Presence and Severity of Cerebral White Matter Lesions and Hypertension, Its Treatment, and Its Control
The ARIC Study
Background and Purpose White matter lesions (WML) may result from cerebral hypoperfusion or ischemia. We investigated the association of WML with blood pressure, hypertension, and its treatment and control.
Methods A random sample of 1920 participants aged 55 to 72 years in the Atherosclerosis Risk in Communities Study (ARIC) was examined. Spin-density 1.5-T MRI scan images were coded from 0 for normal to 9 for most severe WML. Hypertension was defined as systolic or diastolic pressure ≥140/90 mm Hg or use of antihypertensive medication.
Results The percentages of persons with WML grades 0 through 2 and 3 through 9, respectively, were as follows: normotensive, 92.4% and 7.6%, versus all hypertensive subjects, 83% and 17% (P<.001); and treated controlled hypertensive, 86% and 14%, versus treated uncontrolled hypertensive subjects, 76% and 24% (P=.003). Multivariable adjusted odds ratios (95% confidence intervals) for WML grade ≥3 relative to normotensive subjects was 2.34 (1.71 to 3.20) for all hypertensives, 1.99 (1.19 to 3.08) for untreated hypertensives, 1.94 (1.32 to 2.85) for treated controlled hypertensives, and 3.40 (2.30 to 5.03) for treated uncontrolled hypertensives. After additional adjustment for hypertension duration, treatment, and control status, the odds ratios (95% confidence intervals) for a 1 SD increase of systolic and diastolic blood pressure were 1.43 (1.11 to 1.85) and 1.16 (0.94 to 1.43), respectively.
Conclusions Hypertension is associated with increased odds of WML, and treated uncontrolled hypertensive subjects have greater odds of WML than those with treated controlled hypertension. The data suggest that the level of blood pressure, especially systolic blood pressure, is related to WML, additional to the effects of categorically defined hypertension and its treatment and control status.
With the introduction of MRI technology, cerebral WML are frequently detected in elderly individuals.1 2 3 4 5 WML have been characterized as hyperintensity lesions in white matter, but their etiology and clinical and prognostic relevance are not fully understood.2 6 It has been suggested that WML are the consequence of arteriosclerosis, cerebral hypoperfusion, or ischemia,2 4 7 8 and it has been found that individuals with WML have reduced cognitive function.3 9 WML have been found by some investigators to be positively associated with hypertension.1 2 3 9 10 11 However, most published studies have been based on small and highly selected samples. Furthermore, very few studies have included sufficient numbers of African Americans to obtain meaningful information on WML for this group, although African Americans have a high prevalence of hypertension and high risk of stroke.
This study was designed to investigate, at the population level, the association of WML with hypertension and its duration, treatment, and control status, as well as the presence of differences by ethnicity. Additionally, the association of the absolute levels of blood pressure with the prevalence of WML was assessed in both normotensive and hypertensive subjects.
Subjects and Methods
The sample for this report consisted of 1920 individuals who participated in the cerebral MRI study at two study sites of the ARIC Study. ARIC is a longitudinal study of cardiovascular and pulmonary diseases sponsored by the National Heart, Lung, and Blood Institute. It includes community surveillance and cohort components. The ARIC cohort was selected as a probability sample of 15 800 men and women between the ages of 45 to 64 years at four study centers in the United States, three of which enumerated and enrolled community samples (Washington County, Md; Forsyth County, NC; and selected suburbs of Minneapolis, Minn). The fourth quarter of the ARIC cohort was sampled from only African American residents of Jackson, Miss. Details of sampling, study design, and cohort examination procedures have been published.12 Eligible participants were interviewed at home and then invited to a baseline clinical examination conducted in 1987 through 1989. Every 3 years after the baseline examination, participants were invited to a follow-up clinical examination. The first follow-up examination was conducted in 1990 through 1992, and the second follow-up examination in 1993 through 1995.
