A study of periventricular hyperintensity. I. Normal brain aging

Arch. Gerontol. Geriatr., 14 (1992) 183-191

183

© 1992 Elsevier Science Publishers B.V. All rights reserved. 0167-4943/92/$05.00

AGG 00435

A study of periventricular hyperintensity. I. Normal brain aging Kenichi Meguro a, Yasuyoshi Sekita b, Tatsuo Yamaguchi a, Kenji Y a m a d a a, Takashi Hishinuma a and Taiju Matsuzawa a aDepartment of Radiology and Nuclear Medicine, The Research Institute for Tuberculosis and Cancer and bDepartment of Hospital and Medical Care Administration, Tohoku University School of Medicine, 1-1 Seiryo-machi, Aoba-ku, Sendai 980 (Japan) (Received 17 July 1991; revised version received 7 December 1991; accepted 9 December 1991)

Summary Fifty-two patients with cerebrovascular risk factors without neurological abnormalities were examined by magnetic resonance imaging and were evaluated for their periventricular hyperintensity (PVH) on T2-weighted images. We assumed that PVH was not a mere focal finding of the brain but a kind of marker for condition related to arteriosclerosis and cerebral ischemia, and we tried to devise a model screening test, using common parameters available in most ordinary hospitals, to predict PVH. Multiple regression analysis was performed by setting up the PVH% (the volume percentage of PVH to cranial cavity) as a dependent variable and twenty-seven variables associated with general medical examination and brain atrophy as explanatory variables. We found that arteriosclerotic changes in the body as well as brain atrophy were significantly correlated with PVH, and that PVH could be predicted with a high contribution ratio of 0.70. It is clinically important to examine the elderly with our screening test to predict PVH in order to detect the early stages of ischemia. Periventricular hyperintensity (PVH); Magnetic resonance imaging (MRI); Brain atrophy; Arteriosclerosis; Risk factors for stroke

Introduction Periventricular hyperintensity (PVH), which is detected by T2-weighted images of magnetic resonance imaging (MRI), has been found in the brains of the elderly (Awad et al., 1982; Bradley et al., 1984; Brant-Zawadzki et al., 1985; George et al., 1986a, Gerard et al., 1986b; Sarpel et al., 1987; Zimmerman et al., 1986). Previous studies have also suggested that PVH is correlated with cerebral ischemia since patients with cerebral ischemic attack or those with risk factors for stroke have more Correspondence to: Kenichi Meguro, Department of Geriatric Medicine, Tohoku University School of Medicine, 1-1 Seiryo-machi, Aoba-ku, Sendai 980, Japan.

184 severe PVH than healthy subjects (Awad et al., 1986a,b: Rothrock el al., 1987: Lechner et al., 1988). We have previously reported that elderly patients having risk factors for stroke but not suffering cerebral ischemic attack showed various grades of PVH in spite of normal neurological and neuroradiological findings (Meguro et al., 1990). Using positron emission tomography (PET), we also proved that such patients had a norreal range for oxygen metabolism, notwithstanding the correlation between the degree of cerebral circulation impairment and the grade of PVH, because of a compensatory increase of oxygen extraction fraction (Meguro et al.. 1990). As PVH has been found to be an indicator of cerebral ischemia. MRI, a noninvasive method, is useful for examining the brains of the elderly. In this study, we assumed that PVH is not a mere focal brain change but a kind of general marker of condition related to arteriosclerosis and cerebral ischemia, and tried to devise a model screening test for the prediction of PVH using more common parameters. Patients and Methods Patient selection

Fifty-two patients, 25 males and 27 females, were selected from the population participating in a medical screening test at our clinic. Their ages ranged from 65-88 with a mean age of 72. Informed consent was obtained for all patients selected in this study. The patients were interviewed as to medical history and underwent physical and neurological examinations, and the Japanese version of the MiniMental State Examination (MMSE). Blood samples were obtained, and ECG, chest X-ray, brain CT scan and M R ! were performed. The criteria for patient selection were as follows: they had not suffered any cerebral ischemic episodes although they had some risk factors for stroke; their neurological examinations and brain CT scans were normal; no cognitive impairment was evident as shown by the Japanese version of MMSE; there were no diseases other than risk factors for stroke, i.e., no pulmonary, gastrointestinal, motor system, or psychiatric diseases, etc. Risk factors for stroke in this study included hypertension, hyperlipidemia, hyperglycemia, a history of smoking and E C G abnormalities. Hypertension and hyperlipidemia were diagnosed based on World Health Organization (WHO) criteria. Hyperglycemia was defined as 160 mg/dl or over of fasting blood glucose. A positive smoking history was defined as more than 10 cigarettes per day for more than 10 years. ECG abnormalities included left ventricular hypertrophy, bundle branch block, and ST-T changes without symptoms. Patients with atrial fibrillation were excluded. MRI

