- Email: [email protected]
Reduced regional cerebral blood flow in non-psychotic violent offenders
Psychiatry Research: Neuroimaging Section 98 Ž2000. 29]41
Reduced regional cerebral blood flow in non-psychotic violent offenders Henrik Soderstroma,b,U , Mats Tullberg c , Carsten Wikkelso ¨c, d a,b Sven Ekholm , Anders Forsman a Institute of Clinical Neuroscience, Goteborg Uni¨ ersity, Goteborg, Sweden ¨ ¨ National Board of Forensic Medicine, Department of Forensic Psychiatry, Goteborg, Sweden ¨ c Institute of Clinical Neuroscience, Goteborg Uni¨ ersity, Sahlgrenska Uni¨ ersity Hospital, Goteborg, Sweden ¨ ¨ d Institute of Selected Clinical Sciences, Goteborg Uni¨ ersity, Sahlgrenska Uni¨ ersity Hospital, Goteborg, Sweden ¨ ¨ b
Received 26 March 1999; received in revised form 9 November 1999; accepted 5 December 1999
Abstract The present study was designed to replicate previously reported findings of abnormal frontal andror temporal cerebral blood flow in violent offenders and to control for the influence of major mental disorder ŽMMD., substance abuse, and current medication. HMPAO-SPECT-CBF and MRI scans from pretrial forensic psychiatric investigations of 21 subjects convicted of impulsive violent crimes were retrospectively re-evaluated. In 16r21 subjects, visual assessment of SPECT scans showed some hypoperfusion in the temporal andror frontal lobes. MRI showed no corresponding structural damage. Quantified regional cerebral blood flow ŽrCBF. in defined regions of interest was compared between index cases and 11 healthy control subjects. Index subjects had significant reductions in the right angular gyrus and the right medial temporal gyrus, bilaterally in the hippocampus, and in the left white frontal matter, but they had significantly increased rCBF in the parietal association cortex bilaterally. The aberrations were as frequent and severe among the subjects without MMD, substance abuse, and current medication Ž n s 7. as in the entire group of index subjects. Q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Impulsive violence; rCBF; HMPAO-SPECT; MRI; Forensic psychiatry; Substance abuse; Major mental disorder
U
Corresponding author. Tel.: q46-31-559683; fax: q46-31-559671. E-mail address: [email protected] ŽH. Soderstrom.
0925-4927r00r$ - see front matter Q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 9 2 5 - 4 9 2 7 Ž 9 9 . 0 0 0 4 9 - 9
30
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
1. Introduction Brain imaging techniques have made it possible to explore relationships between regional cerebral function and behavior, which is of vital inter-
est in forensic psychiatry. Previous studies in violent offenders have disclosed abnormal electric activity, blood flow, and metabolism in the frontal and temporal lobes ŽTable 1.. The study of neurobiological substrates to violence requires that
Table 1 Previous studies on regional cerebral activity in aggression Publication
Imaging technique
Study population and main findings
Amen et al. Ž1996.
SPECT
In comparison to 40 non-aggressive psychiatric patients, 40 aggressive patients showed decreased activity in the prefrontal cortex, increased activity in antero-medial frontal regions and in the basal ganglia andror limbic system, and focal abnormalities in the left temporal lobe.
Goyer et al. Ž1994.
PET
17 patients with personality disorders had a significant inverse correlation between life history of aggressive impulse problems and regional metabolism in a frontal transaxial region, and those with borderline personality disorder had significant changes frontally as compared to 43 normal control subjects Ž1994.. The association between regional frontal metabolism and aggressive impulse problem was supported by an extension of the study to 10 patients with PTSD Ž1996..
Oder et al. Ž1992.
SPECT
36 patients who had suffered closed head injuries approximately 40 months before the study showed significant correlations between: Ž1. uninhibited behavior and frontal hypoperfusion; Ž2. aggression and right hemisphere hypoperfusion; and Ž3. social isolation and left hemisphere hypoperfusion.
Raine et al. Ž1994.
PET
The 1994 study included 22 perpetrators of murder or attempted murder who had pleaded not guilty by reason of insanity and 22 age- and sexmatched control subjects. Each group contained 3 cases with schizophrenia. In the 1997 study, both groups were extended to 41 subjects. Notes on early psychosocial deprivation were used for comparisons within the index group in 1998a, and 15 predatory murderers were compared to 9 affective murderers in 1998b. Index cases showed significant reduction of prefrontal glucose metabolism, particularly on the left side in the 1994 study. In the larger 1997 sample, reduced metabolism could be demonstrated in the prefrontal cortex, superior parietal gyrus, left angular gyrus, and corpus callosum. The prefrontal hypometabolism was more severe in the index cases without early psychosocial deprivation and in affective murderers.
Seidenwurm et al. Ž1997.
EEG MRI PET
7 violent offenders showed hypometabolism correlating with limbic electrophysiological abnormalities in the medial temporal lobes when compared to 9 healthy control subjects.
Volkow and Tancredi Ž1987.
EEG PET
4 psychiatric patients with histories of irrational violence had blood flow and metabolic changes in the left temporal lobe, 2 had concomitant frontal cortex changes, and CT showed cortical atrophy in 2.
Goyer and Semple Ž1996.
Raine et al. Ž1997.
Raine et al. Ž1998a.
Raine et al. Ž1998b.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
31
Table 1 Ž Continued. Publication
Imaging technique
Study population and main findings
Wong et al. Ž1994.
EEG Ž n s 262. CT Ž n s 77.
In 372 male patients at a special hospital there was an increased frequency of focal temporal abnormalities ŽEEG. and structural temporal changes ŽCT. among the most violent patients.
Wong et al. Ž1997a.
EEG MRI PET
31 Ž1997a. and 39 Ž1997b. repetitive and non-repetitive violent schizophrenic and schizo-affective offenders Žq6 control subjects in 1997b. were investigated with the following findings. EEG: higher incidence of temporal lobe abnormalities in the repetitive offenders. MRI: non-specific white matter changes in both groups. PET: generalized cortical hypometabolism in both groups. Quantitative analysis: significantly lower metabolism in the right anterior inferior temporal regions among non-repetitive offenders.
Wong et al. Ž1997b.
the impact of possible confounding factors frequently encountered in groups of violent offenders, such as psychiatric disorders, substance abuse, organic brain disorders, or current drug treatment, be separated from the possible correlates of the violent behavior per se. Further methodological problems include the difficulty of collecting random samples from the defined populations, of matching control subjects for relevant aspects, and of discerning whether abnormal findings reflect permanent dysfunction or transient states. The comparison of studies is complicated by the use of different imaging techniques, different quantification methods wmaximum, minimum or mean activity in the region of interest ŽROI. in relation to whole brain, contralateral structures, or the cerebellumx, and different anatomical distribution of the ROIs. Even with these limitations, and despite a certain heterogeneity of findings, previously published studies strongly suggest that frontal and temporal lobe dysfunctions are involved in impulsive violence. In a retrospective re-evaluation analysis of hexamethylpropylene-amine oxime-single photon emission computerized tomography ŽHMPAO-SPECT. and magnetic resonance imaging ŽMRI. scans in impulsive violent offenders, we have tested the hypothesis that reductions in frontal andror temporal rCBF can be found in impulsive violent
offenders with or without major mental disorder ŽMMD., substance abuse, or current medication.
