- Email: [email protected]
Hypofrontality in schizophrenia: distributed dysfunctional circuits in neuroleptic-naïve patients
THE LANCET
Early reports
Hypofrontality in schizophrenia: distributed dysfunctional circuits in neuroleptic-naïve patients
Nancy C Andreasen, Daniel S O’Lear y, Michael Flaum, Peg Nopoulos, G Leonard Watkins, Laura L Boles Ponto, Richard D Hichwa
Summary Background There have been repor ts that patients with schizophrenia have decreased metabolic activity in prefrontal cor tex . However, findings have been confounded by medication effects, chronic illness, and difficulties of measurement. We aimed to address these problems by ex amination of cerebral blood flow with positron emission tomography (PET). Methods We studied 17 neuroleptic-naïve patients at the early stages of illness by means of image analysis and statistical methods that can detect abnormalities at the gyral level. Findings An initial omnibus test with a randomisation analysis indicated that patients differed from normal controls at the 0·06 level. In the follow-up analysis, three separate prefrontal regions had decreased per fusion (lateral, orbital, medial), as well as regions in inferior temporal and parietal cor tex that are known to be anatomically connected. Regions with increased per fusion were also identified (eg, thalamus, cerebellum, retrosplenial cingulate), which suggests an imbalance in distributed cor tical and subcor tical circuits. Interpretation These distributed dysfunctional circuits may form the neural basis of schizophrenia through cognitive impairment of the brain, which prevents it from processing input efficiently and producing output effectively, thereby leading to symptoms such as hallucinations, delusions, and loss of volition.
Lancet 1997; 349: 1730–34
Mental Health Clinical Research Centre and PET Imaging Center, University of Iowa College of Medicine and Hospitals and Clinics, Iowa City, IA 52242, USA (Prof N C Andreasen MD, D S O’Lear y PhD, M Flaum MD, P Nopoulos MD, G L Watkins PhD, L L Boles Ponto PhD, R D Hichwa PhD) Correspondence to: Prof N C Andreasen
1730
Introduction In 1974, Ingvar and Franzen coined the term hypofrontality to denote the relative decrease they found in their calculation of a ratio of frontal to post-central blood flow in schizophrenia patients.1 Since then, hypofrontality in schizophrenia has been studied by a range of methods: measurement of regional blood flow with xenon by use of cortical probes;2,3 measurement of cerebral blood flow and glucose metabolism by positron emission tomography (PET);4–7 neuropathological studies of tissue taken at necropsy;8 anatomical measurements by magnetic resonance (MR) imaging;9,10 animal models of working memory;11 and multiple techniques of cognitive psychology and neuropsychology.12,13 Although these studies have broadly suggested prefrontal dysfunction in schizophrenia, the findings are not consistent, and Gur and Gur14 have argued that the concept of hypofrontality should be questioned. There are several difficulties in the study of hypofrontality in schizophrenia15—most importantly, perhaps, the anatomical extent, complexity, and functional diversity of the frontal cortex itself, which includes both neocortex and more primitive limbic cortex, and carries out functions that range from working Region
t max
Relative decreases in patients Right dorsolateral frontal ⫺3·80 ⫺3·37 Left dorsolateral frontal ⫺3·45 Right orbital frontal ⫺4·04 Left orbital frontal ⫺3·35 Medial frontal ⫺3·33 Right inferior temporal ⫺4·18 Left inferior temporal ⫺3·58 Mid cingulate ⫺3·12 ⫺3·40 Precuneus ⫺4·36 Left parietal ⫺3·49 Primar y visual cortex ⫺4·48 Relative increases in patients Left inferior frontal Right thalamus Left thalamus Right retrosplenial cingulate Left retrosplenial cingulate Left parietal/supramarginal Fight fusiform/occipital Left cerebellum
Right cerebellum
⫺3·55 ⫺3·47 ⫺3·17 ⫺4·41 ⫺3·82 ⫺4·12 ⫺4·49 ⫺3·85 ⫺3·41 ⫺3·34 ⫺3·47 ⫺3·41
Volume (N voxels)
x
y
z
79 60 48 211 147 62 320 121 132 129 595 64 356
⫺35 ⫺39 ⫺32 ⫺30 ⫺24 ⫺2 ⫺58 ⫺53 ⫺ 1 ⫺5 ⫺16 ⫺42 ⫺ 4
⫺12 ⫺0 ⫺18 ⫺29 ⫺14 ⫺39 ⫺34 ⫺50 ⫺13 ⫺32 ⫺70 ⫺73 ⫺97
⫺24 ⫺22 ⫺21 ⫺21 ⫺22 ⫺12 ⫺21 ⫺21 ⫺38 ⫺40 ⫺25 ⫺35 ⫺ 0
224 76 69 354 106 238 2608 716 153 151 124 117
⫺31 ⫺13 ⫺7 ⫺16 ⫺14 ⫺40 ⫺15 ⫺7 ⫺24 ⫺28 ⫺10 ⫺23
⫺29 ⫺22 ⫺22 ⫺57 ⫺62 ⫺68 ⫺72 ⫺88 ⫺78 ⫺77 ⫺66 ⫺53
⫺3 ⫺18 ⫺18 ⫺21 ⫺24 ⫺18 ⫺9 ⫺22 ⫺13 ⫺25 ⫺39 ⫺34
Comparison of regional blood flow in drug-naïve schizophrenic patients and matched healthy volunteers
Vol 349 • June 14, 1997
THE LANCET
memory to regulation of eye movements. Through lesion studies, three functional areas have been identified (orbital, dorsolateral, and medial16), though these subdivisions are an oversimplification; the term hypofrontality is thus too general, since it fails to reflect the diversity of specialised regions in frontal cortex. The effects of stage of illness (ie, first episode vs chronically ill) are contested. Early studies of chronically ill patients generally reported decreases in frontal flow in schizophrenia during rest or a specific cognitive task1,4,5,17 (though some were negative).14,18 Studies of first-episode or never-treated patients are less common; some have supported hypofrontality,15 whereas others have found no differences.6,19 A few influential studies have suggested the opposite finding—namely, that first-episode patients have increases in frontal flow.20,21 If true, these findings may
suggest that metabolic activity in the prefrontal cortex decreases over time as a result of a degenerative process, toxic effects of schizophrenia, or effects of medications to treat the illness. Medication effects on cerebral blood flow are a confounding factor that cannot easily be separated from stage of illness. The results of Berman,22 Weinberger,17 and their colleagues suggest that hypofrontality was unrelated to treatment in chronically ill patients. By use of a more powerful longitudinal design in a sample of five drug-naïve patients, Szechtman and colleagues21 found that 1 year of treatment increased metabolism in the basal ganglia, but did not affect frontal metabolism. However, when they used a cross-sectional design to compare a larger group of patients who had been receiving medication for 1 year with a group who had
