atic ‘macrobleeds’ and did not meet the criteria of microbleeds. was 3 days (75th percentile 17 days). Most were submitted as

1058 Vol 9, December 2014, 1057–1064 © 2014 World Stroke Organization

 O. R. Benavente et al. Research
digital images (87%), with the remainder as films. Quality of the Fewer than 1% of the MRIs had nonacute large subcortical
images was rated as good in >99%. infarcts (>15 mm), 3·8% had cortical infarcts, and 1·6% had
asymptomatic intracerebral macrobleeds >10 mm. Patients with
Brain infarcts cortical infarcts (n = 114) and those with asymptomatic intrac-
The qualifying lacunar stroke was visualized by MRI for 99% of erebral bleeds (n = 47) more often had prior symptomatic lacunes
patients utilizing the different imaging criteria (75% by positive and more WMH (Table S1). Patients with cortical infarcts were
DWI) (Fig. S1, Table 1). The median size of maximum dimension also more often older, male, and diabetic. Multiple lacunar
on both DWI and FLAIR/T2 lesions was 12 mm, with 8% of MRIs infarcts were seen on 40% of MRIs, i.e. at least one small subcor-
having lesions estimated to be >20 mm on maximum dimension. tical infarct in addition to the qualifying stroke. WMHs were
The qualifying lacunar infarct was located in the subcortical white none-mild for 50% of patients by the AWRMC score, with 28%
matter 31% (centrum semiovale or corona radiata), thalamus moderate and 22% severe.
26%, brainstem 26%, internal capsule 13%, basal ganglia 4%, and Location of acute lacunar infarct did not vary with age
cerebellum 0·3% (Fig. 1). (Table 2), whereas gender, history of hypertension, diabetes, prior
symptomatic lacunar stroke, and relevant intracranial stenosis
were independently associated with location of the acute index
Table 1 Imaging findings among SPS3 trial participants* event. (Table 3) Males and patients with a history of hypertension
or diabetes were significantly less likely to have an anterior circu-
With
imaging
lation or thalamic lesion vs. posterior lesion, whereas patients with
available† % relevant IC stenosis were more likely to have an anterior circula-
tion or thalamic lesion. Patients with prior symptomatic lacunes
Qualifying infarcts (hierarchial) 3005
more often had posterior circulation lesions vs. thalamic location.
Subcortical infarct on DWI corresponding to 75
symptoms Average maximum dimension of DWI lesions differed by location,
Subcortical infarct (hyperintense lesion) on 23 with symptomatic thalamic lesions being the smallest and basal
FLAIR/T2 corresponding to symptoms when ganglia and subcortical hemispheric lesions the largest. Patients
DWI was not available or delayed MRI with subcortical hemispheric lesions more frequently had WMH.
Multiple hypointense infarcts in cerebral <0·5
Patients with multiple infarcts on MRI were slightly older,
hemispheres on FLAIR/T1 in patients whose
