Structural Brain Abnormalities in Bipolar Affective Disorder Ventricular Victor W.
Enlargement and Focal Signal Hyperintensities
Swayze II, MD; Nancy C. Andreasen, MD, PhD; Randall J. Alliger, PhD; James C. Ehrhardt, PhD; William T. C. Yuh, MD, EE
\s=b\ Structural brain abnormalities were examined in a sample of 48 patients with bipolar affective disorder who were compared with 54 schizophrenic patients and 47 normal controls. As in our previous work using computed tomographic scanning, lateral ventricular enlargement was due to a diagnostic effect. In this study, the effect was more prominent in the schizophrenic men, while a trend was seen in the bipolar men. Women in both groups did not differ significantly from normal subjects. This finding is possibly consistent with the fact that men have a higher frequency of birth anomalies such as hydrocephalus. Since one cause of
such birth anomalies might be periventricular hemorrhage or infarction, we also evaluated all scans for the presence of small focal regions of signal hyperintensity. A significant increase in the number of focal signal hyperintensities was noted in the bipolar patients, in comparison with normal subjects, but not in
the schizophrenics. The bipolar patients with focal signal hyperintensities had a trend toward larger ventricular size than those without. The pathophysiological significance of these findings is unclear. (Aren Gen Psychiatry. 1990;47:1054-1059)
neuroimaging techniques The development unprecedented psychiatrists study of
has pro¬ with an op¬ the structure and function of the brain in portunity to vivo in patients suffering from major mental illnesses. These
vided research
techniques have been widely used for the study of schizophre¬ nia in particular, documenting the presence of both structural and functional abnormalities in at least some patients suffer¬ ing from this disorder.1'11 Ventricular enlargement, as as-
Accepted for publication January 10,1990. From the Department of Psychiatry, College of Medicine, University of Iowa (Drs Swayze, Andreasen, and Alliger), and the Department of Radiology, University of Iowa Hospitals and Clinics (Drs Ehrhardt and Yuh), Iowa City. Reprint requests to Department of Psychiatry, College of Medicine, University of Iowa, 500 Newton Rd, Iowa City, IA 52242 (Dr Andreasen).
sessed with computed tomographic (CT) scanning, has be¬ one of the most replicated findings in biological
come
psychiatry.1"3,12"32
These techniques have also been applied to bipolar illness and major depression, but much less systematically.33jM Work with positron emission tomography has suggested that bipo¬ lar patients have a decreased anterior-posterior ratio of meta¬ bolic rates in the brain, decreased metabolic rates during the depressed state, and decreased metabolic activity in the head of the caudate nucleus during depression.4"9 Studies of re¬ gional cerebral blood flow using the xenon 133 method have
sometimes shown differences between depressed patients and normal controls, but not with any consistent pattern.50"54 Structural imaging studies, conducted with CT scanning, have also shown mixed results. Some investigators have found bipolar or depressed patients to have a mean increase in ventricular size in comparison with normal controls, but oth¬ ers have not.38"13 In our own work, we have observed a signifi¬ cant increase in ventricular size in bipolar men, as compared with depressed patients or normal controls.44 Except for a brief description of increased focal hyperintensities in a small sample of 14 bipolar patients, no studies have been published to date in bipolar patients using the relatively more powerful technique of magnetic resonance imaging (MRI).56 Magnetic resonance imaging has substantial advantages over CT scanning for the evaluation of structural brain abnor¬ malities in major mental illnesses. Since it does not involve any ionizing radiation, it is essentially risk free. Further¬ more, it has a high degree of resolution, and the use of thin slices and multiple imaging planes permits three-dimensional visualization of complex brain structures such as the ventricu¬ lar system. We took advantage of these various strengths to examine brain structure in a sample of patients suffering from bipolar illness. This study had two goals: to determine wheth¬ er bipolar patients display ventricular enlargement with the more precise measurements permitted by MRI, and to deter¬ mine whether we would be able to observe any focal hyperin¬ tensities that might suggest a history of periventricular inju-
Downloaded From: http://archpsyc.jamanetwork.com/ by a Tufts Univ. Hirsh Health Sciences Library User on 10/16/2016
Table 1 .—Previous Course and Treatment in Bipolar and
Schizophrenic Patients* Schizophrenic
Bipolar Previous course, mean : SD Age at onset, y
(median),
No.
