G.D. Kotzalidis, A. Facchi, L. Tarsitani, V. Mantua, P.P. Colombo, P. Pancheri - Vol. 7, Giugno 2001, num.2
Testo Immagini Bibliografia Summary Riassunto Indice
Up-date paper/Articolo di aggiornamento
Thalamic abnormalities in schizophrenia
Anomalie talamiche nella schizofrenia
G.D. KOTZALIDIS, A. FACCHI, L. TARSITANI, V. MANTUA, P.P. COLOMBO, P. PANCHERI
3rd Psychiatric University Clinic, University of Rome
"La Sapienza", Rome, Italy
Parole chiave: Schizophrenia • Thalamus • Mediodorsal
thalamic nucleus • Reticular thalamic nucleus • Anteroventral thalamic
nucleus • Prefrontal cortex • Cognitive dysmetria • Post
mortem studies • Magnetic resonance imaging (MRI) • Positron
emission tomography (PET)
Key words: Schizofrenia • Talamo • Nucleo talamico medio
dorsale • Nucleo talamico reticolare • Nucleo talamico antero ventrale
• Corteccia prefrontale • Dismetria cognitiva • Studi post
mortem • Risonanza magnetica nucleare • Tomografia ad emissione
di positroni Psychiatric manifestations occurring with thalamic stimulation
or ablation were observed as early as 1966 (1)
and the latter has been performed to treat impulsive aggression in psychiatric
disease (2).
Subsequently, thalamic involvement was shown in disturbed emotional behaviour
(3)
and an imbalance between thalamic and neostriatal dopamine and noradrenaline
were believed, on psychophysiological and neurophysiological grounds, to occur
in schizophrenia, with an overstimulation, at least with positive symptoms
(4).
Differences between schizophrenics and controls in thalamic key enzymes in
monoamine production were observed in the late seventies (5),
but a clear demonstration was given by Nancy Andreasen's group in 1994 (6)
using magnetic resonance imaging (MRI). Her group was the first to focus on
subtle differences in thalamic volume between schizophrenic patients and normal
controls, as emerging with computerised tomography (7),
but their 1994 paper heralded a co-ordinated effort to seek for abnormalities
in thalamic structure and function in schizophrenia.
Combining PET and MRI data, the same group reached the conclusion that a basic abnormality resulted in impairment of brain circuitry involving connections between the thalamus and midbrain reticular structures, the corpus callosum, the septum and, most importantly, the prefrontal cortex (8). Abnormal processing in thalamo-cortical circuits could also explain reduced activation in right superior temporal gyrus during dichotic task performance seen with PET in schizophrenic individuals (9). Using the same technique, decreased prefrontal activation was observed in neuroleptic-naïve schizophrenic patients in the early stage of the disease, and this finding correlated with increased thalamic activation, pointing at cortical-subcortical imbalance (10). However, in performance of different types of task, such as active recall, showed decreased right thalamic perfusion, which correlated with left prefrontal cortical hypoperfusion (11). These abnormalities in thalamic-frontocortical-cerebellar integration were considered as basic to some schizophrenic communication and information processing characteristics, defined as "cognitive dysmetria" by Andreasen's Iowa City group, a concept implying a specific aberrant processing of information in the brain resulting in difficulty to establish priorities and hierarchies, to co-ordinate complex information and to respond adequately to external or internal stimuli (12-14), something that matches perfectly to the neuronal "miswiring" concept (15-17), as well as to the hypothesis of a time shift in prenatal neuronal migration (18-20) that could lead to miswiring.
The view of a dysfunctional circuitry in schizophrenia involving the thalamus is supported by the connections this structure establishes with other brain structures. A particularly important connection, which is bidirectional (i.e., both thalamocortical and corticothalamic), is with the prefrontal cortex.