Figure 1: Statistical maps of PET data comparing patients and controls Shows negative peaks in right and left orbital frontal and inferior temporal regions (transaxial and coronal), lateral frontal regions (coronal), and precuneus (sagittal). Positive peaks or regions are seen in cerebellum (transaxial and sagittal), retrosplenial cingulate and fusiform (sagittal), and right thalamus (coronal). Green crosshairs show location of slice. Images show location as if viewer is standing at foot of bed (transaxial views) or facing patient (coronal views). Significantly activated regions superimposed on composite MR image derived by averaging MR scans from patients. Peak maps show small areas where all contiguous voxels exceed predefined threshold for significance (p=0·005). Areas are small because data were not smoothed (filtered); t maps show t value for all voxels in image and provide general over view of landscape of changes in blood flow. Planes illustrate location of relevant activity. Areas of decreased flow in patients in blue tones (“negative” peaks or regions); areas of increased flow in red/yellow/yellow-green tones (”positive” peaks or regions).
Vol 349 • June 14, 1997
1731
THE LANCET
been treated for an average of 7 years, Szechtman and colleagues found relative hypofrontality in the chronically treated patients, which suggests that long-term neuroleptic treatment may produce a decrease in frontal metabolic tone. Cognitive activity during image acquisition may also have substantial impact on the assessment of hypofrontality. Most early studies that supported hypofrontality involved examination of patients in a “resting state”.2,6,7 But the word “rest” is misleading, since the brain does not become inactive or empty of thought in the absence of specific experimental tasks or instructions; on the contrary, patients report after scans that when at “rest” they typically recalled past experiences or made future plans, both of which require prefrontal activity.23 Other investigators tried to “stress test” the prefrontal cortex by use of cognitive challenges.15,17,24,25 Although this approach has the advantage of standardising mental activity, the hypofrontality reported is difficult to interpret because schizophrenic patients typically carry out the tasks less well than healthy volunteers. There is thus a “chicken and egg” problem: does the decrease in frontal blood flow or metabolism result from an intrinsic neural deficit, or from lack of motivation and poor task performance? Other investigators have attempted to answer this question by matching task performance in patients and controls.25 Symptoms may also influence measurements of cerebral blood flow or metabolism. For example, negative symptoms have been related to hypofrontality;1,7,15 and a study of a small group of patients who experienced auditory hallucinations during scanning indicated flow increases in multiple brain regions.26 The concept of hypofrontality raises important questions about the fundamental nature of schizophrenia—its neural substrates, its course, and its treatment. Do specific regions of the prefrontal cortex have abnormal metabolism or blood flow? And, if so, is this abnormality related to symptoms, cognitive state, or treatment effects? Most importantly, are toxic medication or toxic illness effects on the prefrontal cortex a valid concern? To eliminate confounding factors that arise from treatment and chronicity, we focused on a sample of neuroleptic-naïve patients studied early in the course of their illness, and compared them with a group of healthy volunteers closely matched for sex, age, and parental education. To eliminate the confounding effect of poor task performance, we studied them during the resting state. To gauge the relation between symptoms and blood flow, we measured symptoms just before the scan. To assess whether specific subregions of the prefrontal cortex are selectively dysfunctional, we analysed data by image analysis and statistical techniques that enabled us to identify abnormalities at the gyral level.
Methods Patients and cognitive state We studied 17 neuroleptic-naïve patients with recent onset of schizophrenia according to Diagnostic and Statistical Manual (DSM) IV. Ten were male and seven female. Their mean age was 26·2 (SD 7·3) years. Most were experiencing schizophrenia for the first time and had not been admitted to hospital previously. Severity of symptoms was measured just before the PET study with the Scale for Assessment of Positive Symptoms (SAPS) and
1732
the Scale for Assessment of Negative Symptoms (SANS). Healthy volunteers were recruited from the community through a newspaper advertisement. Ten were male and seven female. Their mean age was 26·4 (7·3) years. Controls were similar to patients in their extent of parental education (13·3 [SD 3·4] years and 12·5 [3·9], respectively). All patients gave written informed consent. They were studied during the cognitive state usually referred to as “rest”—that is, they were not given any specific instructions about mental activity. This cognitive state is better characterised as a type of episodic memory.23
PET and MR data acquisition and analysis PET and MR data were acquired by techniques described previously. We used two different methods of statistical analysis. The first was an atlas-based automated technique for dividing the brain into specific lobes; we compared mean flow within the left and right frontal, temporal, parietal, and occipital lobes and the cerebellum by t tests.27 This analysis produced a slightly crude index of hypofrontality and of differences in other brain lobes. The second statistical method was a randomisation analysis, a non-parametric technique particularly appropriate for complex between-group comparisons in PET studies.25,28 We used this method to compare mean blood flow in the two groups, and thereby calculated an index of regional differences at the gyral level. Areas where statistical analysis showed specific regional differences were identified as “peaks”; we have reported the highest t (t max) value (Student’s t test) associated with the peak and the number of adjacent pixels that comprise the peak. Location of the pixel with the highest t-max value was identified with the standard Talairach Atlas coordinates. The region name given to the peak was based on direct visual inspection of registered MR and PET images. Since controls were “subtracted” from patients, regional differences associated with a negative t value indicate that flow was lower in patients than in controls, whereas those with a positive t value indicate that flow was higher in patients than in controls.
