Silent Brain Infarction and Risk of Future Stroke

A Systematic Review and Meta-Analysis

Ajay Gupta, MD; Ashley E. Giambrone, PhD; Gino Gialdini, MD; Caitlin Finn, BS;
Diana Delgado, MLS; Jose Gutierrez, MD, MPH; Clinton Wright, MD, MS;
Alexa S. Beiser, PhD; Sudha Seshadri, MD; Ankur Pandya, PhD; Hooman Kamel, MD

Background and Purpose—Silent brain infarction (SBI) on magnetic resonance imaging has been proposed as a subclinical
risk marker for future symptomatic stroke. We performed a systematic review and meta-analysis to summarize the
association between magnetic resonance imaging–defined SBI and future stroke risk.
Methods—We searched the medical literature to identify cohort studies involving adults with SBI detected by magnetic
resonance imaging who were subsequently followed up for incident clinically defined stroke. Study data and quality
assessment were recorded in duplicate with disagreements in data extraction resolved by a third reader. Strength association
between magnetic resonance imaging–detected SBI and future symptomatic stroke was measured by an hazard ratio.
Results—The meta-analysis included 13 studies (14 764 subjects) with a mean follow-up ranging from 25.7 to 174 months. SBI
predicted the occurrence of stroke with a random effects crude relative risk of 2.94 (95% confidence interval, 2.24–3.86, P<0.001;
Q=39.65, P<0.001). In the 8 studies of 10 427 subjects providing hazard ratio adjusted for cardiovascular risk factors, SBI was
an independent predictor of incident stroke (hazard ratio, 2.08 [95% confidence interval, 1.69–2.56; P<0.001]; Q=8.99; P=0.25).
In a subgroup analysis pooling 9483 stroke-free individuals from large population-based studies, SBI was present in ≈18% of
participants and remained a strong predictor of future stroke (hazard ratio, 2.06 [95% confidence interval, 1.64–2.59]; P<0.01).
Conclusions—SBI is present in ≈1 in 5 stroke-free older adults and is associated with a 2-fold increased risk of future
stroke. Future studies of in-depth stroke risk evaluations and intensive prevention measures are warranted in patients
with clinically unrecognized radiologically evident brain infarctions.    (Stroke. 2016;47:719-725. DOI: 10.1161/
STROKEAHA.115.011889.)
Key Words: brain infarction ◼ infarction ◼ magnetic resonance imaging ◼ risk factors ◼ stroke

S troke is the second-leading cause of death worldwide and a

leading cause of disability.1 Identifying subclinical stroke
risk factors may allow for early and potentially more effective
stroke prevention measures. One such potential stroke risk
factor, silent brain infarction (SBI), is an increasingly detected
abnormality with modern magnetic resonance imaging (MRI)
techniques. Initially described by Fisher,2 mounting epidemio-
logical evidence has shown that SBI can contribute to cogni-
tive dysfunction,3 dementia,4 and increased overall mortality.5
However, the most direct potential sequela of SBI is symp-
tomatic stroke. Despite similar pathophysiologic pathways,
it is unknown whether SBI and stroke have identical mech-
anisms given the heterogeneity of SBI in terms of location,
mechanism, and underlying risk factors. Furthermore, directly
studying the mechanisms of SBI is challenging given that
these lesions are incidentally detected and almost always of
unknown age. The small size of many SBI, their noneloquent
location, and chronic ischemic preconditioning are attractive
theories to explain why certain cerebrovascular events do not
manifest clinically.6
Understanding the extent to which MRI-defined SBI predicts
the occurrence of stroke is important because if these lesions
strongly predict stroke, then their detection might warrant the
initiation of a thorough stroke evaluation and more aggressive
medical management of stroke risk factors. Also, a more precise
quantification of the relative risk (RR) of stroke in the presence
of an SBI may allow the adoption of SBI as a surrogate end
point in clinical trials and thereby potentially reduce the length

Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
Received October 20, 2015; final revision received December 24, 2015; accepted December 29, 2015.
From the Department of Radiology (A.G., C.F.), Feil Family Brain and Mind Research Institute (A.G., G.G., H.K.), Department of Healthcare Policy and
Research (A.E.G.), Samuel J. Wood Library and C.V. Starr Biomedical Information Center (D.D.), Department of Neurology (H.K.), Weill Cornell Medical
College, New York, NY; Department of Neurology, Columbia University Medical Center, New York, NY (J.G.); Department of Neurology, University
of Miami Miller School of Medicine, FL (C.W.); Department of Biostatistics, Boston University School of Public Health, MA (A.S.B.); Department of
Neurology, Boston University School of Medicine, MA (S.S.); and Department of Health Policy and Management, Harvard T.H. Chan School of Public
Health, Boston, MA (A.P.).
Presented in part at the International Stroke Conference of the American Heart Association, Los Angeles, CA, February 17–19, 2016.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.
115.011889/-/DC1.
Correspondence to Ajay Gupta, MD, 525 E 68th St, Starr 8A, Box 141, New York, NY 10065. E-mail ajg9004@med.cornell.edu
© 2016 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.115.011889

719
Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016
720  Stroke  March 2016
and expense of trials that would otherwise have relied on stroke
as a primary outcome measure.7 Although there have been many
individual studies describing the predictive value of SBI, rely-
ing on single-study samples results in wide confidence inter-
vals (CIs) for risk estimates. For these reasons, we performed
a systematic review and meta-analysis evaluating whether MRI
detection of SBI is a predictor of subsequent stroke.
Methods
We performed this study following the guidelines recommended by the
Meta-Analysis of Observational Studies in Epidemiology (MOOSE)
group8 and the Preferred Reporting Items for Systematic Reviews and
Meta-Analyses statement.9 We also prospectively registered our study
protocol on the International Prospective Register of Systematic
Reviews (PROSPERO registration number CRD42014007016).
Data Sources and Searches
A research librarian performed comprehensive searches from data-
base inception on April 3, 2015, in Ovid MEDLINE, Ovid Embase,
and the Cochrane Library. An English language filter was not applied.
The first search was conducted in Ovid MEDLINE. Subject head-
ings and keywords were adapted for the other databases. Additional
records were identified by using the cited by and view references fea-
tures in Scopus on April 17, 2015 (search methodology details are
given in Methods section in the online-only Data Supplement).
Study Selection
We included only studies with MRI characterization of SBI in sub-
jects subsequently followed up for the development of future clini-
cally overt stroke. Specific inclusion criteria were (1) studies of adult
subjects (aged >18 years); (2) at least 100 subjects; (3) MRI deter-
mination of SBI as lesions measuring ≥3 mm with differentiation of
SBI from leukoaraiosis; (4) mean follow-up of >12 months after brain
MRI; and (5) clinical ascertainment of stroke during follow-up. In
cases where the methods for detecting SBI or outcome data were not
clear in the article, we attempted to contact the corresponding author
for additional details. Furthermore, if test data from a cohort were
published more than once, only the article with the largest person-
years of follow-up was included to minimize the analysis of duplicate
or overlapping samples.
Data Extraction and Quality Assessment
A single investigator read the title and abstract of all references pro-
duced by our database search. After preliminary articles were short-
listed as potentially eligible, the articles were read in their entirety by 2
readers to determine eligibility, with disagreements resolved by consen-
sus. Data were extracted from articles meeting the inclusion criteria by
a team of 2 readers using a prespecified data collection template, with
disagreements in data extraction resolved by a third tie-breaking reader.
Study characteristics that were extracted included the first author of
the study, study design (prospective or not), major study inclusion cri-
teria, country of the study, total number of subjects, mean follow-up,
SBI status at baseline, and the prevalence of stroke risk factors in the
studied populations, including age, sex, hypertension, diabetes melli-
tus, chronic kidney disease, atrial fibrillation, coronary artery disease,
hyperlipidemia, and smoking history. Additional study extraction fo-
cused on SBI status and stroke outcomes during follow-up, including
the number of subjects with and without SBI at baseline, the number of
strokes (all strokes, ischemic strokes, and hemorrhagic strokes) that oc-
curred during follow-up in both SBI-positive and SBI-negative groups,
the covariate-adjusted RR measure (hazard ratio [HR] or odds ratio)
relating SBI and future stroke, and the specific vascular risk factors for
which the RR measures were adjusted in multivariate analyses. Data
were also extracted about the imaging definitions of SBI in each study,
including MRI magnet field strength, MRI slice section thickness, MRI
section gap, SBI size classification, SBI MRI signal characteristics, and
means of differentiating SBI from perivascular spaces.
Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016
Because no standardized tool exists to assess the risk of bias in
observational studies, we adapted the bias assessment criteria adapt-
ed from recently published meta-analyses focused on SBI in patients
with atrial fibrillation10 and imaging biomarkers of stroke risk.11,12 A
total of 11 questions were generated to evaluate potential selection,
detection, misclassification, reporting, attrition, and confounding
bias13 (Table I in the online-only Data Supplement). Risks of bias
questions were assessed by 2 readers, with disagreements in assess-
ment resolved by a third tie-breaking evaluator.
Data Synthesis and Analysis
We estimated the prevalence of SBI in the included studies and the
total person-years of follow-up in each study. Meta-analyses of the
individual study crude RRs (ie, RR of stroke in the presence of an
SBI) and covariate-adjusted HRs were conducted with the use of
StatsDirect statistical software (version 2.7.9; 7/9/2012 StatsDirect
Ltd, Cheshire, England). Each pooled risk ratio was calculated using
a random effects (DerSimonian–Laird) model,14 and forest plots were
generated to display the individual study risk ratios and the pooled
risk ratio. We performed all analyses using a random effects model
based on the conservative assumption that included studies did not
have exactly the same effect size given the potential for heterogeneity
between studies in terms of sample size, subject characteristics, and
testing methods. To assess the combinability of the risk ratios, we
calculated the P value from the Cochrane Q statistical heterogene-
ity test. The results of each study were expressed as a risk ratio with
a 95% CI. For each meta-analysis, the presence of publication bias
was evaluated through a Begg–Mazumdar rank correlation test. All P
values <0.05 were considered statistically significant.
We performed subgroup analyses limited to the following patient
samples: (1) stroke-free participants recruited from population-based
or community-dwelling samples; (2) stroke-free participants in stud-
ies with inclusion criteria including at least 1 known stroke risk factor
(such as hypertension, diabetes mellitus, atrial fibrillation, coronary
artery disease, hyperlipidemia, or chronic kidney disease); and (3)
patients with a history of documented previous stroke followed up
for recurrent stroke.
Role of Funding Source
This study received no external funding.
Results
Study Selection
We screened a total of 1654 titles and abstracts from which
we identified 13 articles that met all inclusion criteria for the
systematic review.15–27 One study27 that met our inclusion cri-
teria was published after our search was completed during the
data extraction phase but was included as it provided follow-up
cohort data for an article initially included in our original liter-
ature search.28 Study selection steps are summarized in Figure I
in the online-only Data Supplement. A crude RR expressing
the association between SBI and incident stroke was calculable
in all 13 studies from raw data. The most commonly provided
adjusted risk metric was the HR, with 8 studies15,17,19,20,22,23,25,27
providing covariate-adjusted HRs of SBI as a predictor of
incident stroke. One study provided a covariate-adjusted RR
of SBI as a predictor of stroke,16 and 2 studies18,24 provided
covariate-adjusted or age- and sex-matched odds ratios.
Qualitative Study Characteristics
Of the 13 articles meeting inclusion criteria (Table II in the
online-only Data Supplement), all except one22 were prospec-
tive studies. One study was conducted as an international
 Gupta et al   Silent Brain Infarction and Stroke Risk    721

