Historically it has been emphasized that patients with schizophrenia have a cognitive disorder that may represent the characteristic trait of this illness R1563BABJBJBA, R1563BABHIDFB. Several studies R1563BABDFEJA–R1563BABBHHEJ have shown that this deficit is prominent in specific cognitive domains, such as memory, attention, and executive function.
Identification of impaired cognitive functions is only an initial step in the understanding of the pathophysiology of schizophrenia. The natural second step entails study of the intrinsic neural mechanisms that are associated with these impairments R1563BABCHHCI, R1563BABEJCHB. One approach, used by numerous investigators, seeks to associate individual schizophrenic symptoms and cognitive dysfunctions with specific brain regions R1563BABECDJF–R1563BABCEDBD. More recently, we and others R1563BABDFEJA, R1563BABGGAFI–R1563BABCIEAH have chosen to pursue a strategy that posits the disruption of one fundamental cognitive process that defines the phenomenotype of schizophrenia and affects multiple cognitive domains. If this more parsimonious hypothesis proves to be correct, the heterogeneous array of psychological dysfunctions in schizophrenia may be explained by a disruption in a fundamental cognitive process that can be attributed to a dysfunctional neural circuit R1563BABCIEAH.
In light of the fundamental nature of memory in human cognition and earlier reports of marked memory impairment in schizophrenia in relation to performance on other cognitive tests R1563BABBHHEJ, R1563BABEBJEH, we have focused our efforts on developing an understanding of the neural mechanisms of human memory. Frontal, thalamic, and cerebellar regions may constitute a circuit that is the core network used by the human brain to perform a variety of memory tasks R1563BABEICGD–R1563BABGIACI. Structural and functional abnormalities of the frontal lobe, thalamus, and cerebellum have been demonstrated in schizophrenia R1563BABEJCHB–R1563BABCEDBD, R1563BABDDBFE–R1563BABHBFDI. The emerging body of studies showing the role of the cerebellum in several cognitive processes have led to a great interest in understanding its role in psychiatric illnesses R1563BABEICGD, R1563BABCHBAE, R1563BABCAGJI, R1563BABHFGHB.
We previously reported R1563BABEJCHB that during recall of complex narrative material, schizophrenic patients failed to activate the frontal-cerebellar-thalamic circuitry used by healthy individuals to perform the same memory task. However, these findings may be paradigm-specific and not generalizable to other cognitive processes.
In the present study we used positron emission tomography (PET) to examine the neural circuitry used during recall of unstructured verbal material in schizophrenic patients and healthy volunteer subjects. We hypothesized that recalling a different sort of verbal material might elicit abnormalities in cortical-cerebellar-thalamic circuitry in schizophrenic patients as well. However, because of the different mental mechanisms involved in recalling structured versus unstructured verbal information R1563BABBCCDF, task-specific differences in this circuit were also anticipated.
The subjects were 13 healthy volunteers (six men, seven women) recruited from the community and 14 patients (10 men, four women) suffering from schizophrenia who were evaluated at the University of Iowa Mental Health Clinical Research Center. This is the same sample examined in our previous verbal memory study R1563BABEJCHB. Volunteers were screened to rule out current or past history of psychiatric, neurological, or medical illness by using a short version of the Comprehensive Assessment of Symptoms and History R1563BABDIDCJ, medical history, and physical examination. For the schizophrenic patients, the mean duration of illness was 10.6 years (SD=12.2). Clinical symptoms were rated by using the Scale for the Assessment of Negative SymptomsR1563BABEEADE and the Scale for the Assessment of Positive Symptoms R1563BABDBHAI, and these were used to summarize psychopathology in three dimensions—negative, disorganized, and psychotic; the mean negative symptom dimension score was 2.482 (SD=0.68), the mean psychotic symptom dimension score was 3.357 (SD=0.77), and the mean disorganized symptom dimension score was 1.429 (SD=1.67). The mean age of the volunteers was 28.6 years (SD=7.2), and their mean educational achievement was 14.7 years (SD=1.5). The mean age of the patients was 30.7 years (SD=11.3), and their mean educational achievement was 13.2 years (SD=2.2). The patients either were withdrawn from all medication for a 3-week period prior to study (N=11) or had never been treated (N=3). All patients and volunteers were right-handed. After complete description of the study to the subjects, written informed consent was obtained.
