Previous studies have demonstrated that "ventricular enlargement" is a nonspecific MRI abnormality frequently encountered in a variety of psychiatric conditions.1 In addition, recent ultrasonographic studies in preterm infants have shown that ventricular enlargement caused by pre- or perinatal necrosis of subcortical white matter2,3 is associated with the subsequent development of childhood psychiatric disorders.4 To our knowledge there are no studies that have specifically addressed the issue of incidence and significance of ventricular enlargement in adults with mental retardation who demonstrate violent/destructive behaviors.
Although violent/destructive behaviors of adults with profound mental retardation may not be directly comparable to symptoms of psychiatric disorders in non—mentally retarded psychiatric patients, we nonetheless felt it relevant to examine whether an association of ventricular size with dysfunctional behavior problems in mentally retarded patients could be demonstrated, as it has been in psychiatric patients without mental retardation. We hypothesized that brain ventricular size of adults with profound mental retardation who demonstrate violent/destructive behaviors is greater than brain ventricular size of similar adults with no significant behavior problems.
Brain MRI scans were obtained over a 5-year period from 99 adult residents of Hunterdon Developmental Center, a 650-bed intermediate care facility for mental retardation in Clinton, NJ, and 7 mentally retarded residents of community-based group homes in central New Jersey. Of those scanned, 39 were fully ambulatory and 67 were nonambulatory. Only the 39 fully ambulatory subjects were included in this study because we felt severe motor dysfunction would prevent the full manifestation of violent, aggressive, and destructive behavioral tendencies, which was the focus of our study.
The 39 patients had a mean age of 35.3 years; 19 were male and 20 female. Twenty-three of 39 received medication for seizures. Additional neurological findings included 5 subjects with spasticity (but ambulatory), 1 subject with ataxia, 1 with blindness, and 1 with myopathic muscle weakness. Clinical characteristics of the subjects are listed in t1.
The 39 patients were divided into two groups: 20 patients who did not require psychotropic medication because of the absence of violent/destructive behaviors (the "No Behavior" group in t1) and 19 patients who received psychotropic medication for violent/destructive behaviors (the "Behavior" group in t1). Psychiatric diagnoses, which were recorded in the chart at the time of MRI determination, are shown in t1. It was not always clear from chart review who had determined the psychiatric diagnosis (psychiatrist, psychologist, or other physician or mental health professional). In t1, behavior problems were recorded as "none" if violent/destructive behaviors were not reported in the chart. The types of behaviors are shown in t1. All those in the Behavior group demonstrated either assaultiveness or self-injurious behavior (or both) that was felt to require psychotropic medication. All 19 patients in the Behavior group had received psychotropic medication for at least 10 years. Neuroleptics were the most frequently prescribed medication (17/19).
All patients fell within the profound range of mental retardation (IQ less than 20) as determined by the Slossen Intelligence Test. The most common known etiology of mental retardation was ischemia related to prematurity and/or perinatal injury, but in an even larger number of patients the etiology was unknown (see t1). Etiologies did not differ significantly between the Behavior and No Behavior groups, although there were 3 patients with tuberous sclerosis in the Behavior group and none in the No Behavior group.
In addition to the mentally retarded individuals, 7 non—mentally retarded individuals ("medical control subjects") were included in the study for comparison. These medical control subjects were individuals with no prior history of brain dysfunction, mental retardation, or dementia. They presented to the Robert Wood Johnson Medical School Neurology Clinic with various complaints (see t1) and were subsequently found to be normal. Their MRI scans were also interpreted as "normal."
A 1.5-T MRI scanner was used in all cases. Intravenous sedation was required in most cases. MRI scans were ordered by seven staff physicians of the Hunterdon Developmental Center for various reasons, including loss of motor function, intractable seizures, or persistent bizarre behavior. The rationale for ordering the MRI was not always possible to ascertain from chart review, but it appeared that most frequently MRI was done on the recommendation of a neurology consultant as part of an evaluation for seizures or loss of motor function. Thus, patients with seizures may be overrepresented in our study.
Ventricular Size Determination
Using a computerized system, we evaluated ventricular size in all three groups (see t1): 1) Control: non— mentally retarded medical control subjects; 2) No Behavior: mentally retarded subjects with no behavior problems and not receiving psychotropic medication; and 3) Behavior: mentally retarded subjects receiving psychotropic medication for behavior problems. A ventricular brain ratio (VBR) was determined by dividing the width of the lateral ventricles by the width of the cerebrum in coronal view, at the level of the anterior commissure and amygdala at a point on the septum pellucidum one-third the distance below the corpus callosum and two-thirds of the distance above the fornix. For convenience, the ratio was multiplied by 1,000.
Intraobserver reliability of the VBR was measured by having the first observer measure the VBR on two separate occasions one month apart. The observer was blinded to the results of the first set of results while repeating the measurements. The correlation between the two measurements was extremely strong (r=0.99, P<0.0001), demonstrating excellent reproducibility of these measurements by a single observer. Next, the interobserver reliability was analyzed by having a second observer also measure the data. This observer had no previous experience with measuring VBR and used a different computer system and display program. The correlation between the first measurements of observers A and B was 0.98 (P<0.001). This represents an extremely strong correlation between the two observers.