Results The slightly crude measures of regional blood flow, which involved examination of whole lobes of the brain, did not identify any between-group differences in blood flow, with or without correction for whole-brain blood flow (normalisation). Thus, no specific hypofrontality was identified by this method. The randomisation analysis did, however, indicate differences in frontal and other regions (table and figures 1 and 2). The initial omnibus test showed that patients differed from normal controls at the 0·06 level. In followup analysis, the patients had extensive areas of decreased flow in several regions on both the right and left lateral convexities of the prefrontal cortex. These include Brodmann areas 8, 6, and 46. Area 46 is believed to be active during working memory tasks;29 8 contains the front eye fields, which assist in the regulation of attention; and 6 is a supplementary motor area. The patients also showed bilateral decreases in orbital frontal regions, which were once ablated by leucotomy as a treatment for schizophrenia and are thought to have a vital role in emotion and social judgment. Furthermore, a large area of the medial frontal cortex also showed decreased flow. Bilateral areas in the inferior temporal cortex were also relatively inactive, as were the midcingulate and precuneus areas, left parietal lobe, and visual cortex. Visual inspection of the MR-registered PET scans suggested a specific pattern of inactivity. One cluster of inactivity appeared to extend from primary visual cortex through the inferior temporal cortex
Vol 349 • June 14, 1997
THE LANCET
Figure 2: Statistical maps of PET data in patient y Shows two negative lateral frontal peaks on right (transaxial) and bilaterally (coronal), orbital frontal (sagittal and coronal), precuneus (transaxial) and midcingulate (sagittal). Positive peaks seen in cerebellum (sagittal), left parietal (transaxial), and retrosplenial cingulate (transaxial).
towards the hippocampal complex (which was relatively inactive, but did not reach significance). Another such cluster extended more dorsally, and included the precuneus and parietal association cortex. The dorsal and ventral frontal regions are known to be anatomically linked to these more posterior regions.29 Several of the patients’ brain regions also had increases in flow compared with the same regions in normal controls. Most areas of increased flow were in posterior regions, including occipital and parietal association cortex (secondary visual cortex and the supramarginal gyrus), and the retrosplenial cingulate gyrus. In addition, there were bilateral increases in the thalamus and the cerebellum. Positive correlations were observed between the psychotic dimension of the Scale for Assessment of Positive Symptoms and flow in the thalamus (r=0·90), cerebellum (0·82), and left and right anterior inferior temporal regions (0·73 and 0·83, respectively). Significant negative correlations were found between the negative dimension of the Scale for Assessment of Negative Symptoms, and several anterior prefrontal
Vol 349 • June 14, 1997
regions (r=0·80 and 0·74, respectively).
Discussion This study supports previous investigatons that suggest schizophrenia patients have a dysfunction in prefrontal cortex. The two methods of analysis imply a possible reason for inconsistent results in earlier studies. When the entire frontal lobe was averaged together, hypofrontality was not detected. When we used a more sensitive and specific method of analysis, however, we found many sites with decreased flow in the prefrontal cortex of the schizophrenia patients. Since this study examined a large group of drug-naïve patients early in the course of their illness, flow abnormalities are not secondary to effects of neuroleptic treatment or chronic psychosis. Because areas of decreased flow were found in these patients in three major regions of prefrontal cortex (lateral, orbital, and medial), the results suggest that hypofrontality is a primary abnormality in schizoprenia. These results may 1733
THE LANCET
allay the concern, raised by earlier studies,20,21 that decreased perfusion in regions of prefrontal cortex is a consequence of chronic treatment or toxic effects of chronic psychosis. Our results suggest that the term hypofrontality could be replaced with more specific terms that designate which regions of the prefrontal cortex are dysfunctional. In this study, three different frontal areas were identified. The areas of decreased flow on the lateral convexities include regions that have been implicated as active during tasks that require working memory (both longterm and short-term), attention, and arousal.29,30 The orbital regions contain more primitive cortex and are believed to carry out “limbic” functions such as emotional memory, whereas the medial region of the prefrontal cortex is involved in the initiation of behaviour.16 Moreover, we are not observing hypofrontality alone, but dysfunctional circuitry that is distributed throughout the brain. The associated flow decreases in inferior temporal regions may result from dysfunction, involving orbital frontal regions and temporal-hippocampal areas, in a “limbic network”. The decreases in dorsolateral frontal regions and the precuneus and parietal lobes, however, may reflect a more “neocortical” network.11–29 This study identified many areas that showed increases in flow; these increases may shed light on brain pathology in schizophrenia. One report suggested that patients with active auditory hallucinations show increased flow in several brain regions: thalamus, basal ganglia, anterior cingulate, right and left parahippocampal gyri, and cerebellum.26 Correlational analyses of patients’ symptoms in our study are consistent with that report. The extensive areas of increased flow may indicate that these brain regions are working harder to compensate for decreases in other regions. Increases and decreases are closely related, and may reflect an overall imbalance in cortical and subcortical excitatory (glutamatergic) and gabaergic (inhibitory) circuitry. The net result is that the brain cannot process its input efficiently or produce its output effectively; some parts of the circuits are working too hard, whereas others are underactive. Schizophrenia is characterised by dysfunctional regions in prefrontal cortex, but the illness is clearly more complex than the term hypofrontality allows. Schizophrenia involves an imbalance in circuits distributed throughout the brain—including multiple cortical and subcortical regions—that leads to impairment of the ability to set priorities, process, and produce information, and to turn it into meaningful thoughts and behaviour. This imbalance in circuits is expressed as psychotic or negative symptoms. References 1
