- Email: [email protected]
Computed tomography in Alzheimer's disease: A longitudinal study
a~aL PSVCn~TaY
383
1991~..'~:38~390
Computed Tomography in Alzheimer's Disease: A Longitudinal Study Alistair Bums, Robin Jacoby, and Raymond Levy
SixG'-three patients sati~'ing NINCDS/ADI~DA criteria for A!zheimer's disease (AD) received neuropsychological tests and computed tomography (CT) scans 12 months apart. Significant deterioration occurred in all the cognitive tests and in the CT measures used. that is. lateral ventricular size. third ventricular size. and cortical atrophy. There was a wide variation in the size of the changes taking place; 14 of 63 patients showe¢~ zo significant change and 6 showed a marked increase in ventricular size. However. neither group differed from the others in any demographic, cognitive, or other CT variables which suggested, on the CT measures used. that no clearly identifiable subgroups of AD were present. Change in CT indices was not related to initial severity of disease. An increase in ventricular size was related to deterioration of cognitive function. ?hese results require further replication. The methodological drawbacks of such studies are discussed.
Introduction Cranial computed tomography (CT) scanning has been extensively used in the investigation of Alzbeimer's dis,~ase (AD). Its main use is in the exclusion of intracranial abnormalities such as tumoi~s, hematomata, and normal pressure hydrocephalus (all of which give rise to a dementia syndrome). The characteristic radiological features of AD include cerebral atrophy and ventricular enlargement but neither is diagnostic of the disorder as both are common accompaniments to normal aging (Jacoby et al 1980; Zatz et al 1982; Takeda and Matsuzawa 1984). The CT scan can differentiate between nonde.,,°.,°a a . . . . ,~a .,~,;o.~,o ~,,~.., U~f f a ~l l ~ ,,~ .~wl~t-,~k~O . . . . ~r~ i i l I h c i l i W V l.h.c.V.i l,,,,l~ U V I O and qh~Wlll1~,lIlll,lh~l.ll IJg~.glWIILO I•n IgllIJ~/clxL Vl [ J L ~ l [.3 . ~ J.L.l l ~ J L;~" I . 3 ~' 1k~L al 1983; Jacoby et al 1980). Though a single CT scan is of limited use, serial sczns on individual patients have been performed in an attempt to improve diagnostic accuracy and to follow the course of the disease. These studies are summarized in Table I. Naguib and Levy (1982b) were the first to perform serial scans of patients with senile dementia of the Alzheimer type (SDAT). Their findings (in 10 patients) suggested that there were two subgroups of patients--one with relatively stable ventricular size and little deterioration of cognitive function over time and the second with a marked increase in vc.ntricular size associated with significant cognitive decline. Gado et al (1983) scanned 45 AD patients 1 year apart. They found a significant degree of vennicular enlargement
From the Section of Old Age Psychiatry. Institute of Psychiatry. De Crespigny Park. Denmark Rill. London. England. Address reprint requests to Alistair Bums. MPhil, MRCP. MRCPSych. Institute of Psychiatry. De C,~spigny Park. Der~-nark Hill. London SES. 8AF. England. Received March 19. 1990; revised August 8. 1990. © 1991 Society of Biological Psychiatry
0006-3223/91/$03.50
384
A. Bums et al
BIOL PSYCHIATRY 1991:29:383-390
Table I. l.ongitudinal CT Scan Cha.eges in Dementia of the Alzheimer Type No. of patients
No. of controls
Mean age of InLtients (yr)
10 45
0 !2
76.2 71.0
18-36 months 12 months
Brinkman et al (1984)
5
0
60.6
15-35 months
Bird et al (1986)
0
27
76.0
62-200 weeks
Luxenbetg ct al (1987)
12 M 6F
12
62.8 (M) 70.8 (F)
6--60 months
De Leon et al (1989)
50
45
71.2
36 months
Study Naguib and Levy (~982b) Gado et al (1983)
Follow-up interval
Findings" Two subgroups emerged Volumetic m c ~ u r e s more sensitive than linear measures. Changes greater in patient group of controls 4 of the 5 patients had increases greater than normative values suggested Initial and increasing ventricular size associated with subtle reduction in cognitive function Rate of lateral ventricnlar enlargement discmninate," male patientsfrom conlmls Ventricular volume increase: 9% in patients, 2% in controls
aExpanded in the text,
to occur over that time which did not occur in nondemented controls. They observed that computer-assisted volumetric measures of cerebrospinal fluid (CSF) space were a more sensitive indicator of change than linear measures. Brinkman and Largen (1984) showed, in 5 patients with AD, that the increase in ventricular size that occurred was significantly g~eater than what would be expected by the effects of normal aging. Bird et al (1986), in a unique study, rescanned 27 subjects who, 4 years earlier, had been part of the control group in a CT study (Jacoby et al 1980). A reduction in the Mental Test Score (Hodkinson 1973) of greater than three points was associated with an increase in ventricular brain ratio (VBR) and lower attenuation numbers in the left thalamic region at follow-up. Luxenberg et al (1987) rescanned 18 patients with AD and 12 healthy controls from between 6 months to ~ years after the first scan. The increase in size of the lateral ventricles in male AD patients was such that it completely separated the male AD group from the control group. Finally, De Leon et al (1989) found an annual increase in ventricular size of 9% in AD patients compared with 2% in normal controls. This report presents the results of a study that assessed the changes in CT scan appearance and cognitive function that took place over 12 months in a group of patients with AD, all of whom were par~ of a larger longitudinal study of the natural history of the condition.