qualifying event is clinically hemispheric on more often male, and more often hypertensive compared with
delayed MRI those with a single infarct (i.e. the qualifying event) on MRI
Subcortical hypointense infarct on FLAIR/T1 <0·5 (Table 2). Of those with multiple infarcts on MRI, 18% had symp-
corresponding to symptoms when DWI was tomatic lacunar stroke prior to the qualifying stroke compared
unavailable or MRI was delayed
with 5% of those with only a single infarct on MRI. Participants
None of the above† 1
Other infarcts and hemorrhages on MRI with multiple infarcts on MRI also more often had moderate or
Small subcortical infarcts on FLAIR/T1 2980 39 severe WMH. Of the demographic/clinical characteristics in
(hypointense) in addition to qualifying stroke Table 2, each of increased age (OR 1·3 per 20 years, 95% CI 1·1,
Large subcortical infarcts on FLAIR/T1 (>15 mm 2980 0·8 1·5), male gender (OR 1·5, CI 1·3, 1·7), history of hypertension
diameter)†
(OR 1·5, CI 1·2, 1·8), no diabetes (OR 1·3, CI 1·1, 1·6), increased
Hemorrhage > 10 mm‡ (macrohemorrhage) 2957 1·6
Cortical infarct (>10 mm diameter)† 3004 3·8 systolic BP (OR 1·2 per 20 mmHg, 1·1, 1·3) and prior symptom-
White matter lesions, ARWMC score 2970 atic lacunar stroke (OR 3·8, CI 2·9, 5·0) was independently asso-
0–4 50 ciated with multiple vs. single infarcts on MRI. These results were
5–8 28 essentially unchanged when prior symptomatic lacune was
>9 22
excluded from the model.
Intracranial vascular imaging
Stenosis >50% 2843
Carotid artery 2·1 WMH
Middle cerebral artery (MCA) or anterior 7·7 Increasing age was associated with increasing severity of WMH
cerebral artery (ACA) (Table 4). A history of hypertension was most frequent among
Any of carotid, MCA, ACA 9·5
those with severe white matter disease (84%), and systolic BP
Vertebral artery 1·2
Posterior cerebral artery (PCA) 8·4 averaged 148 mmHg for those with severe WMH vs. 145 for mod-
Basilar artery 1·9 erate and 140 for none-mild. Of the demographic/clinical char-
Any of vertebral, PCA, basilar 11 acteristics in Table 2, each of increased age (OR 4·3 per 20 years,
*Based on central neuroradiologist interpretation.
95% CI 3·4, 5·4), history of hypertension (OR 1·8, CI 1·4, 2·3), no
†
These represent a protocol deviation based on central interpretation diabetes (OR 1·3, CI 1·1, 1·7), increased systolic BP (OR 1·3 per
of MRI. 20 mmHg, 1·1, 1·5), increased diastolic BP (OR 1·2 per 10 mm, CI
‡
These were not symptomatic large intracerebral hemorrhages, but 1·0, 1·3), and prior symptomatic lacunar stroke (OR 3·3, CI 2·3,
did not meet criteria for microbleeds.
4·5) was independently associated with severe vs. none-mild
ARWMC, age-related white matter change; MRI, magnetic resonance
imaging; SPS3, Secondary Prevention of Small Subcortical Stroke.
WMH. Results were essentially unchanged when history of
lacunar stroke was excluded from the model.