Anatomical evidence for prefrontal-thalamic connections
Reciprocal connections between the thalamus and the prefrontal cortex, an area that is critically involved in schizophrenia, were demonstrated in the rat with the combined use of horseradish peroxidase and autoradiography since the mid-seventies (21). The bulk of thalamocortical projections to the prefrontal cortex terminates on dendrites in the 3rd cortical layer (21), whereas neurones from the prefrontal cortex from layers II, V and VI, which are endowed with D1 receptors mainly, but not those where D2 receptors prevail (layer V), give their feedback to the thalamus (22). In the rabbit, the anteroventral thalamic nucleus is connected with the cingulate cortex, whereas the mediodorsal nucleus is linked to the prefrontal cortex (23). In primates, the reciprocal nature of the thalamic-frontocortical connection was confirmed by Preuss and Goldman-Rakic, who used cortical implants of horseradish peroxidase pellets and tetramethyl benzidine histochemistry to show anterograde and retrograde thalamic tracing (24); these investigators showed the existence of both bilateral and ipsilateral corticothalamic projections. Projections to the anteromedial nucleus, the midline nuclei, and the magnocellular part of the mediodorsal nucleus are bilateral, with both arms of about the same density, whereas other bilateral frontocortical projections to the thalamus, such as to the anterior ventral nucleus and to the parvicellular portion of the mediodorsal nucleus have a much thinner portion of fibres that cross to the contralateral thalamus. Exclusively ipsilateral projection from the prefrontal cortex to the thalamus regard the medial pulvinar, the reticular nucleus, the suprageniculate and limitans nuclei (24). In the rat, the prefrontal cortex innervates the thalamus, mainly at the level of the mediodorsal nucleus, where its fibres are distributed both ipsilaterally and contralaterally (25). Projections from the prefrontal cortex to the mediodorsal thalamic nucleus follow a specific distribution pattern. The less architectonically differentiated regions of the prefrontal cortex, i.e., the orbital and medial portions, project to the magnocellular part of the mediodorsal thalamic nucleus, the intermediately differentiated lateral region projects to the parvocellular part, whereas the highly differentiated region of the dorsolateral prefrontal cortex corresponding to Brodmann's area 8 projects to the densocellular (multiform) subdivision, all according to a dorso-ventral topographic pattern (26).
The projections from the thalamus to the prefrontal cortex show the same specificity between origin and distribution; in the primate, the lateral portion of the magnocellular subnucleus of the mediodorsal nucleus innervates the ventromedial prefrontal cortex, whereas its ventral portion innervates the dorsolateral prefrontal cortex (27). On the other hand, the parvocellular portion of the mediodorsal thalamic nucleus projects to the dorsolateral and dorsomedial prefrontal cortices (27). The neurochemical nature of the mediodorsal-prefrontal reciprocal connection appears to be excitatory, in particular glutamatergic; the reverberation of the excitation of this circuit is set-off by dopaminergic inhibition of deep frontocortical pyramidal glutamatergic cells (Fig. 1) (28).
This specificity of connections between the thalamus and the prefrontal cortex is most important when considering alterations in neurone numbers or volume of specific thalamic nuclei occurring in schizophrenic brains, or considering functional modifications in both thalamus and prefrontal cortex when resting or while performing a task. In fact, various specific approaches were developed to evaluate possible thalamic abnormalities in schizophrenia, which may be summarised as mainly structural studies, involving the use of magnetic resonance imaging (MRI), post-mortem studies, offering deeper insights as to the neuropathology and neurochemistry of schizophrenia despite facing many technical problems, and functional studies, comprising single photon emission spectrography (SPECT), proton magnetic spectrometry, and positron emission tomography (PET), that provide clues as to areas dysfunctioning in schizophrenia and the circuitry that underlies such dysfunction.
In vivo structural studies
With the development of MRI, which allowed much higher precision than computerised axonal tomography, many studies focused on the size of various brain nuclei in schizophrenia; some of them measured also the thalamus, reporting interesting data, which are summarised in Table I. The first study was published in 1994 (6) and reported alterations in juxta-laminar thalamic nuclei as the only brain structure that differed between healthy subjects and schizophrenic patients. Subsequent studies did not agree as to the existence or the site of thalamic volume reduction in schizophrenia. Negative or positive findings did not correlate with technique used or population, but two studies which investigated relatives of schizophrenic patients and first-degree relatives reported positive results. Staal et al. (31) compared schizophrenic patients with their healthy siblings and with healthy controls and found schizophrenic thalami of unaffected siblings to be smaller than those of healthy individuals, but significantly larger than those of their affected siblings. The difference in thalamic volume between schizophrenic patients and their age-matched healthy controls was obviously greater. Seidman et al. (35) studied only first-degree relatives of schizophrenic patients and compared them with healthy controls. They reported a significant volume reduction in the thalami of schizophrenic relatives, a finding which is in line with that obtained by Lawrie et al. (34), who used T2-weighted MRI and found decreased thalamic volume in first- or second-degree relatives of schizophrenic patients with respect to normals, but, surprisingly, also with respect to schizophrenics, and failed to detect any difference between schizophrenics and normals.