Two experimental cognitive conditions were used: recall of novel and practiced word lists (two 15-item lists from the Rey Auditory Verbal Learning Test). For the practiced free recall task, the subjects were asked to learn the group of words during an initial training session that occurred 1 week prior to the imaging study. The words were presented auditorily, at the rate of one per second, and the items recalled by the subjects were spoken aloud. In this session, presentation and recall of the word list were repeated until all 15 words were recalled (irrespective of order). On the day before the PET study the subjects were given a "refresher course," during which the group of words was presented again to ensure that recall was 100% accurate. For the condition referred to as "novel recall," the subjects heard the word list 60 seconds before the imaging study and were told that they would be requested to recall as many words as they could during the PET scan. For both conditions the word list was read to the subject immediately prior to the image acquisition, and recall began approximately 10 seconds prior to the arrival of the bolus of [15O]water in the brain, with bolus arrival time determined from an initial low-activity sham study R1563BABGADFG. The subjects were asked to recall as much of the list as possible and to continue to try to recall for the duration of image acquisition, repeating words until image acquisition was complete. Both active recall tasks were compared with a reference condition, lying quietly with eyes closed, which is referred to as REST (random episodic silent thought). Four additional conditions were included in the overall PET experiment R1563BABEICGD.
PET and MR Data Acquisition
Magnetic resonance (MR) scans, to be used for anatomic localization of functional activity, were obtained with a standard T1-weighted three-dimensional gradient-echo sequence on a 1.5-T GE scanner (TE=5 msec, TR=24 msec, flip angle=40°, number of excitations=2, field of view=26 mm, matrix=256×192, slice thickness=1.5 mm).
PET data were acquired with a bolus injection of 75-mCi [15O]water in 5–7 ml saline by using a GE PC4096-plus 15-slice whole-body scanner. The PET acquisition details have been described elsewhere R1563BABEJCHB, R1563BABGADFG. Cerebral blood flow (CBF) was calculated on a voxel-by-voxel basis by using an autoradiographic method R1563BABBGDAA.
The quantitative PET blood flow images and MR images were analyzed by using the locally developed software package BRAINS R1563BABBBBDF, R1563BABGGJBB. The PET image of each individual was then fit to that individual’s MR scan by using a surface-fit algorithm R1563BABCIADF, R1563BABIAGAG. The PET image could be localized on coregistered MR and PET images where the MR image represented the "average brain" of the subjects in this study R1563BABFIIDI. Specific between-group differences in neural activation were examined by a direct statistical comparison of the patients and healthy volunteers using our specific randomization analysis R1563BABGBADG.
Two types of randomization analysis were conducted. One type examined the differences between the two groups by using an initial subtraction analysis (practiced recall minus REST in patients versus control subjects and novel recall minus REST in patients versus control subjects). This analysis reflects differences between the two groups in both conditions and removes differences in head shape or size, but it does not reveal whether the differences were due to the memory task or to REST. To address this latter point we also did randomization comparisons of the two groups for each of the individual conditions with no subtractions of REST.
The selection of a significance threshold for PET studies is always arbitrary. Consistent with our previous studies, we used an uncorrected p value of 0.005. This threshold closely approximates the sizes of peaks defined by t=3.61 when the Montreal method is used R1563BABGBADG. Every peak has been defined by the number of contiguous significant voxels that constitute the peak and by the standard Talairach atlas coordinates R1563BABGBGFH. We report only the peaks that contain more than 50 adjacent voxels. The region name given to each peak was based on direct visual inspection of the coregistered MR and PET images.
The patients and control subjects did not differ significantly in task performance either during the practiced condition (mean=12.2 words, SD=4.0, versus mean=14.2 words, SD=2.9) (t=1.40, df=25, p=0.17) or during the novel condition (mean=5.4 words, SD=1.9, and mean=6.6 words, SD=1.5, respectively) (t=1.76, df=25, p=0.09). Therefore, it is unlikely that between-group differences in blood flow were due to differences in task performance.