No. of previous hospitalizations Months previously hospitalized No. of ECTs (among patients receiving
22.86 ±7.82, 43
21.44 + 4.44,54
(4), 44 9.07± 10.75 (5), 41 15.88 ±13.92 (11.5), 8
(6), 53 (6), 51 13.92 ±5.48 (12), 18
6.62 ±5.70
5.89 ±6.15
ECT)
Previous treatment, % ECT
12.26 ±21.53
20
38
Neuroleptics
49
45
Lithium carbonate
Tranquilizers
91 36
28
Alcohol abuse
27
15
Drug abuse
33
17
Alcohol and drug abuse
15
57
*ECT indicates electroconvulsive therapy.
ry
or
disease and influence the
enlargement.
development of ventricular
SUBJECTS AND METHODS Patient Samples
=
=
32.28 years, while that of the women was 35.39 years. The clinical characteristics of these two samples are summarized in Table 1. Apart from differences in treatment, the two groups did not men was
on most clinical variables; that is, both groups consisted of patients with a relatively early age at onset and a relatively severe course of illness characterized by a substantial number of hospitaliza¬ tions. They also did not differ on factors that might confound results in a study of this type, such as history of substance abuse, apart from the fact that the schizophrenics included a slightly larger number of patients who had been treated wath electroconvulsive therapy.
differ
Control Subjects Control subjects consisted of 47 healthy volunteers recruited from Iowa City and nearby communities. Controls were recruited from nonprofessional hospital personnel or their friends and relatives or newspaper
advertising. Twenty-eight
age of the men was 32.93 years and of the women, 36.63 years. Controls were screened with an abbreviated version of the Compre¬ hensive Assessment of Symptoms and History to rule out diagnoses of major mental illnesses using DSM-III criteria,59 and a medical history was also obtained. Individuals with a history of any major mental illness, medical illness, or neurological disorder were ex¬ cluded; controls were also excluded if they had a history of substance abuse. All control subjects also gave informed consent. mean
Patients were recruited from admissions to the University of Iowra Hospitals and Clinics, Iowa City, during the years 1986 and 1987. A few patients were also recruited from the Iowa City Veterans Admin¬ istration Hospital. All met DSM-III criteria for bipolar affective disorder or schizophrenia. All the bipolar patients were manic at the time they were assessed. Although these patients are quite similar to other samples that we have described in other studies using CT or MRI in terms of age, age at onset, and other characterizing variables, they represent a different sample from those described in either our recent CT study (collected during 1982 through 1985) or our first MRI study (collected during 1985 and 1986).32,57 The schizophrenics in this sample are identical to those described in reports of our second MRI study, however.58 After experimental procedures were described, all subjects gave informed consent. Subjects in this study included both bipolar patients (n 48) and schizophrenic patients (n 54), since we were interested in determin¬ ing whether any of the findings were specific to bipolar illness. Twenty-nine of the bipolar subjects were male and 19 were female. Their mean (±SD) height was 176.90±6.98 and 163.03 ±6.43 cm, respectively; their weight was 83.50± 14.70 and 68.12±9.79 kg. Thirty-six of the schizophrenics were male and 18 female. Their height was 174.35 ±7.63 and 164.38 ±9.36 cm, respectively; weight was 78.24 ± 14.94 and 63.89 ± 10.56 kg. The two groups were educa¬ tionally equivalent. The mean educational level for the bipolar sub¬ jects was 13.77±2.86 years (excluding kindergarten), while that of the schizophrenics was 13.32 ± 2.66 years. The two groups were also similar in age. The mean age of the bipolar men was 33.41 years, while that of the women was 34.63 years. The mean age of the schizophrenic
through
attempted to identify a control sample that would be educationally equivalent to patient samples. Their mean educational achievement was 14.17 + 2.26 years. The height and weight of the male control subjects were 177.50±8.76 cm and 75.98± 10.13 kg; the height and weight of the women were 165.02 ± 5.75 cm and 63.73 ± 8.15 kg. The
were
male and 19
female. This sex ratio was selected to approximate the sex ratio in the bipolar subjects and schizophrenics as closely as possible. We also
Clinical Assessment assessed clinically using the Comprehensive Assess¬ Patients ment of Symptoms and History.59 This structured interview was designed to describe and record current and past symptoms of schizo¬ were
as well as related spectrum condi¬ tions and all varieties of substance abuse. It can be used to make diagnoses with Research Diagnostic Criteria, DSM-III criteria, or DSM-III-R criteria. It includes comprehensive information about onset and course of symptoms, history of treatment, sociodemogra¬ phic descriptors, handedness, and cognitive function. A variety of standard instruments are embedded in it, including the Global As¬ sessment Scale, the Mini-Mental State examination, the Scale of the Assessment of Negative Symptoms, and the Scale for the Assess¬ ment of Positive Symptoms.