Figure. Forest plots of the association between magnetic resonance imaging-determined silent brain infarction and future stroke in all
included studies.15–27 Meta-analysis performed using a random effects model, with crude relative risks pooled in (A) and adjusted-hazard
ratios pooled in (B). Squares represent point estimates for the effect sizes. The size of the squares is proportional to the inverse of the
variance of the estimate. Diamond represents the pooled estimate, and the horizontal lines represent the 95% confidence interval.

multicenter study,24 5 were in Japan,16–18,21,26 4 in the United studies were focused on evaluation of stroke in younger patients
States,15,19,25,27 and 1 each in Finland,20 Canada,22 and the with mean ages of 39.422 and 40.0 years.20 Subjects in all stud-
Netherlands.23 Eleven studies15–19,21,23–27 included cohorts with ies were followed up for at least 25.7 months (range, 25.7–174
mean or median ages >50 years (range, 56.0–75.2), whereas 2 months) for ascertainment of clinically defined stroke.

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016

722  Stroke  March 2016
Definitions of SBI and Stroke Outcomes
All studies defined SBI as lesions ≥3 mm that were hyperin-
tense on T2-weighted images (details are given in Table III in
the online-only Data Supplement). One study provided addi-
tional separate analysis of putative vascular lesions <3 mm
that were too small to definitely characterize as SBI.27 All
studies described the use of additional MRI pulse sequences
to differentiate SBI from adjacent white-matter leukoaraio-
sis, with most studies relying on T1-weighted hypointensity
as a feature suggestive of SBI rather than nonspecific white-
matter leukoaraiosis. There were variable methods used to
distinguish SBI from dilated perivascular spaces, including
the presence of hyperintensity on T2-weighted fluid-atten-
uated inversion recovery images,18,20–22,24,26 proton density
hyperintensity,15,18,27 and lesion morphology/location.19,25
MRI magnet field strengths varied from 0.2 to 3.0 T, with
most studies using a 1.5-T scanner. Most studies used clini-
cally based definitions of ischemic stroke based on a focal
neurological deficit lasting >24 hours and without evidence
of hemorrhage on brain imaging (details are given in Table
IV in the online-only Data Supplement). A majority of stud-
ies used a combination of hospital and outpatient medical
records and telephone interviews to ascertain stroke outcome
events.
Association Between SBI and Future Stroke: Crude
RRs
We were able to obtain sufficient raw data to calculate a
crude RR for future stroke in the presence of SBI for each
of the included 13 studies (Table V in the online-only Data
Supplement). In this analysis of 13 studies (14 764 subjects),
there was a mean follow-up ranging from 25.7 to 174 months
(mean, ≈76.0), yielding a total of 108 360 person-years of
follow-up. There was statistically significant heterogene-
ity (Q=39.65; P<0.001), but no significant publication bias
(Kendall τ score, 0.26; P=0.25) was present in this analy-
sis. We found a significant positive relationship between the
presence of SBI and the risk of stroke with a random effects
crude RR of 2.94 (95% CI, 2.24–3.86; P<0.001; Figure [A]).
Of the total study sample, 3007 subjects (20.4%) had a posi-
tive MRI test for SBI, whereas 11 757 (79.6%) had a nega-
tive test for SBI.
Table 1.  Meta-Analysis Summary for Subgroup Analyses of Crude RR of Stroke in Patients With Magnetic Resonance Imaging–
Defined SBI
Total Person-
No. of Years of
Subgroup Studies Follow-Up
Stroke-free individuals recruited from population- 6 95 449.2
based or community-dwelling studies15,18,19,23,25,28
Stroke-free individuals with at least 1 known stroke 4 5813.4
risk factor16,17,21,26
Known history of stroke followed up for stroke 3 7097.6
recurrence20,22,24
CI indicates confidence interval; RR, relative risk; and SBI, silent brain infarction.
*Publication bias test only performed on >3 studies.
Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016
Association Between SBI and Future Stroke:
Adjusted HRs
Eight of the included studies provided adjusted HRs describ-
ing the strength of association between SBI and future stroke.
In the analysis of these 8 studies (10 427 subjects), there was
a mean follow-up ranging from 25.7 to 174 months (mean,
≈88.8), yielding a total of 85 001 person-years of follow-up.
There was neither significant heterogeneity (Q=8.99; P=0.25)
nor significant publication bias (Kendall τ score, 0.29; P=0.40)
present in this analysis. We found a significant positive rela-
tionship between the presence of SBI and the risk of stroke
with a random effects adjusted HR of 2.08 (95% CI, 1.69–
2.56; P<0.001; Figure [B]). Of the total study sample, 1962
subjects (18.8%) had a positive MRI test for SBI, whereas
8465 (81.2%) had a negative test for SBI.
Subgroup Meta-Analysis Results
Statistically significant random effects crude RR (Table 1)
and adjusted HR (Table 2) were preserved in the following
subgroup analyses: (1) stroke-free patients recruited from
population-based or community-dwelling studies; (2) stroke-
free participants in studies with inclusion criteria requiring
at least 1 known stroke risk factor, including hypertension,16
chronic renal disease requiring hemodialysis,17 diabetes mel-
litus,21 or >1 cardiovascular risk factor26; and (3) patients pre-
senting with first-time symptomatic acute stroke followed up
for recurrent stroke in whom clinically SBI was evident on
baseline imaging performed for the initial acute stroke diag-
nostic evaluation. Of the studies providing an adjusted HR,
the risk of stroke in the presence of SBI conferred ≈2-fold
increased risk of future stroke in both stroke-free patients
(HR, 2.06; 95% CI, 1.64–2.59) and in patients with previous
stroke (HR, 2.00; 95% CI, 1.08–3.71). On the basis of the 5
studies15,19,23,25,27 with adjusted HRs derived from population-
based studies of 9483 stroke-free individuals with mean or
median study ages between 62 and 76, the prevalence of SBI
was 18.7%.
Assessment of the Quality of the Included Studies
The results from the quality assessment questionnaire are
shown in Table VI in the online-only Data Supplement. All
included studies involved >1 investigator independently
Test of
Heterogeneity Publication Bias*
SBI
Prevalence, Pooled RR Cochran Q,
% (95% CI) RR, P Value P Value Kendall τ P Value
17.7 2.81 (2.40–3.28) <0.01 <0.01 0.07 0.99
38.1 4.54 (2.76–7.45) <0.01 0.50 0.33 0.75
27.5 1.99 (1.30–3.04) <0.01 0.21 … …
 Gupta et al   Silent Brain Infarction and Stroke Risk    723
Table 2.  Meta-Analysis Summary for Subgroup Analyses of Risk Factor–Adjusted Hazard Ratios of Stroke in Patients With
Magnetic Resonance Imaging–Defined SBI
Total Person-
No. of Years of
Subgroup Studies Follow-Up Prevalence, %
Stroke-free individuals recruited from population-based 5 78 540
or community-dwelling samples15,19,23,25,27
Known history of stroke followed up for stroke 2 6062.6
recurrence20,22
CI indicates confidence interval; HR, hazard ratio; and SBI, silent brain infarction.
*Publication bias test only performed on >3 studies.
evaluating MRI studies for the presence of SBI. In all but 2
studies,16,17 perivascular spaces were systematically differ-
entiated from SBI. In 7 of the 13 studies,15,18,19,23–26 the risk
of selection bias was minimized by either random selec-
tion of subjects or recruitment from a community-dwelling
population. Six studies explicitly reported that investiga-
tors were blinded to SBI status when determining stroke
outcomes.15,19,20,23,25,27 All but 2 studies21,26 corrected for
covariate risk factors in their analysis of the association
between SBI and incident stroke and provided an adjusted
disease association measure, such as a HR or odds ratio.
Five studies15,19,23,25,27 had the lowest overall risk of bias,
with each of these studies demonstrating low potential for
bias in all questions but one. All 5 of these studies were
of stroke-free individuals recruited from large population-
based studies, including the Cardiovascular Health Study
(CHS),15 the Rotterdam Scan Study,23 the Framingham
Offspring Cohort Study,19 the Northern Manhattan Study
(NOMAS),25 and the Atherosclerosis Risk in Communities
Study (ARIC).27
Discussion
SBI is often incidentally detected in patients during MRI eval-
uation. In this systematic review and meta-analysis of stud-
