Incidence, Manifestations, and Predictors of Worsening White Matter on Serial Cranial Magnetic Resonance Imaging in the Elderly
The Cardiovascular Health Study
Background and Purpose— Magnetic resonance imaging (MRI) scans in the elderly commonly show white matter findings that may raise concerns. We sought to document incidence, manifestations, and predictors of worsening white matter grade on serial imaging.
Methods— The Cardiovascular Health Study is a population-based, longitudinal study of 5888 people aged 65 years and older, of whom 1919 have had extensive initial and follow-up evaluations, including 2 MRI scans separated by 5 years. Scans were read without clinical information in standard side-by-side fashion to determine worsening white matter grade.
Results— Worsening was evident in 538 participants (28%), mostly (85%) by 1 grade. Although similar at initial scan, participants with worsening white matter grade, compared with those without, experienced greater decline on modified Mini-Mental State examination and Digit-Symbol Substitution test (both P≤0.001) after controlling for potential confounding factors, including occurrence of transient ischemic attack or stroke between scans. Independent predictors of worsening white matter grade included cigarette smoking before initial scan and infarct on initial scan. Otherwise, predictors differed according to white matter grade on initial scan. For low initial grade, increased age, increased diastolic blood pressure, increased high-density lipoprotein cholesterol, and decreased low-density lipoprotein cholesterol were associated with increased risk of worsening. For high initial grade, any cardiovascular disease and low ankle–arm index were associated with decreased risk of worsening, whereas use of diuretics and statins were associated with increased risk.
Conclusion— Worsening white matter grade on serial MRI scans in elderly is common, is associated with cognitive decline, and has complex relations with cardiovascular risk factors.
White matter findings on cranial magnetic resonance imaging (MRI) scans are common in the elderly and can be associated with cognitive impairments and increased risk of future stroke.1 Prevention of these white matter findings might reduce risk of these outcomes. Most population-based studies examining clinical manifestations and risk factors for these findings have used cross-sectional designs. Few longitudinal studies with serial MRI exist. In the Cardiovascular Health Study (CHS), a population-based longitudinal study of vascular disease, 1919 participants underwent 2 MRI scans separated by 5 years. We sought to document incidence, manifestations, and predictors of worsening white matter grade, paralleling the approach taken in a CHS article on incident MRI-defined infarcts.2
Members of the CHS cohort were recruited from random samples of people on Medicare eligibility lists for 4 US communities, and 5888 provided informed consent and were enrolled in CHS. Participants had to be 65 years or older and not institutionalized, wheelchair-bound in home, or undergoing active treatment for cancer. Between 1991 and 1994 and again ≈5 years later, participants were invited to undergo MRI scanning. Time between participants’ initial and follow-up scan varied between 3.2 and 7.5 years, with median and mean being 5.0 years. Importantly, 3660 (62%) participants who underwent initial scan were healthier than those who were never scanned, and 2116 (36%) participants who underwent 2 scans were healthier than those who underwent single scan.2
Participants underwent extensive baseline evaluations including standard questionnaires, physical examination, performance measures, and laboratory testing. Parts of baseline evaluations have been repeated annually. Variables considered in these analyses as performance measures and potential risk factors were those from examination closest in time and before initial scan, as detailed previously.2 Performance measures at follow-up scan were those closest in time to scan, either before or after. All participants were screened for history of transient ischemic attack or stroke at baseline and on follow-up. Those with such events were not excluded from these analyses. Participants were also screened for other vascular disease at baseline and during follow-up and were said to have any cardiovascular disease if they had one or more of transient ischemic attack, stroke, angina, myocardial infarction, congestive heart failure, or claudication before initial scan.
Scanning protocol included sagittal T1-weighted localizer images and axial T1, spin-density, and T2-weighted images. Axial images had 5-mm thickness without interslice gaps.3 Without knowledge of any clinical information, neuroradiologists at the reading center estimated white matter, ventricular, and sulcal grades using a 10-point system, from 0 to 9 (most abnormal), using a library of templates.4 An original change score was calculated by subtracting grade on initial scan from grade on follow-up scan, which had been read independently. We reasoned that reliability would improve if pairs of scans were read together, so readers re-read all scans side-by-side without knowing grades from previous readings or order of scans. Based on these re-reads, intrareader reliability for change of 1 or more grades yielded kappa of 0.59, and inter-reader reliability yielded kappa of 0.36. Finally, 2 readers adjudicated samples of scans in final side-by-side comparison to yield measure of change used in these analyses. Samples included all pairs of scans for which revised change score was negative; revised change score was 1 or more but original change score was negative; and revised change score was zero but original change score was 3 or more. Brain infarct was defined as area of abnormal signal intensity in vascular distribution that lacked mass effect.2 Because of technical problems, 197 of original 2116 pairs of scans could not be re-read, leaving 1919 (91%) pairs. Demographics, cardiovascular risk factors, and prevalent cardiovascular disease were similar for these 197 and 1919 participants (data not shown).