phrenia and affective disorders,
MRI Scanning
Both scanner
and controls were examined using a 0.5-T MRI (Picker) in the University of Iowa Department of Radiology.
patients
Three different scanning sequences were used. The first sequence consisted of midsagittal cuts using an inversion recovery technique with an inversion time of 600 milliseconds and a repetition time (TR) of 1600 milliseconds. In some instances, only a single midsagittal slice was obtained, while in others parasagittal slices were obtained as well. Eight coronal cuts were also obtained with this same pulse sequence. Since human head sizes vary, we thought it was important to adopt some technique that would make scans as closely comparable as possible for the various structures we wished to visualize and measure. Therefore, we decided to position our coronal cuts by selecting a reference point in the center of the brain. For this purpose we use the optic chiasm as a landmark. Based on the midsagittal image, a slice that passed through the optic chiasm and was perpendicular to the midsagittal cut was identified as our central reference cut and was designated as CO. Three cuts were then obtained anterior to this reference cut and four posterior to it. These cuts were obtained at 1-cm intervals with no gaps between slices. These inversion recovery sequences were selected to obtain the best resolution between brain and cerebrospinal fluid. In addition, eight transverse cuts were obtained using a spin-echo
Downloaded From: http://archpsyc.jamanetwork.com/ by a Tufts Univ. Hirsh Health Sciences Library User on 10/16/2016
Table 2.—Ventricular Volume in
Bipolar and Schizophrenic Patients and
Normal Controls*
Ventricular Volume, cm3
Schizophrenic,
Bipolar,
Mean ± SD
subjects
All
48
R lateral ventricle L lateral ventricle
M
L lateral ventricle Total ventricular volume
5.48 ±2.52
4.57 ±2.09 5.15 ±2.33
10.82 + 5.21
12.70 ±5.70
9.71 ±4.20
29
35
28
5.86 ±3.19
6.32 + 2.89
4.48 ±1.72
5.91 ±2.69
7.10±3.34
4.93 ±2.11
11.76 ±5.73
13.42 ±5.88
9.42 ±3.67
only 19
17
19
R lateral ventricle
4.57 ±2.03
5.41 ±2.40
4.70 ±2.59
L lateral ventricle
4.82 ±2.13
5.81 ±2.79
5.47 ±2.65
Total ventricular volume
9.38 ±4.03
11.22 + 5.15
10.17 + 4.94
*For statistical
analysis,
see
Table 3.
Table 3.—Analysis of Variance of Ventricular Volume in
Bipolar and Schizophrenic Patients and Normal Controls* Source
Diagnosis Sex
Diagnosis x sex Error Statistical mean
47
6.02 ±2.75 6.68 ±3.20
only R lateral ventricle
F
Mean ± SD
52
5.35 ±2.83
Total ventricular volume
Control,
Mean ± SD
SS 153.39 55.45 70.53 3616.64
df
MS
2
76.70
1
55.45 35.27 25.65
2 141
2.99 2.16 1.37
.0535
.1437
.2562
analysis of values in Table 2. SS indicates sum of squares; MS
square.
Table 4.—Frequency of Focal Signal Hyperintensities in Bipolar and Schizophrenic Patients and Normal Controls Focal Signal
Bipolar Schizophrenic Control
Hyperintensities, No. (%) 9 (19) 5 (9.3) 2
(4.3)
measures
by
segment includes the entire lateral ventricular system, apart from a small portion of the occipital horns in some patients. Measurements
done "blind" to diagnosis; the reliability ofthe measure has been repeatedly checked and consistently exceeds .9. Measurements in this study were done by manual tracing and planimetry as previously described.58 Our second approach to these data was to examine the scans for small focal areas of signal hyperintensity.64 Both inversion recovery and T2-weighted scans were examined by a team of two radiologists under the supervision of one of us (W.T.C.Y.). The scans were evaluated blind to diagnostic class. If both reviewing radiologists agreed on the presence of a focal signal hyperintensity, it was consid¬ ered to be present. Any focal signal hyperintensity was noted if prominent and clearly demarcated, including those at the tips of the were
ventricles.