Results of the randomization analysis comparing the patients and control subjects during the practiced task are shown in T1 and displayed in F1. The patients showed a relative decrease in flow in the left dorsolateral prefrontal cortex (Brodmann area 46), bilateral medial frontal cortex (Brodmann areas 32 and 43), left supplementary motor area (Brodmann area 6), right premotor cortex, left thalamus, left cerebellar regions, anterior vermis, and the right cuneus. Similar results were obtained by using the randomization method without subtracting the REST condition. The patients showed a relative decrease in flow in the right frontal cortex, anterior and posterior cingulate, and left cerebellar regions.
Results for the novel memory task are shown in T1 and F1. Brain regions showing a relative decrease in flow in the patients included the left inferior frontal cortex (Brodmann area 44), right anterior cingulate, right thalamus, and bilateral cerebellum (left greater than right). Prefrontal flow decreases were not observed in this analysis when the p<0.005 threshold was used. In an exploratory post hoc analysis we used one-tailed t tests and repeated this analysis using a p<0.01 threshold. At this threshold, a peak of relatively decreased flow was found in the right prefrontal lobe (Brodmann area 47/11) (x=20, y=22, z=–14). Identical results were found when we used between-group comparisons without subtracting the REST condition. The patients showed a relative decrease in flow in the right anterior cingulate, thalamus, and cerebellum.
We also examined the results for both cognitive tasks for each subject group independently, using the standard Montreal (Worsley) method for subtracting two experimental conditions R1563BABGBADG. In this analysis we examined practiced recall minus REST and novel recall minus REST in the healthy volunteers and in the patients. Comparison of the activation patterns in these analyses revealed that schizophrenic patients fail to activate brain regions that healthy volunteers utilize during performance of the two memory tasks (F2). Thus, convergent statistical analysis using three different methods confirms that patients with schizophrenia have abnormalities in cortical-cerebellar-thalamic-cortical circuitry.
Only one area with a relative increase in blood flow, the left parietal lobe, was found in the patients with schizophrenia during either memory task.
The present study has shown that, compared to healthy individuals, patients with schizophrenia who recall a list of words fail to activate frontal-cerebellar-thalamic circuitry. These findings confirm the hypothesis, based on our previous PET studies, that patients suffering from schizophrenia have a relative decrease in CBF in the same interconnected cerebral regions while performing different types of cognitive tasks R1563BABEJCHB, R1563BABIDEEH. As anticipated, abnormalities in the cortical-cerebellar-thalamic-cortical circuitry are generalized across a variety of memory tasks and are not task-specific.
Lower flow during the practiced task was observed in the left dorsolateral prefrontal cortex (Brodmann area 46), left supplementary motor area (Brodmann area 6), and right premotor cortex. The prefrontal cortex has a pivotal role in carrying out higher cognitive functions by virtue of its reciprocal connections with other brain regions R1563BABDEJIH, R1563BABJFCEH. The results from the present study are consistent with the construct of frontal lobe dysfunction in schizophrenia R1563BABCEDBD, R1563BABGCIGG–R1563BABGECHF. The dorsolateral prefrontal cortex (Brodmann area 46) has been associated with verbal working memory and with the generation of willed actions in healthy individuals R1563BABGIACI, R1563BABJBEGH. It has also been hypothesized that a defect in working memory function or a disorder of willed actions could be at least one of the crucial cognitive impairments in schizophrenia R1563BABDFEJA, R1563BABEAEJH. Taken together, these results allow us to suggest that a basic impairment of the prefrontal cortex may lead to a core cognitive disturbance in schizophrenia that is shown in several cognitive domains (i.e., working memory or generation of willed actions) R1563BABCIEAH.
The lack of differences between patients and control subjects in prefrontal lobe regional CBF during the novel task performance was an unexpected finding. This may be because one strategy used during the novel recall task is the maintenance of the verbal information in an active state (short-term maintenance of verbal information), which has been shown to activate dorsolateral prefrontal cortical areas R1563BABJDJDE. This may be why schizophrenic patients display a close-to-normal pattern of prefrontal cortex activation in response to the challenge of the higher prefrontal requirement allocated to task performance. If so, the two groups would not differ in this region when our standard significance threshold is used, although differences might emerge at a lower one. Because our search was hypothesis-driven by our previous work, we examined results obtained by using a lower threshold. With this lower threshold (p<0.01), the schizophrenic patients did show a relative decrease in regional CBF in the right prefrontal cortex.