CT
in Alzheimer's Disease
BU:NLP s v c m ~ v |99l ;29:383-390
385
Methods Subjects were drawn from the catchment area of the Camberwell Heath Authoriw in South East London. All the patients had been in contact with the only two psychiatric hospitals in the area as outpatients, day patients, inpatients, or were seen ~ domiciliary visits. Patients who had been seen previously at the hospital and were in residential homes or long-stay care were also included. An attempt was made to assess all patients ~ w n to the hospital services between October 1986 and October 1988. All patients were diagnosed in accordance with the criteria for the clinical diagnosis of AD suggested by the National Institute for Neurological and ColTmlum~afiveDisorders aM S ~ e ~ C D S ) and the Alzheimer's Disease and Related Disorders Association (ADRDA) w ~ g r o u p (McKhann et al 1984). These criteria have been well validated by neuropatholo~cal examination (Tierney et al 1988). Subjects with both probable (n = 46) and possible (n = 17) AD were included (separate analyses of the results showed no difference between these groups and so, for the sake of clarity, they are presented for the group as a whole). The 17 cases were placed in the "'possible" category because of hypertension (n = 5); diabetes (n = 4, on oral hypoglycemic agents); atypical presentation of AD with episodes of sudden deterioration postdating a gradual onset of memory loss (n = 5); and associated physical illness (n = 3, i.e., renal failure, hypothyroidism, and chronic anemia of unknown origin). Demographic details of the patients were recorded and a standardized history of the illness was taken with the history section of the CAMDEX (Roth et al 1986).
Cognitive Tests
Each subject completed the cognitive section of the CAMDEX examination (the CAMCOG) (Roth et al 1986), which comprises many subsections including scales assessing memory, language, and praxis. The instrument allows the Mini-Mental State Examination (MMSE) (Folstein et al 1975) to be calculated. In addition, each subject was assigned a global rating of severity on the clinical dementit rating (CDR) (Hughes et al 1982, updated by Berg 1984).
I, o m p m e u • o m o g r u p n y o c u n
All patients underwent a noncontrast CT scan of the head using the GEC 9800 scanner. An average of l0 slices (10 mm thick) were taken through the brain parallel to the orbitomeatal line. The following were performed on each scan. Visual Ratings. These were performed comparing the scans to standards of predetermined severity (selected from a population not in the study). Seven re~ons were assessed on a 4-point scale of atrophy or widening (none, mild, moderate, or severe)--frontal, parietal, temporal, and occipital lobes, right and left syivian fissures, mad interhemispheric fissure (IHF, assessed on a 3-point scale). The scores from these individual regions were summated to give a "total cortical score." Basal ganglia calcification was assessed bilaterally in the caudate, globus pallidus, and putamen. A visual rating of the severity of white matter changes was also made on a 4-point scale (mild-minor periventdcular lucency confined to the anterior horns of the !a'..~-~ ventricles; moderate and severe being more
386
BIOL PSYCHIATRY 1991:29:383-390
A. Bums et al
pronounced and visible alongside the bodies of the lateral ventricles) and the presence or absence of infarcts was noted.
Computer-AssistedRatings. The numerical data from the scans were transfen'ed, using magnetic tapes, to the indeper.dent viewing console of the CT scanner. Using a movable cursor, regions of interest were outlined and the area within the region that fell within predetermined Hounsfield Unit (HU) densities was measured. The following regions were assessed: the third ventricle ,'0-25 HU); the right and left sylvian fissures (0-30 HU); ventricular size on the two slices where the bodies of the lateral ventricles were seen best (0-25 HU); and the VBR was calculated by dividing the area of the brain slice to the inner table of the skull (0-99 HU) by ventricular size and expressing the result as a percentage. In many cases it was difficult to determine the slice in which the lateral ventricle was largest and so the average of two slices was taken. It was decided not to include a third slice as there were very few scans where the lateral ventricles were clearly seen on three consecutive slices. The particular HU densities were chose.,,, because of previous reports and because they corresponded best to the visual impression of the scan. All scans were rated independently by AB and ILl; RJ was blind to whether the scan was baseline or follow-up. The following Kappa scores were computed for each region: frontal 0.84, parietal 0.75, temporal 0.74, occipital 0.88, IHF 0.76, sylvian fissure (riOt) 0.75, sylvian fissure (left) 0.72, basal ganglia calcification 0.85, cerebral infarcts 0.85, and white matter changes (0.85). For the computerized methods, a total of 127 regions were assessed by two raters (from 80 scans) and the Pearson correlation coefficient between the ratings was 0.99. Test/retest reliability was perfoffned by AB on the computerized assessments of 238 regions (from 60 original scans). Scans were selected at random and reassessed 7 days after the initial assessment. A Pearson correlation coefficient of the two ratings was 0.99. A phantom was scanned prior to each CT session to assess the calibration of the scanner. When the phantom scans were analyzed, there appeared to be virtually no change over the 36-month period during which the scans were taken, suggesting that any changes in the CT values were not a result of scanner artifact. All cognitive tests and C;F_scans were c o m p ! e ~ within !4 days of each other. The psychological tests and CT scan were repeated annually. This report is confined to those subjects who succes:fifilv completed the CT scan and cognitive tests at entry to the study and 12 months later. Results The characteristics of the sample are preseote: in Table 2. Of the original sample of 138 who had been scanned at entry to the study, 36 had died and therefore were not rescanned. Two of the repeat scans were disregarded for technical reasons. It was not possible to rescan 37 patients who had been scanned at entry to the study and were still alive after 1 year for the following reasons: uncooperative, aggressive behavior !6 (43%); too physically ill to be scanned 10 (27%); moved too far out of the district making it in,practical to perform a second scan 8 (22%); and rel"tives refusing permission for a second scan 3 (8%). The low proportion of men in the study (16%) reflects the clinical material in the catchment area population and the greater mortality in male AD subjects (of the original sample, 27% were men). Table 3 shows cha.'Iges in ~ wriables and cognitive