During the 1993 and 1994 ARIC clinical examinations, all cohort members aged 55 years and older at the Forsyth County, NC, and Jackson, Miss, study sites were screened for eligibility for the cerebral MRI examination (n=2877). For participant safety, the following criteria were used to exclude individuals as ineligible for the MRI study: (1) prior surgery on an aneurysm in the brain; (2) metal fragments in the eyes, brain, or spinal cord; (3) valvular prosthesis, cardiac pacemaker, cochlear implant, spinal cord stimulator, or other internal electrical device; (4) pregnancy; and (5) occupations associated with exposure to metal fragments. Among 2877 individuals screened, 1.95% of women and 5.90% of men were ineligible. Of those meeting inclusion criteria, 25% of women and 21% of men declined the cerebral MRI examinations. The final sample size was 1920 (611 African American women, 349 African American men, 539 white women, and 421 white men).
Although the MRI examinees were drawn from the first 2 years of the 1993-1995 cohort reexamination, this subgroup is considered a random sample of the full cohort in the two ARIC study sites because examination dates were allocated at baseline through randomly selected induction cycles. Reexamination visits were scheduled according to the anniversary date to the degree possible. Compared with those examined, individuals who refused the MRI examination had similar established cardiovascular risk profiles. The majority of exclusions from the MRI examination originated from occupations associated with exposure to metal fragments.
Cerebral MRI Scan and WML Classification
Details of the MRI scanning and the image interpretation protocols used for this study have been published.10 13 In brief, 1.5-T magnetic resonance scanners (GE and Picker) were used. Axial images were angled to be parallel to the anterior commissure–posterior commissure line. The digitized scan data, including SD/T2-weighted (repetition time, 3000 milliseconds; echo time, 30 and 100 milliseconds) and T1-weighted images (repetition time, 500 milliseconds; echo time, 20 milliseconds), were evaluated at the MRI Reading Center on a Vortech Personal Display System (PDS-4) workstation. The PDS-4 monitors measure 16 in diagonally with 1024×1024 pixel elements and 256 gray-scale intensities. The image evaluation was based on “pattern matching” of individual scans to a library of example images by trained and certified MRI readers, subject to established regular quality-control checks. When evaluating the WML, focal abnormalities (eg, infarctions with low T1-weighted signal intensity) were ignored. Thus, if one or both sides of the brain were focally abnormal, estimates were based on the uninvolved side with normal symmetry assumed. The spin-density images (repetition time, 3000 milliseconds; echo time, 30 seconds) were used to estimate the overall volume of periventricular and subcortical white matter signal abnormality, using a scale from 0 to 9 based on the visual “pattern matching” of the participant's scans to reference standards as presented in Fig 1⇓. The reference standards are briefly described as follows: no white matter signal abnormalities (grade 0); discontinuous periventricular rim or minimal “dots” of subcortical white matter (grade 1); thin continuous periventricular rim or few patches of subcortical WML (grade 2); thicker continuous periventricular rim with scattered patches of subcortical WML (grade 3); thicker shaggier periventricular rim with mild subcortical WML, may have minimal confluent periventricular lesions (grade 4); mild periventricular confluence surrounding the frontal and occipital horns (grade 5); moderate periventricular confluence surrounding the frontal and occipital horns (grade 6); periventricular confluence with moderate involvement of the centrum semiovale (grade 7); periventricular confluence involving most of the centrum semiovale (grade 8); and all white matter involved (grade 9). The intraclass correlation coefficients over the entire course of the study were .68 and .71 for interreader and intrareader reliability, respectively.
Blood Pressure and Hypertension, Its Treatment, and Its Control
These variables were defined based on the information obtained according to standardized protocols common to all ARIC study sites and subject to regular quality-control checks.14 In brief, during each clinical visit, sitting blood pressure was measured after a 5-minute rest three times for each participant with a random-zero sphygmomanometer by trained technicians following a standardized protocol. The average of the second and the third readings of SBP and 5th-phase DBP was used in this report. Hypertension was defined as DBP ≥90 mm Hg, SBP ≥140 mm Hg, or self-reported use of antihypertensive medication. Individuals identified as hypertensive at the third clinical examination were classified as “incident hypertensives” if they were not identified as hypertensive in either of the previous two clinical visits or as “prevalent hypertensives” if they were identified as hypertensive in either of the previous two clinical visits. Consequently, “incident hypertension” in this report is used as a surrogate for relatively shorter duration of exposure to elevated blood pressure, and “prevalent hypertension” is used for relatively longer duration of exposure to elevated blood pressure. Individuals classified as hypertensive at the third clinical examination were defined as treated if they reported use of prescribed antihypertensive medication; otherwise they were classified as untreated. Those with treated hypertension were classified as treated and controlled if their SBP was <140 mm Hg and DBP <90 mm Hg; otherwise they were classified as uncontrolled.