No Focal Signal
Hyperintensities, No. (%) 39 (81) 49 (91) 45 (96)
of lateral ventricular volume (in cubic centimeters)
measuring area of the structure in each cut, multiplying by 1 (since each slice represents 1 cm of tissue), and summing. This 8-cm brain
RESULTS Total 48 54 47
sequence with an echo time (TE) of 80 milliseconds and a TR of 2300 milliseconds. Slice thickness was 1 cm wnth no gaps. This scanning sequence provides much poorer gray-white resolution but was se¬ lected to determine whether any small areas of focal signal hyperintensity suggestive of abnormal tissue could be found (also sometimes referred to as "unidentified bright objects").64 The T2-weighted im¬ ages of this type are particularly useful for identifying small areas of tissue abnormality. Because neuropathological studies of schizophre¬ nia have suggested that periventricular gliosis can be found in some patients, and because the ventricular enlargement observed both in schizophrenia and in bipolar illness could be due to periventricular hemorrhage during birth or to infectious processes, we postulated that use of T2-weighted transverse scanning sequences might provide us with some information that would illuminate the pathophysiological mechanisms producing ventricular enlargement. Measurement and Evaluation of Scans
Lateral ventricular size was measured using eight coronal cuts. The lateral ventricles are seen clearly in multiple cuts, with minimal partial volume effect. The eight 1-cm cuts were used to produce
The coronal cuts were used to compare the three groups using of ventricular volume. Descriptive data concerning ven¬ tricular volume appear in Table 2, with statistical analysis in Table 3. Because our previous work has suggested that ventricular enlarge¬ ment in both schizophrenia and bipolar disorder occurs principally among male patients, the data are shown for the three groups for all subjects pooled within diagnostic subtype and also broken down by sex.44,53 It again appears from the data in Table 2 that male bipolar subjects and schizophrenics have larger absolute ventricular size compared with normal controls, whereas ventricular enlargement is less apparent in female patients. These data were analyzed using a sex-by-diagnosis analysis of variance (ANOVA). A diagnosis effect (F[2,141] 2.99, P
Enlargement and Focal Signal Hyperintensities
Swayze II, MD; Nancy C. Andreasen, MD, PhD; Randall J. Alliger, PhD; James C. Ehrhardt, PhD; William T. C. Yuh, MD, EE
\s=b\ Structural brain abnormalities were examined in a sample of 48 patients with bipolar affective disorder who were compared with 54 schizophrenic patients and 47 normal controls. As in our previous work using computed tomographic scanning, lateral ventricular enlargement was due to a diagnostic effect. In this study, the effect was more prominent in the schizophrenic men, while a trend was seen in the bipolar men. Women in both groups did not differ significantly from normal subjects. This finding is possibly consistent with the fact that men have a higher frequency of birth anomalies such as hydrocephalus. Since one cause of
such birth anomalies might be periventricular hemorrhage or infarction, we also evaluated all scans for the presence of small focal regions of signal hyperintensity. A significant increase in the number of focal signal hyperintensities was noted in the bipolar patients, in comparison with normal subjects, but not in
the schizophrenics. The bipolar patients with focal signal hyperintensities had a trend toward larger ventricular size than those without. The pathophysiological significance of these findings is unclear. (Aren Gen Psychiatry. 1990;47:1054-1059)
neuroimaging techniques The development unprecedented psychiatrists study of
has pro¬ with an op¬ the structure and function of the brain in portunity to vivo in patients suffering from major mental illnesses. These
vided research
techniques have been widely used for the study of schizophre¬ nia in particular, documenting the presence of both structural and functional abnormalities in at least some patients suffer¬ ing from this disorder.1'11 Ventricular enlargement, as as-
Accepted for publication January 10,1990. From the Department of Psychiatry, College of Medicine, University of Iowa (Drs Swayze, Andreasen, and Alliger), and the Department of Radiology, University of Iowa Hospitals and Clinics (Drs Ehrhardt and Yuh), Iowa City. Reprint requests to Department of Psychiatry, College of Medicine, University of Iowa, 500 Newton Rd, Iowa City, IA 52242 (Dr Andreasen).