Other studies as well reported no difference between normal and schizophrenic individuals in thalamic volume. Portas et al. (29), Wolkin et al. (30), and Arciniegas et al. (32) who found no differences had their MRI T1-weighted, whereas Staal et al. (31) and Dasari et al. (33) had also used T2-weighted scans and reported significant thalamic volume reductions in schizophrenic patients. It is not easy to draw any definite conclusion from these studies, since subnuclear characterisation was not carried out and it is possible that subnuclear abnormalities could exist in the thalamus of schizophrenic patients without apparent volume reduction.
Post-mortem studies
Post-mortem studies involve brain sectioning and staining after the death of the individual. The main problem in comparability of obtained brain specimens is different autolytic time, i.e., the time between death of the individual that gives rise to lytic phenomena and may selectively impair some neurotransmitter systems more than others and time of tissue fixation. To ensure sample comparability, autolytic time should either not differ among individuals in the samples or individuals in the samples should be matched for autolytic time, besides other possible confounding factors, such as educational level, social level, age, gender, cause of death, and the like. This limits significantly the possibility to recruit adequate samples for statistical analysis. In fact, most studies in this field (Tab. II) involve samples of small size.
Another possible confounding factor with post-mortem studies, that is also important in functional studies, is neuroleptic/antipsychotic treatment. With drug naïve patients, one may be sure that the differences observed between schizophrenic individuals and other controls could be due to the disease proper, but when patients have been treated until their time of death, one cannot know whether observed differences are due to the disease or to drug treatment. A special case is a drug-free sample. It cannot be considered the same as a never-treated sample, but the time since drug discontinuation may have been sufficient to offset drug effects in one patient and not in another. Carry-over effects are possible even after six months of antipsychotic drug discontinuation, but it is usually admitted that two months are sufficient to get the patient back to an almost drug-naïve state.
Early studies focusing on thalamic size and neuronal number were performed mainly by Bente Pakkenberg and colleagues and showed significant volume reductions in schizophrenia, by about 40% (36), as well as significant reductions in neurones, astrocytes and oligodendroglia in the thalamus and in both shell and core of the nucleus accumbens septi, but not basolateral amygdala or ventral pallidum (37). However, these early studies, to the admission of the same author (64), were flawed by several methodological drawbacks, that weaken their conclusions.
Studies focusing on thalamic nuclear size or neuronal numbers agree that some neuronal clusters from the mediodorsal thalamic nucleus are reduced in schizophrenic patients. Despite the first studies by Pakkenberg (36,37,64), who used innovative technology to count cells in discrete nuclei, did not yield reliable results due to methodological drawbacks, such as haematoxylin/eosin staining, that rendered the definition of the borders of thalamic nuclei somewhat difficult, her conclusions of abnormal thalamic mediodorsal nucleus cytoarchitecture were subsequently confirmed in studies using improved methodology (61,63). In a further study, Pakkenberg observed differences between chronic schizophrenics and leucotomised schizophrenics; she observed a reduction of the number of neurones in the nucleus accumbens shell which was present in both subgroups of schizophrenics and was, therefore, not influenced by leucotomy, and a significant effect of leucotomy on the number of neurons in the mediodorsal thalamic nucleus, in that leucotomised patients had a further reduction in neuronal numbers in this nucleus (38). Her results point at the possibility that although the thalamus is involved in some symptoms of schizophrenia, a lesion of the mediodorsal nucleus could be a consequence and not the primary lesion, in at least some schizophrenic patients.
