Subcortical Silent Brain Infarction as a Risk Factor for Clinical Stroke

Subjects and Methods

Subjects were 933 neurologically normal adults without history of CVD aged 30 to 81 years (mean age, 57.5±9.2 years). There were 536 men (mean age, 57.3±10 years) and 397 women (57.6±8.1 years). The subjects wished to receive health screening of the brain at their own expense. Eighty percent of them were supported by mutual aid associations, such as of public school teachers. This was not a population-based study; therefore, the follow-up period was varied: 217 subjects were followed up for 1 to 2 years, 160 for 2 to 3 years, 164 for 3 to 4 years, 142 for 5 to 6 years, and 250 for more than 6 years. We obtained information about their clinical stroke onset through sending out a questionnaire for subjects who received our health screening of the brain 1 to 7 years before this investigation. We continued this investigation from 1992. This report was based on the results of a 1992 to 1995 survey. We sent questionnaires to 1003 subjects in 1995, and confirmed health information from 93% of these subjects. We confirmed details for subjects with clinical stroke onset by telephone interview or direct examination with CT/MRI as soon as we could.

MRI was performed by using 0.15-T MRI (MT-15, Toshiba Co) (1988 to 1992) and 0.2-T MRI (Magnetom P-8, Siemens Co) (1992 to 1995). We used moderate T2WI (TR, 2000 ms; TE, 50 ms) at the transverse plane parallel to the orbitomeatal line and T1-weighted image (T1WI) (inversion recovery, TR of 2000 ms/TE of 40 ms; inversion time, 400 ms) at coronal slices with 1-cm intervals in the former period. We used T2WI (TR, 2000 ms; TE, 24 ms) and proton density–weighted (TR, 5000 ms; TE, 96 ms) at the transverse plane, and T1WI (TR, 350 ms; TE, 15 ms; flip angle 82°) at coronal slices (7 mm thick) in the latter period. In this series, we could not detect cortical infarction or brain stem or cerebellar infarction. Therefore, all silent brain infarctions were SSBI. We classified silent brain hyperintense lesions on T2WI according to the following: (1) SSBI, (2) FWT2HL, and (3) PVH. We considered a focal and sharply demarcated high intensity on T2WI larger than 3 mm to be SSBI that coincided with low density area of the T1WI. We called small demarcated hyperintensities in basal ganglia and white matter possible lacunes if they measured 3 to 15 mm and had correlative T1 hypointensity. The term “lacune” as we use it is not intended to infer the vascular pathoanatomy (lipohyalinosis versus arteriosclerosis). Most of the SSBIs were less than 10 mm in size, and no SSBI larger than 15 mm was observed in this series. The FWT2HL were defined as focal T2 hyperintensities similar to SSBI but without correlative T1 hypointensity. In addition, we defined the FWT2HL as a T2 hyperintensity lesion with correlative high intensity on proton density image in the latter period, We considered very small punctate (1 to 2 mm) hyperintensities detected in only T2WI to be perivascular enlargements according to histopathological study¹³ and excluded them from statistical analysis. The PVH was classified as grade 0 to 4 as follows: grade 0, almost no or unclear PVH; grade 1, a thin but apparent PVH restricted at the frontal horn; grade 2, smooth PVH surrounding the entire lateral ventricle or horns; grade 3, thick and irregular PVH surrounding the lateral ventricle and horns; and grade 4, marked diffuse PVH. Nonfocal vague T2 hyperintense lesions apart from periventricular were not included in the analysis.