ies involving >14 000 subjects and >100 000 person-years of
follow-up, we found that the presence of SBI conferred an
≈3-fold increased risk of subsequent stroke. After adjustment
for potentially confounding vascular risk factors, the presence
of an SBI was associated with ≈2-fold higher risk of future
stroke. This approximate doubling of the independent risk
of incident stroke was seen in both stroke-free patients and
patients with previous stroke. Approximately 1 in 5 stroke-
free individuals in these studies had SBI.
Our results are particularly applicable to stroke-free, com-
munity-dwelling individuals aged >60 years who are inciden-
tally found to have an SBI on brain MRI. In our subgroup
analysis of 5 large, diverse population-based studies (CHS,
ARIC, NOMAS, Rotterdam Scan Study, and the Framingham
Offspring Cohort Study), the presence of an SBI conferred a
≈2-fold increase in incident stroke risk after adjustment for
vascular risk factors. These 5 studies had the lowest risk of
bias in our assessment and pooled together >9400 subjects
totaling ≈78 000 person-years of follow-up, thereby increas-
ing our confidence in the conclusion from this analysis.
Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016
Test of
Heterogeneity Publication Bias*
SBI Pooled HR Cochran Q,
(95% CI) HR, P Value P Value Kendall τ P Value
18.7 2.06 (1.64–2.59) <0.01 0.24 0 0.82
16.2 2.00 (1.08–3.71) 0.03 0.24 … …
Few guidelines exist on how to manage patients with inci-
dentally found SBI. In contrast, professional societies publish
detailed guidelines on secondary stroke prevention in patients
with clinically overt stroke.29 Given that the likelihood of a
brain lesion presenting clinically depends more on the elo-
quence of the brain region affected rather than the underly-
ing pathogenesis of the lesion,30,31 the distinction between
silent and overt brain ischemia may need reexamination. In
this meta-analysis, we found that SBI confers a similar rela-
tive degree of risk of future stroke as a previous history of
clinically overt stroke in certain settings. In patients with atrial
fibrillation, for example, a previous history of stroke is associ-
ated with an HR of 2.3 for recurrent stroke,32 comparable with
the magnitude of risk we determined in our meta-analysis for
the first-time symptomatic stroke after SBI. Such consider-
ations raise the question of whether SBI should be considered
as a manifestation of cerebrovascular disease that requires
intensive evaluation and secondary prevention, as in clinically
overt stroke, rather than simply being considered an incidental
finding.
Our study has revealed some limitations about the MRI
techniques used to detect SBI in the existing body of literature
summarizing stroke risk after SBI. First, although there was
general agreement on the MRI features of SBI, including size
(≥3 mm, T2 hyperintensity distinguishable from leukoaraio-
sis), we found variability in the methods used to determine the
presence of SBI, including MRI magnet field strength (range,
0.2–3.0 T) and specific MRI sequences obtained for evalua-
tion. Most of the included studies (12 of 13) did not perform
MRI on higher field strength 3-T magnets, which are becom-
ing increasingly common and are more likely to allow detec-
tion of small lesions. The prevalence and predictive value
of SBI detected on these higher field MRI magnets require
additional study. Furthermore, the methods used to distinguish
SBI from perivascular spaces, a potential mimic on imaging
studies, were variably reported and raise the possibility that
some perivascular spaces were misclassified as SBI. Such
misclassification, however, may not significantly affect our
study’s overall conclusions given recent data suggesting that
perivascular spaces are a manifestation of cerebral small ves-
sel disease33 and that small brain lesions <3 mm that are too
small to definitely distinguish from small perivascular spaces
are still predictive of incident stroke.27 Nonetheless, our study
suggests that further standardization of SBI definitions would
724  Stroke  March 2016
allow for the more widespread use of SBI as an imaging risk
biomarker.6,34
Several additional limitations of our study are important to
consider. First, we performed one of our meta-analyses using a
crude RR because the covariate risk factors varied across stud-
ies. However, the relatively higher quality studies included in
our analysis provided adjusted HRs, and although there were
some differences in the specific risk factors for which each study
performed statistical correction, it seems unlikely that these
relatively minor interstudy differences affected our overall con-
clusions. Second, given the variability in length of follow-up
between studies, the existing data are not amenable to the calcu-
lation of absolute stroke rates. Third, there was variable report-
ing of stroke subtypes (ischemic versus hemorrhagic) in studies.
It would seem likely that SBI would more strongly predict isch-
emic rather than hemorrhage stroke risk, although further work
is warranted to investigate this issue. Fourth, it is unclear in a
majority of studies, especially the large population-based stud-
ies, to what extent imaging confirmation of stroke outcomes
occurred. A more standardized definition of stroke outcomes
using modern imaging-based criteria would be helpful in future
studies of SBI and stroke risk. Fifth, stroke prevention measures
have substantially decreased the incidence of stroke1 since the
time many of the cohorts in this study were recruited (some >20
years ago), possibly limiting the validity of using these data to
extrapolate stroke risk in present-day populations. Despite this
reduction in absolute stroke risk in the general population, we
think that the longer term trends in stroke incidence are unlikely
to substantially affect our study’s main conclusions given that
our focus was on RR. Furthermore, to our knowledge, there are
no data to suggest that the recent reductions in stroke incidence
would be different in patients with or without SBI. Finally, the
RR and HR we determined in our meta-analyses are probably
most applicable to older adults given that the majority of par-
ticipants in these studies were middle aged or elderly. Further
studies are warranted to assess whether the future stroke risk
conferred by SBI in young adults is substantially different com-
pared with relatively older individuals.
In summary, our systematic review and meta-analysis sug-
gests that SBI is significantly associated with an increased
risk of symptomatic stroke. Our study specifically suggests
that ≈1 in 5 older, stroke-free adults may be harboring an SBI,
which may in turn confer more than double the risk of future
first-time stroke. These data indicate a need for future studies
of in-depth stroke risk evaluations and intensive prevention
measures, including lifestyle modification and proven medi-
cal therapies for stroke reduction, in patients with these com-
monly discovered brain lesions.
Acknowledgments
We thank Drs Jukka Putaala, Toshitaka Umemura, M. Arfan Ikram,
Marileen L. Portegies, Beverly G. Windham, Michael E. Griswold,
and Wanmei Wang for providing both unpublished data and clarifica-
tion of published data used in this meta-analysis.
Sources of Funding
Dr Gupta reports support in part by the Foundation of the American
Society of Neuroradiology Scholar Award. Dr Wright reports grants
support (National Institutes of Health [NIH] NS 29993) for the
Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016
Northern Manhattan Study (NOMAS). Drs Seshadri and Beiser report
grant support (NIH NS017950 and AG0008122) for the Framingham
Cohort Study. Dr Kamel reports grant support from the National
Institute of Neurological Disorders and Stroke (K23NS082367).
Disclosures
None.
References
1. Mozaffarian D, Benjamin EJ, Go AS, Arnett DK, Blaha MJ, Cushman
M, et al; American Heart Association Statistics Committee and Stroke
Statistics Subcommittee. Heart disease and stroke statistics–2015
update: a report from the American Heart Association. Circulation.
2015;131:e29–e322. doi: 10.1161/CIR.0000000000000152.
2. Fisher CM. Lacunes: small, deep cerebral infarcts. Neurology.
1965;15:774–784.
3. Longstreth WT Jr, Dulberg C, Manolio TA, Lewis MR, Beauchamp NJ
Jr, O’Leary D, et al. Incidence, manifestations, and predictors of brain
infarcts defined by serial cranial magnetic resonance imaging in the