The figures of neuronal reduction in the thalamus are similar in the two studies focusing in mediodorsal nucleus abnormalities, Young et al. (63) found a reduction of about 35% in neuronal number in the mediodorsal thalamic nucleus and Popken et al. (61) went further in pinpointing that the subnuclei involved were the densocellular portion (where a 25% reduction was estimated) and the parvocellular part (where a reduction of about 30% was observed), but not the magnocellular portion. It should be recalled that the former two portions are connected in primates with the cytoarchitectonically most differentiated parts of the prefrontal cortex, i.e., the lateral and dorsolateral parts (26), and that they project to the striatum/accumbens and premotor cortex (densocellular) and dorsolateral and dorsomedial cortices (parvocellular) (61) (Fig. 2), areas which are thought to play an essential part in the aetiopathogenesis and psychopathology of schizophrenia. Neuronal loss in the mediodorsal thalamic nuclei subdivisions that project to the most highly differentiated parts of the prefrontal cortex and that receive innervation from the same structure, independently from being primary or secondary, will impair the thalamocortical and corticothalamic circuitry, thereby resulting in some impairment, such as cognitive deficits in task performance, which are characteristic of schizophrenia. Even though impairment of the prefrontal cortex in schizophrenia appears to involve the totality of this structure (39-42), the functional loss that discriminates schizophrenic behaviour from that of normal individuals resides within its most specialised and differentiated part, such as the dorsolateral prefrontal cortex (43-48).
Immunohistochemical studies focused on the presence of specific molecules in thalamic neurones and found some important differences between schizophrenics and controls. A study by Danos and co-workers (65) used immunocytochemistry to identify parvalbumin-positive neurones in the antero-ventral nucleus of the thalamus and found a trend towards reduction of the number of all anteroventral perikarya, but also a significant reduction in parvalbumin-positive neurones in the same nucleus (Fig. 3). Parvalbumin-positive neuronal dysfunction has been described in schizophrenic cerebral cortex, i.e., in the cingulate (49,50) and in the prefrontal cortices (51), but in another study the latter could not be confirmed (52). This dysfunction mainly regards GABAergic interneurones of the chandelier type which make synaptic boutons on pyramidal neurones at the first axonal segment of the latter; these GABAergic neurones are ectopic in cortical layers and this results in the already mentioned miswiring that gives way to altered feedback from the cortex to the thalamus and the limbic system. In the thalamus, two populations of parvalbumin-positive have been described, one concerning GABAergic interneurones and the other glutamatergic projections to the cortex (53). The subpopulation involved in schizophrenia and shown to be reduced in schizophrenic brains presumably belong to the latter type (65).
Other immunohistochemical studies on thalamic alterations in schizophrenia focused on synaptic proteins, such as rab.3a, synaptophysin and chromogranins. The significance of any abnormalities in such proteins is difficult to understand, but it might relate to loss of synaptic strength and abnormal synaptic processing; these proteins control various steps in exocytotic neurotransmitter release and their decreased expression in various brain areas might imply ineffective synaptic transmission; alternatively, increased expression could be due to an attempt to restore transmission which is impaired. The most thoroughly studied molecule in schizophrenic brain sites is synaptophysin, which was found in schizophrenic patients to be reduced in content at the level of the medial temporal lobe (54), in the prefrontal cortex (55), and in anterior frontal cortex (56), but its mRNA levels were either normal in the prefrontal cortex (57) or reduced only in the occipital cortex (58), consistent with either genetically determined post-transcriptional abnormalities of this molecule in schizophrenic prefrontal cortex or maldevelopmental projection of other neurones in that area from other brain regions (57), probably due to defective neuronal migration from the subplate to the cortex during intrauterine life. Davidson et al. (66) found no difference between schizophrenics and controls in thalamic synaptophysin levels and also in the temporal lobe and cerebellum, but reported a decrease in the cingulate and the hippocampus. Landén et al. also failed to detect any difference in thalamic synaptophysin in schizophrenic patients with respect to controls (69). Chromogranins are proteins associated with large synaptic vesicles in monoamine and peptidergic neurones and were investigated only in the cerebrospinal fluid in schizophrenia, where chromogranin A and B, but not C, were found to be reduced as compared to controls (59). Landén et al., did not observe differences between schizophrenics and controls in thalamic chromogranin content (69). Rab.3a is a protein associated with small vesicles in synapses; it was found to decline with age in the left superior temporal gyrus of schizophrenic, but not control subjects (60). Rab.3a was investigated in two studies in the thalamus and other brain areas, and it was found to be reduced in the left thalamus with respect to controls (66) or only in the left thalamus (68). The first study also found a reduction of Rab.3a in frontal and parietal cortex, in the cingulate and in the hippocampus, but no change in temporal cortex and cerebellum. Results are interpreted as reflecting neurodevelopmental corticolimbic misconnection (60,68).