Age, ethnicity, sex, education levels, and cigarette smoking status were obtained using standardized questionnaires administered by trained and certified interviewers. Body mass index was calculated as weight (kilograms) divided by height (meters) squared. Levels of fasting serum total cholesterol, triglycerides, and HDL cholesterol were measured by enzymatic procedure using a Cobas centrifuge analyzer (Hoffman-La Roche).13 HDL cholesterol was measured after precipitation of apolipoprotein-B containing lipoproteins. LDL cholesterol was calculated in participants with triglyceride levels <400 mg/dL as Total Cholesterol−HDL Cholesterol+1/5 of Triglyceride. Diabetes mellitus was defined as levels of fasting (8 hours) serum glucose >140 mg/dL, glucose >200 mg/dL if fasting <8 hours, history of physician-diagnosed diabetes, or use of an oral hypoglycemic agent or insulin.
The a priori null hypothesis was that prevalence of WML was not associated with hypertension presence, treatment, or control or blood pressure level. The mean values and standard deviations, or proportions of selected population attributes and cardiovascular risk factors, were computed. Differences in the distribution of WML by hypertension status, duration of hypertension, and hypertension treatment and control status were examined using χ2 tests. We classified individuals with WML grade ≥3 as having WML and individuals with WML grade ≤2 as having “normal or little WML.” This cut point corresponded to approximately the 90th percentile of WML grade in the entire sample. Logistic regression models were used to assess the adjusted prevalent ORs and 95% CIs of WML grade ≥3 comparing normotensive to hypertensive, “incident hypertensive,” “prevalent hypertensive,” untreated hypertensive, treated and controlled hypertensive, and treated but uncontrolled hypertensive subjects. The same models were repeated, stratified by ethnicity. To assess the additional effects of the levels of blood pressure on WML in both normotensive and hypertensive subjects, logistic regression models were also used to estimate the ORs and 95% CIs of WML grade ≥3 for 1 SD increase of SBP and DBP, respectively, adjusting for the duration and treatment/control status of hypertension and other potential confounding factors. Predicted probabilities of WML grade ≥3 were calculated on the basis of the fitted logistic regression models. Polytomous logistic regression models were used to repeat the same logistic regression models by treating WML grade as an ordinal variable (WML grades 0, 1, 2, and 3-9). The SAS statistical software package (SAS Institute) was used for the statistical analyses.
The mean age of participants was 62 years; 50% were African Americans, and 40% were men. The characteristics of the sample, both overall and stratified by ethnicity and sex, are presented in Table 1⇓. The profile of cardiovascular and cerebrovascular disease risk factors was similar to that of African American and European American men and women of similar age in the full ARIC cohort from the two study sites.
The distribution of WML grade in this population, by hypertension status, duration of hypertension, and treatment and control of hypertension, is presented in Table 2⇓. The proportions of persons with WML grades 0, 1, 2, and 3-9 were 15%, 53%, 24%, and 7.6%, respectively, among normotensive subjects; this is in contrast to 14%, 47%, 22%, and 17% for hypertensives (P<.001); 15%, 49%, 22%, and 14% for treated and controlled hypertensives; and 14%, 39%, 23%, and 24% for hypertensives treated but uncontrolled (P=.003). The “prevalent hypertension” group had a slightly higher proportion of persons with higher WML grades than the “incident hypertension” group, but the difference was not statistically significant.