CT in Alzheimer's Disease
eIOLPSVCt~Tt¥
387
1991~9:383-390
T a b l e 2. Characteristics o f the S a m p l e (N = 63) Gender
Number
10
Male Female Clinical dementia rating
53
I (mild)
9
11 (moderate) III (severe) Placement Home Hospital Residential care
29 25
36 !9 8 Mean
SD
Range
79.3 74.5 58. I
6.2 6.7 38.0
67-92 62-91 6-240
Age (yr) Age of onset (yr) Duration of illness (months)
function that took place in the sample over 12 months. All were highly s i ~ f i c a n t . It was found that measures of the sylvian fissure area often showed a decrease in size over time which was due to slightly different positioning of the patient at the second scan. These results were therefore not included in the ,nalysis. There was a wide variation in the CT scan changes. Fo:trteen of the 63 showed either no change or an actual decrease in size of the third and lateral ventricles suggesting that no overall change had taken place. A group of 6 patients demonstrated a mean change irt VBR of three or greater. A histogram of the distribution of changes in VBR suggested that these six formed a subgroup. However. neither of ~ese subgroups were different ~om the others in terms of demographic details or initial and follow-up cognitive or CT variables. The presence
T a b l e 3. C h a n g e in C T and C o g n i t i o n o v e r 12 M o n t h s (N = 63)
CT SCANS Ventricular-brain ratio Third ventricle size (cm 2) Total cortical score (Max = 23) Cognitive tests Mini-Mental State Examination (Max - 30) CAMCOG Memory (Max = 27) Language (Max = 30) Praxis (Max = 12) Total score (Max = 107)
Deterioration over 12 months
Initial mean 4- SD
12-mon:n follo~:-up mean 4- SD
Mean _ SD
Mean % change
15.1 --- 5.9 1.8 -4- 0.55 5.9 _4- 2.5
16.5 ± C 2 2.1 --- 0.74 6.3 - 2.8
1 4 ~ 1.9 0.3 : o 6 0.4 __ 0.8
9.3 16.6 6.8
p < 0.@1 p < 0.001 p < 0.0001
10.8 4- 6.8
7.6 _-,2 6.2
- 3 . 2 -+ 4.0
-29.6
p < O.0001
-35.4 -24.2 -26.3 -28.9
p p p p
4.8 14.9 5.7 36.0
4.44-
4.2 8.7 3.7 22.1
3.1 !!.3 4.2 25.6
4-+ 44-
3.5 9.0 3.8 22.4
-!.7 -3.6 -1.5 -10.4
± 444-
2.8 4.7 2.4 11.2
Significance of change (!~tired t-test)
< < < <
0.0001 0.0001 0.0001 0.0001
~88
BIOL PSYCHIATRY 1991:29:383-390
A. Bums et al
or absence of white matter changes did not affect the rate of increase in ventricular size. Changes in the CT indices were not related to initial severity of illness as assessed by the CDR. With regard to white matter changes (periventficular lucencies), 28 had none, 24 (38%) had mild changes, 8 (13%) were rated as moderate changes, and only 3 were regarded as severe. Nine patients (14%) shared progression of these over the follow-up period: 4 developed mild changes, 4 deteriorated from mild to moderate, and I developed severe changes, these having been rated at e n ~ to the study as moderate. The association between decline in cognition and change in CT scan variables was examined. A frequency distribution curve suggested that both the changes in cognitive function and the changes in ventricular size were normally distributed so that a Pearson product moment correlation was used. The correlations between change in ventricular size and the cognitive tests were total CAMCOG score 0.21 (p < 0.05), language 0.19 (p < 0.1), memory 0.24 (p < 0.05), and praxis 0.20 (p < 0.1). Deterioration in cognitive score was associated with an increase in ventricular size. There was no correlation between age of onset of the disease and inc~ase in ventficular size (Pearson correlation - 0.0025). Discussion There are four criticisms that can be leveled at this study. First, because of attrition, the population examined was not unselected and the possibility of bias cannot be ruled out. Thus, only 63 of 100 patients could be rescanned after I year. It is difficult, therefore, to ~ confident about extrapolating these results to a population of AD subjects. However, if serial scans are going to be of help in identifying subtypes of AD or to assess the natural history of the condition, it can obviously only be of use in those patients in whom it is possible to perform serial scans. AD is a progressive condition and deterioration in behavior and therefore diminished ability to cooperate with investigations is to be expected over a 12-month period. It was decided, ab initio for ethical reasons, that no sedation for CT scans would be given in this research project. Second, most of the patients suffered from moderate-to-severe dementia and as such the changes may be different than if the group had mild dementia (due to a CT ':iloo.f' effect). Third, ihere is the problem of repositioning patients after 12 months. Although the scanner employed was the same at both parts of the study, we experienced some difficulties in positioning the patients correctly and, as was shown by the measures of the sylvian fissure, small changes in the position of the patients can produce misleading results. In three of the other studies assessing change in CT (Gado et al 1983; Brinkman and Largen 1984; De Leon et al 1989), different machines were used at entry and at follow-up. The assessment of total cortical score by visual rating and the use of the mean of two slices to measure ventricular size probably help to further minimize these possible artifacts. The effects of technical changes in the scanner itself were felt to be minimal, as shown by the lack of scanner 'drift' over time (Jacobson et al 1985). It remains a possibility that the changes seen in the third and lateral ventricular size and cortical atrophy were due to a scanning artifact. However. the statistically significant results along with these results make this unlikely (the decrease: in sylvian fissure size was small by comparison and not statistically significant). Finally, it was not possib!e in this study to rescan normal adults over 12 months. However, there is sufficient information available from other studies to suggest that changes due to normal aging are minimal and that changes in AD are a result of the
CT in AIzheimer's Disease
BlOt.PSYCHIATRY 199| :29:383~390