Three recent studies focused on specific receptors within the thalamus and found a moderate, but significant decrease of alpha-bungarotoxin-binding nicotinic receptors in the reticular thalamic nucleus only (70), the other studied glutamate receptors; the first could not show any significant difference between controls and elderly schizophrenic patients in the expression of the subtypes 1, 2, 3, 4, 5, 7, and 8 (subtype 6 is confined to the retina) of metabotropic glutamatergic receptors (71), whereas the second showed many ionotropic receptors (NMDA and AMPA/kainate receptors) to be decreased in schizophrenic patients relative to controls in thalamic areas mainly connected with the limbic system, but also, most notably, in the mediodorsal nucleus (72).
Taken together, post-mortem studies suggest an involvement of the thalamus in schizophrenia in the context of abnormal neurodevelopmental connectivity of thalamo-cortico-limbic-cerebellar circuitry.
In vivo functional studies
These studies involve the use of positron emission tomography (PET), single photon emission computerised spectrography or proton magnetic resonance spectroscopy (Tab. III). PET is a technique that studies regional cerebral blood flow (rCBF) or metabolism in various brain regions, giving an approximate measure of activation, but gives a rather rough picture of in vivo brain activity under various stimuli; the advent of MRI allowed for combining data of both techniques to give a more detailed and precise idea of the areas involved; PET scans are superimposed on Talairach and Tournoux standard MRI scans or on the subject's own MRI scan.
The first evidence obtained on the PET of thalamic abnormality in schizophrenia was indirect, as the study by Siegel et al., using [18F]fluorodeoxyglucose PET did not show differences in the metabolism of the thalamus between aged schizophrenics and age-matched controls, but obtained inverse correlations frontocortical and thalamic activity, pointing at a reduction in frontocortical activity and an abnormal thalamic activation, a fact that is in line with impaired thalamocortical circuitry (78). In the same line were the results of the study by O'Leary and co-workers (9), who studied rCBF with [15O]water and identified a failure of schizophrenics to activate the right superior temporal gyrus like normal controls while performing an attentional task, and this was interpreted as reflecting a defect in thalamocortical integration. The same University of Iowa College of Medicine group of authors used the same technique employing [15O]H2O to study rCBF while the subject was instructed to engage in no particular mental activity (resting state) and found increased thalamic perfusion in schizophrenic patients with respect to healthy controls (10,79). Again, combining data with perfusion in other brain areas, the authors concluded that the defect regarded thalamocortical and other associated circuitry. They further demonstrated in their most recent study (79) that chronic schizophrenics had the same type of alteration in thalamic perfusion with acute episode schizophrenics.
With superimposing PET on MRI and using [18F]fluorodeoxyglucose, Buchsbaum et al. found reduced thalamic metabolic rate in the right thalamus and loss of the normal left-to-right asymmetry (77). The same technique was used by Hazlett and co-authors (76) to test schizophrenic individuals vs. subjects with schizotypal personality disorder vs. healthy controls undergoing serial verbal learning task. They found no gross thalamic abnormalities in metabolic rate in schizophrenics, but observed a bilateral reduction in the mediodorsal nucleus and a trend towards reduction in the left anterior thalamic metabolism, while schizotypal individuals showed only a reduction in right anterior thalamic metabolism, as compared to healthy controls. The Iowa group used [15O]H2O to study drug-free or drug-naïve schizophrenic patients and healthy controls undergoing practised and novel recall tasks and again found reduced rCBF in the left thalamus with the practised task and even more important reductions with the novel task when using stories from the Wechsler Memory Scale (13) (Fig. 4) and in the right thalamus with the novel task when using the lists from the Rey Auditory Verbal Learning Test (11). The reason for these discrepancies might reside in the different prefrontocortico-thalamic pathways recruited in these tasks, which could involve alternatively the decussating and ipsilateral fibres (25) from the prefrontal cortex to the mediodorsal thalamus.
The apparent contradiction between studies using PET alone versus combined MRI and PET, which generally found increase in thalamic activity with PET and decrease with the combination, could be due to methodological differences. Taken together, these results confirm the involvement of a prefrontocortical-thalamic-limbic-cerebellar circuitry in schizophrenia and point at a greater involvement of the mediodorsal thalamic nucleus, but leave also open the possibility that the anteroventral nucleus is also involved, much in the same line with post-mortem studies. Further, they stress that fundamental circuitry alterations in schizophrenic patients are not altered through the disease process or with age, supporting the view that schizophrenia is a neurodevelopmental group of disorders.