Continuous variables (age and VBR) were compared by using t-tests and analyses of variance (one-way and two-way), followed by a Student-Newman-Keuls test where appropriate. Proportions were compared by using chi-square and Fisher's exact test. Relationships between continuous variables were tested with Pearson product-moment correlations. All statistical tests were two-tailed, with the level of significance set at 0.05. SAS Software, version 8, was used for all statistical tests (SAS Institute, Cary, NC).
Results of statistical analysis are recorded in t2.
The analysis of variance comparing the VBR among the three groups was highly significant (F=25,38, df=2,43, P<0.0001). The Student-Newman-Keuls multiple-comparison test showed that the mean VBR for the Behavior group (those with a history of behavior problems who were receiving psychotropics, mostly thioridazine) was significantly larger than VBR for both the No Behavior group (those without behavior problems and not receiving psychotropics) and the medical control group (P<0.05). As can be seen in t2, for all three groups the mean VBR was greater in males than females. In a simple comparison between males and females with all groups together, this difference was significant (t=2,23, df=44, P=0.03). However, in a two-way ANOVA controlled for behavior group and gender, the gender difference disappeared (F=1.68, df=1,40, P=0.202) and only the behavior factor was significant (F=20.85, df=2,40, P<0.0001). Finally, in order to fully understand any possible gender differences, we compared males and females on VBR measurements within each of the three groups. Only the normal control subjects showed a trend toward a gender difference (t=2.44, df=5, P=0.0589). Because there were only 7 patients (4 male, 3 female) in the medical control group, there was very low power to detect a gender difference. There were no significant differences in age among the three groups (F=1.99, df=2,43, P=0.15). The difference in proportion of seizures between the Behavior and No Behavior groups (42% vs. 75%) was close to significance (P=0.056). There were no apparent differences between the groups with regard to other neurological conditions (e.g. spasticity, ataxia, blindness).
This pilot study suggests a relationship between ventricular enlargement and violent/destructive behaviors in adults with profound mental retardation; however, certain potential limitations should not go unnoticed.
First, the index of ventricular enlargement that was used in our study (the ventricular brain ratio) is a linear/planar rather than a volumetric measurement. Although we believe our study should be repeated using volumetric techniques, we nonetheless feel that our linear/planar assay may provide information that might not be obtained by volumetric methodology. We specifically chose the coronal plane at the level of the anterior commissure, which also includes the amygdala. It is possible that damage to cortical-subcortical neuronal circuits only at that locality could have been responsible for the psychopathology observed in our patients. If this were true, one might see a localized area of ventricular enlargement that would not significantly contribute to a total volumetric measurement. Possible advantages to our linear/planar method of ventricular size determination are its reproducibility and simplicity, which will make it easy for many centers to repeat the measurements and attempt to confirm our data.
Second, although our data may suggest that ventricular enlargement is frequently present in profoundly mentally retarded individuals with behavior problems, the etiology of ventricular enlargement in our patients who were receiving psychotropic medication was not established by our study. Because most members of the Behavior group had received neuroleptic medication for many years, one might question whether neuroleptic medication itself had caused subcortical atrophy that led to ventricular enlargement. Although this is one possible explanation, others have reported that imaging studies of first-episode schizophrenia patients who had never received neuroleptics also show ventricular enlargement compared with nonschizophrenic control subjects,5,6 and morphometric studies of schizophrenic brains taken before the era of neuroleptics show "degenerative" changes.7 Thus it appears that ventricular enlargement can occur by mechanisms other than drug toxicity. For example, it has been shown that a variety of intrauterine insults can damage subcortical white matter, which can then lead to ventricular enlargement,2—4,8,9 and it has been known for more than 50 years that these insults can be associated with an increased frequency of cognitive, motor, and behavioral problems in childhood.10 In view of these studies that demonstrate ventricular enlargement as a complication of prematurity and perinatal injury, and taking into account that the most common known diagnoses in our study patients were prematurity and perinatal injury, it is plausible that ventricular enlargement in our patients was secondary to prenatal and/or perinatal injury of subcortical white matter.
The lack of relationship of VBR with age in our study may be explained by our relatively young sample; previous studies have shown that the VBR does not significantly increase until after age 50.11 It has been shown that men have larger ventricles than women,12 and our study is consistent with this observation, although the difference did not attain statistical significance. Seizures appeared to be more common in the No Behavior group, who were not receiving psychotropic medications, but the explanation for this is unclear. It is also possible that the seizure patients in the No Behavior group would have demonstrated behavior problems had they not been receiving seizure medications, which do have psychotropic properties. Our small sample prevents us from examining this issue definitively. Neurological symptoms other than seizures, mental retardation, and violent/destructive behavior were relatively uncommon in both groups of patients, although a few had spasticity (see t1). Psychiatric diagnoses varied greatly in the mentally retarded patient groups but usually reflected the impulsive nature of the behavioral disorders.
Ventriculomegalic behavior-disordered mentally retarded patients may differ from non-ventriculomegalic mentally retarded individuals in their responsiveness to neuroleptic medications, as has been suggested by previous reports in nonretarded schizophrenia patients.13,14 If so, then brain MRI determination could become a useful prognostic and diagnostic tool in nonverbal profoundly retarded adults who also display severe destructive behaviors. To address this question, the present study will have to repeated prospectively with larger numbers of patients, looking especially at the correlation between medication responsiveness and ventricular size.
This work was presented in part at the seventh annual meeting of the American Neuropsychiatric Association, October 12—15, 1995, Pittsburgh, PA.