© 2014 World Stroke Organization Vol 9, December 2014, 1057–1064 1059

 Research O. R. Benavente et al.

Fig. 1 Localization of qualifying lacunar infarcts.

Intracranial arterial stenosis Recent lacunar infarcts were localized to areas of the brain
Of 3020 participants, 95% (n = 2866) had intracranial vascular supplied by the small deep penetrating arteries. About one-third
imaging, and 2843 (94%) were deemed of acceptable quality for were in the centrum semiovale/corona radiata, more than half in
interpretation. Of these 2843, MRAs were performed in 2737 the thalamus and brainstem/cerebellum, and fewer than 20% in
patients, CTAs in 106, and conventional cerebral angiograms in the internal capsule and basal ganglia. Our findings in this cohort
13, with 12 participants having multiple types of imaging. Intrac- concur with those found in other series of lacunar strokes diag-
ranial stenosis of ≥50% was found in 17% of all patients and 7% nosed with MRI, except for the low frequency of lacunes in the
when restricted to an artery supplying the vascular territory of the internal capsule and basal ganglia (17,18). At present we are
qualifying stroke. Stenosis was demonstrated on 19% of MRAs, unable to provide a suitable explanation to account for this dif-
8% of CTAs, and 23% of conventional angiograms. ference. While the strict inclusion criteria of the clinical trial may
Black and Hispanic participants each had more brainstem/ have introduced selection bias in our participants, the risk factors
cerebellum index events compared with non-Hispanic White par- and demographic characteristics of our cohort are similar to
ticipants (Table 5). Race/ethnicity was an independent factor for those reported in other lacunar stroke series (19–23), though
location of acute lacunar infarct when added to the model in more ethnically racially diverse than others. Therefore, it is
Table 3: compared with non-Hispanic Whites, Hispanics (OR improbable that the difference in localization can be explained by
0·79, CI 0·64, 0·97) and Blacks (OR 0·62, CI 0·48, 0·81) are less a distinct pathophysiological mechanism. Thalamic lesions were
likely to have an anterior vs. posterior circulation lesion and also considered separate from those in posterior circulation due to
less likely to have a thalamic vs. posterior circulation lesions (His- dual vascular supply of the thalamus.
panics OR 0·59, CI 0·46, 0·76; Blacks vs. non-Hispanic Whites OR Qualifying lacunar infarcts were more likely to be localized to
0·74, CI 0·56, 0·99) (Table 6). the anterior or posterior circulation vs. thalamus if the patient
was male, or with a history of hypertension or prior stroke. On
Discussion the contrary, diabetic patients tended to have the index stroke in
the posterior circulation and thalamic lacunes were significantly
In this large, well-characterized multiethnic cohort of patients associated with relevant intracranial stenosis. Most thalamic
with recent symptomatic lacunar strokes, we quantitate the broad infarcts are due to disease of small penetrating arteries and share
spectrum of MRI findings consistent with cerebral SVD. Lacunar similar risk factors with lacunes in other localizations. Nonethe-
infarcts were verified by MRI, and other plausible mechanisms for less, emboli either artery-to-artery or cardiac have also been
ischemic stroke other than SVD were largely excluded. Conse- implicated as a responsible mechanism for thalamic strokes
quently, it is likely that our cohort represents a relatively pure (24,25). The presence of relevant intracranial stenosis as found
clinical sample of patients with cerebral SVD. in our cohort could be simply coincidence though, so we cannot

1060 Vol 9, December 2014, 1057–1064 © 2014 World Stroke Organization

 Table 2 Patient features associated with location and number of lacunar infarcts*
O. R. Benavente et al.

Location of qualifying infarct Number of infarcts on MRI†

© 2014 World Stroke Organization

Basal ganglia Internal capsule Subcortical hemisphere‡ Thalamus Posterior circulation P value§ Single Multiple P value

n 125 368 924 776 770 1771 1167

Age, years, mean (SD) 62 (11) 63 (10) 63 (11) 62 (11) 63 (11) 0·1 62 (11) 64 (11) <0·001
Men, % 62 62 62 59 69 <0·001 60 67 <0·001
History of hypertension, % 78 71 75 71 80 <0·001 71 80 <0·001
BP at entry, mmHg, mean (SD) 143 (18)/78 (9) 142 (19)/78 (11) 142 (19)/78 (11) 143 (19)/78 (11) 144 (19)/79 (10) 0·2/0·7 142 (18)/78 (10) 145 (20)/79 (11) <0·001/<0·001
Diabetes, % 29 36 30 37 46 <0·001 39 34 0·01
Ischemic heart disease, % 13 8 10 12 11 0·2 10 12 0·07
History of lacunar stroke, % 14 10 11 7 11 0·06 5 18 <0·001
Current tobacco smoker, % 20 26 19 21 19 0·03 20 21 0·2
Maximum diameter of 16 (7) 13 (5) 16 (6) 11 (4) 12 (4) <0·001 13 (5) 13 (5) 0·4
DWI-positive lesion, mm,
mean (SD)
Proximal intracranial artery 3 8 6 11 4 <0·001 7 8 0·3
stenosis >50%¶
ARWMC score
mean (SD) 4·7 (4) 5·6 (4) 6·3 (4) 5·3 (4) 4·7 (4) <0·001 4·4 (4) 7·1 (4) <0·001
ARWMC group, % <0·001 <0·001
0–4 62 45 42 53 57 61 33
5–8 23 35 31 25 25 25 34
>9 16 21 27 22 17 14 33

*Participants without a lacunae meeting MRI criteria (n = 42) are excluded.

