Among clinically stroke-free participants in this mid-life, community-dwelling cohort, 10.7% had SCIs on routine brain MRI. Whereas prior studies have also reported a significant association of hypertension and tHcy with prevalent SCI, we are the first study to demonstrate a significant relationship between AF and SCI. We are also the first study to evaluate the effect of LVM on SCI prevalence. The association of these risk factors for clinical stroke with subclinical SCI is not surprising but reinforces the importance of early detection and treatment of cardiovascular risk factors in midlife. This is especially true since SCI have been associated with an increased risk of incident stroke and cognitive impairment.26
Prior cohort studies have reported a prevalence of SCI on MRI4–6 27
varying from 11% in the younger participants (mean age: 63 years, range: 55–70 years) within the Atherosclerosis Risk in Communities (ARIC) study to 20% in the older participants of the Rotterdam Scan Study (RSS) and 28% in the Cardiovascular Health Study (CHS) sample; mean age was 76 years in both the CHS and RSS. 4, 6
However, autopsy data from 966 brains in the Hisayama Study evaluating asymptomatic community- dwelling adults (mean age = 74 years) disclosed a lower 12.9% prevalence of SCI, perhaps related to failure to count some SCI at autopsy, or to ethnic differences.27
MRI is reliably sensitive in identifying autopsy confirmed cerebral infarcts.28, 29
In our sample, the prevalence of SCI rises with increasing age in both men and women. which is consistent with prior studies‥4, 5, 27
Vermeer and colleagues reported a 30% higher overall prevalence of SCI in women compared to men in all age groups.5
However, we did not find a significant sex difference in the prevalence of SCI; neither did the ARIC and CHS studies.4
Studies have also differed on imaging criteria and technology. MRI scanners used in population studies have varied between field strengths of 1.0 to 1.5 Tesla, slice thicknesses of 5–8 mm and an interslice gap of zero to twenty percent.4, 5, 9
Prior studies share with our study well-defined MRI criteria for infarcts. However, our study was the only one to require a distinct separation between basal ganglial SCI and Circle of Willis vessels perhaps thereby improving diagnostic specificity. CHS additionally reported the the prevalence of lesions <3 mm and designated these as small infarct-like lesions; these lesions had a prevalence of 7.8% and were associated with memory loss.4
We did not assess SCI <3 mms in size.
We observed that more than four-fifth’s of our participants with SCI had a single lesion. In the Hisayama study 70% of participants with SCI had two infarcts or less while Kwon, et al found that 70% had a single lesion.12, 27
Lee et al described single lesion SCIs in 72%.3
We observed a similar distribution of SCI to that noted in prior studies; over half were in the basal ganglia, one-third were subcortical and one-tenth were cortical. In CHS, 72% were subcortical and in the RSS 80% were basal ganglia lesions. 4, 5
Lee, et al note that 38% were in the basal ganglia, a quarter were periventricular, 12 % thalamic and 9% pontine.3
It has been speculated that SCIs are silent (or rather produce subtle neurological damage) because they occur in clinically ineloquent areas of the brain;3
the predominance of lesions in the basal ganglia and subcortically rather than cortically appear to support this hypothesis.
The FSRP has been previously associated with incident clinical stroke, total brain volume and prevalent white matter hyperintensity volume.20, 30
We have now shown that the FSRP score was also associated with an increased risk of prevalent SCI. Additionally, the individual components of the FSRP were evaluated for their impact on SCI prevalence, but with the exception of SBP, hypertension and AF there were no other significant relationships. However, all the components of the FSRP were positively associated with an increased prevalence of SCI, although the relationships failed to reach statistical significance for some components. While prior studies have related the individual components of the FSRP to prevalent SCI, this is the first study to relate the aggregate score to prevalence of silent strokes.
Hypertension has consistently been implicated as a risk factor for SCI.1
. Vermeer demonstrated a more than two-fold increase in SCI in the RSS while Lee and others showed a more than three-fold increase in SCI prevalence in hypertensives.3, 5
At autopsy Shinkawa, et al. found that SCI and non SCI brains differed significantly with regard to diastolic blood pressure measurements.27
In our study we demonstrated that hypertension and elevated SBP are important risk factors for SCI.
In our data, AF increased the risk of prevalent SCI more than two- fold. AF is known to be an important risk factor for clinical stroke, associated with a five-fold increase in stroke incidence.31
In prior studies of persons with AF and no known clinical stroke, the prevalence of SCI on CT scans, in studies that have varied in sample size, has ranged from 13 to 48%.7, 8, 32–34
In the Hisayama autopsy study persons with SCI at autopsy had a significantly higher frequency of AF prior to death, than persons without SCI (10.7% versus 3.9%).27
In a study of 994 healthy middle-aged adults undergoing a screening medical examination with brain MRI, Lee and colleagues did not observe an association between prevalent SCI and AF; this may be explained by the lower mean age (49 years) of their sample, over a decade younger than the Framingham Offspring at the time of MRI.3
The other large community based cohorts did not evaluate AF as an SCI risk factor, or found it non significant.4–6, 35
Risk factors for stroke generally and perhaps SCI in particular are also risk factors for AF. Thus hypertension, prior CVD and MI, and diabetes which predispose to AF also predispose to clinical stroke and, probably to SCI. In this way AF may be a concomitant outcome rather than the mechanism underlying the occurrence of SCI.