Interaction of hypertension variables with age and sex were tested and found not to be statistically significant (P>.15). Consequently, age- and sex-adjusted models were reported instead of age- and sex-specific models. Because preliminary analysis indicated that age, sex, smoking, diabetes mellitus, and total cholesterol were strong candidates as confounding factors in the hypertension-WML association, most of the statistical models were performed with adjustment for these variables.
Age-, sex-, and ethnicity-adjusted, as well as the ethnicity-specific and age- and sex-adjusted, mean levels of SBP and DBP are presented in Table 3⇓, stratified by hypertension, its duration, and its treatment and control status. As can be seen in this table, in almost all categories of hypertension status, African Americans had higher levels of blood pressure than white Americans.
Three sets of logistic regression models are presented in Table 4⇓ to estimate the prevalent ORs and 95% CIs of WML grade ≥3 comparing hypertensive with normotensive subjects: the first set of models was adjusted for basic demographic variables (model A); the second set was additionally adjusted for smoking, diabetes, and total serum cholesterol (model B); and the third set of models included the above covariates as well as SBP and DBP (model C). As shown in Table 4⇓, the first two sets of models (models A and B) indicated that there was a statistically significant increase in the odds of WML grade ≥3 in hypertensive subjects and that the hypertension-WML association was stronger in African Americans than in whites (value for ethnicity interaction with hypertension, P<.15). These associations were independent of age, sex, smoking, diabetes mellitus, and serum total cholesterol. After additional adjustment for SBP and DBP, the ethnicity difference in hypertension and WML association was diminished, as shown in model C (for ethnicity and hypertension interaction term, P>.15). Equivalently adjusted logistic regression models were also fit to estimate the ORs and 95% CIs of WML grade ≥3 comparing “incident hypertensives” and “prevalent hypertensives” with normotensives. The results are also presented in Table 4⇓. As shown in the first two sets of models in this table, both “incident” and “prevalent” hypertensives had greater increased odds of WML than normotensives, and “prevalent” hypertensives had a higher OR of WML than “incident” hypertensives, with the latter not being statistically significant. The association of “prevalent hypertension” to WML was stronger in African Americans than in white Americans. However, this ethnicity difference was greatly reduced after adjusting for blood pressure levels (model C). Table 4⇓ also presents the same three sets of adjusted logistic regression models to estimate the ORs and 95% CIs of WML comparing untreated, treated and controlled, and treated but uncontrolled hypertensive with normotensive subjects. The first two sets of models in this table showed that all three groups of hypertensives had significantly higher odds of WML than normotensives and that treated but uncontrolled hypertensives had the highest odds of WML, followed by untreated hypertensives and then by the treated and controlled hypertensives. African Americans had higher odds of WML than white Americans in all three hypertension treatment and control categories. However, after the adjustment for blood pressure levels (model C), the ethnicity difference was again greatly reduced.
To confirm the findings presented in Table 4⇑, and to assess the dose-response relationship between WML grades and hypertension, its duration, and its treatment and control status, polytomous logistic regression models were used to estimate ORs of WML of grades 1, 2, and 3-9 relative to grade 0 (corresponding to relatively mild, moderate, and more severe WML, respectively). The results (data not shown) confirmed the positive associations between WML and hypertension status and similar patterns of ethnicity differences in these associations. They also indicated a graded (dose-response) relationship of higher WML grades and hypertension, its duration, and its treatment and control.
To investigate the role of elevated blood pressure and WML, binary logistic regression models were used to estimate the ORs and 95% CIs of WML grade ≥3 for 1 SD increase of SBP and DBP in each category of hypertension status. The results are presented in Table 5⇓ for the total sample and stratified by ethnicity. The 1 SD increment of both SBP and DBP was associated with higher odds of WML ≥3, and the association was slightly stronger for SBP (OR, 1.50; 95% CI, 1.31 to 1.72) than for DBP (OR, 1.35; 95% CI, 1.16 to 1.57). These associations were independent of age, ethnicity, sex, smoking, diabetes mellitus, and serum cholesterol levels. In the subgroup analyses, these associations were consistent and of similar magnitude across status of hypertension, its duration, and its treatment and control status, except for the stratum of treated and controlled hypertension. When stratified by ethnicity, the associations between blood pressure levels and WML ≥3 were slightly stronger in African Americans than that in whites, but the differences did not reach a statistically significant level at P<.15.