2. Methods 2.1. Index subjects During 1995 and 1996, SPECT and MRI were included in the pre-trial forensic psychiatric investigations of 21 perpetrators of impulsive violent crimes, 20 men and 1 woman, aged 16]51 Žmedian 27. years. None had a previously known MMD Ždefined as schizophrenia, schizoaffective disorder, delusional disorder, bipolar disorder or other psychotic affective disorders., previously diagnosed brain injury or neurological disease. All were on remand and had been assigned to pretrial investigation at the Department of Forensic Psychiatry in Goteborg, Sweden, by court order, ¨ and all were convicted in the subsequent trials. Fourteen had no psychotropic or CNS-active medication. Subjects with substance abuse were detoxified before admission, and illicit drug use is highly unlikely in a Swedish high-security remand unit. All subjects were investigated by clinical psychologists and forensic psychiatrists using strict
32
Table 2 Index subjects, clinical characteristics, MRI and visual SPECT findings Substance use
DSM-III-R diagnoses
Current medication
Sanctions
MRI findings
Reduced rCBF areas
Case 1 Man, 51 years, attempted manslaughter
Alcohol abuse
0 MMD 305.00
Benzodiazepines
Prison
Overall somewhat widened sulci supratentorially. Some small high-signal lesions in the deep white matter of the cerebral hemispheres on the PDrT2-weighted squares.
Right temporal lobe, anterior infero-lateral parts.
Case 2 Man, 46 years, kidnapping, aggravated rape
0
0 MMD 302.20 302.90
0
Prison
Corpus callosum has a rather thick, uniform configuration without the normal narrowing above the splenium.
Left frontal lobe, basally.
Case 3 Man, 43 years, murder
Mixed substance dependence
Brief reactive pychosis 298.80, 303.90 304.10, 301.70
Antiepileptics SSRI Neuroleptics
Forensic psychiatric treatment
A few small areas of highsignal lesions frontally in the deep white matter on PDrT2weighted images. A similar lesion parietally on the left side.
Right temporal lobe, anterior, lateral, and mesial parts.
Case 4 Man, 42 years, murder
0
0 MMD 300.12 307.00
0
Prison
Slightly widened sulci frontally and to some extent also in the parietal regions.
Frontal lobes bilaterally, basally.
Case 5 Man, 38 years, attempted murder
0
Delusional disorder 297.10
0
Forensic psychiatric treatment
Mild frontal atrophy Žfor age.
Left frontal lobe basally, and left posterior frontal lobe, superior temporal lobe.
Case 6 Man, 38 years, murder
Mixed substance dependence
0 MMD 303.90, 304.90 305.90, 301.83
SSRI Benzodiazepines
Prison
Normal findings.
Left temporal lobe, frontal parts laterally.
Case 7 Man, 36 years, murder
Mixed substance dependence
Delusional disorder 297.10, 304.90 303.90, 301.70
Neuroleptics
Forensic psychiatric treatment
Normal findings.
Frontal lobes bilaterally, basally.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Sex, age at investigation, and index crime
Table 2 Ž Continued. Substance use
DSM-III-R diagnoses
Current medication
Sanctions
MRI findings
Reduced rCBF areas
Case 8 Man, 38 years, aggravated, unlawful threats
Alcohol abuse
0 MMD 305.00
SSRI
Prison
Prominent protruding osseous changes resembling fibrous dysplasia in the frontal paranasal sinusesrorbital region. Secondary to the inward expansion, slight compression of the frontal lobes.
No significant change.
Case 9 Woman, 34 years, murder
Alcohol abuse
Brief reactive psychosis 298.80, 305.00 301.83
SSRI Benzodiazepines
Forensic psychiatric treatment
Generally somewhat widened sulci, but probably within upper limit for age.
Frontal lobes bilaterally, basally, and temporal lobes mesially.
Case 10 Man, 31 years, murder
Mixed substance abuse
Chronic psychosis NOS 298.90, 305.00 305.70, 301.20
0
Forensic psychiatric treatment
Normal findings.
No significant change.
Case 11 Man, 27 years, murder
0
0 MMD 309.30
0
Prison
Normal findings.
No significant change.
Case 12 Man, 23 years, rape, sexual abuse
Mixed drug abuse
0 MMD 301.90 305.90
0
Prison
Normal findings.
Left frontal lobe.
Case 13 Man, 22 years, arson
Mixed substance dependence
0 MMD 303.90, 304.90 301.70, 301.83
Benzodiazepines
Prison
Normal findings.
Frontal lobes bilaterally, basally.
Case 14 Man, 20 years, arson
Alcohol abuse
0 MMD 296.30, 305.00 301.90
0
Forensic psychiatric treatment
Normal findings.
Left frontal lobe basally and right temporal lobe, inferomesial parts.
Case 15 Man, 19 years, attempted rape
0
0 MMD 301.89
0
Prison
Normal findings.
Frontal lobes bilaterally, basally.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Sex, age at investigation, and index crime
33
34
Table 2 Ž Continued. Substance use
DSM-III-R diagnoses
Current medication
Sanctions
MRI findings
Reduced rCBF areas
Case 16 Man, 18 years, murder
Alcohol dependence
0 MMD 303.00, 303.90 301.70
0
Prison
Slight ventricular asymmetry with a somewhat widened left ventricle. Probably a normal variation.
Frontal lobes bilaterally, basally, and right temporal lobe, mesial parts.
Case 17 Man, 18 years, aggravated assault
Alcohol abuse
0 MMD 305.00 301.70
0
Prison
Normal findings.
Frontal lobes bilaterally, basally, and left temporal lobe, anterior mesial parts.
Case 18 Man, 18 years, aggravated assault
0
0 MMD 303.00 301.70
0
Prison
Somewhat accentuated cisterna magna with a mild mass effect. Small arachnoidal cyst?
Left frontal lobe basally, and right temporal lobe, inferior lateral parts.
Case 19 Man, 18 years, arson
Alcohol abuse
No MMD 305.00
0
Prison
Normal findings.
Left temporal lobe, mesial parts.
Case 20 Man, 16 years, murder
0
0 MMD 799.90
0
Prison
Normal findings.
No significant change.
Case 21 Man, 16 years, murder
0
0 MMD 799.90
0
Prison
Signal abnormality within the anterior limb of the right internal capsule, most likely an ischemic lesion.
No significant change.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Sex, age at investigation, and index crime
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
DSM-III-R operational criteria for mental disorders. Attention-deficit disorder and other neuropsychiatric disorders in childhood were not retrospectively diagnosed in a systematic manner. Special care was taken to investigate the true extent of substance abuse or dependence by collateral interviews and file reviews made by specially trained social workers. Official registers are generally excellent in the Scandinavian countries, and the forensic psychiatric authority has full access to medical, social, and court files according to the Swedish legislation. Physical and neurological examinations included heart and lung auscultations, pulse and blood pressure, abdomen palpation, pupil size and reflexes, gross motor functioning and sensibility, tendon reflexes in upper and lower extremities, and Babinsky. The subjects were also observed around the clock by trained nurses on the ward. Table 2 gives a detailed description of the study population and the main brain imaging findings. 2.2. Controls Eleven healthy volunteers, 8 men and 3 women, aged 31]50 Žmedian 40. years, recruited among hospital staff, were investigated with HMPAOSPECT only. None had a history of neurological or psychiatric disorder, head trauma, hypertension, cardiac disease, cancer, metabolic disease, alcohol or drug abuse and none used medication. 2.3. rCBF measurements A SPECT gamma camera system ŽGeneral Electric, Neurocam. equipped with high-resolution collimators was used for the rCBF measurements. A standard dose of 1000 MBq w99mTcxd,l-HMPAO ŽCeretec TM , Amersham, UK. was administered intravenously to the relaxed, painfree subject who was resting with closed eyes. Fifteen minutes after the injection, tomographic acquisition was started. Sixty-four 30-s projections were acquired in a 128 = 128 matrix with a 20% energy discrimination window centered at 140 keV. The projection data were prefiltered with a two-dimensional Hanning filter and a cut-off frequency of 0.9 cmy1 . Conventional attenuation
35
correction with an effective attenuation coefficient of 0.12 cmy1 was used. Transaxial and coronal slices were reconstructed with a zoom factor of 2, giving a pixel size of 0.16= 0.16 cm2 in 0.32-cm thick slices Ž128 = 128 matrix.. The spatial resolution in reconstructed slices was approximately 1.2 cm Žfull-width at half-maximum, FWHM.. The tilt angle of the transaxial slices was corrected to the infra-orbito-meatal line. Two experienced raters ŽMT and CW., blind to all clinical information and status as control or index subjects, visually re-assessed all scans by using the transaxial and coronal reconstructed slices. Changes independently identified by both raters were registered as significant rCBF aberrations. Each rater registered changes as focal or diffuse, hypo- or hyperperfused regional aberrations. The relative quantification of the rCBF was conducted according to the Min]Max method ¨ ŽArlig et al., 1994.. By this method, minimum counts per pixel Žcrp. were registered in the white matter regions and maximum crp in the gray matter regions. The rCBF in gray matter was calculated as the ratio between the maximum crp in the ROIs and the maximum crp in the cerebellar ROI with the highest activity. For white matter regions, the rCBF was calculated as the minimum crp in the ROIs to maximum crp in the cerebellar ROIs. The five transaxial slices used for ROI placements went through: Ž1. the middle of the cerebellum; Ž2. the middle of the temporal lobes with the hippocampus well visualized; Ž3. the middle of the thalamus; Ž4. the middle of the caudate nucleus; and Ž5. through the centrum semiovale above the ventricles ŽFig. 1.. Rectangular ROIs were chosen to cover regions in gray or white matter of special interest. The ROIs were placed automatically by a computer program after defining an origo transaxially through the inferior border of the temporal lobes, sagittally in the center and coronally through the anterior border of the pons. A total of 30 ROIs were placed in the basal frontal cortices, the frontal association cortices, the cingular gyrus, the temporo-parietal cortices corresponding to the angular gyri, the hippocampi, the mid-sections of the middle temporal gyri, the semiovale centers, the fronto-
36
Table 3 Quantitative SPECT analyses Mean rCBFvalue Index subjects Ž n s 21.