We compared change in performance on cognitive and motor tests from initial to follow-up scan across 3 levels of worsening white matter grade (none, 1, or 2 or more) using analysis of covariance with probability value for linear trend, controlling for age, sex, education, performance at initial scan, white matter grade on initial scan, and occurrence of transient ischemic attack or stroke between scans. Total change was difference in scores at initial and follow-up scans. Average annual change was slope of regression line of score on year for each participant, using all scores available at each year from initial to follow-up scan. Variable for time between scans was also included in models for total change.
The list of potential risk factors for worsening white matter that we examined is available online (Table I at http://www.strokeaha.org) and included the same 68 variables examined previously2 and 2 new variables created by averaging all values measured before initial scan, separately, for diastolic and systolic blood pressure. We evaluated bivariate associations with 3 levels of worsening with analysis of variance or χ2, as appropriate. In exploratory multivariable analyses, we used forward stepwise logistic regression (P<0.05 to enter and P>0.10 to remove) to identify predictors of worsening white matter by 1 or more grades, adjusting in all models for age, sex, white matter grade at initial scan, and time between scans. We report results from models rerun with just variables selected to maximize sample size. Interactions between all pairs of predictors in models were tested. We found significant interactions involving white matter score at initial scan with age and averaged diastolic blood pressure before initial scan (both P=0.003); thus, subsequent analyses were stratified by white matter grade at initial scan: 0 or 1 (n=952) versus 2 to 9 (n=967).
SPSS for Windows (version 11; SPSS Inc) was used for analyses, which were based on the updated CHS database incorporating minor corrections through January 2003.
Of 1919 pairs of scans, 538 (28%) showed worsening—mostly (85%) by 1 grade (Figure 1 and online Table II, available at http://www.strokeaha.org). Although initial scores on performance measures were similar across levels of worsening white matter, scores on modified Mini-Mental State examination5 and Digit-Symbol Substitution test6 deteriorated significantly more in those with than without worsening, whether considering total change or average annual change (Table 1). Performances on these 2 tests are shown in Figure 2 for each year of study from initial to follow-up scans. Associations in Table 1 remained significant when presence of infarct on follow-up scan was added to models, and infarct was marginally significant only in the model for total change in modified Mini-Mental State examination (P=0.055). This 100-point scale fell 8 points in 33 participants with worsening by 2 or more grades and infarcts on follow-up scan. In sensitivity analyses, associations in Table 1 remained significant when worsening ventricular grade was added to models, with the exception of average annual change for modified Mini-Mental State examination (P=0.088). For gait speed calculated from time to walk 4.6 m (15 ft), associations were present but not as strong as for cognitive measures (Table 1). Associations were lacking for depression score and finger tapping in either hand (data not shown).
Nine of 70 potential risk factors examined2 were significantly related to worsening (Table 2). Also listed are age, sex, and time between scans. Associations were strongest for cigarette smoking before initial scan and infarcts and white matter grade on initial scan (Figure 1). Report of dizziness on standing was associated but not decrease in systolic or diastolic blood pressure when orthostatic blood pressures were measured. Average diastolic blood pressure before initial scan was associated, although diastolic blood pressure from immediately before initial scan was not. Higher factor VII activity was associated with worsening. Unexpectedly, higher levels of high-density lipoprotein cholesterol, lower levels of low-density lipoprotein cholesterol, and lower body mass index were associated with worsening white matter grade.