elderly: the Cardiovascular Health Study. Stroke. 2002;33:2376–2382.
4. Vermeer SE, Prins ND, den Heijer T, Hofman A, Koudstaal PJ, Breteler
MM. Silent brain infarcts and the risk of dementia and cognitive decline.
N Engl J Med. 2003;348:1215–1222. doi: 10.1056/NEJMoa022066.
5. Bokura H, Kobayashi S, Yamaguchi S, Iijima K, Nagai A, Toyoda
G, et al. Silent brain infarction and subcortical white matter lesions
increase the risk of stroke and mortality: a prospective cohort
study. J Stroke Cerebrovasc Dis. 2006;15:57–63. doi: 10.1016/j.
jstrokecerebrovasdis.2005.11.001.
6. Fanning JP, Wesley AJ, Wong AA, Fraser JF. Emerging spectra of
silent brain infarction. Stroke. 2014;45:3461–3471. doi: 10.1161/
STROKEAHA.114.005919.
7. Vickrey BG, Brott TG, Koroshetz WJ; Stroke Research Priorities
Meeting Steering Committee and the National Advisory Neurological
Disorders and Stroke Council; National Institute of Neurological
Disorders and Stroke. Research priority setting: a summary of the 2012
NINDS Stroke Planning Meeting Report. Stroke. 2013;44:2338–2342.
doi: 10.1161/STROKEAHA.113.001196.
8. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D,
et al. Meta-analysis of observational studies in epidemiology: a proposal
for reporting. Meta-analysis Of Observational Studies in Epidemiology
(MOOSE) group. JAMA. 2000;283:2008–2012.
9. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP,
et al. The PRISMA statement for reporting systematic reviews and meta-
analyses of studies that evaluate health care interventions: explanation
and elaboration. Ann Intern Med. 2009;151:W65–W94.
10. Kalantarian S, Ay H, Gollub RL, Lee H, Retzepi K, Mansour M, et al.
Association between atrial fibrillation and silent cerebral infarctions: a
systematic review and meta-analysis. Ann Intern Med. 2014;161:650–
658. doi: 10.7326/M14-0538.
11. Gupta A, Kesavabhotla K, Baradaran H, Kamel H, Pandya A, Giambrone
AE, et al. Plaque echolucency and stroke risk in asymptomatic carotid
stenosis: a systematic review and meta-analysis. Stroke. 2015;46:91–97.
doi: 10.1161/STROKEAHA.114.006091.
12. Gupta A, Baradaran H, Schweitzer AD, Kamel H, Pandya A, Delgado
D, et al. Carotid plaque MRI and stroke risk: a systematic review
and meta-analysis. Stroke. 2013;44:3071–3077. doi: 10.1161/
STROKEAHA.113.002551.
13. Higgins JPT. Chapter 8: Assessing risk of bias in included studies. In:
Green S, ed. Cochrane Handbook for Systematic Reviews of Interventions
Version 5.1.0 [Updated March 2011]. The Cochrane Collaboration;
2011. http://handbook.cochrane.org/.
14. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin
Trials. 1986;7:177–188.
15. Bernick C, Kuller L, Dulberg C, Longstreth WT Jr, Manolio T,
Beauchamp N, et al; Cardiovascular Health Study Collaborative Reseach
Group. Silent MRI infarcts and the risk of future stroke: the cardiovascu-
lar health study. Neurology. 2001;57:1222–1229.
16. Kario K, Shimada K, Schwartz JE, Matsuo T, Hoshide S, Pickering
TG. Silent and clinically overt stroke in older Japanese subjects
with white-coat and sustained hypertension. J Am Coll Cardiol.
2001;38:238–245.
17. Naganuma T, Uchida J, Tsuchida K, Takemoto Y, Tatsumi S,
Sugimura K, et al. Silent cerebral infarction predicts vascular events
 Gupta et al   Silent Brain Infarction and Stroke Risk    725
in hemodialysis patients. Kidney Int. 2005;67:2434–2439. doi:
10.1111/j.1523-1755.2005.00351.x.
18. Bokura H, Kobayashi S, Yamaguchi S, Iijima K, Nagai A, Toyoda
G, et al. Silent brain infarction and subcortical white matter lesions
increase the risk of stroke and mortality: a prospective cohort
study. J Stroke Cerebrovasc Dis. 2006;15:57–63. doi: 10.1016/j.
jstrokecerebrovasdis.2005.11.001.
19. Debette S, Beiser A, DeCarli C, Au R, Himali JJ, Kelly-Hayes M,
et al. Association of MRI markers of vascular brain injury with inci-
dent stroke, mild cognitive impairment, dementia, and mortality: the
Framingham Offspring Study. Stroke. 2010;41:600–606. doi: 10.1161/
STROKEAHA.109.570044.
20. Putaala J, Haapaniemi E, Kurkinen M, Salonen O, Kaste M, Tatlisumak
T. Silent brain infarcts, leukoaraiosis, and long-term prognosis in young
ischemic stroke patients. Neurology. 2011;76:1742–1749. doi: 10.1212/
WNL.0b013e31821a44ad.
21. Umemura T, Kawamura T, Umegaki H, Mashita S, Kanai A, Sakakibara T,
et al. Endothelial and inflammatory markers in relation to progression
of ischaemic cerebral small-vessel disease and cognitive impairment:
a 6-year longitudinal study in patients with type 2 diabetes mellitus.
J Neurol Neurosurg Psychiatry. 2011;82:1186–1194. doi: 10.1136/
jnnp.2010.217380.
22. Gioia LC, Tollard É, Dubuc V, Lanthier S, Deschaintre Y, Chagnon M,
et al. Silent ischemic lesions in young adults with first stroke are asso-
ciated with recurrent stroke. Neurology. 2012;79:1208–1214. doi:
10.1212/WNL.0b013e31826aacac.
23. Poels MM, Steyerberg EW, Wieberdink RG, Hofman A, Koudstaal PJ,
Ikram MA, et al. Assessment of cerebral small vessel disease predicts
individual stroke risk. J Neurol Neurosurg Psychiatry. 2012;83:1174–
1179. doi: 10.1136/jnnp-2012-302381.
24. Weber R, Weimar C, Wanke I, Möller-Hartmann C, Gizewski ER,
Blatchford J, et al; PRoFESS Imaging Substudy Group. Risk of recurrent
stroke in patients with silent brain infarction in the Prevention Regimen
for Effectively Avoiding Second Strokes (PRoFESS) imaging substudy.
Stroke. 2012;43:350–355. doi: 10.1161/STROKEAHA.111.631739.
25. Di Tullio MR, Jin Z, Russo C, Elkind MS, Rundek T, Yoshita M, et al.
Patent foramen ovale, subclinical cerebrovascular disease, and ischemic
stroke in a population-based cohort. J Am Coll Cardiol. 2013;62:35–41.
doi: 10.1016/j.jacc.2013.03.064.
Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016
26. Miwa K, Tanaka M, Okazaki S, Furukado S, Sakaguchi M, Mochizuki H,
et al. Association between interleukin-6 levels and first-ever cerebrovas-
cular events in patients with vascular risk factors. Arterioscler Thromb
Vasc Biol. 2013;33:400–405. doi: 10.1161/ATVBAHA.112.300350.
27. Windham BG, Deere B, Griswold ME, Wang W, Bezerra DC, Shibata
D, et al. Small brain lesions and incident stroke and mortality: a cohort
study. Ann Intern Med. 2015;163:22–31. doi: 10.7326/M14-2057.
28. Windham BG, Griswold ME, Shibata D, Penman A, Catellier DJ,
Mosley TH Jr. Covert neurological symptoms associated with
silent infarcts from midlife to older age: the Atherosclerosis Risk
in Communities study. Stroke. 2012;43:1218–1223. doi: 10.1161/
STROKEAHA.111.643379.
29. Kernan WN, Ovbiagele B, Black HR, Bravata DM, Chimowitz MI,
Ezekowitz MD, et al; American Heart Association Stroke Council,
Council on Cardiovascular and Stroke Nursing, Council on Clinical
Cardiology, and Council on Peripheral Vascular Disease. Guidelines for
the prevention of stroke in patients with stroke and transient ischemic
attack: a guideline for healthcare professionals from the American Heart
Association/American Stroke Association. Stroke. 2014;45:2160–2236.
doi: 10.1161/STR.0000000000000024.
30. Leary MC, Saver JL. Annual incidence of first silent stroke in the United
States: a preliminary estimate. Cerebrovasc Dis. 2003;16:280–285.
31. Brott T, Tomsick T, Feinberg W, Johnson C, Biller J, Broderick J,
et al. Baseline silent cerebral infarction in the Asymptomatic Carotid
Atherosclerosis Study. Stroke. 1994;25:1122–1129.
32. Diener HC, Eikelboom J, Connolly SJ, Joyner CD, Hart RG, Lip GY, et
al; AVERROES Steering Committee and Investigators. Apixaban versus
aspirin in patients with atrial fibrillation and previous stroke or transient
ischaemic attack: a predefined subgroup analysis from AVERROES,
a randomised trial. Lancet Neurol. 2012;11:225–231. doi: 10.1016/
S1474-4422(12)70017-0.
33. Potter GM, Doubal FN, Jackson CA, Chappell FM, Sudlow CL, Dennis
MS, et al. Enlarged perivascular spaces and cerebral small vessel disease.
Int J Stroke. 2015;10:376–381. doi: 10.1111/ijs.12054.
34. Wardlaw JM, Smith EE, Biessels GJ, Cordonnier C, Fazekas F, Frayne
R, et al; STandards for ReportIng Vascular changes on nEuroimaging
(STRIVE v1). Neuroimaging standards for research into small vessel
disease and its contribution to ageing and neurodegeneration. Lancet
Neurol. 2013;12:822–838. doi: 10.1016/S1474-4422(13)70124-8.
 Silent Brain Infarction and Risk of Future Stroke: A Systematic Review and
Meta-Analysis
Ajay Gupta, Ashley E. Giambrone, Gino Gialdini, Caitlin Finn, Diana Delgado, Jose Gutierrez,
Clinton Wright, Alexa S. Beiser, Sudha Seshadri, Ankur Pandya and Hooman Kamel