disease process and not a consequence of aging (Brinkman and I~rgen 1984; Gado et al 1983). Despite these drawbacks in method, useful results were obt~ned in this study. First, it was shown that significant changes in CT scan appearance occurred in the AD group as a whole over 12 months. Both ventricular size and cortical atrophy increased. degree of lateral ventricular enlargement was remarkably similar (9% per annum) to that described by De Leon et al (1989) who estimated total ventricular volume. Luxenberg et al (1987) assessed both third and lateral ventricular enlargement and found the increases were on the order of 20%-50%; however, the numbers studied were small iN = 12). As far as we know, no other work has looked at the change of sulcal size as assessed by visual rating over time. The second finding, similar to that of Naguib and Levy (1982b), was that a significant minority of patients (14 of 63) showed no deterioration in the CT scan. The reason for this may be ~ a t two subgroups do exist, as suggested by Naguib and Levy ( 1982b)--one with patients who had minor or no change in CT indices and another ruth patients who had marked changes. A further possibility is that the patients who did not change had reached a predetermined level of brain atrophy and so the lack of continuing atrophy was due to a "floor effect." However, it is noteworthy that the patients whose atrophy did not progress were no more cognitively impaired than those who did show further CT changes. The CT methods used in this study may also be criticized, but the same applies to other work in this field. We believe it is unlikely that any shortcomings in the CT method contributed significantly to the results of this study. Test/retest, interrater, and concurrent validity have been proven with this current technique (Bums et al, in press) and it has also been shown that there has been no significant alteration in the CI" attenuation numbers over time. The third main finding was a relationship between cognitive deterioration and increasing ventricular size. This accords with other studies which have shown an association between cognitive and radiological deterioration (Naguib and Levy 1982b; Luxenberg et al 1987; De Leon et al 1989). The specific association between increasing ventricular size and apraxia just failed to reach statistical significance. Other workers, for example, McDonald (1969) and Hare (1978), have stressed that parietal lobe involvement was a poor prognostic sign, and Naguib and Levy (1982a, 1982b) provided radiological evidence to support this. In conclusion, this work has shown that significant changes in CT scan indices ~tke place over 12 months in patients with moderate-to-seve:e dementia. However, the results should be interpreted with caution because of the problems inherent with longitudinal studies. Although there is wide variation in the changes seen over time, the evidence for the presence of specific subgroups of AD is small as the changes were not related to other indices of the disease. This study has the advantage of being the largest to date (as far as we know) and has employed the sam~ scan.-_erover the follow-up period. Further work with earlier cases and longer follow-up will be required to fully evaluate the significance of serial CT changes in AD. We are grateful to Mr~ MargaretReith, Dr. Sh6n Lewis, Dr. lan Harvey, and "he staff of the Radiology Department at ~e Mau~,sleyHospital for help with the CT scans. Also, to Margaret Derrick for secretarial help. The work was supportedby a Medical ResearchCouncilproject grant.
References Bird J, Levy R, Jacoby R (1986): C.~mputed tomography in t~heelderly: Changes over time in a normal population. Br J Psychiatry. 148:80-85.
390
BIOLPSYCHIATRY 1991;29:383-390
A. Bums et al
Brinkman S, Largen J (1984): Changes in brain ventricular size with repeated CAT scans in suspected Alzheimer's disease. Am J Psychiatry 141:81-83. Bums A, Jacoby R, Phiipot M, and Levy R (in press): Computed Tomography in AIzheimer's disea~methods of scan analysis, comparison with normal controls and clinico-mdiological correlations. British Journal of Psychiatry. Damasio H, Eslinger P, Dmpa._sioA, Rizzo M, Huang H, Demeter S (1983): Quantitative C'T analysis in the diagnosis of dementia. Arch Neuroi 40:715-719. De Leon M, Geo, ge A, Reisberg B, et al (1989): Alzheimer's disease: Longitudinal ~ studies of ventricular change. AIR 152:1257-1262. Folstein M, Folstein S, McHugh P (1975): Mini Mental State Examination J PsychiawRes 12:189-. 198. Gado M, Hughes C, Danziger W, Chi D (! 983): Ageing dementia and brain atrophy: A longitudinal CT study. Am J Neuroradiol 4:699-702. Hare M (1978): Clinical checklist for diagnosis of dementia. Br MedJ ii:266-267. Hodkinson H (1973): Mental impairment in the elderly. J R CoU Physicians Lond 7:305-317. Hughes C, Berg L, Danziger W, Cohen L, Martin R (1982): A new clinical scale for the staging of dementia. BrJ Psychiatry 140:566-572 [updated by Berg L (1984): BrJPsychiatry 145:339]. Jacobson R, Turner S, Baldy R, Lishman W (1985): Densitometric analysis of scans: Important sources of artefact. Psychol Med 15:879--889. Jacoby R, Levy R, Dawson J ( 1980): Computed tomography in the elderly. I. The normal population. Br J Psychiatry 136:249-255. Luxenberg J, Haxby J, Creasey H, Sundaram M, Rapopon S (198"/): Rate of ventricular enlargement in dementia of the Alzheimer type correlates with rate of neuropsychological deterioration. Neurology 37: ! 135-1140. McDonald C ( 1969): Clinical heterogeneity in senile dementia. Br J Psychiatry ! 15:26"/-271. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan E (1984): Clinical diagnosis of Alzheimer's disease: Report of the NINCDSIADRDA work group under the auspices of the Department of Health and Human Services Task Force on Alzheimer's disease. Neurology 34:939-944. Naguib M, Levy R ( 1982a): Prediction of outcome in senile dementia--a CT study. Br J Psychiatry 140:263-267. Naguib M, Levy R (1982b): CT scanning in senile dementia. A follow-up of survivors. Br J Psychiatry 141:618-620. Roth M, iym E, Mountjoy C, et al (1986): C MDEX. Br J Psychiatry 149:698-709. Takeda S, Matsuzawa T (1984): Brain atrophy during ageing. A quantitative study using CT. J Am Geriatr Soc 32:520-524. Tierney M, Fisher R, Lewis A, et al (1988): The NINCDS/ADRDA workgroup criteria for the clinical diagnosis of probable Alzheimer's disease. Neurology 38:359-364. Zatz L, Jernigan T, Ahumada A (1982): Changes on computed CT with ageing: lntracranial fluid volume. Am J Neuroradiol 3:!-10.