In this study carotid artery stenosis ≥25% was associated with an increased prevalence of SCI. In the CHS study, participants with carotid stenosis ≥75% had nearly twice the likelihood of prevalent SCI on MRI.36
Uehara and others evaluated MRIs in 219 stroke-free participants who had presented to the hospital with nonspecific neurological complaints. Carotid artery stenosis >25% on MR angiography was associated with a three fold increased risk of basal ganglial SCIs on multivariate analysis.37
It is known that carotid artery stenosis may serve as an embolic source for cerebrovascular lesions; Norris and Zhu report a direct relationship between degree of stenosis and the frequency of both overt and silent stroke in a population that included participants with carotid distribution transient ischemic attacks.38
The pathogenetic mechanism, however, is unclear. In 108 participants, convergent color Doppler revealed greater turbulence in internal carotid arteries with stenosis greater than 50% if the participants also had an SCI on brain MRI than in those without an SCI.39
Uehara however found that there was no statistically significant difference in the prevalence of SCIs in the territory served by the stenosed carotid as compared to that of the nonstenosed artery.37
This is similar to prior data from Framingham40
and to data from the Asymptomatic Carotid Atherosclerosis Study suggesting that carotid artery disease may be a marker for atherosclerotic disease rather than the direct proximate cause of SCI 41
In our study, common carotid artery intima media thickness (CCAIMT) was significantly associated with prevalence of SCI as was ICAIMT. CCAIMT is known to predict clinical vascular outcomes including stroke and myocardial infarction.42, 43
Prior studies have related CCAIMT and ICAIMT to clinical stroke.44
A single Japanese study observed an association between CCAIMT and prevalent MRI infarcts in 28 adults, but it was not clear whether the observed MRI infarcts were silent since no history regarding prior stroke had been obtained.11
In the CHS study, prevalent SCI prevalence was associated with both CCAIMT and ICAIMT.36
Two prior studies have evaluated the effect of elevated circulating tHcy concentrations on prevalent SCI. Matsui and colleagues examined 153 elderly community dwelling participants with cranial MRI. Individuals in the highest and middle tertiles of plasma tHcy had more than four times and nearly three times the risk of SCI, respectively, compared to persons in the lowest tertile.9
In the RSS, participants in the upper quintile had approximately 2.5 times the risk of SCI compared to persons in the lowest quintile. 10
The results of our study are consistent with the literature. We demonstrate an increased prevalence of SCI in association with the plasma homocysteine level; Elevated tHcy is known to be associated with an increased incidence of stroke and MI.44
Pathogenetic hypotheses for SCIs have included dysregulated cerebral blood flow from hypertension and vascular disease, large vessel atherothrombosis, as well as from cerebral emboli.1, 45
Risk factors that proved significant for prevalent SCI remain those that fit these mechanisms: hypertension, atrial fibrillation, hyperhomocysteinemia, and carotid artery disease.
The strengths of evaluating the prevalence and risk factors for SCI using data from the Framingham Offspring Study include the use of previously validated MRI imaging to assess the presence or absence of MRI infarcts and the well-validated ongoing surveillance for clinical stroke which minimizes the probability of clinically symptomatic events being misclassified as SCI. In the past, studies have used differing methods to determine participants’ clinical history with regard to stroke symptomatology. Some have relied on subject self report,3, 6
others both on self report and evaluation of participants’ medical records5
while a few have used periodic visits and surveillance of hospital admission.4, 27
In our study surveillance techniques include systematic history taking for symptoms suggestive of stroke or transient ischemic attack at each biennial health evaluation and active daily surveillance of admissions to the local hospital.
Participants in the Framingham Study are largely of European descent. This limits the generalizability of our findings to other ethnic groups. Ethnic differences have been described in other multi-ethnic cohorts such as the Northern Manhattan Study.46
A second limitation of this study is the lack of incidence data given that the participants underwent a single cranial MRI.
In conclusion, we document a 10.7% prevalence of SCI in a community- based sample. This is concerning since SCI have been related both to the risk of incident stroke and to cognitive impairment26
. The significant relationship between hypertension, elevated serum homocysteine as well as carotid artery disease and prevalent SCI underscores the importance of current guidelines for the early diagnosis and prevention of hypertension and atherosclerosis and their risk factors. AF is also associated with SCI although our observational data cannot show if screening for and appropriately treating AF would reduce the population burden of SCI.