†
At least one small subcortical infarct on MRI in addition to the qualifying lacunar infarct; 25 participants without FLAIR/T1 sequences are excluded.
‡
834 in corona radiate and 90 in centrum semiovale.
§
644 pons, 55 midbrain, 61 medulla, and 10 cerebellum.
¶
For basal ganglia, capsular and hemispheric lacunes, stenosis ipsilateral to the qualifying lacune in intracranial carotid, middle cerebral, or anterior cerebral arteries; for posterior circulation lacunes, any
stenosis in the vertebral or basilar arteries; for thalamic lacunes, any stenosis in the vertebral, basilar, or posterior cerebral arteries.
ARWMC, age-related white matter change; BP, blood pressure; DWI, diffusion-weighted imaging; MRI, magnetic resonance imaging.

Vol 9, December 2014, 1057–1064

Research

1061
 Research O. R. Benavente et al.

exclude SVD as primarily responsible for most of the thalamic

Table 3 Independent demographic/clinical variables associated with
lacunes.
acute lacunar infarct location*
We found that race-ethnicity was an independent factor for
OR (95% CI) OR (95% CI) location of the index event. Hispanics were more likely than non-
anterior vs. thalamic vs. Hispanic Whites to have an anterior or posterior circulation
posterior lesion posterior lesion
lesion vs. thalamic, and Blacks vs. non-Hispanic Whites more
Male 0·71 (0·58, 0·86) 0·60 (0·48, 0·74) likely to have a posterior vs. thalamic lesion. Several studies
History of hypertension 0·76 (0·61, 0·95) 0·59 (0·46, 0·75) reported differences in stroke subtype, vascular risk factors,
Diabetes 0·56 (0·47, 0·68) 0·73 (0·59, 0·90) mechanisms, and outcome in Blacks and Hispanics. While the
History of lacunar stroke 1·0 (0·78, 1·4) 0·66 (0·46, 0·95)
Relevant IC stenosis 1·5 (1·0, 2·3) 2·9 (1·9, 4·5)
cause is not fully understood, it is likely multifactorial, and race-
ethnicity by itself, lifestyle, and environmental factors may play a
*Participants without a lacune meeting MRI criteria (n = 42) are role in the differences in burden of disease (26–33). We speculate
excluded.
that similar factors may have contributed to the different distri-
IC, intracranial; MRI, magnetic resonance imaging.
bution of lacunes in Blacks, Hispanic, and non-Hispanic Whites
in SPS3.

Table 4 Patient features according to age-related white matter change (ARWMC) score for WMHs

White matter hyperintensities (ARWMC Score)

None-mild (0–4) Moderate (5–8) Severe (9+) P value (for trend)

n 1490 834 646

Age, years, mean (SD) 60 (10) 65 (10) 67 (11) <0·001
Male, % 64 64 59 0·03
History of hypertension, % 69 78 84 <0·001
Screening BP, mmHg, mean (SD) 140 (17)/78 (10) 145 (20)/78 (11) 148 (20)/79 (11) (11) <0·001/0·007
Diabetes, % 38 38 32 0·01
Ischemic heart disease, % 10 12 11 0·2
History of lacunar stroke, % 6 13 16 <0·001
Current tobacco smoker, % 21 22 18 0·3

ARWMC, age-related white matter change; BP, blood pressure.