2
3
4 5
Ingvar DH, Franzen G. Abnormalities of cerebral blood flow distribution in patients with chronic schizophrenia. Acta Psychiatr Scand 1974; 50: 425–62. Gur RE, Gur RC, Skolnick BE, et al. Brain function in psychiatric disorders: regional cerebral blood flow in unmedicated schizophrenics. Arch Gen Psychiatry 1985; 42: 329–34. Berman KF, Illowsky BP, Weinberger DR. Physiological dysfunction of dorsolateral prefrontal cortex in schizophrenia: further evidence for regional and behavioral specificity. Arch Gen Psychiatry 1988; 45: 616–22. Buchsbaum MS, Ingvar DH, Kessler R, et al. Cerebral glucography with positron tomography. Arch Gen Psychiatry 1982; 39: 251–59. Volkow ND, Wolf AP, Van Gelder P, et al. Phenomenological correlates
1734
6
7
8 9
10
11
12
13 14 15
16
17
18
19
20
21
22
23
24
25
26 27
28
29
30
of metabolic activity in 18 patients with chronic schizophrenia. Am J Psychiatry 1987; 144: 151–58. Early TS, Reiman EM, Raichle ME, Spitznagel EL. Left globus pallidus abnormality in never-medicated patients with schizophrenia. Proc Natl Acad Sci USA 1987; 84: 561–63. Liddle PF, Friston KJ, Frith CD, Hirsch SR, Jones T, Frackowiak RSJ. Patterns of cerebral blood flow in schizophrenia. Br J Psychiatry 1992; 60: 179–86. Benes FM. Is there a neuroanatomic basis for schizophrenia? An old question revisited. Neuroscientist 1995; 1: 104–15. Andreasen NC, Nasrallah HA, Dunn VD, Olson SC, Grove WM, Ehrhardt JC, et al. Structural abnormalities in the frontal system in schizophrenia: a magnetic resonance imaging study. Arch Gen Psychiatry 1986; 43: 136–44. Andreasen NC, Flashman L, Flaum M, et al. Regional brain abnormalities in schizophrenia measured with magnetic resonance imaging. JAMA 1994; 272: 1763–69. Goldman-Rakic PS, Selemon LD, Schwartz ML. Dual pathways connecting the dorsolateral prefrontal cortex with the hippocampal formation and parahippocampal cortex in the rhesus monkey. Neuroscience 1984; 12: 719–43. Bilder RM, Turkel E, Lipschutz-Borch L, Lieberman JA. Antipsychotic medication effects on neuropsychological functions. Psychopharmacol Bull 1992; 28: 353–66. Frith CD. The cognitive neuropsychology of schizophrenia. East Sussex: Lawrence Erlbaum, 1992. Gur RC, Gur RE. Hypofrontality in schizophrenia: RIP. Lancet 1995; 345: 1383–85. Andreasen NC, Rezai K, Alliger R, et al. Hypofrontality in neuroleptic-naive and chronic schizophrenic patients: assessment with xenon-133 single-photon emission computed tomography and the Tower of London. Arch Gen Psychiatry 1992; 49: 943–58. Mega MS, Cummings JL. Frontal-subcortical circuits and neuropsychiatric disorders. J Neuropsychiatry Clin Neurosci 1994; 6: 358–70. Weingerger DR, Berman KF, Zec RF. Physiological dysfunction of dorsolateral prefrontal cortex in schizophrenia I: regional cerebral blood flow (rCBF) evidence. Arch Gen Psychiatry 1986; 43: 114–24. Wiesel FA, Wik G, Sjogren I, Blomqvist G, Greitz T. Altered relationships between metabolic rates of glucose in brain regions of schizophrenic patients. Acta Psychiatr Scand 1987; 76: 642–47. Gur RE, Mozley PD, Resnick SM, et al. Resting cerebral glucose metabolism in first-episode and previously treated patients with first episode schizophrenia relates to clinical features. Arch Gen Psychiatry 1995; 52: 657–67. Cleghorn JM, Garnett ES, Nahmias C, et al. Increased frontal and reduced parietal glucose metabolism in acute untreated schizophrenia. Psychiatry Res 1989; 28: 119–33. Szechtman H, Nahmias C, Garnett S, et al. Effect of neuroleptics on altered cerebral glucose metabolism in schizophrenia. Arch Gen Psychiatry 1988; 45: 523–32. Berman KF, Zec RF, Weinberger DR. Physiologic dysfunction of dorsolateral prefrontal cortex in schizophrenia: role of neuroleptic treatment, attention, and mental effort. Arch Gen Psychiatry 1986; 43: 126–35. Andreasen NC, O’Leary DS, Cizadlo T, et al. Remembering the past: two facets of episodic memory explored with positron emission tomography. Am J Psychiatry 1995; 152: 1576–85. Buchsbaum MS, Neuchterlein KH, Haier RJ, et al. Glucose metabolic rate in normals and schizophrenics during the continuous performance test assessed by positron emission tomography. Br J Psychiatry 1990; 156: 216–27. Andreasen NC, O’Leary DS, Cizadlo T, et al. Schizophrenia and cognitive dysmetria: a PET study of dysfunctional prefrontalthalamic-cerebellar circuitry. Proc Natl Acad Sci USA 1996; 93: 9985–90. Silbersweig DA, Stern E, Frith C, et al. A functional neuroanatomy of hallucinations in schizophrenia. Nature 1995; 378: 176–79. Andreasen NC, Rajarethinam R, Cizadlo T, et al. Automatic atlasbased volume estimation of human brain regions from the MR images. J Comput Assist Tomogr 1996; 20: 98–106. Arndt S, Cizadlo T, Andreasen NC, Heckel D, Gold S, O’Leary DS. Tests for comparing images based on randomization and permutation methods. J Cereb Blood Flow Metab 1996; 16: 1271–79. Goldman-Rakic PS. Circuitry of the prefrontal cortex and the regulation of behavior by representational knowledge. In: Plum F, Mountcastle V, eds. Handbook of physiology, vol 5. Bethesda: American Physiological Society, 1987: 373–417. Andreasen NC, O’Leary DS, Arndt S, et al. Short-term and longterm verbal memory: a positron emission tomography study. Proc Natl Acad Sci USA 1995; 92: 5111–15.