Mean rCBFvalue Control subjects Ž n s 11.
Ž A.. ROIs placed as suggested by pre¨ ious studies Basal frontal cortex Right 0.80 0.81 Left 0.77 0.79 Frontal association cortex Right 0.84 0.84 Left 0.82 0.85 Angular gyrus right 0.79 0.86 left 0.81 0.85 Hippocampus right 0.77 0.83 left 0.80 0.84 Medial temporal gyrus Right 0.79 0.83 Left 0.80 0.83 White frontal matter Right 0.55 0.59 Left 0.51 0.57
t-value: all index vs. all control subjects
Two-tailed P-values Žonly indicated when - 0.1.
Two-tailed P-values after Bonferroni correction Žfrontal and temporal ROIs only.
Mean rCBFvalue in index subjects Ž n s 7. without MMD, substance abuse, or medication
t-value: index subjects without MMD, substance abuse, or medication vs. control subjects
y0.79 y0.67
NS NS
NS NS
0.81 0.78
0.11 y0.29
NS NS
0.11 y1.90
NS 0.068
NS NS
0.85 0.82
0.41 y0.88
NS NS
y3.34 y1.82
0.002UU 0.079
0.024U NS
0.78 0.79
y3.37 y2.32
0.004UU 0.034U
y3.30 y2.05
0.002UU 0.049U
0.024U NS
0.77 0.80
y2.31 y1.59
0.050 NS
y2.33 y1.86
0.026U 0.072
NS NS
0.78 0.79
y2.34 y1.9
0.033U 0.076
y1.83 y2.54
0.078 0.016U
NS NS
0.53 0.52
y2.03 y2.08
0.059 0.054
Two-tailed P-values Žonly indicated when - 0.1.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Region
Table 3 Ž Continued. Region
Mean rCBFvalue Index subjects Ž n s 21.
U
P- 0.05;
UU
t-value: all index vs. all control subjects
Two-tailed P-values Žonly indicated when - 0.1.
0.55 0.54
y1.58 y0.81
NS NS
0.50 0.52
y2.85 y1.39
0.83 0.84
1.17 y0.50
NS NS
0.86 0.85
1.23 0.09
NS NS
0.92 0.90
0.89 0.90
NS NS
0.92 0.90
0.22 y1.10
NS NS
0.95 0.96
y1.74 y1.54
0.091 NS
0.89 0.93
y1.47 y0.90
NS NS
0.77 0.77
2.83 2.63
0.009UU 0.013U
0.86 0.84
3.07 2.32
0.96 0.95
0.65 1.79
NS 0.083
0.95 0.96
y0.23 0.37
NS NS
0.54 0.50 0.92
y1.38 0.49 y2.03
NS NS 0.051
0.51 0.49 0.90
y1.61 y0.60 y0.66
NS NS NS
0.84
y0.74
NS
0.84
0
NS
Two-tailed P-values after Bonferroni correction Žfrontal and temporal ROIs only.
Mean rCBFvalue in index subjects Ž n s 7. without MMD, substance abuse, or medication
t-value: index subjects without MMD, substance abuse, or medication vs. control subjects
Two-tailed P-values Žonly indicated when - 0.1.
0.012U NS
0.007UU 0.034U
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Ž B . All other ROIs Semiovale center Right 0.52 Left 0.53 Frontoparietal cortex Right 0.85 Left 0.83 Caudate nucleus Right 0.94 Left 0.93 Occipital cortex Right 0.90 Left 0.92 Parietal association cortex Right 0.83 Left 0.83 Thalamus Right 0.98 Left 1.00 White parietal matter Right 0.51 Left 0.51 Cingulate 0.88 gyrus Mesen0.82 cephalon
Mean rCBFvalue Control subjects Ž n s 11.
P- 0.01
37
38
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Fig. 1. Anatomical placement of ROIs.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
parietal cortices, the caudate nuclei, the right and left thalamus, the mesencephalon, the parietal association cortices, the occipital cortices, the frontal and parietal white matter, and the right and left lobes of the cerebellum ŽFig. 1 and Table 3.. Since the two cerebellar ROIs were used only for calculation of the relative rCBF, 28 ROIs were used for the comparisons.
39
for other possible perfusion abnormalities. Finally, mean rCBF for the subjects without MMD, substance abuse, and current medication Ž n s 7. was calculated in all ROIs and compared to the control subjects to investigate whether the findings would remain similar when only index subjects without these possible confounding factors were included in the analysis.
2.4. MRI 3. Results The examinations were performed on either a 0.5 tesla ŽT. ŽGyroscan NT5. or a 1.5 T ŽGyroscan ACS II. magnet. The protocol for all examinations included an initial T1-weighted spin echosequence Žscout sequence including all three main projections; transverse, sagittal, and coronal. followed by transverse conventional spin echo w0.5 T: time of repetitionrtime of echo ŽTRrTE. s 2813r40, 120 and 1.5 T:TRrTEs 2458r30, 90x and a coronal turbo spin echo Ž0.5T:TRrTEs 3000r120 and 1.5 T: TRrTEs 2387r120. of the entire brain using a slice thickness of 6 mm. When an abnormality was suspected, additional sequences were investigated. The choice of additional sequences and projections depended on the initial finding. All examinations were re-evaluated by one neuro-radiologist ŽSE. to discern any specific patterns in this study population. 2.5. Statistical analyses Since the statistical techniques employed in this study are included in the SPSS PC v. 8.0 software, further references are omitted. All tests of significance are two-tailed. The relative value for tracer uptake in each ROI as compared to the cerebellar uptake was approximately normally distributed, and there was no non-equal variance according to Levene’s test. In the initial analysis, aimed at testing hypotheses based on the published literature, differences between index subjects and control subjects in frontal and temporal ROIs were calculated by t-tests. In these tests, the significance levels were corrected for multiple comparisons by the Bonferroni method. Multiple t-tests were then calculated between index subjects and control subjects for all ROIs to screen
3.1. Clinical e¨ aluations Data on comprehensive DSM-III-R Axes I and II diagnoses, including substance abuse and MMD, and current medication are represented in Table 2. Normal neurology was seen in 19r21 subjects Žone showed ataxia in the finger]nose test and one had a nonspecific reflex asymmetry.. Standard EEG registration, performed in 13r21 subjects, showed slow rhythms in two cases but no consistent pattern of pathology. 3.2. Visual re-e¨ aluation of MRI and SPECT scans Overall, the MRI examinations revealed subtle, although most likely nonspecific, changes in a higher frequency than expected in this age group. Three subjects Žcases 3, 8, and 21. had more distinct abnormalities ŽTable 2.. On the SPECT scans, hypoperfused areas were noted in the frontal lobes of 13 index subjects, 7 of whom had additional hypoperfusion in the temporal lobes. Temporal lobe hypoperfusion alone was found in 3, and 5 scans were normal. Comprehensive results of the visual re-evaluations of MRI and SPECT scans are given in Table 2. 3.3. Quantitati¨ e rCBF measurements Frontal and temporal ROIs, regarded as hypothetical target locations on the basis of previously reported findings, were compared between index subjects and control subjects with repeated t-tests which yielded significantly lower tracer uptake in the hippocampus bilaterally, the left white frontal
40
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
matter, and the right angular gyrus and mediotemporal area. The hypoperfusions of the right hippocampus and angular gyrus remained significant after Bonferroni correction. When only the index subjects without MMD, substance abuse, or current medication Ž n s 7. were compared to control subjects, the hypoperfusions of the angular gyrus bilaterally and the hippocampus remained significant, while those in the right hippocampus and the white frontal matter bilaterally tended towards statistical significance Ž P-values between 0.06 and 0.05.. Index subjects, also those without MMD, substance abuse, or current medication, had significantly higher rCBF in the parietal association cortex.