The M R I used was the Bruker 1000 J system (Karlsruhe, Germany) with a resistive magnet operating at 0.14 T and a resonance of 6 MHz for proton imaging. Pulse sequence used was the Carr-Purcell-Meiboom-Gill (CPMG) method (Yamada et al., 1989), which is a spin echo method. It consisted of 12 echos and the first echo was used for the T 1-weighted image and the summation of the 1st through 12th echo

185

Fig. 1. T2-weighted magnetic resonance images of the patient with 0.6% of PVH Index.

was used for the T2-weighted image because PVH was clearly detectable in this image. The actual T R and T E values were 600 ms (recovery time) and 34 ms, respectively. T2-weighted images o f O M + 50, +70 m m were used for detecting PVH; the P V H index, the volume percentage o f P V H (Yamada et al., 1987) to cranial cavity

Fig. 2. T2-weighted magnetic resonance images of the patient with 6.6% PVH Index.

186

Fig. 3. T2-weighted magnetic resonance images of the patient with 18.0".~,PVH Index.

was calculated by computer. Figures 1-3 demonstrate the various grades of PVH in these images. Similarly, in order to evaluate brain atrophy, the volume percentages of brain (BVI%, Brain Volume Index), sulci (Sulcus%) and lateral ventricles (Ventricle"/,,) to the cranial cavity were calculated from Tl-weighted images.

Hypothesis We assumed that PVH is not a mere focal brain change but a kind of marker of general condition related to arteriosclerosis and cerebral ischemia because PVH always emerges bilaterally and symmetrically and increases with age. Therefore we investigated whether PVH can be predicted by some of the more common parameters associated with arteriosclerosis and cerebral ischemia such as routine medical examinations including ECG, cerebrovascular risk factors, arterioscrelosis, and brain atrophy; we tried to devise a model screening test which could facilitate detection of cerebral ischemia in the early stages and thus facilitate more effective treatment.

Statistical analyses First of all, we set up the parameters described below for explanatory variables and PVH% as a dependent variable based on the hypothesis described above. After examining the correlation between each explanatory variable and PVH%, the significant variables were set for the next step, i.e., multiple regression analysis.

Dependent variable and explanatory variables. Statistical analyses were performed by setting up the PVH°/,, as a dependent

187

variable and the variables described below as explanatory variables: 1, age (year); 2, sex; 3, height (cm); 4, weight (kg); 5, cardio-thoracic ratio, CTR > 50°/,,; 6, calcification of the wall of the aortic arch; 7, hypertension; 8, anti-hypertensive drugs (types); 9, anti-hypertensive drugs (numbers); 10, hyperglycemia; 1 l, hyperlipidemia; 12, TC (total cholesterol); 13, HDL (high density lipoprotein); 14, LDL (low density lipoprotein); 15, beta-lipoprotein; 16, TG (triglycerid); 17, ECG abnormality; 18, smoking (numbers of cigarettes per day); 19, smoking years; 20, smoking (the product of factor 17 by factor 18); 21, number of risk factors; 22, haemoglobin (Hb); 23, haemoconcentration (Hct); 24, Pit (platelet number); 25, BVI%; 26, Sulcus%; 27, Ventricle%. Variables 5 and 6 were obtained from chest X-ray film, variable 25 from CT, and variables 26 and 27 from MRI images. Variables 1, 3, 4, 9, 12-16, and 18-27 were absolute quantities. Variables 2, 5-8, 10, 11, 17 were not absolute quantities, so they were assigned a positive number such as 1, 2 and 3. For variable 2, males were assigned a value of 1 and females a value of 2. For variable 5, CTR > 50% was assigned a value of 1 and CTR < 50% was assigned a value of 2. For variable 7, hypertension was assigned a value of I and no hypertension (WHO criteria described above) was assigned a value of 2. Correlation between PVH% and each explanatory variable. First of all, the correlation coefficients of PVH% and each explanatory variable described above were calculated and significant variables were chosen for the next step in the analysis. Multiple regression analysis. After determining the significant explanatory variables~ multiple regression analysis was performed. Log transformation was performed for factor 25 and logistic transformation was performed for PVH% and the same statistical analysis was used.