383
1991~..'~:38~390
Computed Tomography in Alzheimer's Disease: A Longitudinal Study Alistair Bums, Robin Jacoby, and Raymond Levy
SixG'-three patients sati~'ing NINCDS/ADI~DA criteria for A!zheimer's disease (AD) received neuropsychological tests and computed tomography (CT) scans 12 months apart. Significant deterioration occurred in all the cognitive tests and in the CT measures used. that is. lateral ventricular size. third ventricular size. and cortical atrophy. There was a wide variation in the size of the changes taking place; 14 of 63 patients showe¢~ zo significant change and 6 showed a marked increase in ventricular size. However. neither group differed from the others in any demographic, cognitive, or other CT variables which suggested, on the CT measures used. that no clearly identifiable subgroups of AD were present. Change in CT indices was not related to initial severity of disease. An increase in ventricular size was related to deterioration of cognitive function. ?hese results require further replication. The methodological drawbacks of such studies are discussed.
Introduction Cranial computed tomography (CT) scanning has been extensively used in the investigation of Alzbeimer's dis,~ase (AD). Its main use is in the exclusion of intracranial abnormalities such as tumoi~s, hematomata, and normal pressure hydrocephalus (all of which give rise to a dementia syndrome). The characteristic radiological features of AD include cerebral atrophy and ventricular enlargement but neither is diagnostic of the disorder as both are common accompaniments to normal aging (Jacoby et al 1980; Zatz et al 1982; Takeda and Matsuzawa 1984). The CT scan can differentiate between nonde.,,°.,°a a . . . . ,~a .,~,;o.~,o ~,,~.., U~f f a ~l l ~ ,,~ .~wl~t-,~k~O . . . . ~r~ i i l I h c i l i W V l.h.c.V.i l,,,,l~ U V I O and qh~Wlll1~,lIlll,lh~l.ll IJg~.glWIILO I•n IgllIJ~/clxL Vl [ J L ~ l [.3 . ~ J.L.l l ~ J L;~" I . 3 ~' 1k~L al 1983; Jacoby et al 1980). Though a single CT scan is of limited use, serial sczns on individual patients have been performed in an attempt to improve diagnostic accuracy and to follow the course of the disease. These studies are summarized in Table I. Naguib and Levy (1982b) were the first to perform serial scans of patients with senile dementia of the Alzheimer type (SDAT). Their findings (in 10 patients) suggested that there were two subgroups of patients--one with relatively stable ventricular size and little deterioration of cognitive function over time and the second with a marked increase in vc.ntricular size associated with significant cognitive decline. Gado et al (1983) scanned 45 AD patients 1 year apart. They found a significant degree of vennicular enlargement
From the Section of Old Age Psychiatry. Institute of Psychiatry. De Crespigny Park. Denmark Rill. London. England. Address reprint requests to Alistair Bums. MPhil, MRCP. MRCPSych. Institute of Psychiatry. De C,~spigny Park. Der~-nark Hill. London SES. 8AF. England. Received March 19. 1990; revised August 8. 1990. © 1991 Society of Biological Psychiatry
0006-3223/91/$03.50
384
A. Bums et al
BIOL PSYCHIATRY 1991:29:383-390
Table I. l.ongitudinal CT Scan Cha.eges in Dementia of the Alzheimer Type No. of patients
No. of controls
Mean age of InLtients (yr)
10 45
0 !2
76.2 71.0
18-36 months 12 months
Brinkman et al (1984)
5
0
60.6
15-35 months
Bird et al (1986)
0
27
76.0
62-200 weeks
Luxenbetg ct al (1987)
12 M 6F
12
62.8 (M) 70.8 (F)
6--60 months
De Leon et al (1989)
50
45
71.2
36 months
Study Naguib and Levy (~982b) Gado et al (1983)
Follow-up interval
Findings" Two subgroups emerged Volumetic m c ~ u r e s more sensitive than linear measures. Changes greater in patient group of controls 4 of the 5 patients had increases greater than normative values suggested Initial and increasing ventricular size associated with subtle reduction in cognitive function Rate of lateral ventricnlar enlargement discmninate," male patientsfrom conlmls Ventricular volume increase: 9% in patients, 2% in controls
aExpanded in the text,
to occur over that time which did not occur in nondemented controls. They observed that computer-assisted volumetric measures of cerebrospinal fluid (CSF) space were a more sensitive indicator of change than linear measures. Brinkman and Largen (1984) showed, in 5 patients with AD, that the increase in ventricular size that occurred was significantly g~eater than what would be expected by the effects of normal aging. Bird et al (1986), in a unique study, rescanned 27 subjects who, 4 years earlier, had been part of the control group in a CT study (Jacoby et al 1980). A reduction in the Mental Test Score (Hodkinson 1973) of greater than three points was associated with an increase in ventricular brain ratio (VBR) and lower attenuation numbers in the left thalamic region at follow-up. Luxenberg et al (1987) rescanned 18 patients with AD and 12 healthy controls from between 6 months to ~ years after the first scan. The increase in size of the lateral ventricles in male AD patients was such that it completely separated the male AD group from the control group. Finally, De Leon et al (1989) found an annual increase in ventricular size of 9% in AD patients compared with 2% in normal controls. This report presents the results of a study that assessed the changes in CT scan appearance and cognitive function that took place over 12 months in a group of patients with AD, all of whom were par~ of a larger longitudinal study of the natural history of the condition.