The entire study population was divided into persons who reported using antihypertensive medication and persons not using antihypertensive medication. Logistic regression models were fit separately in these two groups to estimate the predicted probability of WML grade ≥3 in relation to levels of SBP, adjusted for age, ethnicity, sex, smoking, diabetes, and total cholesterol. The predicted probabilities of WML ≥3 from the logistic regression models are presented in Fig 2⇓. There was a trend of a curvilinear relationship between the levels of SBP and the probability of WML ≥3 in persons using antihypertensive medication (for quadratic term of SBP, P=.07), in contrast to a linear relationship in persons not using any antihypertensive medication (for quadratic term of SBP, P=.22). The slopes of these two lines were similar at higher levels of SBP (after approximately SBP >125 mm Hg).
The ORs (95% CI) of WML ≥3 were also estimated for a 1 SD increase in SBP and DBP in hypertensive subjects after additional adjustment for the estimated duration and treatment and control status of hypertension. An increase of 1 SD in SBP corresponded to a 1.43 (1.11 to 1.85)–fold increase in the odds of WML ≥3 and a 1.16 (0.94 to 1.43)–fold higher odds for 1 SD increase in DBP.
Thirty-four individuals were diagnosed as having a history of transient ischemic attack and/or stroke. Adjusted models were rerun excluding these individuals, with results almost identical to those in which participants with transient ischemic attack and/or stroke were included (data not shown).
WML are hyperintensity areas in the white matter seen on MRI scan.1 2 3 4 5 6 The term “leukoaraiosis,” indicating thinning of cerebral white matter, has been proposed to describe these lesions.2 WML can be the result of many pathophysiological processes, and their etiology and clinical importance are not fully understood at present. It has been suggested that WML are the consequence of arteriosclerosis, cerebral hypoperfusion, or ischemia.2 4 7 8 Reports of the prevalence of WML have varied from 8% to 92%,5 15 16 17 18 based on selected samples of healthy volunteers/patients and different criteria for WML. The prevalence of WML was 27% in 100 individuals aged 65 to 85 years examined by the Rotterdam Study,9 87% in individuals aged 65 years and older examined by the Cardiovascular Health Study,10 and 85% in this sample.19 It was observed that WML were associated with age, ethnicity, cigarette smoking, positive history of stroke, elevated blood pressure, and hypertension.10 19 Our study represents the first population-based study designed to investigate the association of WML with hypertension, its duration, and its treatment and control status; to investigate the effects of blood pressure level with WML in both normotensive and hypertensive subjects; to examine ethnicity differences in these associations; and to identify possible etiologic mechanisms underlying the ethnicity differences.
Association of WML to Hypertension, Its Duration, and Its Treatment and Control
In this population sample, the prevalence of mild to severe WML (grade 1 or above) was 85% among individuals aged 55 to 72 years, and the prevalence of more severe WML (grade 3 or above) was 12%. As shown in Table 2⇑, compared with normotensive subjects, a higher proportion of hypertensives had higher WML grades, regardless of hypertension duration or treatment and control status. After adjustment for several potential confounding factors, such as age, ethnicity, sex, smoking, diabetes, and serum cholesterol levels, compared with normotensive subjects, all categories of hypertensives carried significantly elevated ORs of WML grade ≥3. When comparing across categories of hypertension status, our data show a consistent trend of higher ORs in prevalent hypertensive than in incident hypertensive subjects, indicating that individuals with an estimated longer duration of elevated high blood pressure are more likely to have WML. Our results also indicate a consistent trend of higher ORs in untreated hypertensives and treated but uncontrolled hypertensives than in treated and controlled hypertensives. These findings are consistent with the hypothesis that control of elevated blood pressure levels may lead to lower risk of WML at the population level. This conclusion is also supported by the results presented in model C in Table 4⇑, in which the ORs of WML in all categories of hypertension are greatly reduced after adjusting for the levels of SBP and DBP. Our data also suggest slightly lower odds of WML in untreated than treated hypertensive subjects (Table 4⇑). In this cross-sectional study, hypertension treatment status may be a marker of the severity and of the duration of elevated blood pressure. Thus, untreated hypertensives may have had less severe hypertension and consequently were at lower risk of having WML.