Table 5 Infarct location, white matter abnormalities, and intracranial stenosis according to race-ethnicity and region

Race-ethnicity Region

MRI findings (n = 3005) White Hispanic Black Other P value NA LA Spain P value

n 1531 911 489 74 1953 689 363

Number of infarcts*†, % 0·05 <0·001
Single 62 57 60 63 64 53 56
Multiple 38 43 40 37 36 47 44
Location of acute lacunar infarct*, % <0·001 <0·001
Basal ganglia 4 5 4 3 4 5 3
Internal capsule 13 10 14 15 13 10 14
Subcortical hemisphere 32 34 23 30 29 37 33
Thalamus 29 21 29 23 29 19 26
Brainstem/cerebellum 22 30 30 30 25 30 23
WMHs ARWMC score,% 0·02 0·4
0–4 52 51 43 51 51 48 50
5–8 28 27 31 27 28 28 31
9+ 20 22 26 22 21 24 20
Proximal intracranial stenosis‡ 6 7 8 12 0·3 7 5 7 0·1

ARWMC, age-related white matter change; LA, Latin America; NA, North America; MRI, magnetic resonance imaging; WMH, white matter
hyperintensities.
*Excludes those with no acute lacunar infarct on MRI (n = 42).
†
At least one small subcortical infarct on MRI in addition to the qualifying lacune; 25 participants without FLAIR/T1 sequences are excluded.
‡
For basal ganglia, capsular and hemispheric lacunes, stenoses ipsilateral to the qualifying stroke in intracranial carotid, middle cerebral, or anterior
cerebral arteries; for posterior lacunes, any stenosis in the vertebral or basilar arteries; for thalamic lacunes, any stenosis in the vertebral, basilar, or
posterior cerebral arteries.

1062 Vol 9, December 2014, 1057–1064 © 2014 World Stroke Organization

 O. R. Benavente et al. Research
Table 6 Location of acute infarct by independently associated clinical characteristics

History of Proximal intracranial

Female gender Diabetes Hypertension lacunar stroke Hispanic Black stenosis

Anterior circulation + * + + + * +
Thalamic ++ + * * * * ++
Posterior circulation * ++ ++ + ++ + *

The ‘+’ indicates a significantly higher odds ratio than the ‘*’ group. A ‘++’ indicates a significantly higher odds ratio than the ‘+’ group. There is no
significant difference between groups with the same symbol.