Stroke. 2016;47:719-725; originally published online February 17, 2016;

doi: 10.1161/STROKEAHA.115.011889
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2016 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/47/3/719

Data Supplement (unedited) at:

http://stroke.ahajournals.org/content/suppl/2016/02/17/STROKEAHA.115.011889.DC1.html

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Stroke is online at:

http://stroke.ahajournals.org//subscriptions/

Downloaded from http://stroke.ahajournals.org/ at VA MED CTR BOISE on February 28, 2016

Supplemental Material
Supplemental Methodology

Literature Search Details

Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations and Ovid MEDLINE(R) 1946 to Present - Search ran 04/03/2015

1. exp Brain Infarction/

2. ((brain or cerebral or circulation or covert or lacunar or subcortical or venous) adj3 infarct*).tw.
3. 1 or 2
4. silent.tw.
5. 3 and 4
6. (SBI or SLI or SSBI).tw.
7. 5 or 6
8. exp Magnetic Resonance Imaging/
9. magnetic resonance.tw.
10. (MR or MRA or MRDTI or MRI or MRIs or NMR).tw.
11. or/8-10
12. exp stroke/
13. stroke*.tw.
14. Cerebrovascular.tw.
15. ((brain or cerebral or isch?emic or lacunar or vascular or venous) adj2 (accident* or attack* or event* or infarct*)).tw.
16. (cva or cvas).tw.
17. or/14-16
18. 7 and 11 and 17

The English language filter was not applied. The first search was conducted in Ovid MEDLINE. Subject headings and key words were adapted for the other databases.
Supplemental Table I: Risk of Bias Questions to Assess the Quality of Included Studies

Type of Bias Question Answers

Was the study sample randomly selected or a community-dwelling population to minimize the risk of selection bias? Yes (+) or no (-)
Selection Were the inclusion and exclusion criteria adequately described? Yes (+) or no (-)
Was the study's primary objective to assess whether SBI is predictive of clinically overt stroke? Yes (+) or no (-)
Was the study prospective in nature? Yes (+) or no (-)
Were the investigators blinded to the clinical history of patients during ascertainment of SBI? Yes (+) or no (-)
Detection
Did more than one investigator assess for the presence of SBI? Yes (+) or no (-)
Were investigators blinded to SBI status when determining stroke outcomes? Yes (+) or no (-)
Misclassification Did the investigators describe a method by which SBI was differentiated from dilated perivascular spaces? Yes (+) or no (-)
Reporting Did more than one investigator assess for stroke outcomes? Yes (+) or no (-)
Attrition Were losses to follow up systematically recorded and reported? Yes (+) or no (-)
Minimal (+) if studies controlled for 4 of the following 6 potential stroke risk factors which are
potential confounders: age, sex, hypertension, diabetes mellitus, coronary artery disease, and
smoking history and hyperlipidemia by including these variables in the multivariate model or by
Confounding Were data adjusted for covariate risk factors to minimize the risk of confounding bias?
ensuring that patients with and without SBI were similar or matched on these variables. Relatively
higher risk (-) if studies did not provide adjusted risk metric demonstrating strength of association
between SBI and incident stroke.
Note if data not provided or not specified, recorded as no (-).
 Supplemental Table II. Overview of Patient Characteristics in Studies Evaluating Risk of Stroke in Patients with MRI defined Silent Brain Infarct

Mean Chronic Atrial

Study Study First Author SBI status at Number of Mean Age (+/- Women, no. (% Diabetes Coronary Artery Hyperlipidemia
Study Design Major Inclusion Criteria Study Name Country Follow-Up Hypertension (%) Kidney Fibrillation Smoking History (%)
Number and Year baseline Subjects SD) female) Mellitus (%) Disease (%) (%)
(Mo) Disease (%) (%)

stroke-free subjects,
prospective Cardiovascular
1 Bernick 20011 randomly sampled, United States 50.4 SBI positive 923 76* 543 (58.8) 526 (57) 123 (13.3) No data 22 (2.4) 216 (23.4) 208.4† 1.2‡
cohort Health Study
population-based cohort
SBI negative 2401 73.1* 1437 (59.9) 1118 (46.6) 317 (13.2) No data 65 (2.7) 411 (17.1) 207.1† 0.5‡

prospective stroke-free, ambulatory Total sample with SBI

2 Kario 2001§2 / Japan 42 585 72.0 +/- 9.9 592 (61.8) 811 (84.6) 111 (11.6) No data No data No data 187 (19.5) 197 (20.6)
cohort subjects with hypertension status known

prospective stroke-free, ambulatory

3 Naganuma 20053 / Japan 40.2 SBI positive 59 62.9+/-8.1 18(30.5) 51 (86.4) 24 (40.6) 59 (100) No data 17 (28.8) 11 (18.6) 22 (37.2)
cohort subjects on hemodialysis

SBI negative 60 49.2+/-12.8 17(28.3) 51 (85) 10 (16.7) 60 (100) No data 3 (5) 4 (6.6) 8 (13.3)

prospective stroke-free subjects, healthy

4 Bokura 20064 / Japan 75.6 SBI positive 380 62 +/- 7.2 142 (37.4) 248 (65.3) 42 (11.1) No data 6 (1.6) No data 124 (32.6) 142 (37.4)
cohort volunteers