Similar findings from the polytomous logistic regression approach (data not shown in this report) not only confirm the findings from the binary logistic regression models but also reassure that the statistically significant associations of WML with hypertension, its estimated duration, and its treatment and control found in this study are unlikely to be attributable to the scale and/or cut point used in defining WML.
Ethnicity Difference in the Associations of WML with Hypertension, Its Duration, and Its Treatment and Control
This study is the first to include sufficient numbers of African Americans in the study sample to report a pattern of ethnic difference in these associations. Because of the much higher prevalence of hypertension in African Americans than in whites, it is important from a public health perspective to examine the ethnicity differences in these associations and to investigate the possible etiologic mechanism(s) underlying these differences. In these data, the associations of WML with hypertension, its duration, and its treatment and control were stronger in African Americans than in whites, even after adjustment for age, sex, smoking, diabetes mellitus, and serum cholesterol levels. Two approaches were taken to determine whether the underlying etiologic mechanism of such ethnicity differences is largely due to the ethnic difference in the levels of blood pressure within categories of hypertension. First, we compared ethnic differences in SBP and DBP within each category of hypertension. As shown in Table 3⇑, in almost all the categories of hypertension classification, African Americans had higher levels of both SBP and DBP compared with whites. These ethnicity differences in SBP and DBP in all categories of hypertension classification suggest that the stronger associations with WML in African Americans may be attributable to the differences in the levels of blood pressure between the two ethnic groups. We further tested the role of SBP and DBP in explaining the ethnic differences in this population by also adjusting for the levels of SBP and DBP in the multivariable logistic regression models. As can be seen in model C in Table 4⇑, previously observed ethnic differences in the association of WML with hypertension, duration of hypertension, and the treatment and control of hypertension (in models A and B) were greatly reduced after additional adjustment for SBP and DBP. These results suggest that the ethnic differences in the association of hypertension status with WML in this population are largely attributable to the differences in the levels of blood pressure and that there is no meaningful ethnic difference at similar levels of SBP and DBP. It should also be noted that because the majority of African American examinees were sampled from the Jackson, Miss, study site and all European Americans were sampled from the Forsyth, NC, site, the results attributed to ethnic differences in this report might be confounded by geography. However, it is unlikely that such geographic differences could account fully for the ethnic differences.
Effect of Blood Pressure Level on the Presence and Severity of WML
It is biologically plausible that the associations of WML with hypertension, its duration, and its treatment and control are attributable to the levels of blood pressure. To examine the effects of SBP and DBP on WML in both normotensive and hypertensive subjects, we estimated standardized ORs and 95% CIs of WML grade ≥3 for a 1 SD increase in SBP and DBP. As can be seen in Table 5⇑, in almost all the strata of hypertension classification, higher levels of both SBP and DBP were significantly associated with higher odds of WML, and most of these associations were statistically significant. No meaningful ethnic differences were detected in these patterns. Comparing the standardized ORs for SBP and DBP, it is apparent that SBP is stronger in its association with WML than is DBP. We also estimated the standardized ORs and 95% CIs of WML in hypertensives, adjusted for the estimated duration and treatment and control of hypertension. For a 1 SD increase in SBP, the associated odds of WML are higher by 43% in contrast to 16% higher odds of WML for 1 SD increase in DBP. The above associations of blood pressure and WML were independent of age, ethnicity, sex, smoking, diabetes mellitus, total cholesterol, and the duration, treatment, and control of hypertension. These findings provide compelling analytic evidence that the level of blood pressure, especially SBP level, plays an important role in the development of WML.