Vol 349 • June 14, 1997
Early reports
Hypofrontality in schizophrenia: distributed dysfunctional circuits in neuroleptic-naïve patients
Nancy C Andreasen, Daniel S O’Lear y, Michael Flaum, Peg Nopoulos, G Leonard Watkins, Laura L Boles Ponto, Richard D Hichwa
Summary Background There have been repor ts that patients with schizophrenia have decreased metabolic activity in prefrontal cor tex . However, findings have been confounded by medication effects, chronic illness, and difficulties of measurement. We aimed to address these problems by ex amination of cerebral blood flow with positron emission tomography (PET). Methods We studied 17 neuroleptic-naïve patients at the early stages of illness by means of image analysis and statistical methods that can detect abnormalities at the gyral level. Findings An initial omnibus test with a randomisation analysis indicated that patients differed from normal controls at the 0·06 level. In the follow-up analysis, three separate prefrontal regions had decreased per fusion (lateral, orbital, medial), as well as regions in inferior temporal and parietal cor tex that are known to be anatomically connected. Regions with increased per fusion were also identified (eg, thalamus, cerebellum, retrosplenial cingulate), which suggests an imbalance in distributed cor tical and subcor tical circuits. Interpretation These distributed dysfunctional circuits may form the neural basis of schizophrenia through cognitive impairment of the brain, which prevents it from processing input efficiently and producing output effectively, thereby leading to symptoms such as hallucinations, delusions, and loss of volition.
Lancet 1997; 349: 1730–34
Mental Health Clinical Research Centre and PET Imaging Center, University of Iowa College of Medicine and Hospitals and Clinics, Iowa City, IA 52242, USA (Prof N C Andreasen MD, D S O’Lear y PhD, M Flaum MD, P Nopoulos MD, G L Watkins PhD, L L Boles Ponto PhD, R D Hichwa PhD) Correspondence to: Prof N C Andreasen
1730
Introduction In 1974, Ingvar and Franzen coined the term hypofrontality to denote the relative decrease they found in their calculation of a ratio of frontal to post-central blood flow in schizophrenia patients.1 Since then, hypofrontality in schizophrenia has been studied by a range of methods: measurement of regional blood flow with xenon by use of cortical probes;2,3 measurement of cerebral blood flow and glucose metabolism by positron emission tomography (PET);4–7 neuropathological studies of tissue taken at necropsy;8 anatomical measurements by magnetic resonance (MR) imaging;9,10 animal models of working memory;11 and multiple techniques of cognitive psychology and neuropsychology.12,13 Although these studies have broadly suggested prefrontal dysfunction in schizophrenia, the findings are not consistent, and Gur and Gur14 have argued that the concept of hypofrontality should be questioned. There are several difficulties in the study of hypofrontality in schizophrenia15—most importantly, perhaps, the anatomical extent, complexity, and functional diversity of the frontal cortex itself, which includes both neocortex and more primitive limbic cortex, and carries out functions that range from working Region
t max
Relative decreases in patients Right dorsolateral frontal ⫺3·80 ⫺3·37 Left dorsolateral frontal ⫺3·45 Right orbital frontal ⫺4·04 Left orbital frontal ⫺3·35 Medial frontal ⫺3·33 Right inferior temporal ⫺4·18 Left inferior temporal ⫺3·58 Mid cingulate ⫺3·12 ⫺3·40 Precuneus ⫺4·36 Left parietal ⫺3·49 Primar y visual cortex ⫺4·48 Relative increases in patients Left inferior frontal Right thalamus Left thalamus Right retrosplenial cingulate Left retrosplenial cingulate Left parietal/supramarginal Fight fusiform/occipital Left cerebellum
Right cerebellum
⫺3·55 ⫺3·47 ⫺3·17 ⫺4·41 ⫺3·82 ⫺4·12 ⫺4·49 ⫺3·85 ⫺3·41 ⫺3·34 ⫺3·47 ⫺3·41
Volume (N voxels)
x
y
z
79 60 48 211 147 62 320 121 132 129 595 64 356
⫺35 ⫺39 ⫺32 ⫺30 ⫺24 ⫺2 ⫺58 ⫺53 ⫺ 1 ⫺5 ⫺16 ⫺42 ⫺ 4
⫺12 ⫺0 ⫺18 ⫺29 ⫺14 ⫺39 ⫺34 ⫺50 ⫺13 ⫺32 ⫺70 ⫺73 ⫺97
⫺24 ⫺22 ⫺21 ⫺21 ⫺22 ⫺12 ⫺21 ⫺21 ⫺38 ⫺40 ⫺25 ⫺35 ⫺ 0
224 76 69 354 106 238 2608 716 153 151 124 117
⫺31 ⫺13 ⫺7 ⫺16 ⫺14 ⫺40 ⫺15 ⫺7 ⫺24 ⫺28 ⫺10 ⫺23
⫺29 ⫺22 ⫺22 ⫺57 ⫺62 ⫺68 ⫺72 ⫺88 ⫺78 ⫺77 ⫺66 ⫺53
⫺3 ⫺18 ⫺18 ⫺21 ⫺24 ⫺18 ⫺9 ⫺22 ⫺13 ⫺25 ⫺39 ⫺34
Comparison of regional blood flow in drug-naïve schizophrenic patients and matched healthy volunteers
Vol 349 • June 14, 1997
THE LANCET
memory to regulation of eye movements. Through lesion studies, three functional areas have been identified (orbital, dorsolateral, and medial16), though these subdivisions are an oversimplification; the term hypofrontality is thus too general, since it fails to reflect the diversity of specialised regions in frontal cortex. The effects of stage of illness (ie, first episode vs chronically ill) are contested. Early studies of chronically ill patients generally reported decreases in frontal flow in schizophrenia during rest or a specific cognitive task1,4,5,17 (though some were negative).14,18 Studies of first-episode or never-treated patients are less common; some have supported hypofrontality,15 whereas others have found no differences.6,19 A few influential studies have suggested the opposite finding—namely, that first-episode patients have increases in frontal flow.20,21 If true, these findings may
suggest that metabolic activity in the prefrontal cortex decreases over time as a result of a degenerative process, toxic effects of schizophrenia, or effects of medications to treat the illness. Medication effects on cerebral blood flow are a confounding factor that cannot easily be separated from stage of illness. The results of Berman,22 Weinberger,17 and their colleagues suggest that hypofrontality was unrelated to treatment in chronically ill patients. By use of a more powerful longitudinal design in a sample of five drug-naïve patients, Szechtman and colleagues21 found that 1 year of treatment increased metabolism in the basal ganglia, but did not affect frontal metabolism. However, when they used a cross-sectional design to compare a larger group of patients who had been receiving medication for 1 year with a group who had
Figure 1: Statistical maps of PET data comparing patients and controls Shows negative peaks in right and left orbital frontal and inferior temporal regions (transaxial and coronal), lateral frontal regions (coronal), and precuneus (sagittal). Positive peaks or regions are seen in cerebellum (transaxial and sagittal), retrosplenial cingulate and fusiform (sagittal), and right thalamus (coronal). Green crosshairs show location of slice. Images show location as if viewer is standing at foot of bed (transaxial views) or facing patient (coronal views). Significantly activated regions superimposed on composite MR image derived by averaging MR scans from patients. Peak maps show small areas where all contiguous voxels exceed predefined threshold for significance (p=0·005). Areas are small because data were not smoothed (filtered); t maps show t value for all voxels in image and provide general over view of landscape of changes in blood flow. Planes illustrate location of relevant activity. Areas of decreased flow in patients in blue tones (“negative” peaks or regions); areas of increased flow in red/yellow/yellow-green tones (”positive” peaks or regions).