4. Discussion Structural as well as functional brain images were explored in our subjects. The MRI scans revealed no consistent pattern of major structural pathology but showed a higher frequency of nonspecific minor changes than is generally expected in the age range of our index subjects. SPECT demonstrated functional pathology by means of both visual and quantitative assessments. The visual method, which gives a general but subjective picture, is advantageous for discovering diffuse rCBF aberrations and for detecting small focal areas with rCBF changes. The quantitative methods are objective and more appropriate for statistical evaluation of rCBF changes, but require predefined ROI placements that must not be adjusted to the visual assessment of the images. The ROI distribution for this study was determined in accordance with previous publications, and a large number of ROIs were placed in the frontal and temporal regions. However, no ROI actually covered the orbital gyrus, where the visual assessments revealed a high proportion of hypoperfusions. A forthcoming prospective study will include the basal frontal cortex in the quantitative analysis, but to add more ROIs in further retrospective analyses of the present sample would introduce bias by the visual guidance gained. The following hypotheses are compatible with our findings and with those of previous authors:
v
v
v
Reductions of prefrontal blood flow may correspond to executive dysfunction and poor impulse control. Dysfunction in the temporal lobes may be linked to violent behavior and aggression. Increased rCBF in the parietal association cortex may be compatible with frequently described dishabituation for sensory input in patients with attention deficits.
The difference in perfusion between index and control cases cannot be ascribed to the higher age in the control group since rCBF generally declines with age ŽMozley et al., 1997.. The finding that the rCBF pathology did not differ between subjects with and without MMD, substance abuse, and current medication is an important step in associating the findings with the violent behavior but requires replication in a sufficiently large sample to permit true multivariate statistical models. SPECT imaging has, as no other diagnostic tool used in forensic psychiatry today, demonstrated a consistent pattern of aberration from the normal in non-psychotic perpetrators of crimes of extreme aggression and lack of impulse control. Do these aberrations reflect innate or fixed alterations of brain function as trait, or do they demonstrate temporary states of depression, psychological decompensation or diminished impulse control? Follow-up studies are required to determine if the SPECT pathologies in this category of forensic patients progress, remain stable, or even diminish with time and recovery from the acute stress induced by the violent act and the subsequent judicial process. Should the brain imaging findings of fronto-temporal abnormalities in violent offenders be further replicated and shown to remain stable over time, as well as being closely related to violent criminal behavior, research on functional consequences and possible treatment strategies would be of great interest to society.
Acknowledgements This study was supported by grants from the
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
National Board of Forensic Medicine in Sweden. We thank Agneta Brimse, BA, and Anders Manhem, MD, for critical discussions and encouraging collaboration at the Department of Forensic Psychiatry, Goteborg. ¨ References Amen, D., Stubblefield, M., Carmichael, B., Thisted, R., 1996. Brain SPECT findings and aggressiveness. Annals of Clinical Psychiatry 8, 129]137. ¨ ˚ Larsson, A., Bergh, A.-C., Jacobsson, L., Wikkelso, Arlig, A., ¨ C., 1994. A new method for the relative quantification of rCBF examined by 99Tcm-HMPAO SPECT. Nuclear Medicine Communications 15, 814]823. Goyer, P., Andreason, P., Semple, W., Clayton, A., King, A., Compton-Toth, B., Schultz, S., Cohen, R., 1994. Positronemission tomography and personality disorders. Neuropsychopharmacology 10, 21]28. Goyer, P., Semple, W., 1996. PET studies of aggression in personality disorder and other nonpsychotic patients. In: Stoff, D.M., Cairns, R.B. ŽEds.., Aggression and Violence: Genetic, Neurobiological, and Biosocial Perspectives. Lawrence Erlbaum Associates Inc, Mahwah, pp. 219]235. Mozley, P.D., Sadek, A.M., Alavi, A., Gur, R.C., Muenz, L.R., Bunow, B.J., Kim, H.-J., Stecker, M.H., Jolles, P., Newberg, A., 1997. Effects of aging on the cerebral distribution of technetium-99m hexamethylpropylene amine oxime in healthy humans. European Journal of Nuclear Medicine 24, 754]761. Oder, W., Goldenberg, G., Spatt, J., Podreka, I., Binder, H., Deecke, L., 1992. Behavioural and psychosocial sequelae of severe closed head injury and regional cerebral blood flow: a SPECT study. Journal of Neurology, Neurosurgery and Psychiatry 55, 475]480.
41
Raine, A., Buchsbaum, M., Stanley, J., Lottenberg, S., Abel, L., Stoddard, J., 1994. Selective reductions in prefrontal glucose metabolism in murderers. Biological Psychiatry 36, 365]373. Raine, A., Buchsbaum, M., LaCasse, L., 1997. Brain abnormalities in murderers indicated by positron emission tomography. Biological Psychiatry 42, 495]508. Raine, A., Stoddard, J., Bihrle, S., Buchsbaum, M., 1998a. Prefrontal glucose deficits in murderers lacking psychosocial deprivation. Neuropsychiatry, Neuropsychology and Behavioral Neurology 11, 1]7. Raine, A., Meloy, J.R., Bihrle, S., Stoddard, J., LaCasse, L., Buchsbaum, M.S., 1998b. Reduced prefrontal and increased subcortical brain functioning assessed using positron emission tomography in predatory and affective murderers. Behavioral Science and the Law 16, 319]332. Seidenwurm, D., Pounds, T., Globus, A., Valk, P., 1997. Abnormal temporal lobe metabolism in violent subjects: correlation of imaging and neuropsychiatric findings. American Journal of Neuroradiology 18, 625]631. Volkow, N., Tancredi, L., 1987. Neural substrates of violent behaviour. A preliminary study with positron emission tomography. British Journal of Psychiatry 151, 668]673. Wong, M., Lumsden, J., Fenton, G.W., Fenwick, P.B., 1994. Electroencephalography, computed tomography and violence ratings of male patients in a maximum security mental hospital. Acta Psychiatrica Scandinavica 90, 97]101. Wong, M., Fenwick, P., Fenton, G., Lumsden, J., Maisey, M., Stevens, J., 1997a. Repetitive and non-repetitive violent offending behaviour in male patients in a maximum security mental hospital } clinical and neuroimaging findings. Medical Science and the Law 37, 150]160. Wong, M., Fenwick, P., Lumsden, J., Fenton, G., Maisey, M., Lewis, P., Badawi, R., 1997b. Positron emission tomography in male violent offenders with schizophrenia. Psychiatry Research: Neuroimaging 68, 111]123.