sessed with computed tomographic (CT) scanning, has be¬ one of the most replicated findings in biological
come
psychiatry.1"3,12"32
These techniques have also been applied to bipolar illness and major depression, but much less systematically.33jM Work with positron emission tomography has suggested that bipo¬ lar patients have a decreased anterior-posterior ratio of meta¬ bolic rates in the brain, decreased metabolic rates during the depressed state, and decreased metabolic activity in the head of the caudate nucleus during depression.4"9 Studies of re¬ gional cerebral blood flow using the xenon 133 method have
sometimes shown differences between depressed patients and normal controls, but not with any consistent pattern.50"54 Structural imaging studies, conducted with CT scanning, have also shown mixed results. Some investigators have found bipolar or depressed patients to have a mean increase in ventricular size in comparison with normal controls, but oth¬ ers have not.38"13 In our own work, we have observed a signifi¬ cant increase in ventricular size in bipolar men, as compared with depressed patients or normal controls.44 Except for a brief description of increased focal hyperintensities in a small sample of 14 bipolar patients, no studies have been published to date in bipolar patients using the relatively more powerful technique of magnetic resonance imaging (MRI).56 Magnetic resonance imaging has substantial advantages over CT scanning for the evaluation of structural brain abnor¬ malities in major mental illnesses. Since it does not involve any ionizing radiation, it is essentially risk free. Further¬ more, it has a high degree of resolution, and the use of thin slices and multiple imaging planes permits three-dimensional visualization of complex brain structures such as the ventricu¬ lar system. We took advantage of these various strengths to examine brain structure in a sample of patients suffering from bipolar illness. This study had two goals: to determine wheth¬ er bipolar patients display ventricular enlargement with the more precise measurements permitted by MRI, and to deter¬ mine whether we would be able to observe any focal hyperin¬ tensities that might suggest a history of periventricular inju-
Downloaded From: http://archpsyc.jamanetwork.com/ by a Tufts Univ. Hirsh Health Sciences Library User on 10/16/2016
Table 1 .—Previous Course and Treatment in Bipolar and
Schizophrenic Patients* Schizophrenic
Bipolar Previous course, mean : SD Age at onset, y
(median),
No.
No. of previous hospitalizations Months previously hospitalized No. of ECTs (among patients receiving
22.86 ±7.82, 43
21.44 + 4.44,54
(4), 44 9.07± 10.75 (5), 41 15.88 ±13.92 (11.5), 8
(6), 53 (6), 51 13.92 ±5.48 (12), 18
6.62 ±5.70
5.89 ±6.15
ECT)
Previous treatment, % ECT
12.26 ±21.53
20
38
Neuroleptics
49
45
Lithium carbonate
Tranquilizers
91 36
28
Alcohol abuse
27
15
Drug abuse
33
17
Alcohol and drug abuse
15
57
*ECT indicates electroconvulsive therapy.
ry
or
disease and influence the
enlargement.
development of ventricular
SUBJECTS AND METHODS Patient Samples
=
=
32.28 years, while that of the women was 35.39 years. The clinical characteristics of these two samples are summarized in Table 1. Apart from differences in treatment, the two groups did not men was
on most clinical variables; that is, both groups consisted of patients with a relatively early age at onset and a relatively severe course of illness characterized by a substantial number of hospitaliza¬ tions. They also did not differ on factors that might confound results in a study of this type, such as history of substance abuse, apart from the fact that the schizophrenics included a slightly larger number of patients who had been treated wath electroconvulsive therapy.