Results The results of multiple regression analysis are noted in Table I. The contribution ratio (R 2) was 0.40 only when the explanatory variable was variable 25 (BVI%). R 2 became 0.67 when variables 1 (age), 6 (calcification of the wall of the aortic arch), 25 and 27 (Ventricle%) were defined as explanatory variables. R 2 increased up to 0.70 when the following variables were considered as explanatory variables: variables 1, 4 (CTR > 50%), 6, 7 (hypertension), 11 (hyperlipidemia), 17 (ECG abnormality), 20 (smoking), 25 and 27. The summary of the findings was shown in Table II. Almost the same results were obtained after log transformation of variable 25 and logistic transformation of PVH% (data not shown).

Discussion In this study we found that PVH percentages could be predicted with a contribution ratio of 0.67 by 4 of the explanatory variables, and a contribution ratio of 0.70 by 9 variables. The contribution ratio of 0.70 is statistically very high because the

188 TABLE l Multiple regression analysis Contribution ratio R2

0.40 0.42 0.42 0.42 0.41 0.67 0.67 0.67 0.68 0.68 0.68 0.69 0.70

Multiple correlation coefficient R*

R

0.64 0.64 0.65 0.65 0.64 0.82 0.82 0.82 0.82 0.82 0.83 0.83 0.83

0.63 0.63 0.62 0.61 0.59 0.81 0.81 0.80 0.80 0.79 0.79 0.79 0.79

Explanatory variables

F-ratio

DOE

25 b 1.25 a 1,2a,27

33.99 17.44 I 1.4(t 8.38 6.00 45.83 30.87 22.64 I 7.89 14.69 12.51 10.99 9.68

51 51 51 51 48 48 48 48 48 48 48 47 47

1,7,25a,27 1,5,7,25~t,27 6b,25 a 1,6b,25 '~ [,6b,25a,27 1,6b,7,25a,27 1,5,6b,7,25~k27 [,5,6,7,20,25d.27

1,5,6b,7,11,20,25",27 1,5,6b,7,11,17.20,25~L27

aSignificant at 0.05 level, bSignificant at 0.01 level. DOF, degree of freedom.

T A B L E 11 Summary of the results. Table II demonstrates that the PVH Index can be predicted by nine variables: age, calcification of the wall of the aortic arch, cardio-thoracic ratio > 50%, hypertension, hyperlipidemia, ECG abnormality, a history of smoking, Brain Volume Index, and Ventricle Index (see text and Table l).

Chest Xl) f illdillgs

f~ ~_~

J C a ] c i f i l a l i o n iIf the wail ] l,, u ...... -t~ ..... h 1

[Ta,d,o~i[o,-aD,7-at,o>50,.-] I.... ]

for strroke

Bvah~ atrophy

f ~ - i 77

[ Brain Vo}unte Index ] { Ventric]e

index]