The magnetic field strength of the MRI that was used in the former period was low and had lower spacial resolution than that used in the latter period. Therefore, most of the perivascular enlargements were likely to be undetected. The MRI in the latter period was also of low magnetic field strength, but it had enough spatial resolution to detect the perivascular enlargements 1 to 2 mm in diameter. We confirmed these findings by comparing them with scans made with 1.5-T MRI. Evaluation of the lesions on MRI was made by the same two authors independently in a blinded fashion. The interobserver agreement was high (94%). We discussed the diagnosis and made a group decision when there was disagreement. Blood pressure was measured by the standard sphygmomanometric method at supine position after 15 minutes of rest. Hematocrit, serum lipids (total cholesterol, triglyceride level, and HDL-C), fasting blood sugar, and ECG were measured on the same day. Sclerotic changes of retinal arteries were also examined with an ophthalmoscopic camera, and we defined retinal artery sclerosis as positive if judged to be Keith-Wegener’s grade 1 or more. We defined a history of hypertension and diabetes that was diagnosed by physician with or without treatment according to the subject’s information. One-time hypertension was not included the history. We obtained information of a family history of CVD from personal interview by a neurologist. We defined a smoker as any subject whose smoking index (number of cigarettes/day × number of years) was more than 200, and alcohol habits were graded as none or occasional drinker, 28 to 57 g of alcohol/d, 58 to 85 g/d, and 86 g/d or more. Informed consent for this study was obtained from all subjects according to our institutional guidelines.

Results

Incidence of Clinical Stroke Onset

Incidence of clinical stroke onset during the observation was significantly higher in the SSBI group (10 of 99; 10.1%) and the FWT2HL group (3 of 56, 5.4%) than that in the nonfocal lesion group (6 of 779; 0.77%) (P=.0001). Annual incidence of clinical stroke was 2.79% in the SSBI group, 1.50% in the FWT2HL group, and 0.21% in those without either focal lesion (0.55% in totals).

In these patients with clinical stroke onset, there were 14 brain infarction, 4 brain hemorrhages, and 1 SAH. The earliest onset of clinical stroke occurred 2.5 months and latest 7.3 years after the brain examination. Details of these cases are shown in Table 5⇓.

View this table:

• View inline
• View popup

Table 5.

Comparison of Potential Risk Factors for PVH Grade 3 vs PVH Grades 0 to 2 Estimated by Multiple Logistic Regression Analysis

We could confirm MRI or CT findings after clinical stroke onset in 15 of 19 cases. Details are shown in Table 4⇑. Lacunar infarctions in basal ganglia and deep white matter were commonly observed in the patients with brain infarction, but 1 patient had a 2.5-cm–sized corona radiata lesion associated with ipsilateral internal carotid artery occlusion. In the cases of brain hemorrhage, 3 patients showed hypertensive putaminal hemorrhage and 1 showed temporo-occipital lobar hemorrhage. The case of lobar hemorrhage was clinically suspected to be associated with amyloid angiopathy. The subject showed a low level of cystatin C in cerebrospinal fluid, which indicated amyloid angiopathy with deposition of cystatin C as we reported before.¹⁴ In these 15 subjects, as confirmed by CT or MRI, SSBI increased in 1 subject and confluent lesions increased in 2. All 3 of these subjects had SSBI and hypertension, and 2 of them also had diabetes. The rate of diabetes complication was significantly higher in these 3 subjects than in another 12 subjects (P=.037). Intervals from examination to stroke were 1044±731 days in the former period and 1356±579 days in the latter (NS). Clinical diagnosis of another 4 subjects without imaging data was confirmed by information from the physicians who treated these patients.

Lifetable analysis showed that occurrence of the clinical stroke was earlier and most frequent in the SSBI group among all subjects (P<.0001) (see the Figure⇓). The FWT2HL group had the second most frequent occurrence as compared with the nonfocal lesion group. In this table, stroke incidence was rather amplified for later periods because the number of follow-up subjects was decreased.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Graph showing the probability that, given survival, the subjects will remain free from stroke onset, stratified by SSBI, FWT2HL, and nonfocal lesion.

Discussion

The incidence of SSBI was 10.6% in this study, and this result was coincidently almost the same as in our previous report³ and as the incidence of small infarction in a community based autopsy study in Hisayama.¹⁰ Lindgren et al¹⁵ reported that 62.3% of 77 adults (mean age, 65 years) without a history of focal brain lesion in the community showed white matter hyperintensities on 0.2 T MRI. Their incidence included all white matter lesions. We defined SSBI as a focal and demarcated T2 hyperintensity larger than 3 mm that had correlative T1 hypointensity. Therefore, most of them could be considered to be lacunes. As regards silent deep infarcts on MRI, Ylikoski et al⁷ reported that the incidence of deep infarct was 2.6% in young-old (n=76) and 15.4% in old-old (n=52) by community based study using 0.02 T MRI. The Japanese community based studies using MRI revealed more higher incidence of lacunes as 47% in the elderly ^{4 5}