The only study to date that used SPECT to study differences in perfusion between schizophrenic patients while on classical neuroleptics vs. the same patients 6 months after having switched to clozapine compared them with healthy controls from a database, a procedure that might limit its methodological validity. The atypical antipsychotic clozapine induced a trend towards normalisation in responders, whereas the same patients showed lower thalamic, striatal and prefrontocortical perfusion with respect to the cohort of normals while on classical neuroleptics (75). However, albeit these results point at a dysfunctional thalamo-cortico-limbic circuitry in schizophrenia, it is not possible to state whether and to what extent baseline differences between patients and controls were due to the disease or to drug treatment.
Proton magnetic resonance spectroscopy is a technique that was applied relatively recently to study differences between schizophrenics and normals (80). It involves the measurement of resonance of proton-labelled N-Acetylaspartate (NAA), phosphocreatine-creatine (PCr-Cr), choline, inositol, glutamate, glutamine and other metabolites; in most studies, the ratio between N-acetylaspartate/creatine compounds (NAA/Cr), N-acetylaspartate/choline derivatives (NAA/Ch) and choline derivatives/creatine compounds (Ch/Cr) are measured. Reduced NAA/Cr were found in the hippocampus (81,82), in the basal ganglia (83), in the pons (84), in the temporal (85,86), cingulate (87), frontal (86,88,89), medial (90) and dorsolateral prefrontal (82,91) cortices, although one study failed to find differences in NAA in the latter region (92). This study, however, found differences in the glutamine metabolite of glutamate after acute treatment of chronic schizophrenics, a finding that is difficult to interpret. Glutamine and glutamate measured in discrete brain areas provide a measure of glutamatergic activity (93), while NAA/Cr correlates with striatal dopaminergic activity (94). Since decreased NAA was observed in a study only in men, it could be justified that studies using this technique use selected patient populations taking care to take into account gender effects (88). In fact, one of the studies that focused on the thalamus (74) used only male schizophrenics and controls. This study found a bilateral thalamic reduction in NAA/Cr, while Ch/Cr did not differ between controls and schizophrenics. Another study (73) found both NAA/Cr and Ch/Cr to be reduced in the thalamus, but not in the frontal cortex. These studies, in spite of variance in obtained results, show co-ordinated deficits of glutamatergic and possibly dopaminergic function in both dorsolateral prefrontal cortex and thalamus.
Hence, functional in vivo studies point at the existence of some abnormality in the cerebellar-thalamo-cortico-limbic circuitry with specific thalamic correlates. The abnormality becomes increasingly apparent with increasing demands on highly differentiated brain structures, as when higher brain functions are brought into play, for example, when performing complex cognitive tasks that require increased capacity for working memory (43,95,96). The consequence of impaired cerebellar-thalamo-cortico-limbic circuitry would be uncoordinated flow of information within the brain and disrupted synchrony (97), that according to Andreasen's group gives rise to "cognitive dysmetria", i.e., reduced ability to establish priorities, to co-ordinate complex internal or external information and to respond to various stimuli (12-14), thereby giving rise to a multitude of symptoms occurring in schizophrenia (97); the thalamus could play the role of filter in this context, allowing for information overflow to the prefrontal cortex when tasks increase in complexity.
Final considerations and conclusions
The data available from structural and functional studies indicate an involvement of the thalamus in schizophrenic aetiopathogenesis. In particular, these alterations would regard mainly the parvocellular and densocellular portions of the mediodorsal thalamic nucleus, which are connected with the cytoarchitectonically most differentiated portions of the prefrontal cortex. The anteroventral nucleus could be involved as well, but to a lesser extent; this structure also is connected with cortical structures which are believed to be important in the control of behaviour that is disrupted in schizophrenia. Decreased number of neurones tend to reduce the volume of the thalamus, that is, however not always detected in structural studies. It is hoped that improved technology will offer some clues in the future. At this regard, a new, fully automatic technique for detecting structural brain differences has been recently developed; this relies on intensity-based nonlinear registration routine deformation fields analysis. Using a general multivariate statistical approach to analyse deformation fields in the brains of 85 schizophrenic patients and 75 healthy volunteers, a German group of the University of Jena examined low frequency deformations to detect regional deviations in the brains of both groups, finding bilateral volume reductions in the thalamus and the anterior temporal gyrus (98). This novel technique has as yet to find wider applications, but appears to be reliable and promising.