The role of SBP in the development of WML was examined in the entire study population, stratified by the usage of antihypertensive medication status (Fig 2⇑). The similar slopes in higher levels of SBP (>125 mm Hg) indicate that the level of blood pressure is one of the key factors in relation to the risk of WML regardless of antihypertensive medication usage status. It is evident from these results that the relationship of SBP and WML is continuous across the entire range of blood pressure levels. These results also suggest that lowering blood pressure may reduce the risk of WML at the population level. A significant trend of curvilinear relationship between SBP and WML was found in individuals using antihypertensive medication, with a flat relationship at a level of SBP <125 mm Hg. In addition to the lack of detailed information on the duration, dosage, and types of antihypertensive medication used, this finding needs to be replicated by other studies, and tested by prospective studies, before any definitive conclusions can be made. Several factors may account for such a curvilinear relationship. First, it is possible that individuals using antihypertensive medication to lower their blood pressure to low levels have very intense pharmacological intervention, and the relatively flat relationship between blood pressure and WML may indicate the potential side effects of some antihypertensive medications. Second, in this population-based cross-sectional study, it cannot be ruled out that some individuals with elevated blood pressure took extra doses or a higher dose of antihypertensive medication before their ARIC clinic visits to lower their blood pressure to a level that was lower than their usual blood pressure level. Therefore, their true levels of blood pressure may have been systematically underestimated and an “artificial” flat relationship introduced. The greater predicted probability of WML at all levels of attained blood pressure among individuals using antihypertensive medication compared with those not using medication may represent the longer duration and/or untreated higher level of blood pressure among those on medication, ie, the possibility of greater severity and/or duration of hypertension among those treated.
This study is based on a random sample from an ongoing population-based cohort study, and the sample size is the largest of the published studies. More important, this is the first study to include sufficient numbers of African Americans in the study sample to investigate any meaningful patterns of ethnic differences in the associations of WML with hypertension, its estimated duration, and its treatment and control, and the first to investigate possible etiologic mechanisms of the ethnic difference. In the context of a cross-sectional survey, one needs to consider that the African Americans included in this study are survivors of a population characterized by a higher cerebrovascular disease risk profile than European Americans and a higher mortality in younger age than comparably aged European Americans.20 Thus, the ethnic differences found in this study may be underestimated.
Our results are based on a cross-sectional sample; therefore, the temporal relationship between elevated blood pressure and WML cannot be assessed, nor can the effects of selective survival be ruled out. It is unlikely, however, that the selection of individuals through refusal to participate in the examination would play an appreciable role in this middle-aged population and that such selection forces would be differentiated by WML. Conservative exclusion criteria were used to screen the MRI study population for this study for reasons of participants' safety. Consequently, the generalizability of our results to individuals with these exclusion criteria is in doubt. In this study, a scale of 0 to 9 was used to classify WML, and because of the relatively young age of the participants, few participants had WML grade >4. In analyzing these data, we reclassified participants into four WML categories (relatively normal, mild, moderate, and severe) to evaluate the dose-response relationships between hypertension, its duration, and its treatment and control and WML. The relatively severe WML in this study may not be equivalent to the severity of cases of WML seen in symptomatic patients in clinical settings. However, it is unlikely that the significant statistical associations for WML found in our data are the result of scaling, since similar patterns were present when analyses were repeated with a different choice of cut points.
In summary, this population-based study indicates that MRI-defined WML are associated with hypertension, its duration, and its treatment and control; these associations are attributable to the levels of blood pressure, especially SBP. On the basis of preliminary studies available to date, it can be speculated that a possible prevention of the development of WML, by the treatment and/or prevention of hypertension, might reduce stroke. Verification of this speculation must await replication of our cross-sectional findings, follow-up study of cohorts, and combination of MRI studies with clinical trials.