Reduced regional cerebral blood flow in non-psychotic violent offenders Henrik Soderstroma,b,U , Mats Tullberg c , Carsten Wikkelso ¨c, d a,b Sven Ekholm , Anders Forsman a Institute of Clinical Neuroscience, Goteborg Uni¨ ersity, Goteborg, Sweden ¨ ¨ National Board of Forensic Medicine, Department of Forensic Psychiatry, Goteborg, Sweden ¨ c Institute of Clinical Neuroscience, Goteborg Uni¨ ersity, Sahlgrenska Uni¨ ersity Hospital, Goteborg, Sweden ¨ ¨ d Institute of Selected Clinical Sciences, Goteborg Uni¨ ersity, Sahlgrenska Uni¨ ersity Hospital, Goteborg, Sweden ¨ ¨ b
Received 26 March 1999; received in revised form 9 November 1999; accepted 5 December 1999
Abstract The present study was designed to replicate previously reported findings of abnormal frontal andror temporal cerebral blood flow in violent offenders and to control for the influence of major mental disorder ŽMMD., substance abuse, and current medication. HMPAO-SPECT-CBF and MRI scans from pretrial forensic psychiatric investigations of 21 subjects convicted of impulsive violent crimes were retrospectively re-evaluated. In 16r21 subjects, visual assessment of SPECT scans showed some hypoperfusion in the temporal andror frontal lobes. MRI showed no corresponding structural damage. Quantified regional cerebral blood flow ŽrCBF. in defined regions of interest was compared between index cases and 11 healthy control subjects. Index subjects had significant reductions in the right angular gyrus and the right medial temporal gyrus, bilaterally in the hippocampus, and in the left white frontal matter, but they had significantly increased rCBF in the parietal association cortex bilaterally. The aberrations were as frequent and severe among the subjects without MMD, substance abuse, and current medication Ž n s 7. as in the entire group of index subjects. Q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Impulsive violence; rCBF; HMPAO-SPECT; MRI; Forensic psychiatry; Substance abuse; Major mental disorder
U
Corresponding author. Tel.: q46-31-559683; fax: q46-31-559671. E-mail address: [email protected] ŽH. Soderstrom.
0925-4927r00r$ - see front matter Q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 9 2 5 - 4 9 2 7 Ž 9 9 . 0 0 0 4 9 - 9
30
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
1. Introduction Brain imaging techniques have made it possible to explore relationships between regional cerebral function and behavior, which is of vital inter-
est in forensic psychiatry. Previous studies in violent offenders have disclosed abnormal electric activity, blood flow, and metabolism in the frontal and temporal lobes ŽTable 1.. The study of neurobiological substrates to violence requires that
Table 1 Previous studies on regional cerebral activity in aggression Publication
Imaging technique
Study population and main findings
Amen et al. Ž1996.
SPECT
In comparison to 40 non-aggressive psychiatric patients, 40 aggressive patients showed decreased activity in the prefrontal cortex, increased activity in antero-medial frontal regions and in the basal ganglia andror limbic system, and focal abnormalities in the left temporal lobe.
Goyer et al. Ž1994.
PET
17 patients with personality disorders had a significant inverse correlation between life history of aggressive impulse problems and regional metabolism in a frontal transaxial region, and those with borderline personality disorder had significant changes frontally as compared to 43 normal control subjects Ž1994.. The association between regional frontal metabolism and aggressive impulse problem was supported by an extension of the study to 10 patients with PTSD Ž1996..
Oder et al. Ž1992.
SPECT
36 patients who had suffered closed head injuries approximately 40 months before the study showed significant correlations between: Ž1. uninhibited behavior and frontal hypoperfusion; Ž2. aggression and right hemisphere hypoperfusion; and Ž3. social isolation and left hemisphere hypoperfusion.
Raine et al. Ž1994.
PET
The 1994 study included 22 perpetrators of murder or attempted murder who had pleaded not guilty by reason of insanity and 22 age- and sexmatched control subjects. Each group contained 3 cases with schizophrenia. In the 1997 study, both groups were extended to 41 subjects. Notes on early psychosocial deprivation were used for comparisons within the index group in 1998a, and 15 predatory murderers were compared to 9 affective murderers in 1998b. Index cases showed significant reduction of prefrontal glucose metabolism, particularly on the left side in the 1994 study. In the larger 1997 sample, reduced metabolism could be demonstrated in the prefrontal cortex, superior parietal gyrus, left angular gyrus, and corpus callosum. The prefrontal hypometabolism was more severe in the index cases without early psychosocial deprivation and in affective murderers.
Seidenwurm et al. Ž1997.
EEG MRI PET
7 violent offenders showed hypometabolism correlating with limbic electrophysiological abnormalities in the medial temporal lobes when compared to 9 healthy control subjects.
Volkow and Tancredi Ž1987.
EEG PET
4 psychiatric patients with histories of irrational violence had blood flow and metabolic changes in the left temporal lobe, 2 had concomitant frontal cortex changes, and CT showed cortical atrophy in 2.
Goyer and Semple Ž1996.
Raine et al. Ž1997.
Raine et al. Ž1998a.
Raine et al. Ž1998b.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
31
Table 1 Ž Continued. Publication
Imaging technique
Study population and main findings
Wong et al. Ž1994.
EEG Ž n s 262. CT Ž n s 77.
In 372 male patients at a special hospital there was an increased frequency of focal temporal abnormalities ŽEEG. and structural temporal changes ŽCT. among the most violent patients.
Wong et al. Ž1997a.
EEG MRI PET
31 Ž1997a. and 39 Ž1997b. repetitive and non-repetitive violent schizophrenic and schizo-affective offenders Žq6 control subjects in 1997b. were investigated with the following findings. EEG: higher incidence of temporal lobe abnormalities in the repetitive offenders. MRI: non-specific white matter changes in both groups. PET: generalized cortical hypometabolism in both groups. Quantitative analysis: significantly lower metabolism in the right anterior inferior temporal regions among non-repetitive offenders.
Wong et al. Ž1997b.
the impact of possible confounding factors frequently encountered in groups of violent offenders, such as psychiatric disorders, substance abuse, organic brain disorders, or current drug treatment, be separated from the possible correlates of the violent behavior per se. Further methodological problems include the difficulty of collecting random samples from the defined populations, of matching control subjects for relevant aspects, and of discerning whether abnormal findings reflect permanent dysfunction or transient states. The comparison of studies is complicated by the use of different imaging techniques, different quantification methods wmaximum, minimum or mean activity in the region of interest ŽROI. in relation to whole brain, contralateral structures, or the cerebellumx, and different anatomical distribution of the ROIs. Even with these limitations, and despite a certain heterogeneity of findings, previously published studies strongly suggest that frontal and temporal lobe dysfunctions are involved in impulsive violence. In a retrospective re-evaluation analysis of hexamethylpropylene-amine oxime-single photon emission computerized tomography ŽHMPAO-SPECT. and magnetic resonance imaging ŽMRI. scans in impulsive violent offenders, we have tested the hypothesis that reductions in frontal andror temporal rCBF can be found in impulsive violent
offenders with or without major mental disorder ŽMMD., substance abuse, or current medication.
2. Methods 2.1. Index subjects During 1995 and 1996, SPECT and MRI were included in the pre-trial forensic psychiatric investigations of 21 perpetrators of impulsive violent crimes, 20 men and 1 woman, aged 16]51 Žmedian 27. years. None had a previously known MMD Ždefined as schizophrenia, schizoaffective disorder, delusional disorder, bipolar disorder or other psychotic affective disorders., previously diagnosed brain injury or neurological disease. All were on remand and had been assigned to pretrial investigation at the Department of Forensic Psychiatry in Goteborg, Sweden, by court order, ¨ and all were convicted in the subsequent trials. Fourteen had no psychotropic or CNS-active medication. Subjects with substance abuse were detoxified before admission, and illicit drug use is highly unlikely in a Swedish high-security remand unit. All subjects were investigated by clinical psychologists and forensic psychiatrists using strict
32
Table 2 Index subjects, clinical characteristics, MRI and visual SPECT findings Substance use
DSM-III-R diagnoses
Current medication
Sanctions
MRI findings
Reduced rCBF areas
Case 1 Man, 51 years, attempted manslaughter
Alcohol abuse
0 MMD 305.00
Benzodiazepines
Prison
Overall somewhat widened sulci supratentorially. Some small high-signal lesions in the deep white matter of the cerebral hemispheres on the PDrT2-weighted squares.