Because of significant interactions involving white matter score at initial scan with age and averaged diastolic blood pressure before initial scan (both P=0.003), 2 multivariable models, both controlling for age, sex, and time between scans, are presented in Table 3: one for low and one for high white matter grade on initial scan. Only cigarette smoking and infarct on initial scan were common to both models. When grade on initial scan was low, increased age and increased average diastolic blood pressure were associated with increased risk of worsening white matter grade. Again, higher high-density lipoprotein cholesterol and lower low-density lipoprotein cholesterol were also both associated with increased risk. Having any epsilon 4 alleles of apolipoprotein E genotype trended toward increased risk (P=0.052). When grade on initial scan was high, use of diuretics and use of statins before initial scan were associated with increased risk of worsening, whereas having any cardiovascular disease and low ankle–arm index (<0.9) were associated with reduced risk of worsening. Results were largely unchanged when infarct on initial scan was not allowed to enter multivariable models or when ordinal logistic regression was used considering worsening in 3 rather than 2 levels. Excluding participants who experienced transient ischemic attack or stroke between scans (n=100, including 59 with incident and 7 with recurrent strokes) did not substantially alter results in Table 2 or Table 3.
In this cohort of elderly participants willing to undergo 2 cranial MRI scans separated by 5 years, worsening white matter by 1 or more grades occurred commonly. Over 5 years between scans, 3% experienced stroke, whereas 28% had worsening white matter grade. These numbers suggest that the burden of cerebrovascular disease in the elderly is far greater than indicated by stroke alone. In addition, worsening white matter cannot be considered benign or silent. Cognitive decline was greater in those with than without worsening, despite controlling in analyses for several potential confounding factors. Assuming linear decline in average change of modified Mini-Mental State examination scores, participants with worsening white matter would experience in only 5 years the same cognitive decline that those without worsening white matter would experience over 10 years (worsening by 1 grade) or 19 years (2 or more grades).
Relations between potential risk factors and worsening white matter were complex. In multivariable models, associations with worsening white matter differed based on low or high white matter grade on initial scan, except for cigarette smoking and infarct on initial scan—the only variables that entered both models. Constituents of cigarette smoke, such as nicotine, carbon monoxide, and oxidant gases, could injure white matter by adversely affecting blood pressure, oxygen availability, and clotting.7 For stratum with low white matter grade, increased age and increased diastolic blood pressure were associated with worsening. Otherwise, associations were found in unanticipated directions for levels of high-density lipoprotein cholesterol and low-density lipoprotein cholesterol before initial scan. Risk factors for small vessel disease of brain may differ from those for large vessel disease.
For stratum with high white matter grade, similarly difficult to explain associations were found for any cardiovascular disease, diuretic use, statin use, and ankle–arm index. Bias could result from differential survival if participants with high white matter grade on initial scan and with any cardiovascular disease or low ankle–arm index were unlikely to survive 5 years if they also had worsening white matter. Any cardiovascular disease or low ankle–arm index before initial scan paradoxically would seem protective against worsening white matter. Use of diuretics or statins might increase chances of surviving 5 years to follow-up scan and thus be associated with increased risk of worsening. Alternatively, participants with recognized cardiovascular disease before initial scan, compared with those without, might have had more aggressive treatment between scans and thus less worsening. Additional studies will be needed to explore these possibilities. Finally, findings may reflect some other artifact resulting from select nature of participants studied, being healthier in general than other participants in CHS.2
Comparisons are difficult among studies that have examined worsening white matter with serial MRI scans because of differing sizes (n=10 to 366) and differing methods to quantify worsening, manifestations, and potential risk factors.8–17 With respect to clinical manifestations, 3 were unable8–10 but 111 was able to find significant associations with cognitive decline. Another suggested associations with worsening gait and balance,12 as we did. Potential risk factors have been sought in the Austrian Stroke Prevention Study, which included 296 volunteers who had up to 3 MRI scans over 6 years.9,13 Age and arterial hypertension were important predictors of worsening, but white matter lesion grade on initial scan dominated multivariable models. Two studies found that increased diastolic blood pressure was associated with worsening,9,14 as in the current study, although another suggested stronger associations for systolic blood pressure.15 In the National Heart, Lung, and Blood Insitute Twin Study, age and brain infarct were associated with worsening white matter,16 whereas in 2 other studies, age was.17,18 In one of these studies,18 self-reported diabetes was associated with increased risk of worsening, whereas self-reported heart disease was associated with reduced risk, as we found. Although formal statistical interactions have not been described previously, one study in patients with history of transient ischemic attack or stroke suggested associations differed by initial white matter severity,19 as we found. In this substudy of 226 patients in randomized trial of blood pressure lowering, protection from worsening with active treatment was greater in those with more severe initial white matter lesions.