Traditionally, lacunar infarcts were described in pathological the high inclusion of a significant proportion of patients with
studies as a cavitation, with a maximum dimension of 15 mm. It different race-ethnicities.
was postulated that larger lacunes (>15 mm) would have a differ- Intracranial stenosis was infrequent in our cohort, present in
ent mechanism than SVD, i.e. microatheroma (9,34). In our 17% of all patients and 7% when confined to the arteries supply-
series, the mean size of the recent lacunes on DWI or FLAIR was ing the vascular territory of the index event. This finding is con-
<15 mm in maximum dimension. This finding coincides with sistent with that found in other series of lacunar stroke patients
prior data and supports the hypothesis that the likely mechanism (40). It is possible, however, that the low prevalence could be
of infarction is damage of the small penetrating arteries, so-called explained by the use of clinical MRA and not high-resolution
SVD. The ARIC study attempted to sub-classify according to the scanners (41). However, currently the use of high-resolution scan-
mechanism of lacunar strokes based on vascular risk factors and ners is mainly limited to research, therefore most of the clinical
size of the MRI lesion. The study found that lesions of ≤7 mm management decisions are based on the use of traditional vascular
were more likely to be due to lipohyalinosis than those between 8 images.
and 20 mm which could probably be attributed to micro- Several limitations of this study warrant comment. First, the
atheroma. However, infarcts were vaguely defined based only on cohort is that of a large randomized clinical trial; therefore, it is
the presence of asymptomatic hyperintensities on FLAIR possible that selection biases could have been introduced. Second,
sequences (35). Therefore, it is possible that many of these lesions the MRIs were clinical studies performed on multiple scanners
do not represent true lacunar infarcts, and despite the fact that the without a prespecified acquisition protocol. It is possible that this
mean size in SPS3 was >7 mm, it is likely that the mechanism is methodological caveat precluded us from having a more accurate
due to disease of the small penetrating arteries. estimate of the findings. However, one can argue that having the
Multiple brain infarcts were present in 40% of participants and data from clinical scanners would help to generalize our results to
∼80% were covert or asymptomatic, also called silent brain clinical practice. Third, all the readings were done by a single
infarcts (SBIs). As in previous studies (36), we found the tradi- neuroradiologist which may have introduced potential biases.
tional vascular risk factors of prior stroke and age independently Finally, these were not prespecified, hypothesis-driven analyses.
associated with multiple. SBIs are often present on MRIs, with a One of the strengths of this study was the large sample size of
prevalence ranging from 10% up to 60% (8,37). This wide range patients with recent lacunar infarct verified by MRI, the largest to
of SBI depends on the vascular risk factors of the population our knowledge. Significant proportions were from different
studied and the adopted definition on MRI. A recent systematic ethnic groups and geographic regions, which likely improves gen-
review of MRI criteria used to diagnose SBI showed a great incon- eralizability of these results.
sistence in definition (i.e. hyperintensity or hypointensity) and In this large, multiethnic cohort, the previously unrecognized
only a few studies excluded EPVS (38). In our study, old infarcts correlations of infarct size and location with vascular risk factors
were defined as hypointensities in FLAIR and/or T1 measuring and race-ethnicity demonstrate the complex pathogenesis of
>3 mm, and EPVS were not considered lacunar infarcts. The cerebral SVD. For example, thalamic lacunes, which have multiple
adopted definition is more conservative and precise than the ones mechanisms and are not a homogeneous entity due to the varying
based only on the presence of hyperintensities on FLAIR or T2. vascular blood supply of the thalamus, are more independently
Moderate to severe WMHs were present in 50% of patients and associated with female gender, whereas diabetes is significantly
were independently associated with age, vascular risk factors, more frequent with brainstem lacunes (Table 6). The implications
prior stroke, and multiple infarcts on MRI. In our study, the of these observations warrant further research and new con-
association between severity of WMH and multiple infarcts sup- structs to illuminate the pathogenesis of this common subtype of
ports the notion that both conditions are expressions of the same ischemic stroke.
vascular mechanism. This is reinforced by findings on
pathological-radiological series where the pathophysiology of References
WMH was attributed to SVD (39). However, our cohort has a 1 Inzitari D, Pracucci G, Poggesi A et al. Changes in white matter as
determinant of global functional decline in older independent outpa-
lower proportion of severe WMHs than expected in patients with tients: three year follow-up of LADIS (leukoaraiosis and disability)
established cerebrovascular disease. This observation may be the study cohort. BMJ 2009; 339:b2477. [Multicenter StudyResearch
result of the relatively younger age of the SPS3 participants and/or Support, Non-U.S. Gov’t].

© 2014 World Stroke Organization Vol 9, December 2014, 1057–1064 1063

 Research O. R. Benavente et al.