Vol 349 • June 14, 1997
1731
THE LANCET
been treated for an average of 7 years, Szechtman and colleagues found relative hypofrontality in the chronically treated patients, which suggests that long-term neuroleptic treatment may produce a decrease in frontal metabolic tone. Cognitive activity during image acquisition may also have substantial impact on the assessment of hypofrontality. Most early studies that supported hypofrontality involved examination of patients in a “resting state”.2,6,7 But the word “rest” is misleading, since the brain does not become inactive or empty of thought in the absence of specific experimental tasks or instructions; on the contrary, patients report after scans that when at “rest” they typically recalled past experiences or made future plans, both of which require prefrontal activity.23 Other investigators tried to “stress test” the prefrontal cortex by use of cognitive challenges.15,17,24,25 Although this approach has the advantage of standardising mental activity, the hypofrontality reported is difficult to interpret because schizophrenic patients typically carry out the tasks less well than healthy volunteers. There is thus a “chicken and egg” problem: does the decrease in frontal blood flow or metabolism result from an intrinsic neural deficit, or from lack of motivation and poor task performance? Other investigators have attempted to answer this question by matching task performance in patients and controls.25 Symptoms may also influence measurements of cerebral blood flow or metabolism. For example, negative symptoms have been related to hypofrontality;1,7,15 and a study of a small group of patients who experienced auditory hallucinations during scanning indicated flow increases in multiple brain regions.26 The concept of hypofrontality raises important questions about the fundamental nature of schizophrenia—its neural substrates, its course, and its treatment. Do specific regions of the prefrontal cortex have abnormal metabolism or blood flow? And, if so, is this abnormality related to symptoms, cognitive state, or treatment effects? Most importantly, are toxic medication or toxic illness effects on the prefrontal cortex a valid concern? To eliminate confounding factors that arise from treatment and chronicity, we focused on a sample of neuroleptic-naïve patients studied early in the course of their illness, and compared them with a group of healthy volunteers closely matched for sex, age, and parental education. To eliminate the confounding effect of poor task performance, we studied them during the resting state. To gauge the relation between symptoms and blood flow, we measured symptoms just before the scan. To assess whether specific subregions of the prefrontal cortex are selectively dysfunctional, we analysed data by image analysis and statistical techniques that enabled us to identify abnormalities at the gyral level.
Methods Patients and cognitive state We studied 17 neuroleptic-naïve patients with recent onset of schizophrenia according to Diagnostic and Statistical Manual (DSM) IV. Ten were male and seven female. Their mean age was 26·2 (SD 7·3) years. Most were experiencing schizophrenia for the first time and had not been admitted to hospital previously. Severity of symptoms was measured just before the PET study with the Scale for Assessment of Positive Symptoms (SAPS) and
1732
the Scale for Assessment of Negative Symptoms (SANS). Healthy volunteers were recruited from the community through a newspaper advertisement. Ten were male and seven female. Their mean age was 26·4 (7·3) years. Controls were similar to patients in their extent of parental education (13·3 [SD 3·4] years and 12·5 [3·9], respectively). All patients gave written informed consent. They were studied during the cognitive state usually referred to as “rest”—that is, they were not given any specific instructions about mental activity. This cognitive state is better characterised as a type of episodic memory.23
PET and MR data acquisition and analysis PET and MR data were acquired by techniques described previously. We used two different methods of statistical analysis. The first was an atlas-based automated technique for dividing the brain into specific lobes; we compared mean flow within the left and right frontal, temporal, parietal, and occipital lobes and the cerebellum by t tests.27 This analysis produced a slightly crude index of hypofrontality and of differences in other brain lobes. The second statistical method was a randomisation analysis, a non-parametric technique particularly appropriate for complex between-group comparisons in PET studies.25,28 We used this method to compare mean blood flow in the two groups, and thereby calculated an index of regional differences at the gyral level. Areas where statistical analysis showed specific regional differences were identified as “peaks”; we have reported the highest t (t max) value (Student’s t test) associated with the peak and the number of adjacent pixels that comprise the peak. Location of the pixel with the highest t-max value was identified with the standard Talairach Atlas coordinates. The region name given to the peak was based on direct visual inspection of registered MR and PET images. Since controls were “subtracted” from patients, regional differences associated with a negative t value indicate that flow was lower in patients than in controls, whereas those with a positive t value indicate that flow was higher in patients than in controls.