CT
in Alzheimer's Disease
BU:NLP s v c m ~ v |99l ;29:383-390
385
Methods Subjects were drawn from the catchment area of the Camberwell Heath Authoriw in South East London. All the patients had been in contact with the only two psychiatric hospitals in the area as outpatients, day patients, inpatients, or were seen ~ domiciliary visits. Patients who had been seen previously at the hospital and were in residential homes or long-stay care were also included. An attempt was made to assess all patients ~ w n to the hospital services between October 1986 and October 1988. All patients were diagnosed in accordance with the criteria for the clinical diagnosis of AD suggested by the National Institute for Neurological and ColTmlum~afiveDisorders aM S ~ e ~ C D S ) and the Alzheimer's Disease and Related Disorders Association (ADRDA) w ~ g r o u p (McKhann et al 1984). These criteria have been well validated by neuropatholo~cal examination (Tierney et al 1988). Subjects with both probable (n = 46) and possible (n = 17) AD were included (separate analyses of the results showed no difference between these groups and so, for the sake of clarity, they are presented for the group as a whole). The 17 cases were placed in the "'possible" category because of hypertension (n = 5); diabetes (n = 4, on oral hypoglycemic agents); atypical presentation of AD with episodes of sudden deterioration postdating a gradual onset of memory loss (n = 5); and associated physical illness (n = 3, i.e., renal failure, hypothyroidism, and chronic anemia of unknown origin). Demographic details of the patients were recorded and a standardized history of the illness was taken with the history section of the CAMDEX (Roth et al 1986).
Cognitive Tests
Each subject completed the cognitive section of the CAMDEX examination (the CAMCOG) (Roth et al 1986), which comprises many subsections including scales assessing memory, language, and praxis. The instrument allows the Mini-Mental State Examination (MMSE) (Folstein et al 1975) to be calculated. In addition, each subject was assigned a global rating of severity on the clinical dementit rating (CDR) (Hughes et al 1982, updated by Berg 1984).
I, o m p m e u • o m o g r u p n y o c u n
All patients underwent a noncontrast CT scan of the head using the GEC 9800 scanner. An average of l0 slices (10 mm thick) were taken through the brain parallel to the orbitomeatal line. The following were performed on each scan. Visual Ratings. These were performed comparing the scans to standards of predetermined severity (selected from a population not in the study). Seven re~ons were assessed on a 4-point scale of atrophy or widening (none, mild, moderate, or severe)--frontal, parietal, temporal, and occipital lobes, right and left syivian fissures, mad interhemispheric fissure (IHF, assessed on a 3-point scale). The scores from these individual regions were summated to give a "total cortical score." Basal ganglia calcification was assessed bilaterally in the caudate, globus pallidus, and putamen. A visual rating of the severity of white matter changes was also made on a 4-point scale (mild-minor periventdcular lucency confined to the anterior horns of the !a'..~-~ ventricles; moderate and severe being more
386
BIOL PSYCHIATRY 1991:29:383-390
A. Bums et al
pronounced and visible alongside the bodies of the lateral ventricles) and the presence or absence of infarcts was noted.
Computer-AssistedRatings. The numerical data from the scans were transfen'ed, using magnetic tapes, to the indeper.dent viewing console of the CT scanner. Using a movable cursor, regions of interest were outlined and the area within the region that fell within predetermined Hounsfield Unit (HU) densities was measured. The following regions were assessed: the third ventricle ,'0-25 HU); the right and left sylvian fissures (0-30 HU); ventricular size on the two slices where the bodies of the lateral ventricles were seen best (0-25 HU); and the VBR was calculated by dividing the area of the brain slice to the inner table of the skull (0-99 HU) by ventricular size and expressing the result as a percentage. In many cases it was difficult to determine the slice in which the lateral ventricle was largest and so the average of two slices was taken. It was decided not to include a third slice as there were very few scans where the lateral ventricles were clearly seen on three consecutive slices. The particular HU densities were chose.,,, because of previous reports and because they corresponded best to the visual impression of the scan. All scans were rated independently by AB and ILl; RJ was blind to whether the scan was baseline or follow-up. The following Kappa scores were computed for each region: frontal 0.84, parietal 0.75, temporal 0.74, occipital 0.88, IHF 0.76, sylvian fissure (riOt) 0.75, sylvian fissure (left) 0.72, basal ganglia calcification 0.85, cerebral infarcts 0.85, and white matter changes (0.85). For the computerized methods, a total of 127 regions were assessed by two raters (from 80 scans) and the Pearson correlation coefficient between the ratings was 0.99. Test/retest reliability was perfoffned by AB on the computerized assessments of 238 regions (from 60 original scans). Scans were selected at random and reassessed 7 days after the initial assessment. A Pearson correlation coefficient of the two ratings was 0.99. A phantom was scanned prior to each CT session to assess the calibration of the scanner. When the phantom scans were analyzed, there appeared to be virtually no change over the 36-month period during which the scans were taken, suggesting that any changes in the CT values were not a result of scanner artifact. All cognitive tests and C;F_scans were c o m p ! e ~ within !4 days of each other. The psychological tests and CT scan were repeated annually. This report is confined to those subjects who succes:fifilv completed the CT scan and cognitive tests at entry to the study and 12 months later. Results The characteristics of the sample are preseote: in Table 2. Of the original sample of 138 who had been scanned at entry to the study, 36 had died and therefore were not rescanned. Two of the repeat scans were disregarded for technical reasons. It was not possible to rescan 37 patients who had been scanned at entry to the study and were still alive after 1 year for the following reasons: uncooperative, aggressive behavior !6 (43%); too physically ill to be scanned 10 (27%); moved too far out of the district making it in,practical to perform a second scan 8 (22%); and rel"tives refusing permission for a second scan 3 (8%). The low proportion of men in the study (16%) reflects the clinical material in the catchment area population and the greater mortality in male AD subjects (of the original sample, 27% were men). Table 3 shows cha.'Iges in ~ wriables and cognitive