Another striking finding of our study is the relative decrease in CBF in a large area that encompasses the left lateral and medial rostral supplementary motor area (Brodmann area 6). The rostral supplementary motor area is critically involved in internal representation of time and in internal selection of movement, and it is activated during "complex" tasks requiring selection of response R1563BABCJICJ, R1563BABCJABD. Although little is known about changes in the supplementary motor area in schizophrenia, relatively decreased activation of the supplementary motor area in schizophrenic patients during motor tasks has been reported R1563BABIGCCE.
Decreased thalamic regional CBF in schizophrenic patients was observed on the right for the novel task and on the left for the practiced task. Neuropathological R1563BABBEDHJ, R1563BABJGEJB, in vivo anatomical R1563BABFIIDI, R1563BABFABAD, and functional R1563BABEJCHB, R1563BABFHAGG studies have shown thalamic abnormalities in schizophrenia. A defect in information processing capacity and in filtering input stimuli has been proposed as one component of the core underlying deficit in schizophrenia R1563BABBHHBD, R1563BABGFIJJ. Because of the role of the thalamus in modulating attention and in filtering external information in order to exclude redundant or distracting stimuli, it has been suggested that thalamic dysfunction might play a crucial role in the etiopathological model of schizophrenia R1563BABFIIDI, R1563BABEBCJB.
The patients with schizophrenia also showed a relative decrease in flow in the right anterior cingulate during performance of the recall task involving the novel word list. The anterior cingulate has an important role in emotional and attentional mechanisms R1563BABFBGII, which are also altered in schizophrenia R1563BABFJGEE, R1563BABDDEDI. The healthy individuals showed greater neural activation in the anterior cingulate cortex during performance of the novel task, which has a higher attentional demand than do practiced tasks R1563BABCHBAE, R1563BABCGAEC. Low flow or activity in the anterior cingulate has been reported in schizophrenic patients during several cognitive tasks R1563BABEJCHB, R1563BABEBAEB, R1563BABCFIDD. Our results showing a decreased flow in the right anterior cingulate, right prefrontal cortex, and right thalamus only during the novel condition suggest that schizophrenic patients may have an inability to activate a right attentional circuit during memory tasks for which the attentional demands are also high.
Multiple cerebellar regions were found to have a relative decrease in flow in schizophrenic patients during both conditions. Our group R1563BABEJCHB, R1563BABIDEEH has previously demonstrated that schizophrenic patients display relatively lower blood flow in diverse cerebellar regions during the recall of complex narratives and during recognition memory for words. Intriguingly, neuroleptic-naive schizophrenic patients had a relative increase of CBF in the cerebellum when they were examined in a REST condition R1563BABFCCJD.
It has been proposed that a cerebellar dysfunction in schizophrenia may lead to "poor mental coordination," resembling abnormalities in motor coordination and sequencing R1563BABBFAJJ. Results from the present study also show that patients with schizophrenia have a relatively lower blood flow during recall of a practiced word list in the anterior vermis. Interestingly, some anatomical studies R1563BABBGIJI, R1563BABHEGBD have shown a smaller anterior vermis in schizophrenia.
Historically, dysmetria has been defined as the inability of the individual to time the control at onset and offset of activity in the appropriate pairs of muscles. Dysfunction in the central timing process has been attributed to the lateral cerebellum, the putamen, and the supplementary motor area R1563BABBFAJJ, R1563BABIAGDE. It seems possible that these brain regions related to the timing and programming of movement may also be involved in timing and controlling the fluid coordination of mental activity R1563BABCBFCD–R1563BABGBBAI.