differ
Control Subjects Control subjects consisted of 47 healthy volunteers recruited from Iowa City and nearby communities. Controls were recruited from nonprofessional hospital personnel or their friends and relatives or newspaper
advertising. Twenty-eight
age of the men was 32.93 years and of the women, 36.63 years. Controls were screened with an abbreviated version of the Compre¬ hensive Assessment of Symptoms and History to rule out diagnoses of major mental illnesses using DSM-III criteria,59 and a medical history was also obtained. Individuals with a history of any major mental illness, medical illness, or neurological disorder were ex¬ cluded; controls were also excluded if they had a history of substance abuse. All control subjects also gave informed consent. mean
Patients were recruited from admissions to the University of Iowra Hospitals and Clinics, Iowa City, during the years 1986 and 1987. A few patients were also recruited from the Iowa City Veterans Admin¬ istration Hospital. All met DSM-III criteria for bipolar affective disorder or schizophrenia. All the bipolar patients were manic at the time they were assessed. Although these patients are quite similar to other samples that we have described in other studies using CT or MRI in terms of age, age at onset, and other characterizing variables, they represent a different sample from those described in either our recent CT study (collected during 1982 through 1985) or our first MRI study (collected during 1985 and 1986).32,57 The schizophrenics in this sample are identical to those described in reports of our second MRI study, however.58 After experimental procedures were described, all subjects gave informed consent. Subjects in this study included both bipolar patients (n 48) and schizophrenic patients (n 54), since we were interested in determin¬ ing whether any of the findings were specific to bipolar illness. Twenty-nine of the bipolar subjects were male and 19 were female. Their mean (±SD) height was 176.90±6.98 and 163.03 ±6.43 cm, respectively; their weight was 83.50± 14.70 and 68.12±9.79 kg. Thirty-six of the schizophrenics were male and 18 female. Their height was 174.35 ±7.63 and 164.38 ±9.36 cm, respectively; weight was 78.24 ± 14.94 and 63.89 ± 10.56 kg. The two groups were educa¬ tionally equivalent. The mean educational level for the bipolar sub¬ jects was 13.77±2.86 years (excluding kindergarten), while that of the schizophrenics was 13.32 ± 2.66 years. The two groups were also similar in age. The mean age of the bipolar men was 33.41 years, while that of the women was 34.63 years. The mean age of the schizophrenic
through
attempted to identify a control sample that would be educationally equivalent to patient samples. Their mean educational achievement was 14.17 + 2.26 years. The height and weight of the male control subjects were 177.50±8.76 cm and 75.98± 10.13 kg; the height and weight of the women were 165.02 ± 5.75 cm and 63.73 ± 8.15 kg. The
were
male and 19
female. This sex ratio was selected to approximate the sex ratio in the bipolar subjects and schizophrenics as closely as possible. We also
Clinical Assessment assessed clinically using the Comprehensive Assess¬ Patients ment of Symptoms and History.59 This structured interview was designed to describe and record current and past symptoms of schizo¬ were
as well as related spectrum condi¬ tions and all varieties of substance abuse. It can be used to make diagnoses with Research Diagnostic Criteria, DSM-III criteria, or DSM-III-R criteria. It includes comprehensive information about onset and course of symptoms, history of treatment, sociodemogra¬ phic descriptors, handedness, and cognitive function. A variety of standard instruments are embedded in it, including the Global As¬ sessment Scale, the Mini-Mental State examination, the Scale of the Assessment of Negative Symptoms, and the Scale for the Assess¬ ment of Positive Symptoms.
phrenia and affective disorders,
MRI Scanning
Both scanner
and controls were examined using a 0.5-T MRI (Picker) in the University of Iowa Department of Radiology.
patients
Three different scanning sequences were used. The first sequence consisted of midsagittal cuts using an inversion recovery technique with an inversion time of 600 milliseconds and a repetition time (TR) of 1600 milliseconds. In some instances, only a single midsagittal slice was obtained, while in others parasagittal slices were obtained as well. Eight coronal cuts were also obtained with this same pulse sequence. Since human head sizes vary, we thought it was important to adopt some technique that would make scans as closely comparable as possible for the various structures we wished to visualize and measure. Therefore, we decided to position our coronal cuts by selecting a reference point in the center of the brain. For this purpose we use the optic chiasm as a landmark. Based on the midsagittal image, a slice that passed through the optic chiasm and was perpendicular to the midsagittal cut was identified as our central reference cut and was designated as CO. Three cuts were then obtained anterior to this reference cut and four posterior to it. These cuts were obtained at 1-cm intervals with no gaps between slices. These inversion recovery sequences were selected to obtain the best resolution between brain and cerebrospinal fluid. In addition, eight transverse cuts were obtained using a spin-echo
Downloaded From: http://archpsyc.jamanetwork.com/ by a Tufts Univ. Hirsh Health Sciences Library User on 10/16/2016
Table 2.—Ventricular Volume in
Bipolar and Schizophrenic Patients and
Normal Controls*
Ventricular Volume, cm3
Schizophrenic,
Bipolar,
Mean ± SD
subjects
All
48
R lateral ventricle L lateral ventricle
M
L lateral ventricle Total ventricular volume
5.48 ±2.52
4.57 ±2.09 5.15 ±2.33
10.82 + 5.21
12.70 ±5.70
9.71 ±4.20
29
35
28
5.86 ±3.19
6.32 + 2.89
4.48 ±1.72
5.91 ±2.69
7.10±3.34
4.93 ±2.11
11.76 ±5.73
13.42 ±5.88
9.42 ±3.67
only 19
17
19
R lateral ventricle
4.57 ±2.03
5.41 ±2.40
4.70 ±2.59
L lateral ventricle
4.82 ±2.13
5.81 ±2.79
5.47 ±2.65
Total ventricular volume
9.38 ±4.03
11.22 + 5.15
10.17 + 4.94
*For statistical
analysis,
see
Table 3.