2 Norrving B. Lacunar infarcts: no black holes in the brain are benign. 26 Cantu-Brito C, Ruiz-Sandoval JL, Murillo-Bonilla LM et al. The first
Pract Neurol 2008; 8:222–8. [Review]. Mexican multicenter register on ischaemic stroke (the PREMIER
3 Pantoni L. Cerebral small vessel disease: from pathogenesis and clini- study): demographics, risk factors and outcome. Int J Stroke 2011;
cal characteristics to therapeutic challenges. Lancet Neurol 2010; 6:93–4. [Letter].
9:689–701. [Review]. 27 Del Brutto OH, Dong C, Rundek T, Elkind MS, Del Brutto VJ, Sacco
4 Sudlow CL, Warlow CP. Comparable studies of the incidence of stroke RL. Cardiovascular health status among caribbean hispanics living in
and its pathological types: results from an international collaboration. northern Manhattan and Ecuadorian natives/mestizos in rural coastal
International Stroke Incidence Collaboration. Stroke 1997; 28:491–9. Ecuador: a comparative study. J Community Health 2013; 38:634–41.
[Comparative StudyMulticenter StudyResearch Support, Non-U.S. 28 Del Brutto OH, Mosquera A, Sanchez X, Santos J, Noboa CA. Stroke
Gov’t]. subtypes among Hispanics living in Guayaquil, Ecuador. Results from
5 Cordonnier C, Al-Shahi Salman R, Wardlaw J. Spontaneous brain the Luis Vernaza Hospital Stroke Registry. Stroke 1993; 24:1833–6.
microbleeds: systematic review, subgroup analyses and standards for 29 Hajat C, Dundas R, Stewart JA et al. Cerebrovascular risk factors and
study design and reporting. Brain 2007; 130(Pt 8):1988–2003. stroke subtypes: differences between ethnic groups. Stroke 2001;
6 Doubal FN, MacLullich AM, Ferguson KJ, Dennis MS, Wardlaw JM. 32:37–42. [Research Support, Non-U.S. Gov’t].
Enlarged perivascular spaces on MRI are a feature of cerebral small 30 Lavados PM, Hennis AJ, Fernandes JG et al. Stroke epidemiology,
vessel disease. Stroke 2010; 41:450–4. prevention, and management strategies at a regional level: Latin
7 Rost NS, Rahman RM, Biffi A et al. White matter hyperintensity America and the Caribbean. Lancet Neurol 2007; 6:362–72. [Review].
volume is increased in small vessel stroke subtypes. Neurology 2010; 31 Lavados PM, Sacks C, Prina L et al. Incidence, case-fatality rate, and
75:1670–7. prognosis of ischaemic stroke subtypes in a predominantly Hispanic-
8 Vermeer SE, Longstreth WT Jr, Koudstaal PJ. Silent brain infarcts: a Mestizo population in Iquique, Chile (PISCIS project): a community-
systematic review. Lancet Neurol 2007; 6:611–9. based incidence study. Lancet Neurol 2007; 6:140–8. [Comparative
9 Fisher CM. Lacunes: small, deep cerebral infarcts. Neurology 1965; Study Research Support, Non-U.S. Gov’t].
15:774–84. 32 Lavados PM, Sacks C, Prina L et al. Incidence of lobar and non-lobar
10 Benavente OR, White CL, Pearce L et al. The Secondary Prevention of spontaneous intracerebral haemorrhage in a predominantly Hispanic-
Small Subcortical Strokes (SPS3) study. Int J Stroke 2011; 6:164–75. Mestizo population–the PISCIS stroke project: a community-based
11 White CL, McClure LA, Wallace PM et al. The correlates and course of prospective study in Iquique, Chile. Neuroepidemiology 2010; 34:214–
depression in patients with lacunar stroke: results from the Secondary 21. [Research Support, Non-U.S. Gov’t].
Prevention of Small Subcortical Strokes (SPS3) study. Cerebrovasc Dis 33 Saposnik G, Del Brutto OH. Stroke in South America: a systematic
2011; 32:354–60. review of incidence, prevalence, and stroke subtypes. Stroke 2003;
12 White CL, Szychowski JM, Roldan A et al. Clinical features and Racial/ 34:2103–7. [Review].
Ethnic differences among the 3020 participants in the Secondary Pre- 34 Fisher CM. Lacunar strokes and infarcts: a review. Neurology 1982;