A dysfunctional circuitry linking brain regions involved in timing and sequencing mental functions (i.e., cerebellum and rostral supplementary motor area) to regions involved in high-level cognitive processes (i.e., prefrontal lobe) provide strong support for a "cognitive dysmetria" model of schizophrenia. This model posits that a neural misconnection in cortical-cerebellar-thalamic-cortical circuitry in schizophrenic patients may lead to difficulty in coordinating and sequencing mental processes, such as receiving, processing, retrieving, and expressing information, that are needed to achieve desired cognitive acts. While this circuit is hypothesized to be impaired across a broad range of cognitive tasks, specific differences in regional CBF patterns across different types of cognitive tasks are also expected, since different interconnected brain regions could be recruited according to specific features of the cognitive task. On the basis of these findings it is possible to speculate that the clinical heterogeneity of schizophrenia can be understood in the context of dynamic shifts in this disturbed circuitry, whereas the disturbed circuitry represents a basic neural phenomenon common to patients suffering from schizophrenia.
Presented at the 4th International Conference on Functional Mapping of the Human Brain, Montreal, June 7–12, 1998. Received April 7, 1998; revision received Sept. 8, 1998; accepted Oct. 19, 1998. From the Mental Health Clinical Research Center, Department of Psychiatry, College of Medicine, University of Iowa Hospitals and Clinics; and the Instituto de Salud "Carlos III," Madrid. Address reprint requests to Dr. Andreasen, Department of Psychiatry, University of Iowa, MHCRC, 2911 JPP, 200 Hawkins Dr., Iowa City, IA 52242-1057. Supported by NIMH grants MH-31593 and MH-40856 and NIMH Clinical Research Center grant MH-43271.
Brain Images Showing Differences in Regional Cerebral Blood Flowa Between 13 Healthy Volunteers and 14 Schizophrenic Patients During Recall of (a) Practiced and (b) Novel Word Listsb, c
aThree orthogonal views are shown: axial (top), sagittal (middle), and coronal (bottom). Crosshairs are used to show the location of the slice. Statistical (randomization) maps of the PET data are superimposed on a composite magnetic resonance (MR) image derived by averaging the MR scans from the subjects. Within each condition, the "peak map" (left) provides a descriptive picture of areas where all contiguous voxels exceed the predefined threshold for statistical significance. The "t map" (right) represents the value of t for all voxels in the image and shows the general geography of the activations; the color bar at the right shows the t statistic values. Only regions with a relative decrease in flow for schizophrenic patients are shown.
bFor recall of the practiced word list (part a), regions in blue/purple tones indicate lower flow in patients. The crosshairs point to the left thalamus. The view of the sagittal plane indicates that blood flow in the left medial frontal, left thalamus, and left cerebellum is significantly lower in schizophrenic patients.
cFor recall of the novel word list (part b), regions in blue/purple tones indicate lower flow in patients during the novel recall condition. The crosshairs point to the right thalamus. The view of the sagittal plane indicates that blood flow in the right anterior cingulate, right thalamus, and several regions in the cerebellum is significantly lower in schizophrenic patients.
Brain Images Showing Relativea Regional Cerebral Blood Flowb Separately in 13 Healthy Volunteers and 14 Schizophrenic Patients During Recall of (a) Practiced and (b) Novel Word Listsc, c
aBlood flow during random episodic silent thought (REST) was subtracted from blood flow during word list recall. The images show the relative activations in normal volunteers and patients with schizophrenia for each recall condition. These images indicate that the patients have markedly decreased or absent activations in key areas of the cortical-cerebellar-thalamic-cortical circuitry.
bThree orthogonal views are shown: axial (top), sagittal (middle), and coronal (bottom). Crosshairs are used to show the location of the slice. Statistical (Worsley) maps of the PET data are superimposed on a composite magnetic resonance (MR) image derived by averaging the MR scans from the subjects. For each subject group in each condition, the "peak map" (left) provides a descriptive picture of areas where all contiguous voxels exceed the predefined threshold for statistical significance. The "t map" (right) represents the value of t for all voxels in the image and shows the general geography of the activations; the color bar at the far right shows the t statistic values. Only regions with a relative decrease in flow for schizophrenic patients are shown.
cDuring recall of the practiced word list (part a), healthy volunteers use the frontal lobe, thalamus, and cerebellum. Schizophrenic patients display a much smaller increase in flow in these regions, and peaks exceeding the significance threshold are sometimes completely absent.
dDuring recall of the novel word list (part b), schizophrenic patients show less increase in flow in the brain regions that are activated in the volunteers.