This study has its strengths but also weaknesses. Participants in CHS, and especially those included in these analyses, are healthier than the general population of elderly people. Consequently, these findings may not apply to all elderly people. MRI–pathologic correlations are lacking. Risk factor status could have changed between initial scan and worsening, which happened at some unknown time between scans. Change in risk factor status between scans could be examined, but more frequent brain imaging than in the current study would be needed to assure that change in risk factor status came before and not after change in white matter grade. Automated techniques, not used in this study, may have resulted in more precise measures of change in white matter grade. Finally, multiple comparisons were performed in these exploratory analyses, and findings may be attributable to chance. Nonetheless, this study is one of the largest to address these issues, and its participants are characterized with wealth of prospectively collected information on many cardiovascular risk factors.
In this population of generally health elderly people, worsening white matter grade was common and associated with accelerated cognitive decline. Risk factors for worsening differed based on white matter grade on initial scan, except for cigarette smoking and infarct on initial scan. Unexpectedly, presence of several factors associated with atherosclerosis was associated with reduced risk of worsening. Additional studies of serial MRI scans, ideally with quantitative assessment in large numbers of elderly participants, will be needed to clarify these findings and to seek means to prevent worsening white matter grade and associated cognitive decline. Investigators should consider including serial MRI and cognitive assessment in trials aiming to control risk factors and prevent vascular disease of heart and brain.
Contracts N01-HC-85079 through N01-HC-85086, N01-HC-35129, and N01-HC-15103 from the National Heart, Lung, and Blood Institute supported the research reported in this article. For a full list of participating investigators and institutions in the Cardiovascular Health Study, see About CHS: Principal Investigators and Study Sites at: http://chs3.chs.biostat.washington.edu/chs/.
- Received June 19, 2004.
- Revision received October 6, 2004.
- Accepted October 12, 2004.
Longstreth WT Jr, Dulberg C, Manolio TA, Lewis MR, Beauchamp NJ Jr, O’Leary D, Carr J, Furberg CD. 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.
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. Am J Neuroradiol. 1994; 15: 1625–1633.
Longstreth WT Jr, Manolio TA, Arnold A, Burke GL, Bryan N, Jungreis CA, Enright PL, O’Leary D, Fried L. Clinical correlates of white matter findings on cranial magnetic resonance imaging of 3301 elderly people. The Cardiovascular Health Study. Stroke. 1996; 27: 1274–1282.
Salthouse TA. The role of memory in the age decline in digit-symbol substitution performance. J Gerontol. 1978; 33: 232–238.
Schmidt R, Fazekas F, Kapeller P, Schmidt H, Hartung HP. MRI white matter hyperintensities: three-year follow-up of the Austrian Stroke Prevention Study. Neurology. 1999; 53: 132–139.
Mascalchi M, Moretti M, Della Nave R, Lolli F, Tessa C, Carlucci G, Bartolini L, Pracucci G, Pantoni L, Filippi M, Inzitari D. Longitudinal evaluation of leukoaraiosis with whole brain ADC histograms. Neurology. 2002; 59: 938–940.
Daffertshofer M, Sedlaczek O, Oster M, Hennerici M. MR quantification of white matter lesions in follow-up study of patients with subcortical vascular encephalopathy. Cerebrovasc Dis. 2001; 11 (Suppl 4): 2. (Abstract).
Whitman GT, Tang Y, Lin A, Baloh RW, Tang T. A prospective study of cerebral white matter abnormalities in older people with gait dysfunction. Neurology. 2001; 57: 990–994.
Veldink JH, Scheltens P, Jonker C, Launer LJ. Progression of white matter hyperintensities on MRI is related to diastolic blood pressure. Neurology. 1998; 51: 319–320.
DeCarli C, Swan GE, Wolf PA, Carmelli D. Longitudinal changes in brain and white matter hyperintensity volumes among elderly male twins from the NHLBI Twin Study. Neurology. 2002; 58 (Suppl 3): A399. (Abstract).
Dufouil C, Mazoyer B, Chalmers J, Tzourio C. Effects of blood pressure lowering on incidence of cerebral white matter lesions in patients with stroke. Neurology. 2004; 62 (Suppl 5): A172. (Abstract).