SBI negative 2304 57.1 +/- 7.1 1069 (46.4) 894 (38.8) 221 (9.6) No data 14 (0.6) No data 733 (31.8) 740 (32.1)
community-based,
Framingham
prospective prospective study, offspring of
5 Debette 20105 Offspring Cohort United States 67.2 SBI positive 253 65+/-9 126 (49.8) 146 (57.7) 40 (15.8) No data No data No data No data 33 (13.0)
cohort original Framingham Heart
Study
Study participants
SBI negative 1975 62 +/- 9 923 (46.7) 770 (39.0) 215 (10.9) No data No data No data No data 234 (12.0)

prospective subjects with history of any Total sample with SBI

6 Putaala 2011§6 / Finland 104.4 655 40+/-8.0 270 (41.2) 244 (37.3) 66(10.1) No data 19 (2.9) 46 (7.0) 368 (56.2) 264 (40.3)
cohort ischemic stroke status known

prospective stroke-free, ambulatory

7 Umemura 20117 / Japan 72 SBI positive 46 62.7 +/-8.1 102 (53.7) 96 (50.5) 46 (100) No data No data No data 95 (50) 38 (20.0)
cohort subjects with DM

SBI negative 144 62.7 +/-8.1 102 (53.7) 96 (50.5) 144 (100) No data No data No data 95 (50) 38 (20.0)

retrospective subjects with history of any

8 Gioia 20128 / Canada 25.7 SBI positive 48 42.5+/-7.0 27(56.2) 20 (41.7) 10 (20.8) No data No data 6 (12.5) 36 (75) 24 (50)
cohort ischemic stroke

SBI negative 122 38.2+/-8.7 54(43.8) 32 (26.2) 12 (9.8) No data No data 5 (4.1) 74 (60.7) 56 (45.9)
stroke-free subjects,
prospective The Rotterdam Total sample with SBI
9 Poels 2012§9 randomly sampled, Netherlands 120 210 || 72+/-7.4 522 (51.8) 332 (33.0) 66 (6.6) No data 28 (2.8) 96 (9.5) No data 176 (17.5)
cohort Scan Study status known
population-based cohort
prospective subjects with recent ischemic Prevention
Multicenter
cohort data stroke within 120 days of Regimen for
trial (35
10 Weber 201210 from a study, SBI positive and Effectively 30 SBI positive 207 66.2 +/-8.5 57 (27.5) 162 (78.2) 67 (32.3) No data 8 (3.8) No data 99 (47.8) 135
countries or
randomized negative patients matched by Avoiding Second
regions)
controlled trial age and sex Strokes trial

SBI negative 207 66.2 +/-8.5 57 (27.5) 155 (74.7) 63 (30.4) No data 6 (2.9) No data 111 (53.8) 130
 prospective stroke-free subjects, Northern Total sample with SBI
11 Di Tullio 2013 ||, §11 United States 85.2 1287 71+/-9 779 (61) 934 (73) 291 (23) No data 31 (2.4) 160 (12.4) 504 (39) 119 (9)
cohort population-based cohort Manhattan Study status known

Osaka Follow-up
stroke-free, ambulatory
prospective Study for Carotid
12 Miwa 201312 subjects with >1 Japan 57.6 SBI positive 130 68.8+/-8.6 237 (51) 315(68) 79 (17) No data No data No data 246(53) 70 (15)
cohort Atherosclerosis,
cardiovascular risk factor
part 2

SBI negative 334 68.8+/-8.6 237 (51) 315(68) 79 (17) No data No data No data 246(53) 70 (15)

The
Atherosclerosis
prospective stroke-free subjects,
13 Windham 2015¶13 Risk in United States 174 SBI positive 220 64.0+/-4.5 134 (61) 143 (65) 50 (23) No data No data 17 (8) No data 57 (26)
cohort population-based cohort
Communities
Study
SBI negative 1611 62.1+/-4.5 972 (60) 722 (45) 264 (17) No data No data 82 (5) No data 277 (17)

* weighted median value

†Weighted median value, total cholesterol (mg/dL)

‡Weighted median value, pack-years of smoking

§cohort characteristics refer to entire patient sample as data stratified by SBI status is not available

|| Di Tullio et al. presented SBI results and stroke outcomes for a subset of those subjects enrolled in the MRI substudy of the Northern Manhattan Study. The study characteristics in this table for this study were obtained after direct correspondence with study authors who were preparing a study focused on SBI and incident stroke risk
at the time of data extraction.

¶Windham et al. studied small brain lesions less than 3 mm as putative vascular lesions, in addition to lesions greater than 3 mm which were considered to be SBI. The data extracted from this study, including clinical characteristics, are focused on SBI defined>3mm versus no lesions.
Supplemental Table III: Silent Brain Infarction Definitions

Section
Study Study First Author Magnet Field Section Size
Thickness SBI MRI Signal Characteristics Means of differentiating SBI from perivascular spaces Means of detecting SBI in patients with documented prior stroke
Number and Year Strength Gap (mm) classification
(mm)

brighter lesions on spin density and T2 sequences than normal gray matter (for
0.35 and 1.5
1 Bernick 20011 5 mm 0 3 mm or greater cortical and deep grey matter); brighter at spin density and T1 hypointense (for spin density brigthness used to distinguish SBI from perivascular spaces NA
Tesla
white matter)
not low signal intensity area on T1-weighted images that was also visible as a
2 Kario 20012 1.5 Tesla 7.8 to 8.0 mm 3 to 15 mm not specified NA
specified hyperintense lesion on T2-weighted images
focal area on both T1 and T2-weighted images that was visible as low-intensity
not
3 Naganuma 20053 1.5 Tesla 10 mm >3 mm areas on T1 weighted image and as high signal intensity area on T2 weighted not specified NA
specified
images.
0.15, 0.2 and 1.5 not focal hyperintensity lesion on T2WI corresponding to a hypointensity lesion on proton density weighted or FLAIR images used to distinguish infarcts
4 Bokura 20064 10 mm, 7 mm >3 mm NA
Tesla specified T1WI from dilated perivascular spaces

area of abnormal signal intensity in a vascular distribution, at least 3 mm in size

5 Debette 20105 1 or 1.5 Tesla
6 Putaala 20116 1.0 to 1.5 Tesla not specified
7 Umemura 20117 1.5 Tesla
8 Gioia 20128 1.5 T or 3 Tesla not specified
not size, location, shape, and tissue contrast to distinguish SBI from dilated
4 mm >3 mm with a cerebrospinal fluid density on the subtraction image and, for lesions in the NA
specified perivascular spaces
basal ganglia area, distinct separation from the circle of Willis vessels
MRI of the brain studies acquired at the initial presentation for acute ischemic
not focal hyperintensity on T2-weighted images without a corresponding history of simultaneous hyperintensity on T2-weighted images and hypointensity stroke were reinterpreted by study stroke neurologists and a senior
≥ 3 mm
specified neurologic symptoms or signs on FLAIR images for perivascular spaces as opposed to SBI neuroradiologist. SBI classification required appropriate imaging criteria as well
as no corresponding history of neurologic symptoms or signs.
lesions less than 3 mm in diameter or with a signal intensity similar to
areas of focal hyperintensity larger than 3 mm in diameter detected on T2-
that of cerebrospinal fluid on FLAIR images excluded because of the
5 mm 2 mm >3 mm weighted images, hypointensity areas on T1-weighted images and areas of NA
high possibility of enlarged perivascular spaces, even if hyperintensity
hypointensity surrounded by hyperintense rim on FLAIR images.
on T2-weighted images and hypointensity on T1-weighted images
focal hyperintensities on T2-weighted and FLAIR-weighted sequences, 3 mm in
diameter, without corresponding neurologic symptoms; leukoaraiosis defined as SBI determined on imaging performed during the patient’s initial workup for
not hyperintensity of T2-FLAIR images used to distinguish SBI from dilated
≥ 3 mm multifocal or confluent hyperintensities located in periventicular or subcortical acute ischemic stroke; consensus of 2 neurologists needed to establish a lesion
specified perivascular space
regions or in the pontine white matter on T2-weighted or FLAIR sequences. as an asymptomatic brain infarction using all available clinical data
Differentiated from SBIs based on lesion morphology and localization

evidence of one or more infarcts on MRI, without a history of (corresponding)