CT in Alzheimer's Disease
eIOLPSVCt~Tt¥
387
1991~9:383-390
T a b l e 2. Characteristics o f the S a m p l e (N = 63) Gender
Number
10
Male Female Clinical dementia rating
53
I (mild)
9
11 (moderate) III (severe) Placement Home Hospital Residential care
29 25
36 !9 8 Mean
SD
Range
79.3 74.5 58. I
6.2 6.7 38.0
67-92 62-91 6-240
Age (yr) Age of onset (yr) Duration of illness (months)
function that took place in the sample over 12 months. All were highly s i ~ f i c a n t . It was found that measures of the sylvian fissure area often showed a decrease in size over time which was due to slightly different positioning of the patient at the second scan. These results were therefore not included in the ,nalysis. There was a wide variation in the CT scan changes. Fo:trteen of the 63 showed either no change or an actual decrease in size of the third and lateral ventricles suggesting that no overall change had taken place. A group of 6 patients demonstrated a mean change irt VBR of three or greater. A histogram of the distribution of changes in VBR suggested that these six formed a subgroup. However. neither of ~ese subgroups were different ~om the others in terms of demographic details or initial and follow-up cognitive or CT variables. The presence
T a b l e 3. C h a n g e in C T and C o g n i t i o n o v e r 12 M o n t h s (N = 63)
CT SCANS Ventricular-brain ratio Third ventricle size (cm 2) Total cortical score (Max = 23) Cognitive tests Mini-Mental State Examination (Max - 30) CAMCOG Memory (Max = 27) Language (Max = 30) Praxis (Max = 12) Total score (Max = 107)
Deterioration over 12 months
Initial mean 4- SD
12-mon:n follo~:-up mean 4- SD
Mean _ SD
Mean % change
15.1 --- 5.9 1.8 -4- 0.55 5.9 _4- 2.5
16.5 ± C 2 2.1 --- 0.74 6.3 - 2.8
1 4 ~ 1.9 0.3 : o 6 0.4 __ 0.8
9.3 16.6 6.8
p < 0.@1 p < 0.001 p < 0.0001
10.8 4- 6.8
7.6 _-,2 6.2
- 3 . 2 -+ 4.0
-29.6
p < O.0001
-35.4 -24.2 -26.3 -28.9
p p p p
4.8 14.9 5.7 36.0
4.44-
4.2 8.7 3.7 22.1
3.1 !!.3 4.2 25.6
4-+ 44-
3.5 9.0 3.8 22.4
-!.7 -3.6 -1.5 -10.4
± 444-
2.8 4.7 2.4 11.2
Significance of change (!~tired t-test)
< < < <
0.0001 0.0001 0.0001 0.0001
~88
BIOL PSYCHIATRY 1991:29:383-390
A. Bums et al
or absence of white matter changes did not affect the rate of increase in ventricular size. Changes in the CT indices were not related to initial severity of illness as assessed by the CDR. With regard to white matter changes (periventficular lucencies), 28 had none, 24 (38%) had mild changes, 8 (13%) were rated as moderate changes, and only 3 were regarded as severe. Nine patients (14%) shared progression of these over the follow-up period: 4 developed mild changes, 4 deteriorated from mild to moderate, and I developed severe changes, these having been rated at e n ~ to the study as moderate. The association between decline in cognition and change in CT scan variables was examined. A frequency distribution curve suggested that both the changes in cognitive function and the changes in ventricular size were normally distributed so that a Pearson product moment correlation was used. The correlations between change in ventricular size and the cognitive tests were total CAMCOG score 0.21 (p < 0.05), language 0.19 (p < 0.1), memory 0.24 (p < 0.05), and praxis 0.20 (p < 0.1). Deterioration in cognitive score was associated with an increase in ventricular size. There was no correlation between age of onset of the disease and inc~ase in ventficular size (Pearson correlation - 0.0025). Discussion There are four criticisms that can be leveled at this study. First, because of attrition, the population examined was not unselected and the possibility of bias cannot be ruled out. Thus, only 63 of 100 patients could be rescanned after I year. It is difficult, therefore, to ~ confident about extrapolating these results to a population of AD subjects. However, if serial scans are going to be of help in identifying subtypes of AD or to assess the natural history of the condition, it can obviously only be of use in those patients in whom it is possible to perform serial scans. AD is a progressive condition and deterioration in behavior and therefore diminished ability to cooperate with investigations is to be expected over a 12-month period. It was decided, ab initio for ethical reasons, that no sedation for CT scans would be given in this research project. Second, most of the patients suffered from moderate-to-severe dementia and as such the changes may be different than if the group had mild dementia (due to a CT ':iloo.f' effect). Third, ihere is the problem of repositioning patients after 12 months. Although the scanner employed was the same at both parts of the study, we experienced some difficulties in positioning the patients correctly and, as was shown by the measures of the sylvian fissure, small changes in the position of the patients can produce misleading results. In three of the other studies assessing change in CT (Gado et al 1983; Brinkman and Largen 1984; De Leon et al 1989), different machines were used at entry and at follow-up. The assessment of total cortical score by visual rating and the use of the mean of two slices to measure ventricular size probably help to further minimize these possible artifacts. The effects of technical changes in the scanner itself were felt to be minimal, as shown by the lack of scanner 'drift' over time (Jacobson et al 1985). It remains a possibility that the changes seen in the third and lateral ventricular size and cortical atrophy were due to a scanning artifact. However. the statistically significant results along with these results make this unlikely (the decrease: in sylvian fissure size was small by comparison and not statistically significant). Finally, it was not possib!e in this study to rescan normal adults over 12 months. However, there is sufficient information available from other studies to suggest that changes due to normal aging are minimal and that changes in AD are a result of the
CT in AIzheimer's Disease
BlOt.PSYCHIATRY 199| :29:383~390
