Methods

SUBJECTS

Thirteen patients with alcohol induced persisting amnestic disorder (Korsakoff's syndrome; DSM-IV 291.114), 13 patients with alcohol dependence (DSM-IV 303.9014), and 13 healthy control subjects participated in the study. These groups are referred to as the Korsakoff group, alcoholic group, and control group, respectively. Patients were diagnosed by a multidisciplinary team consisting of a neuropsychiatrist, a neuropsychologist, and a neurologist. The patients were recruited from specialised departments for patients with Korsakoff's syndrome or chronic alcoholism from the Vincent van Gogh Institute for Mental Health in Venray, the Netherlands. All patients were abstinent for at least 1 month. The control subjects were recruited via newspaper advertisements. Exclusion criteria for all subjects were: age older than 56 years (because of the interaction between age and alcohol related brain damage15), an intelligence quotient below 80 (to exclude subjects with generalised cognitive impairment), the use of psychotropic medication, the presence of depression, psychotic disorders, anxiety disorders, dementia, diabetes mellitus, liver disease, other CNS diseases, and cardiac, pulmonary, or endocrine diseases that could affect cognitive functioning. Exclusion criteria for control subjects also included a history of alcohol dependence or a current intake or a history (longer than 1 month) of alcohol intake of 28 or more units a week. Subjects with Korsakoff's syndrome who were not amnestic on neuropsychological testing were also excluded. Written informed consent was obtained from all subjects.

The three groups were matched for age, sex, and education. Subject characteristics are shown in table 1. The duration of alcohol misuse was estimated from the history of the patient and an informed other person, and by examining medical charts. No differences existed in the duration of alcohol misuse between the subjects with Korsakoff's syndrome and the subjects with chronic alcoholism (table1).

NEUROPSYCHOLOGICAL METHODOLOGY

The neuropsychological assessment has been described in detail elsewere.16 Anterograde amnesia was assessed with the delayed recall of the auditory verbal learning test.17Fifteen words were presented five times and after each presentation the subject was asked to reproduce as many words as possible. After 20 minutes, during which non-verbal tests were performed, the delayed recall was tested. The data are expressed as z scores. The z score is the number of SDs that the score deviates from the expected score in a normal population of a given age, sex, and education. The z scores were based on a reference population of 1870 normal and healthy subjects randomly selected from a registry of general practitioners.18 19 As expected, the Korsakoff group performed significantly worse than the alcoholic group and the control group on the delayed recall task (table 1). No differences were found between the alcoholic group and the control group.

The shortened form of the Wechsler adult intelligence scale20 was administered to the patient groups to obtain a measure of general intelligence, and the shortened form of an equivalent Dutch intelligence test, the Groningen intelligence test,21 was administered to the control group for the same purpose. No significant differences existed between the three groups (table 1).

MRI METHODOLOGY

A 3D volumetric scan (T1 weighted, fast field echo, TR 24 ms, TE 7 ms, flip angle 30°, number of averages=2, FOV 230 mm, resolution 256×154), and an inversion recovery (IR) scan (TR 2107 ms, TE 18 ms, turbofactor=3, flip angle 90°, number of averages=2, FOV 230 mm, resolution 256×177) were made on a 1.5 Tesla scanner (Gyroscan ACS-II, Philips). The slice thickness of the 3D volumetric scan was 1.5 mm and the scan axis was coronal, perpendicular to the intercommissural line. The slice thickness of the IR scan was 3 mm and the scan axis was coronal, perpendicular to the long axis of the hippocampus. The thalamus was measured on the IR scan and all other structures on the 3D volume scan.

Data were transferred to a SUN workstation and the regions of interest were measured with ShowImage (developed at the Department of Clinical Physics and Informatics, VrijeUniversiteit, Amsterdam, The Netherlands). A seed function was used to measure the third ventricle. The cut off level between CSF and brain was visually adjusted by the rater on each slice. The other structures were manually traced with a mouse driven cursor. All measurements were taken in a rostrocaudal direction. The volumes of the left side and right side were added in all analyses as the pathological changes in Korsakoff syndrome and alcoholism are bilateral.2 22 The volume of the brain structures was calculated by multiplying the surface area of each region of interest by the slice thickness and summing the volumes of all slices on which the structure was measured. Measurements were done with reference to several anatomical atlases.23-25 Each structure was measured by one rater who was blinded to all clinical information.