At the beginning of this study, we expected that some cortical infarcts due to atherothrombosis or embolism may be included in silent brain infarctions, but we could detect only SSBI in this series. The reason may be that most of our subjects were healthy and relatively young people without severe complications that were often observed in symptomatic stroke patients. For example, only six subjects had atrial fibrillation in this series. We also evaluated the clinical significance of nonspecific focal T2 hyperintensities. Clinical significance of nonspecific T2 hyperintenisties, except very small punctate lesions (suspected of being mostly perivascular enlargement), was not fully elucidated. Our results indicate that the most important and common underlying mechanism for SSBI, FWT2HL, and PVH is hypertensive small-vessel vasculopathy. The second most important risk factors for SSBI were diabetes and alcohol habits. On the other hand, only hypertension and age were significant risk factors for FWT2HL and PVH. Kase et al¹ emphasized that diabetes was the sole risk factor for silent stroke in the Framingham study, but these silent lesions included cortical infarcts in 7 of 15 lesions. Initial stroke subtype in these patients with silent stroke was mainly atherothrombotic infarcts and embolism (10/13). On the other hand, Boon et al² reported that advanced age and hypertension were the only risk factors for silent infarcts in their patients with initial ischemic stroke. This discrepancy may be caused by a difference in the type of silent infarction, because 82% of silent infarcts were small and deep in the study of Boon et al. Tuszynski et al⁹ reported that hypertension (64%) and diabetes (34%) were the most important risk factors in pathologically verified lacunar infarctions. It has been known that lacunar infarction caused not only by hypertensive small-vessel vasculopathy but also by atherosclerotic stenosis of the middle cerebral artery or embolism.¹⁶ However, hypertensive small-vessel vasculopathy was considered to be the most common cause of lacunar infarction.^{17 18} Shimada et al⁵ reported that silent lacunar infarctions were detected in 47% of 73 normal elderly subjects (40 of them were hypertensive), and SBP during sleep was significantly associated to silent lacunar infarctions. Our results showed heavy alcohol habits was also a significant risk factor for SSBI. A J-shaped relationship between alcohol intake and risk of stroke has been reported.¹⁹ The Hisayama study showed that hypertensive heavy drinkers were in a high-risk group for hemorrhagic and ischemic strokes.²⁰ They suggested that the synergistic effect of heavy alcohol use and hypertension may be related to SSBI.

Boon et al² reported that silent infarcts were found in 27% of 755 ischemic stroke cases on CT and that these lesions were significantly more strongly associated with a lacunar than an atherothrombotic infarction.