vention of Small Subcortical Strokes (SPS3) trial. J Stroke Cerebrovasc 32:871–6.
Dis 2013; 22:764–74. 35 Bezerra DC, Sharrett AR, Matsushita K et al. Risk factors for lacune
13 Tamraz JC, Comair YG. Atlas of Regional Anatomy of the Brain Using subtypes in the Atherosclerosis Risk in Communities (ARIC) Study.
MRI: with functional correlations. Springer-Verlag Berlin Heidelberg Neurology 2012; 78:102–8.
New York. 2000. 36 Mast H, Thompson JL, Lin IF et al. Cigarette smoking as a determi-
14 Ghika JA, Bogousslavsky J, Regli F. Deep perforators from the carotid nant of high-grade carotid artery stenosis in Hispanic, black, and
system. Template of the vascular territories. Arch Neurol 1990; white patients with stroke or transient ischemic attack. Stroke 1998;
47:1097–100. 29:908–12. [Multicenter Study Research Support, U.S. Gov’t, P.H.S.].
15 Fox AJ. How to measure carotid stenosis. Radiology 1993; 186:316–8. 37 Kovacs KR, Czuriga D, Bereczki D, Bornstein NM, Csiba L. Silent
16 Wahlund LO, Barkhof F, Fazekas F et al. A new rating scale for age- brain infarction – a review of recent observations. Int J Stroke 2013;
related white matter changes applicable to MRI and CT. Stroke 2001; 8:334–47.
32:1318–22. 38 Zhu YC, Dufouil C, Tzourio C, Chabriat H. Silent brain infarcts: a
17 De Reuck J, De Groote L, Van Maele G. The classic lacunar syndromes: review of MRI diagnostic criteria. Stroke 2011; 42:1140–5.
clinical and neuroimaging correlates. Eur J Neurol 2008; 15:681–4. 39 Fazekas F, Kleinert R, Offenbacher H et al. Pathologic correlates of
18 Hommel M, Besson G, Le Bas JF et al. Prospective study of lacunar incidental MRI white matter signal hyperintensities. Neurology 1993;
infarction using magnetic resonance imaging. Stroke 1990; 21:546–54. 43:1683–9.
19 Marti-Vilalta JL, Arboix A. The barcelona stroke registry. Eur Neurol 40 Wardlaw JM, Doubal FN, Eadie E, Chappell F, Shuler K, Cvoro V.
1999; 41:135–42. Little association between intracranial arterial stenosis and lacunar
20 Salgado AV, Ferro JM, Gouveia-Oliveira A. Long-term prognosis of stroke. Cerebrovasc Dis 2011; 31:12–8.
first-ever lacunar strokes. A hospital-based study. Stroke 1996; 41 Okuchi S, Okada T, Ihara M et al. Visualization of lenticulostriate
27:661–6. [Research Support, Non-U.S. Gov’t]. arteries by Flow-Sensitive Black-Blood MR angiography on a 1.5T
21 Toni D, Fiorelli M, De Michele M et al. Clinical and prognostic corre- MRI system: a comparative study between subjects with and without
lates of stroke subtype misdiagnosis within 12 h from onset. Stroke stroke. AJNR Am J Neuroradiol 2013; 34:780–4.
1995; 26:1837–40. [Comparative Study Research Support, Non-U.S.
Gov’t].
22 Turin TC, Kita Y, Rumana N et al. Ischemic stroke subtypes in a Supporting information
Japanese population: takashima stroke registry, 1988–2004. Stroke
2010; 41:1871–6. [Research Support, Non-U.S. Gov’t]. Additional Supporting Information may be found in the online
23 Yip PK, Jeng JS, Lee TK et al. Subtypes of ischemic stroke. A hospital- version of this article at the publisher’s web-site:
based stroke registry in Taiwan (SCAN-IV). Stroke 1997; 28:2507–12.
24 Bogousslavsky J, Regli F, Uske A. Thalamic infarcts: clinical syn- Fig. S1 Study profile.
dromes, etiology, and prognosis. Neurology 1988; 38:837–48.
25 Castaigne P, Lhermitte F, Buge A, Escourolle R, Hauw JJ, Lyon-Caen
O. Paramedian thalamic and midbrain infarct: clinical and neu- Table S1 Patient features in relation to presence of cortical infarct
ropathological study. Ann Neurol 1981; 10:127–48. Case Reports]. and asymptomatic bleed.

1064 Vol 9, December 2014, 1057–1064 © 2014 World Stroke Organization