METHODOLOGY OF BRAIN MEASUREMENTS

Hippocampus

Measurements started with the slice on which both the semianular sulcus and a notch between the amygdala and the hippocampus in the medial wall of the lateral ventricle were visible.26 Then every second slice was measured. The last slice was the slice before the slice on which the crura of the fornices were visible. On average 10 (range 8–13) slices on each side were measured. The measurements included the hippocampus proper, the dentate gyrus, the alveus, and the portion of the subiculum which lies directly underneath the hippocampus.26

Parahippocampal gyrus

The same slices on which the hippocampus was measured were used, except for the last slice in order not to include the isthmus of the cingulate gyrus. The upper boundary was the hippocampus or the transverse fissure and the lateral boundary was the collateral sulcus. If this sulcus was not visible, a straight line was drawn perpendicular to the temporal stem through the centre of the first gyrus at the medial site of the temporal stem. The brain tissue medial to this line was considered as the parahippocampal gyrus. The lower and medial boundaries consisted of CSF or the tentorium cerebelli at the posterior slices.

Mamillary bodies

The mamillary bodies were measured on four or five consecutive slices. On the first slice the mamillary bodies appeared as a bulge in the floor of the third ventricle. On the next two slices the mamillary bodies had an ovoid shape and on the last slices they were small thickenings in the floor of the third ventricle or the medial wall of the hypothalamus. The left and right mamillary body were in most cases measured together.

Third ventricle

Measurements started with the slice on which the optic chiasma was connected to the hypothalamus. The last slice was the slice on which the posterior commissure was visible. On average 17 (range 15–20) slices were measured. The upper boundary on the slices on which the fornix was visible consisted of a straight line from the most superior point of the left and right side of the thalamus to the fornix. When the fornix was not present, a horizontal line was drawn between the upper medial border of the left and the right part of the thalamus where the distance between the two parts was shortest. The lateral and inferior boundaries consisted of the the thalamus or hypothalamus.

Thalamus

Measurements started one slice after the anterior commissure and stopped when the thalamus could no longer be detected. On average 10 (range 8–12) slices on each side were measured. The lateral boundary consisted of the internal capsule, the upper boundary of the lateral ventricles, and the medial boundary of the third ventricle. The columns of the fornix, the mamillary bodies, and the nucleus ruber were excluded. The medial and lateral geniculate bodies were included only when there was no white matter tract between these structures and the thalamus.

Intracranial area

The cranial area was measured on three slices: on the first slice on which the third ventricle was measured, on the second slice on which the mamillary bodies were measured, and on the last slice on which the third ventricle was measured. The volume of each section was derived by tracing the outline of the supratentorial compartment, following the dural and tentorial margins.

Intraobserver variability

The structures on 10 scans were remeasured. Pearson's correlation coefficient between the first and second measurements was 0.91 for the hippocampus, 0.91 for the parahippocampal gyrus, 0.98 for the third ventricle, 0.90 for the whole thalamus, 0.86 for the mamillary bodies, and 0.99 for the intracranial area. These correlations indicate a high level of intrarater agreement for all measurements.

Correction for age, sex, intracranial area, number of slices, and years of alcohol misuse

The years of alcohol misuse did not correlate with any of the brain volumes in the Korsakoff or alcoholic group and therefore we did not correct for it. The number of slices correlated with the total volume of the hippocampus, the parahippocampal gyrus, the third ventricle, and the thalamus. The number of slices on which the hippocampus and parahippocampal gyrus were measured did not differ between the Korsakoff group, alcoholic group, or control group. To reduce the variance of the data we corrected for the number of slices. This correction reduced the SD in the total group by 21% (parahippocampal gyrus) and 33% (hippocampus), but the average volume in both the total group and the subgroups remained the same as that without correction for the number of slices. Because the thalamus and the third ventricle could be measured on significantly fewer slices in the Korsakoff group (p=0.04) than in the control group, which possibly resulted from atrophy of the thalamus, we did not correct for the number of slices for these structures.

The correction for age, sex, intracranial area, and number of slices was based on a population of 60 healthy people ranging in age from 21 to 82 years (average age 56 (SD 15.9) years). These subjects were recruited by newspaper advertisements and included the subjects who were control subjects in the present study. Regression was performed with the brain structure as dependent variable and intracranial area, age, number of slices (only with hippocampus and parahippocampal gyrus), sex, the interaction term age by sex (because of reported differences in aging between men and women27), and age square (only with the third ventricle) as independent variables. Variables and interaction terms that were significant at the p=0.05 level were included in the final regression model. An expected volume for each individual was calculated on the basis of the constant and the β coefficients from the final regression equation. The expected volume was subtracted from the observed volume. The difference (residue) was used for all analyses. The data are presented in table 2as follows. We calculated an expected brain volume for a subject with the average study age (46 years), male sex, average number of slices on which the hippocampus and parahippocampus were measured, and average intracranial area. To this expected volume we added the residue of the study subjects.

STATISTICAL ANALYSIS

The data were analysed using SPSS for the Macintosh 4.0 (SPSS Inc, Chigaco, IL, USA). Categorical data were analysed by χ²test. Group comparisons of continuous data were analysed with at test. Linear regression analysis was used to analyse the relation between memory score and brain volume. All tests were two tailed, and the significance level was set at 0.05.