189

correlation coefficient, which is the square root of the contribution ratio, is over 0.83. Such variables as variables 1, 5, 6, 7, 11, 17, and 20 were obtained from general medical examinations. Variables 25 and 27 in this study were obtained by MRI but ventricular volume as well as brain volume as shown by MRI were similar to values obtained by CT scan (Yamada et al., 1987). Therefore, this study indicated that the findings of general medical examinations and CT scan, available in most ordinary hospitals, can be used to predict PVH percentages as shown by MRI, and thus demonstrated that the screening test based on the above parameters is clinically important. Because PVH is manifested by physiological as well as neuroradiological findings, the hypothesis is proved. Namely, PVH is not a mere focal brain change but a kind of marker of the general condition related to arteriosclerosis and brain atrophy. This is supported by the fact that PVH always appears bilaterally and symmetrically and increases with age. The relationships between PVH and the above factors are discussed below in turn. Variable 1, age, is a commonly recognized factor related to PVH (Awad et al., 1982; Bradley et al., 1984; Brant-Zawadzki et al., 1985; George et al., 1986; Gerard et al., 1986; Sarpel et al., 1987; Zimmerman et al., 1986). Variables 7 (hypertension), 11 (hyperlipidemia), 17 (ECG abnormality), and 20 (history of smoking) are risk factors for stroke. It is well known that patients with hypertension show PVH (Awad et al., 1986; MacQuinn and O'Leary, 1987). Although other variables have not been reported individually as affecting PVH, it is reasonable that the sum of such risk factors for stroke decreases cerebral blood flow (smoking (Rogers et al., 1983), hyperlipidemia (Meyer et al., 1987), heart disease (Meyer et al., 1987), etc.) and that PVH reflects such ischemia (Lechner et al., 1988). Atherosclerosis might be related to PVH as indicated by variable 5 (CTR > 50°/,,) and variable 6 (calcification of the wall of the aortic arch). One previous theory (Babikian and Ropper, 1987) have shown that atherosclerosis is correlated with PVH, and other theories (Roman, 1987) have disclosed that arteriosclerosis of the penetrating long medullary arteries supplying the white matter around the lateral ventricles caused PVH. In our population, both of these conditions might be related to PVH. As far as variable 25 (Brain Volume Index) and variable 27 (Ventricle Index) are concerned, both of these variables in this study are obtained from MRI but each of these variables can also be obtained from CT scan. These variables suggested that impaired cerebrospinal fluid circulation might also be related to PVH pathogenesis. Variables not correlated with PVH in this study are variables 2 (sex), 3 (height), 4 (weight), 10 (hyperglycemia), 22 (Hb), 23 (Hct) and 24 (Pit no.). There have been no previous reports on gender difference for PVH. No patients in this study showed obesity, so variables 3 and 4 might be masked. No definitive relationship was found between PVH and hyperglycemia nor between PVH and diabetes mellitus. No findings on PVH and haemoconcentration have been reported. In addition to changes due to aging, arteriosclerosis as well as atherosclerosis could be important in PVH pathogenesis. In a comparative study between MRI and postmortem histology by Awad et al. (Awad et al., 1986c), white matter lesions detected by MRI were shown to be caused not by infarction but rather by

190

extracellular fluid retention due to arteriosclerosis, axonal loss, etc. Risk factors for stroke would facilitate such changes and a sum of all such variables might impair cerebral circulation. In this study, we have shown that PVH can be explained by the various markers of arteriosclerosis and/or atherosclerosis, as well as brain atrophy. Therefore, the grade of PVH is an important index for showing such changes. Rowe and Kahn (Rowe and Kahn, 1987) suggested that normal human aging may be subdivided into usual aging (no overt neurologic symptoms) and successful aging (minimal physiologic loss even when compared with younger individuals). Drayer (Drayer, 1988) also suggested that MRI may facilitate the distinction between usual (no neurologic dysfunction) and successful (no brain or vascular changes) aging and that the former has more severe white matter lesions on MRI or more impaired cerbral circulation and metabolism.

Acknowledgement We are grateful to Professors H. Sasaki, K. Kogure and T. Yoshimoto for valuable suggestions.