Right temporal lobe, anterior infero-lateral parts.
Case 2 Man, 46 years, kidnapping, aggravated rape
0
0 MMD 302.20 302.90
0
Prison
Corpus callosum has a rather thick, uniform configuration without the normal narrowing above the splenium.
Left frontal lobe, basally.
Case 3 Man, 43 years, murder
Mixed substance dependence
Brief reactive pychosis 298.80, 303.90 304.10, 301.70
Antiepileptics SSRI Neuroleptics
Forensic psychiatric treatment
A few small areas of highsignal lesions frontally in the deep white matter on PDrT2weighted images. A similar lesion parietally on the left side.
Right temporal lobe, anterior, lateral, and mesial parts.
Case 4 Man, 42 years, murder
0
0 MMD 300.12 307.00
0
Prison
Slightly widened sulci frontally and to some extent also in the parietal regions.
Frontal lobes bilaterally, basally.
Case 5 Man, 38 years, attempted murder
0
Delusional disorder 297.10
0
Forensic psychiatric treatment
Mild frontal atrophy Žfor age.
Left frontal lobe basally, and left posterior frontal lobe, superior temporal lobe.
Case 6 Man, 38 years, murder
Mixed substance dependence
0 MMD 303.90, 304.90 305.90, 301.83
SSRI Benzodiazepines
Prison
Normal findings.
Left temporal lobe, frontal parts laterally.
Case 7 Man, 36 years, murder
Mixed substance dependence
Delusional disorder 297.10, 304.90 303.90, 301.70
Neuroleptics
Forensic psychiatric treatment
Normal findings.
Frontal lobes bilaterally, basally.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Sex, age at investigation, and index crime
Table 2 Ž Continued. Substance use
DSM-III-R diagnoses
Current medication
Sanctions
MRI findings
Reduced rCBF areas
Case 8 Man, 38 years, aggravated, unlawful threats
Alcohol abuse
0 MMD 305.00
SSRI
Prison
Prominent protruding osseous changes resembling fibrous dysplasia in the frontal paranasal sinusesrorbital region. Secondary to the inward expansion, slight compression of the frontal lobes.
No significant change.
Case 9 Woman, 34 years, murder
Alcohol abuse
Brief reactive psychosis 298.80, 305.00 301.83
SSRI Benzodiazepines
Forensic psychiatric treatment
Generally somewhat widened sulci, but probably within upper limit for age.
Frontal lobes bilaterally, basally, and temporal lobes mesially.
Case 10 Man, 31 years, murder
Mixed substance abuse
Chronic psychosis NOS 298.90, 305.00 305.70, 301.20
0
Forensic psychiatric treatment
Normal findings.
No significant change.
Case 11 Man, 27 years, murder
0
0 MMD 309.30
0
Prison
Normal findings.
No significant change.
Case 12 Man, 23 years, rape, sexual abuse
Mixed drug abuse
0 MMD 301.90 305.90
0
Prison
Normal findings.
Left frontal lobe.
Case 13 Man, 22 years, arson
Mixed substance dependence
0 MMD 303.90, 304.90 301.70, 301.83
Benzodiazepines
Prison
Normal findings.
Frontal lobes bilaterally, basally.
Case 14 Man, 20 years, arson
Alcohol abuse
0 MMD 296.30, 305.00 301.90
0
Forensic psychiatric treatment
Normal findings.
Left frontal lobe basally and right temporal lobe, inferomesial parts.
Case 15 Man, 19 years, attempted rape
0
0 MMD 301.89
0
Prison
Normal findings.
Frontal lobes bilaterally, basally.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Sex, age at investigation, and index crime
33
34
Table 2 Ž Continued. Substance use
DSM-III-R diagnoses
Current medication
Sanctions
MRI findings
Reduced rCBF areas
Case 16 Man, 18 years, murder
Alcohol dependence
0 MMD 303.00, 303.90 301.70
0
Prison
Slight ventricular asymmetry with a somewhat widened left ventricle. Probably a normal variation.
Frontal lobes bilaterally, basally, and right temporal lobe, mesial parts.
Case 17 Man, 18 years, aggravated assault
Alcohol abuse
0 MMD 305.00 301.70
0
Prison
Normal findings.
Frontal lobes bilaterally, basally, and left temporal lobe, anterior mesial parts.
Case 18 Man, 18 years, aggravated assault
0
0 MMD 303.00 301.70
0
Prison
Somewhat accentuated cisterna magna with a mild mass effect. Small arachnoidal cyst?
Left frontal lobe basally, and right temporal lobe, inferior lateral parts.
Case 19 Man, 18 years, arson
Alcohol abuse
No MMD 305.00
0
Prison
Normal findings.
Left temporal lobe, mesial parts.
Case 20 Man, 16 years, murder
0
0 MMD 799.90
0
Prison
Normal findings.
No significant change.
Case 21 Man, 16 years, murder
0
0 MMD 799.90
0
Prison
Signal abnormality within the anterior limb of the right internal capsule, most likely an ischemic lesion.
No significant change.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Sex, age at investigation, and index crime
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
DSM-III-R operational criteria for mental disorders. Attention-deficit disorder and other neuropsychiatric disorders in childhood were not retrospectively diagnosed in a systematic manner. Special care was taken to investigate the true extent of substance abuse or dependence by collateral interviews and file reviews made by specially trained social workers. Official registers are generally excellent in the Scandinavian countries, and the forensic psychiatric authority has full access to medical, social, and court files according to the Swedish legislation. Physical and neurological examinations included heart and lung auscultations, pulse and blood pressure, abdomen palpation, pupil size and reflexes, gross motor functioning and sensibility, tendon reflexes in upper and lower extremities, and Babinsky. The subjects were also observed around the clock by trained nurses on the ward. Table 2 gives a detailed description of the study population and the main brain imaging findings. 2.2. Controls Eleven healthy volunteers, 8 men and 3 women, aged 31]50 Žmedian 40. years, recruited among hospital staff, were investigated with HMPAOSPECT only. None had a history of neurological or psychiatric disorder, head trauma, hypertension, cardiac disease, cancer, metabolic disease, alcohol or drug abuse and none used medication. 2.3. rCBF measurements A SPECT gamma camera system ŽGeneral Electric, Neurocam. equipped with high-resolution collimators was used for the rCBF measurements. A standard dose of 1000 MBq w99mTcxd,l-HMPAO ŽCeretec TM , Amersham, UK. was administered intravenously to the relaxed, painfree subject who was resting with closed eyes. Fifteen minutes after the injection, tomographic acquisition was started. Sixty-four 30-s projections were acquired in a 128 = 128 matrix with a 20% energy discrimination window centered at 140 keV. The projection data were prefiltered with a two-dimensional Hanning filter and a cut-off frequency of 0.9 cmy1 . Conventional attenuation
35
correction with an effective attenuation coefficient of 0.12 cmy1 was used. Transaxial and coronal slices were reconstructed with a zoom factor of 2, giving a pixel size of 0.16= 0.16 cm2 in 0.32-cm thick slices Ž128 = 128 matrix.. The spatial resolution in reconstructed slices was approximately 1.2 cm Žfull-width at half-maximum, FWHM.. The tilt angle of the transaxial slices was corrected to the infra-orbito-meatal line. Two experienced raters ŽMT and CW., blind to all clinical information and status as control or index subjects, visually re-assessed all scans by using the transaxial and coronal reconstructed slices. Changes independently identified by both raters were registered as significant rCBF aberrations. Each rater registered changes as focal or diffuse, hypo- or hyperperfused regional aberrations. The relative quantification of the rCBF was conducted according to the Min]Max method ¨ ŽArlig et al., 1994.. By this method, minimum counts per pixel Žcrp. were registered in the white matter regions and maximum crp in the gray matter regions. The rCBF in gray matter was calculated as the ratio between the maximum crp in the ROIs and the maximum crp in the cerebellar ROI with the highest activity. For white matter regions, the rCBF was calculated as the minimum crp in the ROIs to maximum crp in the cerebellar ROIs. The five transaxial slices used for ROI placements went through: Ž1. the middle of the cerebellum; Ž2. the middle of the temporal lobes with the hippocampus well visualized; Ž3. the middle of the thalamus; Ž4. the middle of the caudate nucleus; and Ž5. through the centrum semiovale above the ventricles ŽFig. 1.. Rectangular ROIs were chosen to cover regions in gray or white matter of special interest. The ROIs were placed automatically by a computer program after defining an origo transaxially through the inferior border of the temporal lobes, sagittally in the center and coronally through the anterior border of the pons. A total of 30 ROIs were placed in the basal frontal cortices, the frontal association cortices, the cingular gyrus, the temporo-parietal cortices corresponding to the angular gyri, the hippocampi, the mid-sections of the middle temporal gyri, the semiovale centers, the fronto-
36
Table 3 Quantitative SPECT analyses Mean rCBFvalue Index subjects Ž n s 21.