disease process and not a consequence of aging (Brinkman and I~rgen 1984; Gado et al 1983). Despite these drawbacks in method, useful results were obt~ned in this study. First, it was shown that significant changes in CT scan appearance occurred in the AD group as a whole over 12 months. Both ventricular size and cortical atrophy increased. degree of lateral ventricular enlargement was remarkably similar (9% per annum) to that described by De Leon et al (1989) who estimated total ventricular volume. Luxenberg et al (1987) assessed both third and lateral ventricular enlargement and found the increases were on the order of 20%-50%; however, the numbers studied were small iN = 12). As far as we know, no other work has looked at the change of sulcal size as assessed by visual rating over time. The second finding, similar to that of Naguib and Levy (1982b), was that a significant minority of patients (14 of 63) showed no deterioration in the CT scan. The reason for this may be ~ a t two subgroups do exist, as suggested by Naguib and Levy ( 1982b)--one with patients who had minor or no change in CT indices and another ruth patients who had marked changes. A further possibility is that the patients who did not change had reached a predetermined level of brain atrophy and so the lack of continuing atrophy was due to a "floor effect." However, it is noteworthy that the patients whose atrophy did not progress were no more cognitively impaired than those who did show further CT changes. The CT methods used in this study may also be criticized, but the same applies to other work in this field. We believe it is unlikely that any shortcomings in the CT method contributed significantly to the results of this study. Test/retest, interrater, and concurrent validity have been proven with this current technique (Bums et al, in press) and it has also been shown that there has been no significant alteration in the CI" attenuation numbers over time. The third main finding was a relationship between cognitive deterioration and increasing ventricular size. This accords with other studies which have shown an association between cognitive and radiological deterioration (Naguib and Levy 1982b; Luxenberg et al 1987; De Leon et al 1989). The specific association between increasing ventricular size and apraxia just failed to reach statistical significance. Other workers, for example, McDonald (1969) and Hare (1978), have stressed that parietal lobe involvement was a poor prognostic sign, and Naguib and Levy (1982a, 1982b) provided radiological evidence to support this. In conclusion, this work has shown that significant changes in CT scan indices ~tke place over 12 months in patients with moderate-to-seve:e dementia. However, the results should be interpreted with caution because of the problems inherent with longitudinal studies. Although there is wide variation in the changes seen over time, the evidence for the presence of specific subgroups of AD is small as the changes were not related to other indices of the disease. This study has the advantage of being the largest to date (as far as we know) and has employed the sam~ scan.-_erover the follow-up period. Further work with earlier cases and longer follow-up will be required to fully evaluate the significance of serial CT changes in AD. We are grateful to Mr~ MargaretReith, Dr. Sh6n Lewis, Dr. lan Harvey, and "he staff of the Radiology Department at ~e Mau~,sleyHospital for help with the CT scans. Also, to Margaret Derrick for secretarial help. The work was supportedby a Medical ResearchCouncilproject grant.
References Bird J, Levy R, Jacoby R (1986): C.~mputed tomography in t~heelderly: Changes over time in a normal population. Br J Psychiatry. 148:80-85.
390
BIOLPSYCHIATRY 1991;29:383-390
A. Bums et al
Brinkman S, Largen J (1984): Changes in brain ventricular size with repeated CAT scans in suspected Alzheimer's disease. Am J Psychiatry 141:81-83. Bums A, Jacoby R, Phiipot M, and Levy R (in press): Computed Tomography in AIzheimer's disea~methods of scan analysis, comparison with normal controls and clinico-mdiological correlations. British Journal of Psychiatry. Damasio H, Eslinger P, Dmpa._sioA, Rizzo M, Huang H, Demeter S (1983): Quantitative C'T analysis in the diagnosis of dementia. Arch Neuroi 40:715-719. De Leon M, Geo, ge A, Reisberg B, et al (1989): Alzheimer's disease: Longitudinal ~ studies of ventricular change. AIR 152:1257-1262. Folstein M, Folstein S, McHugh P (1975): Mini Mental State Examination J PsychiawRes 12:189-. 198. Gado M, Hughes C, Danziger W, Chi D (! 983): Ageing dementia and brain atrophy: A longitudinal CT study. Am J Neuroradiol 4:699-702. Hare M (1978): Clinical checklist for diagnosis of dementia. Br MedJ ii:266-267. Hodkinson H (1973): Mental impairment in the elderly. J R CoU Physicians Lond 7:305-317. Hughes C, Berg L, Danziger W, Cohen L, Martin R (1982): A new clinical scale for the staging of dementia. BrJ Psychiatry 140:566-572 [updated by Berg L (1984): BrJPsychiatry 145:339]. Jacobson R, Turner S, Baldy R, Lishman W (1985): Densitometric analysis of scans: Important sources of artefact. Psychol Med 15:879--889. Jacoby R, Levy R, Dawson J ( 1980): Computed tomography in the elderly. I. The normal population. Br J Psychiatry 136:249-255. Luxenberg J, Haxby J, Creasey H, Sundaram M, Rapopon S (198"/): Rate of ventricular enlargement in dementia of the Alzheimer type correlates with rate of neuropsychological deterioration. Neurology 37: ! 135-1140. McDonald C ( 1969): Clinical heterogeneity in senile dementia. Br J Psychiatry ! 15:26"/-271. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan E (1984): Clinical diagnosis of Alzheimer's disease: Report of the NINCDSIADRDA work group under the auspices of the Department of Health and Human Services Task Force on Alzheimer's disease. Neurology 34:939-944. Naguib M, Levy R ( 1982a): Prediction of outcome in senile dementia--a CT study. Br J Psychiatry 140:263-267. Naguib M, Levy R (1982b): CT scanning in senile dementia. A follow-up of survivors. Br J Psychiatry 141:618-620. Roth M, iym E, Mountjoy C, et al (1986): C MDEX. Br J Psychiatry 149:698-709. Takeda S, Matsuzawa T (1984): Brain atrophy during ageing. A quantitative study using CT. J Am Geriatr Soc 32:520-524. Tierney M, Fisher R, Lewis A, et al (1988): The NINCDS/ADRDA workgroup criteria for the clinical diagnosis of probable Alzheimer's disease. Neurology 38:359-364. Zatz L, Jernigan T, Ahumada A (1982): Changes on computed CT with ageing: lntracranial fluid volume. Am J Neuroradiol 3:!-10.