References Awad, I.A., Modic, M., Little, J.R., Furlan, A.J. and Weinsteim M. (1986a): Focal parenchymal lesions in transient ischemic attacks: correlation of computed tomography and magnetic resonance imaging. Stroke, 17, 399-403. Awad, I.A., Spetzler, R.F., Hodak, J.A., Awad, C.A. and Carey, R. (1986b): Incidental subcortical lesions identified on magnetic resonance imaging in the elderly. Correation with age and cerebrovascular risk factors. Stroke, 17, 1084-1089. Awad, I.A., Johnson, P.C., Spetzler, R.F. and Hodak, J.A. (1986c): Incidental subcortical lesions identified on magnetic resonance imaging in the elderly. II. Postmortem pathological correlations. Stroke, 17, 1090-1097. Awad, I.A., Spetzler, R.F., Hodak, J.A., Awad, C.A. and Williams, F. Jr. (1982): Incidental lesions noted on magnetic resonance imaging of the brain: prevalence and clinical significance in various age groups. Neurosurgery, 20, 222-227. Babikian, V. and Ropper, A.H.(1987): Binswanger's disease: a review. Stroke, 18, 2-12. Bradley, W.G., Waluch, V., Brant-Zawadzki, M., Vadley, R.A. and Wycoff, R.R. (1984): Patchy, periventricular white matter lesions in the elderly: a common observation during NMR imaging. Noninvas. Med. Imaging, 1, 35-41. Brant-Zawadzki, M., Fein, G., Van Dyke, C., Kiernan, R. Davenport, L. and deGroot, J. (1985): MR imaging of the aging brain: patchy white matter lesions and dementia. Am. J. Neuroradiol., 6, 675-682. Drayer, B.P. (1988): Imaging of the aging brain. Part I. Normal findings. Radiology, 166, 785-796. Drayer, B.P. (1988): Imaging of the aging brain. Part l I. Pathologic conditions. Radiology, 166, 797-906. George, A.E., de Leon, M.J., Gentes, C.I., Miller, J., London, E., Budzilovich, G.M., Ferris, S. and Chase, N. (1986a): Leukoencephalopathy in normal and pathologic aging: I. CT of brain lucencies. Am. J. Neuroradiol., 7, 561-566. George, A.E., de Leon, M.J., Kalnin, A., Rosner, L., Goodgold, A. and Chase, N. (1986b): Leukoencephalopathy in normal and pathologic aging: 2. MRI of brain lucencies. Am. J. Neuroradiol., 7, 567-570. Gerard, G. and Weisberg, L.A. (1986): MRI periventricular lesions in adults. Neurology, 36, 998-1000. Lechner, H., Schmidt, R., Bertha, G., Justich, E. and Offenbacher, H. ( 1988): Nuclear magnetic resonance imaging white matter lesions and risk factors for stroke in normal individuals. Stroke, 19. 263-265.

191

MacQuinn, B.A. and O'Leary D.H. (1987): White matter lucencies on computed tomography. Subacute arteriosclerotic encephalopathy (Binswanger's disease) and blood pressure. Stroke, 18, 900-905. Meguro, K., Hatazawa, J., Yamaguchi, T., ltoh, M., Matsuzawa, T., Ono, S., Miyazawa, H. Hishinuma, T., Yanai, K. Sekita, Y. and Yamada, K. (1990): Cerebral circulation and oxygen metabolism associated with subclinical periventricular hyperintensity as shown by magnetic resonance imaging. Ann. Neurol., 28, 378-383. Meyer, J.S., Rogers, R.L., Mortel, K.F. and Judd, B.W. (1987): Hyperlipidemia is a risk factor for decreased cerebral perfusion and stroke. Arch. Neurol., 44, 418-422. Rogers, R.L., Meyer, J.S., Shaw, T.G., Mortel, K.F., Hardenberg, J.P. and Zaid, R.R. (1983): Cigarette smoking decreases cerebral blood flow suggesting increased risk for stroke. J. Am. Med. Assoc., 250, 2796-2800. Roman, G.C.(1987): Senile dementia of the Binswanger type. A vascular form of dementia in the elderly. J. Am. Med. Assoc., 258, 1782-1788. Rothrock, J.F., Lyden, P.D., Hesselink, J.R., Brown, J.J. and Healy, M.E. (1987): Brain magnetic resonance imaging in the evaluation of lacunar stroke. Stroke, 18, 7981-7986. Rowe, J.W. and Kahn, R.L. (1987): Human aging: usual and successful. Science, 237, 143-149. Sarpel, G., Chaudry, F. and Hindo, W. (1987): Magnetic resonance imaging of periventricular hyperintensity in a veterans administration hospital population. Arch. Neurol., 44, 725-728. Yamada, A., Matsuzawa, T., Yamada, K., Yoshioka, S., Ono, S. and Hishinuma, T. (1989): A modified signal intensity equation of Carr-PurcelI-Meiboom-Gill pulse sequence for MR imaging. Tohoku J. Exp. Med., 158, 203-209. Yamada, L., Matsuzawa, T., Ono, S., Kawashima, R., Matsui, H., Yamada, S. and Hishinuma, T. ( 1987): Age-related changes in volumes of the ventricles, sulci and periventricular hyperintensity area: a q uantitative study using magnetic resonance imaging. Jpn. J. Geriatr., 24, 575-579. Zimmerman, R.D., Fleming, C.A., Lee, B.C.P., St-Louis, LA. and Deck, M.D.F. (1986): Periventricular hyper-intensity as seen by magnetic resonance: prevalence and significance. Am. J. Radiol., 146, 443-450.