Mean rCBFvalue Control subjects Ž n s 11.
Ž A.. ROIs placed as suggested by pre¨ ious studies Basal frontal cortex Right 0.80 0.81 Left 0.77 0.79 Frontal association cortex Right 0.84 0.84 Left 0.82 0.85 Angular gyrus right 0.79 0.86 left 0.81 0.85 Hippocampus right 0.77 0.83 left 0.80 0.84 Medial temporal gyrus Right 0.79 0.83 Left 0.80 0.83 White frontal matter Right 0.55 0.59 Left 0.51 0.57
t-value: all index vs. all control subjects
Two-tailed P-values Žonly indicated when - 0.1.
Two-tailed P-values after Bonferroni correction Žfrontal and temporal ROIs only.
Mean rCBFvalue in index subjects Ž n s 7. without MMD, substance abuse, or medication
t-value: index subjects without MMD, substance abuse, or medication vs. control subjects
y0.79 y0.67
NS NS
NS NS
0.81 0.78
0.11 y0.29
NS NS
0.11 y1.90
NS 0.068
NS NS
0.85 0.82
0.41 y0.88
NS NS
y3.34 y1.82
0.002UU 0.079
0.024U NS
0.78 0.79
y3.37 y2.32
0.004UU 0.034U
y3.30 y2.05
0.002UU 0.049U
0.024U NS
0.77 0.80
y2.31 y1.59
0.050 NS
y2.33 y1.86
0.026U 0.072
NS NS
0.78 0.79
y2.34 y1.9
0.033U 0.076
y1.83 y2.54
0.078 0.016U
NS NS
0.53 0.52
y2.03 y2.08
0.059 0.054
Two-tailed P-values Žonly indicated when - 0.1.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Region
Table 3 Ž Continued. Region
Mean rCBFvalue Index subjects Ž n s 21.
U
P- 0.05;
UU
t-value: all index vs. all control subjects
Two-tailed P-values Žonly indicated when - 0.1.
0.55 0.54
y1.58 y0.81
NS NS
0.50 0.52
y2.85 y1.39
0.83 0.84
1.17 y0.50
NS NS
0.86 0.85
1.23 0.09
NS NS
0.92 0.90
0.89 0.90
NS NS
0.92 0.90
0.22 y1.10
NS NS
0.95 0.96
y1.74 y1.54
0.091 NS
0.89 0.93
y1.47 y0.90
NS NS
0.77 0.77
2.83 2.63
0.009UU 0.013U
0.86 0.84
3.07 2.32
0.96 0.95
0.65 1.79
NS 0.083
0.95 0.96
y0.23 0.37
NS NS
0.54 0.50 0.92
y1.38 0.49 y2.03
NS NS 0.051
0.51 0.49 0.90
y1.61 y0.60 y0.66
NS NS NS
0.84
y0.74
NS
0.84
0
NS
Two-tailed P-values after Bonferroni correction Žfrontal and temporal ROIs only.
Mean rCBFvalue in index subjects Ž n s 7. without MMD, substance abuse, or medication
t-value: index subjects without MMD, substance abuse, or medication vs. control subjects
Two-tailed P-values Žonly indicated when - 0.1.
0.012U NS
0.007UU 0.034U
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Ž B . All other ROIs Semiovale center Right 0.52 Left 0.53 Frontoparietal cortex Right 0.85 Left 0.83 Caudate nucleus Right 0.94 Left 0.93 Occipital cortex Right 0.90 Left 0.92 Parietal association cortex Right 0.83 Left 0.83 Thalamus Right 0.98 Left 1.00 White parietal matter Right 0.51 Left 0.51 Cingulate 0.88 gyrus Mesen0.82 cephalon
Mean rCBFvalue Control subjects Ž n s 11.
P- 0.01
37
38
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
Fig. 1. Anatomical placement of ROIs.
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
parietal cortices, the caudate nuclei, the right and left thalamus, the mesencephalon, the parietal association cortices, the occipital cortices, the frontal and parietal white matter, and the right and left lobes of the cerebellum ŽFig. 1 and Table 3.. Since the two cerebellar ROIs were used only for calculation of the relative rCBF, 28 ROIs were used for the comparisons.
39
for other possible perfusion abnormalities. Finally, mean rCBF for the subjects without MMD, substance abuse, and current medication Ž n s 7. was calculated in all ROIs and compared to the control subjects to investigate whether the findings would remain similar when only index subjects without these possible confounding factors were included in the analysis.
2.4. MRI 3. Results The examinations were performed on either a 0.5 tesla ŽT. ŽGyroscan NT5. or a 1.5 T ŽGyroscan ACS II. magnet. The protocol for all examinations included an initial T1-weighted spin echosequence Žscout sequence including all three main projections; transverse, sagittal, and coronal. followed by transverse conventional spin echo w0.5 T: time of repetitionrtime of echo ŽTRrTE. s 2813r40, 120 and 1.5 T:TRrTEs 2458r30, 90x and a coronal turbo spin echo Ž0.5T:TRrTEs 3000r120 and 1.5 T: TRrTEs 2387r120. of the entire brain using a slice thickness of 6 mm. When an abnormality was suspected, additional sequences were investigated. The choice of additional sequences and projections depended on the initial finding. All examinations were re-evaluated by one neuro-radiologist ŽSE. to discern any specific patterns in this study population. 2.5. Statistical analyses Since the statistical techniques employed in this study are included in the SPSS PC v. 8.0 software, further references are omitted. All tests of significance are two-tailed. The relative value for tracer uptake in each ROI as compared to the cerebellar uptake was approximately normally distributed, and there was no non-equal variance according to Levene’s test. In the initial analysis, aimed at testing hypotheses based on the published literature, differences between index subjects and control subjects in frontal and temporal ROIs were calculated by t-tests. In these tests, the significance levels were corrected for multiple comparisons by the Bonferroni method. Multiple t-tests were then calculated between index subjects and control subjects for all ROIs to screen
3.1. Clinical e¨ aluations Data on comprehensive DSM-III-R Axes I and II diagnoses, including substance abuse and MMD, and current medication are represented in Table 2. Normal neurology was seen in 19r21 subjects Žone showed ataxia in the finger]nose test and one had a nonspecific reflex asymmetry.. Standard EEG registration, performed in 13r21 subjects, showed slow rhythms in two cases but no consistent pattern of pathology. 3.2. Visual re-e¨ aluation of MRI and SPECT scans Overall, the MRI examinations revealed subtle, although most likely nonspecific, changes in a higher frequency than expected in this age group. Three subjects Žcases 3, 8, and 21. had more distinct abnormalities ŽTable 2.. On the SPECT scans, hypoperfused areas were noted in the frontal lobes of 13 index subjects, 7 of whom had additional hypoperfusion in the temporal lobes. Temporal lobe hypoperfusion alone was found in 3, and 5 scans were normal. Comprehensive results of the visual re-evaluations of MRI and SPECT scans are given in Table 2. 3.3. Quantitati¨ e rCBF measurements Frontal and temporal ROIs, regarded as hypothetical target locations on the basis of previously reported findings, were compared between index subjects and control subjects with repeated t-tests which yielded significantly lower tracer uptake in the hippocampus bilaterally, the left white frontal
40
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
matter, and the right angular gyrus and mediotemporal area. The hypoperfusions of the right hippocampus and angular gyrus remained significant after Bonferroni correction. When only the index subjects without MMD, substance abuse, or current medication Ž n s 7. were compared to control subjects, the hypoperfusions of the angular gyrus bilaterally and the hippocampus remained significant, while those in the right hippocampus and the white frontal matter bilaterally tended towards statistical significance Ž P-values between 0.06 and 0.05.. Index subjects, also those without MMD, substance abuse, or current medication, had significantly higher rCBF in the parietal association cortex.