Results The slightly crude measures of regional blood flow, which involved examination of whole lobes of the brain, did not identify any between-group differences in blood flow, with or without correction for whole-brain blood flow (normalisation). Thus, no specific hypofrontality was identified by this method. The randomisation analysis did, however, indicate differences in frontal and other regions (table and figures 1 and 2). The initial omnibus test showed that patients differed from normal controls at the 0·06 level. In followup analysis, the patients had extensive areas of decreased flow in several regions on both the right and left lateral convexities of the prefrontal cortex. These include Brodmann areas 8, 6, and 46. Area 46 is believed to be active during working memory tasks;29 8 contains the front eye fields, which assist in the regulation of attention; and 6 is a supplementary motor area. The patients also showed bilateral decreases in orbital frontal regions, which were once ablated by leucotomy as a treatment for schizophrenia and are thought to have a vital role in emotion and social judgment. Furthermore, a large area of the medial frontal cortex also showed decreased flow. Bilateral areas in the inferior temporal cortex were also relatively inactive, as were the midcingulate and precuneus areas, left parietal lobe, and visual cortex. Visual inspection of the MR-registered PET scans suggested a specific pattern of inactivity. One cluster of inactivity appeared to extend from primary visual cortex through the inferior temporal cortex
Vol 349 • June 14, 1997
THE LANCET
Figure 2: Statistical maps of PET data in patient y Shows two negative lateral frontal peaks on right (transaxial) and bilaterally (coronal), orbital frontal (sagittal and coronal), precuneus (transaxial) and midcingulate (sagittal). Positive peaks seen in cerebellum (sagittal), left parietal (transaxial), and retrosplenial cingulate (transaxial).
towards the hippocampal complex (which was relatively inactive, but did not reach significance). Another such cluster extended more dorsally, and included the precuneus and parietal association cortex. The dorsal and ventral frontal regions are known to be anatomically linked to these more posterior regions.29 Several of the patients’ brain regions also had increases in flow compared with the same regions in normal controls. Most areas of increased flow were in posterior regions, including occipital and parietal association cortex (secondary visual cortex and the supramarginal gyrus), and the retrosplenial cingulate gyrus. In addition, there were bilateral increases in the thalamus and the cerebellum. Positive correlations were observed between the psychotic dimension of the Scale for Assessment of Positive Symptoms and flow in the thalamus (r=0·90), cerebellum (0·82), and left and right anterior inferior temporal regions (0·73 and 0·83, respectively). Significant negative correlations were found between the negative dimension of the Scale for Assessment of Negative Symptoms, and several anterior prefrontal
Vol 349 • June 14, 1997
regions (r=0·80 and 0·74, respectively).
Discussion This study supports previous investigatons that suggest schizophrenia patients have a dysfunction in prefrontal cortex. The two methods of analysis imply a possible reason for inconsistent results in earlier studies. When the entire frontal lobe was averaged together, hypofrontality was not detected. When we used a more sensitive and specific method of analysis, however, we found many sites with decreased flow in the prefrontal cortex of the schizophrenia patients. Since this study examined a large group of drug-naïve patients early in the course of their illness, flow abnormalities are not secondary to effects of neuroleptic treatment or chronic psychosis. Because areas of decreased flow were found in these patients in three major regions of prefrontal cortex (lateral, orbital, and medial), the results suggest that hypofrontality is a primary abnormality in schizoprenia. These results may 1733
THE LANCET
allay the concern, raised by earlier studies,20,21 that decreased perfusion in regions of prefrontal cortex is a consequence of chronic treatment or toxic effects of chronic psychosis. Our results suggest that the term hypofrontality could be replaced with more specific terms that designate which regions of the prefrontal cortex are dysfunctional. In this study, three different frontal areas were identified. The areas of decreased flow on the lateral convexities include regions that have been implicated as active during tasks that require working memory (both longterm and short-term), attention, and arousal.29,30 The orbital regions contain more primitive cortex and are believed to carry out “limbic” functions such as emotional memory, whereas the medial region of the prefrontal cortex is involved in the initiation of behaviour.16 Moreover, we are not observing hypofrontality alone, but dysfunctional circuitry that is distributed throughout the brain. The associated flow decreases in inferior temporal regions may result from dysfunction, involving orbital frontal regions and temporal-hippocampal areas, in a “limbic network”. The decreases in dorsolateral frontal regions and the precuneus and parietal lobes, however, may reflect a more “neocortical” network.11–29 This study identified many areas that showed increases in flow; these increases may shed light on brain pathology in schizophrenia. One report suggested that patients with active auditory hallucinations show increased flow in several brain regions: thalamus, basal ganglia, anterior cingulate, right and left parahippocampal gyri, and cerebellum.26 Correlational analyses of patients’ symptoms in our study are consistent with that report. The extensive areas of increased flow may indicate that these brain regions are working harder to compensate for decreases in other regions. Increases and decreases are closely related, and may reflect an overall imbalance in cortical and subcortical excitatory (glutamatergic) and gabaergic (inhibitory) circuitry. The net result is that the brain cannot process its input efficiently or produce its output effectively; some parts of the circuits are working too hard, whereas others are underactive. Schizophrenia is characterised by dysfunctional regions in prefrontal cortex, but the illness is clearly more complex than the term hypofrontality allows. Schizophrenia involves an imbalance in circuits distributed throughout the brain—including multiple cortical and subcortical regions—that leads to impairment of the ability to set priorities, process, and produce information, and to turn it into meaningful thoughts and behaviour. This imbalance in circuits is expressed as psychotic or negative symptoms. References 1