Table 3.—Analysis of Variance of Ventricular Volume in
Bipolar and Schizophrenic Patients and Normal Controls* Source
Diagnosis Sex
Diagnosis x sex Error Statistical mean
47
6.02 ±2.75 6.68 ±3.20
only R lateral ventricle
F
Mean ± SD
52
5.35 ±2.83
Total ventricular volume
Control,
Mean ± SD
SS 153.39 55.45 70.53 3616.64
df
MS
2
76.70
1
55.45 35.27 25.65
2 141
2.99 2.16 1.37
.0535
.1437
.2562
analysis of values in Table 2. SS indicates sum of squares; MS
square.
Table 4.—Frequency of Focal Signal Hyperintensities in Bipolar and Schizophrenic Patients and Normal Controls Focal Signal
Bipolar Schizophrenic Control
Hyperintensities, No. (%) 9 (19) 5 (9.3) 2
(4.3)
measures
by
segment includes the entire lateral ventricular system, apart from a small portion of the occipital horns in some patients. Measurements
done "blind" to diagnosis; the reliability ofthe measure has been repeatedly checked and consistently exceeds .9. Measurements in this study were done by manual tracing and planimetry as previously described.58 Our second approach to these data was to examine the scans for small focal areas of signal hyperintensity.64 Both inversion recovery and T2-weighted scans were examined by a team of two radiologists under the supervision of one of us (W.T.C.Y.). The scans were evaluated blind to diagnostic class. If both reviewing radiologists agreed on the presence of a focal signal hyperintensity, it was consid¬ ered to be present. Any focal signal hyperintensity was noted if prominent and clearly demarcated, including those at the tips of the were
ventricles.
No Focal Signal
Hyperintensities, No. (%) 39 (81) 49 (91) 45 (96)
of lateral ventricular volume (in cubic centimeters)
measuring area of the structure in each cut, multiplying by 1 (since each slice represents 1 cm of tissue), and summing. This 8-cm brain
RESULTS Total 48 54 47
sequence with an echo time (TE) of 80 milliseconds and a TR of 2300 milliseconds. Slice thickness was 1 cm wnth no gaps. This scanning sequence provides much poorer gray-white resolution but was se¬ lected to determine whether any small areas of focal signal hyperintensity suggestive of abnormal tissue could be found (also sometimes referred to as "unidentified bright objects").64 The T2-weighted im¬ ages of this type are particularly useful for identifying small areas of tissue abnormality. Because neuropathological studies of schizophre¬ nia have suggested that periventricular gliosis can be found in some patients, and because the ventricular enlargement observed both in schizophrenia and in bipolar illness could be due to periventricular hemorrhage during birth or to infectious processes, we postulated that use of T2-weighted transverse scanning sequences might provide us with some information that would illuminate the pathophysiological mechanisms producing ventricular enlargement. Measurement and Evaluation of Scans
Lateral ventricular size was measured using eight coronal cuts. The lateral ventricles are seen clearly in multiple cuts, with minimal partial volume effect. The eight 1-cm cuts were used to produce
The coronal cuts were used to compare the three groups using of ventricular volume. Descriptive data concerning ven¬ tricular volume appear in Table 2, with statistical analysis in Table 3. Because our previous work has suggested that ventricular enlarge¬ ment in both schizophrenia and bipolar disorder occurs principally among male patients, the data are shown for the three groups for all subjects pooled within diagnostic subtype and also broken down by sex.44,53 It again appears from the data in Table 2 that male bipolar subjects and schizophrenics have larger absolute ventricular size compared with normal controls, whereas ventricular enlargement is less apparent in female patients. These data were analyzed using a sex-by-diagnosis analysis of variance (ANOVA). A diagnosis effect (F[2,141] 2.99, P