4. Discussion Structural as well as functional brain images were explored in our subjects. The MRI scans revealed no consistent pattern of major structural pathology but showed a higher frequency of nonspecific minor changes than is generally expected in the age range of our index subjects. SPECT demonstrated functional pathology by means of both visual and quantitative assessments. The visual method, which gives a general but subjective picture, is advantageous for discovering diffuse rCBF aberrations and for detecting small focal areas with rCBF changes. The quantitative methods are objective and more appropriate for statistical evaluation of rCBF changes, but require predefined ROI placements that must not be adjusted to the visual assessment of the images. The ROI distribution for this study was determined in accordance with previous publications, and a large number of ROIs were placed in the frontal and temporal regions. However, no ROI actually covered the orbital gyrus, where the visual assessments revealed a high proportion of hypoperfusions. A forthcoming prospective study will include the basal frontal cortex in the quantitative analysis, but to add more ROIs in further retrospective analyses of the present sample would introduce bias by the visual guidance gained. The following hypotheses are compatible with our findings and with those of previous authors:
v
v
v
Reductions of prefrontal blood flow may correspond to executive dysfunction and poor impulse control. Dysfunction in the temporal lobes may be linked to violent behavior and aggression. Increased rCBF in the parietal association cortex may be compatible with frequently described dishabituation for sensory input in patients with attention deficits.
The difference in perfusion between index and control cases cannot be ascribed to the higher age in the control group since rCBF generally declines with age ŽMozley et al., 1997.. The finding that the rCBF pathology did not differ between subjects with and without MMD, substance abuse, and current medication is an important step in associating the findings with the violent behavior but requires replication in a sufficiently large sample to permit true multivariate statistical models. SPECT imaging has, as no other diagnostic tool used in forensic psychiatry today, demonstrated a consistent pattern of aberration from the normal in non-psychotic perpetrators of crimes of extreme aggression and lack of impulse control. Do these aberrations reflect innate or fixed alterations of brain function as trait, or do they demonstrate temporary states of depression, psychological decompensation or diminished impulse control? Follow-up studies are required to determine if the SPECT pathologies in this category of forensic patients progress, remain stable, or even diminish with time and recovery from the acute stress induced by the violent act and the subsequent judicial process. Should the brain imaging findings of fronto-temporal abnormalities in violent offenders be further replicated and shown to remain stable over time, as well as being closely related to violent criminal behavior, research on functional consequences and possible treatment strategies would be of great interest to society.
Acknowledgements This study was supported by grants from the
H. Soderstrom et al. r Psychiatry Research: Neuroimaging 98 (2000) 29]41
National Board of Forensic Medicine in Sweden. We thank Agneta Brimse, BA, and Anders Manhem, MD, for critical discussions and encouraging collaboration at the Department of Forensic Psychiatry, Goteborg. ¨ References Amen, D., Stubblefield, M., Carmichael, B., Thisted, R., 1996. Brain SPECT findings and aggressiveness. Annals of Clinical Psychiatry 8, 129]137. ¨ ˚ Larsson, A., Bergh, A.-C., Jacobsson, L., Wikkelso, Arlig, A., ¨ C., 1994. A new method for the relative quantification of rCBF examined by 99Tcm-HMPAO SPECT. Nuclear Medicine Communications 15, 814]823. Goyer, P., Andreason, P., Semple, W., Clayton, A., King, A., Compton-Toth, B., Schultz, S., Cohen, R., 1994. Positronemission tomography and personality disorders. Neuropsychopharmacology 10, 21]28. Goyer, P., Semple, W., 1996. PET studies of aggression in personality disorder and other nonpsychotic patients. In: Stoff, D.M., Cairns, R.B. ŽEds.., Aggression and Violence: Genetic, Neurobiological, and Biosocial Perspectives. Lawrence Erlbaum Associates Inc, Mahwah, pp. 219]235. Mozley, P.D., Sadek, A.M., Alavi, A., Gur, R.C., Muenz, L.R., Bunow, B.J., Kim, H.-J., Stecker, M.H., Jolles, P., Newberg, A., 1997. Effects of aging on the cerebral distribution of technetium-99m hexamethylpropylene amine oxime in healthy humans. European Journal of Nuclear Medicine 24, 754]761. Oder, W., Goldenberg, G., Spatt, J., Podreka, I., Binder, H., Deecke, L., 1992. Behavioural and psychosocial sequelae of severe closed head injury and regional cerebral blood flow: a SPECT study. Journal of Neurology, Neurosurgery and Psychiatry 55, 475]480.
41
Raine, A., Buchsbaum, M., Stanley, J., Lottenberg, S., Abel, L., Stoddard, J., 1994. Selective reductions in prefrontal glucose metabolism in murderers. Biological Psychiatry 36, 365]373. Raine, A., Buchsbaum, M., LaCasse, L., 1997. Brain abnormalities in murderers indicated by positron emission tomography. Biological Psychiatry 42, 495]508. Raine, A., Stoddard, J., Bihrle, S., Buchsbaum, M., 1998a. Prefrontal glucose deficits in murderers lacking psychosocial deprivation. Neuropsychiatry, Neuropsychology and Behavioral Neurology 11, 1]7. Raine, A., Meloy, J.R., Bihrle, S., Stoddard, J., LaCasse, L., Buchsbaum, M.S., 1998b. Reduced prefrontal and increased subcortical brain functioning assessed using positron emission tomography in predatory and affective murderers. Behavioral Science and the Law 16, 319]332. Seidenwurm, D., Pounds, T., Globus, A., Valk, P., 1997. Abnormal temporal lobe metabolism in violent subjects: correlation of imaging and neuropsychiatric findings. American Journal of Neuroradiology 18, 625]631. Volkow, N., Tancredi, L., 1987. Neural substrates of violent behaviour. A preliminary study with positron emission tomography. British Journal of Psychiatry 151, 668]673. Wong, M., Lumsden, J., Fenton, G.W., Fenwick, P.B., 1994. Electroencephalography, computed tomography and violence ratings of male patients in a maximum security mental hospital. Acta Psychiatrica Scandinavica 90, 97]101. Wong, M., Fenwick, P., Fenton, G., Lumsden, J., Maisey, M., Stevens, J., 1997a. Repetitive and non-repetitive violent offending behaviour in male patients in a maximum security mental hospital } clinical and neuroimaging findings. Medical Science and the Law 37, 150]160. Wong, M., Fenwick, P., Lumsden, J., Fenton, G., Maisey, M., Lewis, P., Badawi, R., 1997b. Positron emission tomography in male violent offenders with schizophrenia. Psychiatry Research: Neuroimaging 68, 111]123.