9 Poels 20129
stroke or TIA (focal hyperintensities on T2 weighted images); white matter
proton density scans were used to distinguish infarcts from dilated
1.5 Tesla 5 or 6 mm 1 or 2 mm at least 3 mm lesions (rather than SBIs) were considered to be present if hyperintensities were NA
perivascular spaces
visible on proton-density and T2-weighted images, without prominent
hypointensities on T1-weighted scans

focal hyperintense lesion on T2-weighted images and/or fluid-attenuated Two study investigators defined SBI on the baseline imaging performed for
inversion recovery with no corresponding symptoms in the clinical history of the acute ischemic stroke as chronic lesions with no corresponding symptoms in the
not hyperintensity of T2-FLAIR images used to distinguish SBI from dilated
10 Weber 201210 Not specified not specified ≥ 3 mm patient that could be attributed to the lesion; SBI were distinguished from clinical history of the patient that could be attributed the presumed SBI;
specified perivascular space
nonspecific subcortical and periventricular white matter lesions by the presence Information about symptoms of the qualifying ischemic stroke was collected
of a corresponding hypointense lesion on T1-weighted images using baseline case report forms

(1) CSF density on the subtraction image and (2) if the stroke was in the basal
FLAIR=3 mm;
11 Di Tullio 201311 1.5 Tesla 0 mm ≥ 3 mm ganglia area, distinct separation from the circle of Willis vessels and perivascular lesion morphology used to distinguish SBI from perivascular spaces NA
T1=1.3 mm
spaces.

hypointense lesion and hyperintense rim on FLAIR images when located

not supratentorially, according to the corresponding hyperintensity and hyperintensity of T2-FLAIR images used to distinguish SBI from dilated
12 Miwa 201312 Not specified Not specified >3 and <15 mm NA
specified hypointensity on T2- and T1-weighted images, respectively, without perivascular space
stroke history.

lesions 3 mm in size and visible on both T1- and proton-density/T2-weighted

images were classified as infarcts; an additional analysis was performed on
13 Windham 201513 1.5 Tesla 5 mm 0 ≥ 3 mm spin density brigthness used to distinguish SBI from perivascular spaces NA
putative vascular lesions <3 mm which were too small to definitely characterize
as SBI

MRI = magnetic resonance imaging; SBI = silent brain infarction; FLAIR = fluid attenuated inversion recovery; NA = not applicable
Supplemental Table IV: Definitions of stroke
Study First Author
Ischemic Stroke Definition
and Year
clinical event of rapid onset consisting of neurological deficit lasting
Bernick 20011 more than 24 hours or if less than 24 hours, associated with an
appropriate brain lesion not due to hemorrhage
evidence of brain infarction or embolism with neurological deficit
Kario 20012
lasting more than 24 hours
Naganuma 20053 standard clinical criteria, not further specified
Bokura 20064 not specified
acute-onset focal neurological deficit of presumed vascular etiology
lasting ≥24 hours with imaging showing no hemorrhage and an
Debette 20105
infarct correlating with the clinical deficit, or an infarct documented at
autopsy
an episode of focal neurologic deficits with acute onset and lasting
more than 24 hours or if lasting less than 24 hours, with imaging
Putaala 20116
evidence of stroke corresponding with current symptoms plus
corresponding ischemic lesion on MRI
Umemura 20117 not specified
sudden onset of a persistent neurologic deficit or a transient
Gioia 20128
neurologic deficit associated with a new infarct on brain imaging
rapidly developing clinical signs of focal disturbance of cerebral
Poels 20129 function with no apparent cause other than a vascular origin, with a
duration of more than 24 hours
focal neurological deficit of vascular origin lasting more than 24 h, or
where there is evidence of a new brain infarct upon brain imaging;
Weber 201210 brain imaging must rule out hemorrhagic stroke, but it does not need
to confirm the presence of a brain infarct if clinical symptomatology is
sufficient to diagnose stroke
stroke defined by Trial of Org 10172 in Acute Stroke Treatment
criteria. Diagnosis of ischemic stroke was determined by two
Di Tullio 201311
neurologists independently, and Northern Manhattan Study principal
investigators adjudicated disagreements
an acute disturbance of focal neurological dysfunction with symptoms
Miwa 201312 lasting >24 hours (or resulting in earlier death) and thought to be a
result of either cerebral infarction
evidence of sudden or rapid onset of neurologic symptoms that
Windham 201513 persisted for more than 24 hours or led to death with no other
apparent cause, such as trauma, tumor, infection, or anticoagulation
CT = computed tomography; MRI = magnetic resonance imaging
Hemorrhagic Stroke Definition
clinical event of rapid onset consisting of neurological deficit lasting more
than 24 hours or if less than 24 hours, with evidence of blood in the
subarachnoid space, ventricles, or brain parenchyma
parenchymal or subarachnoid hemorrhage at brain imaging
standard clinical criteria, not further specified
not specified
acute-onset focal neurological deficit of presumed vascular etiology
lasting ≥24 hours with imaging showing intracranial hemorrhage
not specified
not specified
not specified, obtained from database and chart review
not specified
not studied
not studied
an acute disturbance of focal neurological dysfunction with symptoms
lasting >24 hours (or resulting in earlier death) and thought to be a result
of either cerebral hemorrhage.
any one of the following criteria: CT or MRI with intraparenchymal
hematoma; demonstration at autopsy or surgery; or at least 1 major or 2
minor neurologic deficits, bloody spinal fluid on lumbar puncture, no CT
or MRI with or without cerebral angiography demonstrating an avascular
mass effect, and no evidence of aneurysm or arteriovenous
malformation
Stroke Outcome Adjudication Process
follow up yearly examinations and at interim 6 months phone contacts; additional
review from all non stroke hospitalizations for ICD codes 403 through 438 identifying
cerebrovascular disease; events adjudicated by stroke adjudication committee made
up of study neurologists, neuroradiologist, and a clinician from the Coordinating Center
intermittent review of medical records and telephone interviews confirmed by physician
caring for the patient at the time of the event
review of medical records with diagnosis made by attending physician blinded to study
purpose; family telephone interview in events that happened out of hospital
annual questionnaire sent to the patient's home and, if needed, telephone interview and
interview of attending physician
panel adjudication of Framingham investigators reviewing all relevant medical records
and clinical data
telephone interview or letter sent to the patient and all patient records available from
hospitals and primary care; outcome events were classified by study stroke
neurologists blinded to other clinical data with consensus agreement
not specified
patient follow-up was performed by outpatient hospital visits with stroke neurologists;
clinical outcomes were obtained from the database and confirmed by retrospective
chart review
review of medical records of all participants at the general practitioner’s office and
additional information from hospital records and municipal health authorities; to verify
all diagnoses, investigators discussed information on all potential strokes and TIA with
an experienced stroke neurologist
all strokes are reviewed by at least 2 Prevention Regimen for
Effectively Avoiding Second Strokes Trial study adjudicators
follow-up annually by telephone; any vascular event or neurological or cardiac
symptoms triggered an in-person assessment; active hospital surveillance of admission
and discharge) codes was performed.
outpatient periodic visits and, in case of missed appointment follow-up, telephone
interview; original medical records were also reviewed to determine the occurrence of
stroke; all possible events were audited independently by 3 physicians.
annual follow-up interviews and community surveillance, including medical record
reviews; cases were reviewed separately using a computerized algorithm and a
physician reviewer
Supplemental Table V: Silent Brain Infarction MRI Test Results and Stroke Outcomes

Magnitude of
Unknown
Number of All Adjusted All
Number of Unknown or or Adjusted
Subjects Strokes Ischemic All Ischemic Hemorrhagic Stroke Risk
Subjects with Hemorrhagic Unclassifed Unclassifed All Adjusted Adjusted All
Study Mean Follow- without MRI in SBI Strokes in Strokes Strokes in Strokes in Metric (HR
Study First Author and Year MRI evidence Strokes in Stroke Stroke Stroke Adjustments for which covariates All Stroke Stroke p-
Number Up (Mo) evidence of Positive SBI in no SBI no SBI no SBI or OR) in
of SBI at SBI Group Subtype in Subtype in risk 95% CI value
SBI at Test Group Group Group Group presence of
baseline SBI Group no SBI metric
baseline Group SBI at
Group
baseline
age, sex, systolic blood pressure, diastolic blood
pressure, atrial fibrillation, left ventricular hypertrophy,
1 Bernick 20011 50.4 923 2401 67 57 10 92 74 13 5 HR 1.52 1.10-2.10 0.051
fasting insulin, common carotid artery wall thickness,
myocardial infarction history
not not not specified, adjusted relative risks and 95% CI were
2 Kario 20012 42 282 303 38 not available 7 not available RR 4.63 2.04-10.5 0.003
available available calculated using Cox regression analysis