Selected Abbreviations and Acronyms
|ARIC||=||Atherosclerosis Risk in Communities|
|DBP||=||diastolic blood pressure|
|SBP||=||systolic blood pressure|
|WML||=||white matter lesion(s)|
Support for this study was provided by National Heart, Lung, and Blood Institute contracts N01-HC-55018, N01-HC-55019, N01-HC-55021, and N01-HC-35126. The authors wish to acknowledge the valuable contributions made by the ARIC staff at the collaborating institutions: The University of North Carolina at Chapel Hill (Chapel Hill)—Louis Wijnberg, Tom G. Goodwin, Myra A. Carpenter, Doris L. Jones, Catherine Paton, and Phyllis H. Johnson; The University of Mississippi Medical Center (Jackson)—Bobbie Alliston and Jane G. Johnson; The Johns Hopkins University (Baltimore, Md)—Melanie M. Hawkins and Patrice Holtz; and Bowman-Gray School of Medicine (Winston-Salem, NC)—Jane Jensen and Dawn Scott.
Reprint requests to Dr Duanping Liao, Department of Epidemiology, School of Public Health, University of North Carolina at Chapel Hill, 137 E Franklin St, NationsBank Plaza, Suite 306, Chapel Hill, NC 27514. E-mail Duanping_liao@unc.edu.
- Received May 22, 1996.
- Revision received August 16, 1996.
- Accepted August 22, 1996.
- Copyright © 1996 by American Heart Association
Awad IA, Johnson PC, Spetzler RF, Hodak JA, Awad CA, Carey R. Incidental subcortical lesions identified on magnetic resonance imaging in the elderly, I: correlation with age and cerebrovascular risk factors. Stroke. 1986;17:1084-1089.
George AE, de Leon MJ, Kalnin A, Rosner L, Goodgold A, Chase N. Leucencephalopathy in normal and pathologic aging, II: MRI of brain lucencies. AJNR Am J Neuroradiol. 1986;7:567-570.
Forsting M, Hacke W, Sartor K. The spectrum of subcortical lesions in MRI sensitivity and specificity. J Neural Transm. 1991;33:21-26.
van Swieten JC, van den Hout JH, van Ketel BA, Hijdra A, Wokke JH, van Gijn J. Periventricular lesions in the white matter on magnetic resonance imaging in the elderly: a morphometric correlation with arteriolosclerosis and dilated perivascular spaces. Brain. 1991;114:761-774.
Breteler MM, van Swieten JC, Bots ML, Grobbee DE, Claus JJ, van den Hout JH, van Harskamp F, Tanghe HL, de Jong PT, van Gijn J, Hofman A. Cerebral white matter lesions, vascular risk factors, and cognitive function in a population-based study: the Rotterdam Study. Neurology. 1994;44:1246-1252.
Manolio TA, Kronmal RA, Burke GL, Poirier V, O'Leary DH, Gardin JM, Fried LP, Steinberg EP, Bryan RN. Magnetic resonance abnormalities and cardiovascular disease in older adults: the Cardiovascular Health Study. Stroke. 1994;25:318-327.
ARIC Investigators. The atherosclerosis risk in the communities (ARIC) study: design and objectives. Am J Epidemiol. 1989;129:687-702.
Bryan RN, Manolio TA, Schertz LD, Jungreis C, Poirier VC, Elster AD, Kronmal RA. A method for using MR to evaluate the effects of cardiovascular disease on the brain: the Cardiovascular Health Study. AJNR Am J Neuroradiol. 1994;15:1625-1633.
National Heart, Lung, and Blood Institute, National Institutes of Health. The ARIC Manuals of Operation. Chapel Hill, NC: ARIC Coordinating Center, School of Public Health, University of North Carolina.
Bradley WG, Waluch V, Brant-Zawadzki MN. Patchy periventricular white matter lesions in the elderly: a common observation during NMR imaging. Noninvas Med Imaging. 1984;1:35-41.
Gerard G, Weisberg LA. MRI periventricular lesions in adults. Neurology. 1986;36:998-1001.
Liao D, Cooper L, Cai J, Toole J, Bryan N, Shahar E, Mosley T, Nieto J, Heiss G. The prevalence and severity of white matter lesions, their relationship with age, ethnicity, gender, and cardiovascular disease risk factors: the ARIC Study. Neuroepidemiology. In press.
Cooper R, Sempos C, Hsieh SC, Kovar MG. Slowdown in the decline of stroke mortality in the United States, 1978-1986. Stroke. 1990;21:1274-1279.