Chamorro et al¹¹ reported that silent infarctions on MRI were observed in 33% of 249 ischemic or hemorrhagic stroke patients. Okada et al¹² reported that the incidence of silent lacunes was 89% in 65 patients with hypertensive brain hemorrhage on MRI. However, no prospective study has been performed to investigate the rate of clinical stroke onset in neurologically normal subjects with SSBI. We have examined more than 1400 neurologically normal adults without any stroke history since 1988. We investigated clinical stroke onset of these subjects by questionnaire every year since 1992, and confirmed clinical stroke onset in 19 of 933 subjects during the observation period. Annual incidence of clinical stroke in our total subjects was 0.55% per year. Stroke incidence was estimated as 0.1% to 0.2% per year in the general population.²¹ It was reported that the age- and sex-adjusted average annual incidence rate of lacunar cerebral infarction was 13.4/100 000 persons.²² Our results demonstrated that the subjects with SSBI showed an apparently higher annual incidence of clinical stroke, 2.79% compared with 0.21% in those without SSBI. Kaplan-Meier lifetable analysis showed early and frequent clinical stroke onset in the subjects with SSBI. The subjects with FWT2HL showed an incidence median between those with SSBI and those without focal lesions. Retinal artery sclerosis was significantly related to SSBI but not to FWT2HL. This result suggested that SSBI was related to more long-standing hypertension than that in FWT2HL. We suppose that noncystic infarcts may be involved in FWT2HL to some extent, because focal subcortical T2 hyperintensity lesions with proton hyperintensity were possibly noncystic infarction if they had no correlative T1 hypointensity.¹³ The FWT2HL may have included so-called “unidentified bright objects” (UBO), which has been proposed by Kertesz et al.²³ They showed that UBO was significantly related to hypertension, and they commented that UBO should be differentiated from the white matter confluent lesions. Therefore, FWT2HL should be included in the analysis of prospective study for stroke onset. Although the number of cases with clinical stroke onset was relatively small, our results indicate that the subjects with SSBI should be considered to be at high risk for clinical stroke. Seventy-four percent of the cases who developed clinical stroke were ischemic stroke, however, hemorrhagic stroke was also observed in 26% of subjects in our study. Most cases of brain infarction were lacunar infarction except 1. In cases of brain hemorrhage, 3 of 4 cases showed typical hypertensive putaminal hemorrhage similar to SSBI. This type of hemorrhage is considered to be caused by fibrinoid necrosis of a small perforating artery.²⁴ It has been known that a common cause of lacunes is lipohyalinosis of the perforating arteries.²⁵ However, fibrinoid necrosis of perforating arteries also is often observed in hypertensive elderly people with not only brain hemorrhage but also lacunar stroke.²⁶ Therefore, it must be considered that SSBI puts one not only at high risk for subcortical brain infarction but also for hypertensive brain hemorrhage. On the other hand, PVH was not a significant risk for clinical stroke onset in this study. Our results showed that hypertension was the most significant risk factor for PVH. Although detailed pathogenesis of PVH or leukoaraiosis was not fully elucidated, it is supposed to be caused by chronic ischemia in the terminal zones of the medullary arteries related to hypertensive arteriolosclerosis.²⁷ A large cooperative study with MRI in 3301 elderly showed that white matter findings correlated with age, hypertension, silent stroke, and impaired cognitive function.²⁸ We have reported that frontal white matter lesions are significantly related to cognitive impairment in patients with multiple lacunar infarction.²⁹ Silent diffuse white matter lesions are supposed to be a risk factor for mental deterioration rather than clinical stroke.

One of our hemorrhagic cases showed lobar hemorrhage associated with probable amyloid angiopathy. It is well known that amyloid angiopathy causes lobar hemorrhage in elderly people.³⁰ Moreover, Vinters³⁰ reported that amyloid angiopathy was often observed in dementia of Binswanger’s type. We also reported a case of amyloid angiopathy with marked leukoaraiosis found on autopsy of an elderly subject who had been diagnosed by low level of cystatin C in cerebrospinal fluid.³¹ Therefore, antiplatelet therapy is not recommended for SSBI in elderly subjects with apparent PVH.

Thirty years ago in Japan the death rate from brain hemorrhage was one of the highest in the world, but it has dramatically decreased with management of hypertension and improvement of nutrition.³²

Reed et al³³ reported that Japanese men living in Japan more often had intracerebral small artery lesions than those in Hawaii. This result suggests that Japanese living in Japan have small cerebral artery disease more often than people in Western countries.

The incidence of clinical stroke differed according to distribution of age, sex, and other risk factors for stroke such as hypertension. As regards the recurrence rate of stroke, it was much higher than that of initial stroke. Alter et al³⁴ reported that the recurrence rate of stroke was significantly higher in hypertensive patients (43% compared with 19%) than in normotensives during 24 months in a population-based cohort of 662 stroke patients. Lai et al³⁵ also showed that the recurrence rate of stroke for 4 years was 24% in hypertensives and 9% in normotensives. On the other hand, primary stroke prevention by antihypertensive drugs has been proved in a large clinical trial.³⁶ Our results showed that SSBI was the most important risk factor for subcortical clinical stroke, OR of 10.48, suggesting that SSBI may be considered to be a prestroke stage if these subjects leave risk factors, such as hypertension, as they are, untreated.

In conclusion, the subjects with SSBI should be considered to be a high-risk group for clinical stroke not only for subcortical brain infarction but also for brain hemorrhage. Further prospective, controlled studies are needed to clarify the effect of management of risk factors on prevention of clinical stroke onset in the subjects with SSBI.

Selected Abbreviations and Acronyms