not specified, adjusted HR obtained with a multivariate

3 Naganuma 20053 40.2 59 60 10 6 4 0 0 0 HR Cox proportional hazards taking into account 11 7.33 1.27-42.25 0.026
clinical variables
age, sex hypertension, diabetes mellitus, dyslipidemia,
4 Bokura 20064 75.6 380 2304 48 32 15 1 43 24 17 2 OR 3.66 2.28-5.89 <0.0001
smoking, alcohol use, family history of stroke

age, sex, systolic blood pressure, current smoking,

5 Debette 20105 110.4* 224* 1862* 16* 15* not available 51 42 not available HR 2.3* 1.30-4.06* 0.004*
diabetes, history of cerebrovascular disease

age, sex, hypertension, history of transient ischemic

6 Putaala 20116 104.4 86 569 19* 18 1* 62* 54 8* HR* attack, diabetes, stroke etiology, silent brain infarcts, 1.62* 0.93-2.82* 0.087*
and leukoaraiosis
not not not not
7 Umemura 20117 72 46 144 7* 6* 1 6* 6* 0* not applicable
provided applicable applicable applicable
age, sex, hypertension, diabetes, dyslipidemia,
8 Gioia 20128 25.7 48 122 11 11 0 8 8 0 HR coronary artery disease, tobacco use, alcohol use, and 3.2 1.2-8.7 0.02
chronic renal failure
age, sex, smoking, systolic blood pressure,
antihypertensive treatment, systolic blood pressure X
9 Poels 20129 120 210 797 42 25* 5* 12* 57 34* 7* 16* HR antihypertensive treatment, diabetes, atrial fibrillation, 2.5 1.7-3.9 not provided
left ventricular hypertrophy, and coronary heart
disease
patients with and without SBI were matched on age
10 Weber 201210 30 207 207 27 24 3 19 17 2 OR 1.42 0.79-2.56 0.24
and sex
age, sex, race-ethnicity, education, medical insurance
status, body mass index, smoking, physical activity,
not not
11 Di Tullio 2013 (Northern Manhattan Study)11 85.2 192* 1043* 22* not available 49* not available HR moderate alcohol drinking, hypertension, diabetes, 1.9 1.1-3.3 0.014
available available
hypercholesterolemia, history of atrial fibrillation,
coronary artery disease, and myocardial infarction
not not not not
12 Miwa 201312 57.6 130* 334* 12* 9* 3* 9* 7* 2* not applicable
provided applicable applicable applicable
age, sex, race–center, education level, body mass
index, smoking, alcohol use, diabetes, systolic and
not not
13 Windham 201513 174 220 1611 40 not available 95 not available HR diastolic blood pressures, antihypertensive medication 2.54 1.70-3.79 <.001
available available
use, heart disease, statin use, high- and low-density
lipoprotein cholesterol levels, and triglyceride level.
*data obtained via direct correspondence with study authors

MRI = magnetic resonance imaging; SBI = silent brain infarction; CI = confidence interval; HR = hazard ratio; OR = odds ratio;
 Supplemental Table VI: Results of Risk of Bias Questions to Assess the Quality of Included Studies

Type of Bias: Selection Detection Misclassification Reporting Attrition Confounding

Was the study Were the Did the
Was the study's Were Were data
sample randomly Were the investigators investigators Were losses to
primary objective Did more than investigators Did more than adjusted for
selected or a inclusion and Was the study blinded to the describe a method follow up
to assess whether one investigator blinded to SBI one investigator covariate risk
Question: community-dwelling exclusion criteria prospective in clinical history of by which SBI was systematically
SBI is predictive of assess for the status when assess for stroke factors to minimize
population to adequately nature? patients during differentiated from recorded and
clinically overt presence of SBI? determining outcomes? the risk of
minimize the risk of described? ascertainment of dilated perivascular reported?
stroke? stroke outcomes? confounding bias?
selection bias? SBI? spaces?
Answer: Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-) Yes (+) or no (-)* Yes (+) or no (-) Yes (+) or no (-)*

Bernick 20011 + + + + + + + + + - +
Kario 20012 - + - + + + - - + + +
Naganuma 20053 - + + + + + - - + + +
Bokura 20064 - - + + - + - + - + +
Debette 20105 + + + + + + + + + - +
Putaala 20116 - + + + - + + + + + +
Umemura 20117 - + - + + + - + - + -
Gioia 20128 - + + - + + + + - + +
Poels 20129 + + + + + + - + + + +
Weber 201210 - + - + + + + + + + +
Di Tullio 201311 + + - + + + + + + + +
Miwa 201312 - - - + + + - + + - -
Windham 201513 + + + + + + - + + + +

*Minimal (+) if studies controlled for 4 of the following 6 potential stroke risk factors which are potential confounders: age, sex, hypertension, diabetes mellitus, coronary
artery disease, and smoking history and hyperlipidemia by including these variables in the multivariate model or by ensuring that patients with and without SBI were
similar or matched on these variables. Relatively higher risk (-) if studies did not provide adjusted risk metric demonstrating strength of association between SBI and
incident stroke.
Supplemental Figure I. Study selection flow diagram.

SBI = silent brain infarction

Supplemental References

1. Bernick C, Kuller L, Dulberg C, Longstreth Jr WT, Manolio T, Beauchamp N, et al. Silent mri infarcts and the risk of future stroke: The cardiovascular health study. Neurology. 2001;57:1222-1229
2. Kario K, Shimada K, Schwartz JE, Matsuo T, Hoshide S, Pickering TG. Silent and clinically overt stroke in older japanese subjects with white-coat and sustained hypertension. Journal of the American College of Cardiology. 2001;38:238-
245
3. Naganuma T, Uchida J, Tsuchida K, Takemoto Y, Tatsumi S, Sugimura K, et al. Silent cerebral infarction predicts vascular events in hemodialysis patients. Kidney international. 2005;67:2434-2439
4. Bokura H, Kobayashi S, Yamaguchi S, Iijima K, Nagai A, Toyoda G, et al. Silent brain infarction and subcortical white matter lesions increase the risk of stroke and mortality: A prospective cohort study. J Stroke Cerebrovasc Dis.
2006;15:57-63
5. Debette S, Beiser A, DeCarli C, Au R, Himali JJ, Kelly-Hayes M, et al. Association of mri markers of vascular brain injury with incident stroke, mild cognitive impairment, dementia, and mortality: The framingham offspring study. Stroke; a
journal of cerebral circulation. 2010;41:600-606
6. Putaala J, Haapaniemi E, Kurkinen M, Salonen O, Kaste M, Tatlisumak T. Silent brain infarcts, leukoaraiosis, and long-term prognosis in young ischemic stroke patients. Neurology. 2011;76:1742-1749
7. Umemura T, Kawamura T, Umegaki H, Mashita S, Kanai A, Sakakibara T, et al. Endothelial and inflammatory markers in relation to progression of ischaemic cerebral small-vessel disease and cognitive impairment: A 6-year longitudinal
study in patients with type 2 diabetes mellitus. J Neurol Neurosurg Psychiatry. 2011;82:1186-1194
8. Gioia LC, Tollard E, Dubuc V, Lanthier S, Deschaintre Y, Chagnon M, et al. Silent ischemic lesions in young adults with first stroke are associated with recurrent stroke. Neurology. 2012;79:1208-1214
9. Poels MM, Steyerberg EW, Wieberdink RG, Hofman A, Koudstaal PJ, Ikram MA, et al. Assessment of cerebral small vessel disease predicts individual stroke risk. J Neurol Neurosurg Psychiatry. 2012;83:1174-1179
10. Weber R, Weimar C, Wanke I, Möller-Hartmann C, Gizewski ER, Blatchford J, et al. Risk of recurrent stroke in patients with silent brain infarction in the prevention regimen for effectively avoiding second strokes (profess) imaging
substudy. Stroke; a journal of cerebral circulation. 2012;43:350-355
11. Di Tullio MR, Jin Z, Russo C, Elkind MS, Rundek T, Yoshita M, et al. Patent foramen ovale, subclinical cerebrovascular disease, and ischemic stroke in a population-based cohort. Journal of the American College of Cardiology.
2013;62:35-41
12. Miwa K, Tanaka M, Okazaki S, Furukado S, Sakaguchi M, Mochizuki H, et al. Association between interleukin-6 levels and first-ever cerebrovascular events in patients with vascular risk factors. Arterioscler Thromb Vasc Biol.
2013;33:400-405
13. Windham BG, Deere B, Griswold ME, Wang W, Bezerra DC, Shibata D, et al. Small brain lesions and incident stroke and mortality: A cohort study. Ann Intern Med. 2015;163:22-31