2
3
4 5
Ingvar DH, Franzen G. Abnormalities of cerebral blood flow distribution in patients with chronic schizophrenia. Acta Psychiatr Scand 1974; 50: 425–62. Gur RE, Gur RC, Skolnick BE, et al. Brain function in psychiatric disorders: regional cerebral blood flow in unmedicated schizophrenics. Arch Gen Psychiatry 1985; 42: 329–34. Berman KF, Illowsky BP, Weinberger DR. Physiological dysfunction of dorsolateral prefrontal cortex in schizophrenia: further evidence for regional and behavioral specificity. Arch Gen Psychiatry 1988; 45: 616–22. Buchsbaum MS, Ingvar DH, Kessler R, et al. Cerebral glucography with positron tomography. Arch Gen Psychiatry 1982; 39: 251–59. Volkow ND, Wolf AP, Van Gelder P, et al. Phenomenological correlates
1734
6
7
8 9
10
11
12
13 14 15
16
17
18
19
20
21
22
23
24
25
26 27
28
29
30
of metabolic activity in 18 patients with chronic schizophrenia. Am J Psychiatry 1987; 144: 151–58. Early TS, Reiman EM, Raichle ME, Spitznagel EL. Left globus pallidus abnormality in never-medicated patients with schizophrenia. Proc Natl Acad Sci USA 1987; 84: 561–63. Liddle PF, Friston KJ, Frith CD, Hirsch SR, Jones T, Frackowiak RSJ. Patterns of cerebral blood flow in schizophrenia. Br J Psychiatry 1992; 60: 179–86. Benes FM. Is there a neuroanatomic basis for schizophrenia? An old question revisited. Neuroscientist 1995; 1: 104–15. Andreasen NC, Nasrallah HA, Dunn VD, Olson SC, Grove WM, Ehrhardt JC, et al. Structural abnormalities in the frontal system in schizophrenia: a magnetic resonance imaging study. Arch Gen Psychiatry 1986; 43: 136–44. Andreasen NC, Flashman L, Flaum M, et al. Regional brain abnormalities in schizophrenia measured with magnetic resonance imaging. JAMA 1994; 272: 1763–69. Goldman-Rakic PS, Selemon LD, Schwartz ML. Dual pathways connecting the dorsolateral prefrontal cortex with the hippocampal formation and parahippocampal cortex in the rhesus monkey. Neuroscience 1984; 12: 719–43. Bilder RM, Turkel E, Lipschutz-Borch L, Lieberman JA. Antipsychotic medication effects on neuropsychological functions. Psychopharmacol Bull 1992; 28: 353–66. Frith CD. The cognitive neuropsychology of schizophrenia. East Sussex: Lawrence Erlbaum, 1992. Gur RC, Gur RE. Hypofrontality in schizophrenia: RIP. Lancet 1995; 345: 1383–85. Andreasen NC, Rezai K, Alliger R, et al. Hypofrontality in neuroleptic-naive and chronic schizophrenic patients: assessment with xenon-133 single-photon emission computed tomography and the Tower of London. Arch Gen Psychiatry 1992; 49: 943–58. Mega MS, Cummings JL. Frontal-subcortical circuits and neuropsychiatric disorders. J Neuropsychiatry Clin Neurosci 1994; 6: 358–70. Weingerger DR, Berman KF, Zec RF. Physiological dysfunction of dorsolateral prefrontal cortex in schizophrenia I: regional cerebral blood flow (rCBF) evidence. Arch Gen Psychiatry 1986; 43: 114–24. Wiesel FA, Wik G, Sjogren I, Blomqvist G, Greitz T. Altered relationships between metabolic rates of glucose in brain regions of schizophrenic patients. Acta Psychiatr Scand 1987; 76: 642–47. Gur RE, Mozley PD, Resnick SM, et al. Resting cerebral glucose metabolism in first-episode and previously treated patients with first episode schizophrenia relates to clinical features. Arch Gen Psychiatry 1995; 52: 657–67. Cleghorn JM, Garnett ES, Nahmias C, et al. Increased frontal and reduced parietal glucose metabolism in acute untreated schizophrenia. Psychiatry Res 1989; 28: 119–33. Szechtman H, Nahmias C, Garnett S, et al. Effect of neuroleptics on altered cerebral glucose metabolism in schizophrenia. Arch Gen Psychiatry 1988; 45: 523–32. Berman KF, Zec RF, Weinberger DR. Physiologic dysfunction of dorsolateral prefrontal cortex in schizophrenia: role of neuroleptic treatment, attention, and mental effort. Arch Gen Psychiatry 1986; 43: 126–35. Andreasen NC, O’Leary DS, Cizadlo T, et al. Remembering the past: two facets of episodic memory explored with positron emission tomography. Am J Psychiatry 1995; 152: 1576–85. Buchsbaum MS, Neuchterlein KH, Haier RJ, et al. Glucose metabolic rate in normals and schizophrenics during the continuous performance test assessed by positron emission tomography. Br J Psychiatry 1990; 156: 216–27. Andreasen NC, O’Leary DS, Cizadlo T, et al. Schizophrenia and cognitive dysmetria: a PET study of dysfunctional prefrontalthalamic-cerebellar circuitry. Proc Natl Acad Sci USA 1996; 93: 9985–90. Silbersweig DA, Stern E, Frith C, et al. A functional neuroanatomy of hallucinations in schizophrenia. Nature 1995; 378: 176–79. Andreasen NC, Rajarethinam R, Cizadlo T, et al. Automatic atlasbased volume estimation of human brain regions from the MR images. J Comput Assist Tomogr 1996; 20: 98–106. Arndt S, Cizadlo T, Andreasen NC, Heckel D, Gold S, O’Leary DS. Tests for comparing images based on randomization and permutation methods. J Cereb Blood Flow Metab 1996; 16: 1271–79. Goldman-Rakic PS. Circuitry of the prefrontal cortex and the regulation of behavior by representational knowledge. In: Plum F, Mountcastle V, eds. Handbook of physiology, vol 5. Bethesda: American Physiological Society, 1987: 373–417. Andreasen NC, O’Leary DS, Arndt S, et al. Short-term and longterm verbal memory: a positron emission tomography study. Proc Natl Acad Sci USA 1995; 92: 5111–15.
Vol 349 • June 14, 1997