Both types of approach, structural and functional, do not provide evidence as to whether altered thalamic structure is primary or consequent to alterations in other brain structures. At present, the neurodevelopmental hypothesis of schizophrenia postulates that a temporal shift in migration from the subplate to the cortex (47,99) during pregnancy is crucial for the generation of ectopic cell placement in the cortex and subsequent misconnection. This is supported by findings of abnormal GABAergic cells in the prefrontal cortex (100,101); such neurones provide in the cortex interconnection from terminals that reach the cortex from subcortical structures to the first segment of deep pyramidal neurones (102), which in turn provide feedback to the same subcortical structures. Misplacement of such GABAergic interneurones result in miswiring and defective feedback. However, both spatial and temporal cues are necessary for the thalamo-cortical/cortico-thalamic laminar distribution pattern to establish and for specificity to be ensured (103) as well as for lattice structures to provide for convergence, divergence and mirror reversals to enhance connective plasticity (104); whether it is the cortex or the thalamus that provides the cue is not as yet established, but it should be recalled that axons providing guidance to thalamic axons projecting to the cortex are preplate (105), whereas most differentiated cortical migration is guided by subplate cells (106) that persist abnormally in prefrontal cortical layer I in schizophrenia (107) and are preceded in neurogenesis by pioneer preplate neurones (108); on the other hand, pioneer preplate and Cajal-Retzius cells preorganise the neocortical primordium before the arrival of thalamocortical neurones (108,109). It is likely that some processes of developmental thalamo-cortical and cortico-thalamic connectivity proceed concomitantly (110), so it is hard to speculate as to where the primary lesion resides. It should be mentioned that should the intrauterine/neonatal prefrontal lesion be primary in schizophrenia, a compensatory increase in medio-dorsal thalamic neuronal density and decrease in mediodorsal thalamic size and neuronal number should be expected, as these changes were elicited by prefrontal lesions in neonatally-lesioned rats, but not adult-lesioned ones (111), this finding is in line with data obtained with structural and functional studies performed on schizophrenic patients.
Another interesting findings of studies on the thalamus in schizophrenia regard the possible genetic/hereditary nature of the disease. The fact that offspring of schizophrenic people has a greater risk for developing schizophrenia than other offspring (112) might be due to the fact that schizophrenia is a disorder of behaviour and behavioural patterns. Besides relying on the genetically-determined brain structure, it may also be learned during the first years of life through the interaction with parents or relatives who relate with the patient-to-be in a psychotic style. The adoption studies of twins reared apart did not resolve the issue, as adoptive families of twins who developed schizophrenia were psychopathologically more disturbed than families that adopted the twin who was discordant for schizophrenia (113). The same could be said of studies that investigated clinical clusters which are shared by schizophrenics and their relatives, since all such clusters, at the exception of neurological soft signs, consist in behavioural symptoms. Structural, functional and neuropsychological studies have a better chance to show a possible genetic load on schizophrenia. These studies usually find that the most significant changes regard patients, but their relatives most often behave midway between the patients and healthy controls; furthermore, patients in the schizophrenic spectrum, such as schizoaffectives or schizotypals, show also alterations half-way between normals and schizophrenic patients (114-118). Although it is possible that some neuropsychological defects in working memory, startling response (pre-pulse inhibition), eye tracking and smooth pursuit, P300, hypoperfusion in certain brain nuclei and ventricle/brain ratio could be the necessary accompanying feature of psychotic behaviour, whether clinical or subclinical, nevertheless they add credibility to the view that some components of some schizophrenias are genetically determined (119,120). Thalamic abnormalities are also found consistently in schizophrenic relatives (31,34,35). The fact that sometimes thalamic MRI abnormalities are found only in relatives (34) rather than in the schizophrenic patients could be taken to mean that thalamic abnormality conveys vulnerability and predisposition and codes for psychotic behaviour. Although adoption studies present inherent problems when psychopathology is dealt with (121), and particularly when it comes to interpreting data on schizophrenia (122), nevertheless they point at a complex environmental-genetic interaction (123,124) in expression of schizophrenic spectrum disorders (119).
The mediodorsal thalamic nucleus has complex connections with the cortex, in particular with the prefrontal cortex (vide infra). Lesions of this nucleus result in impairment of learning acquisition (125), retardation of the processes underlying abstract thinking and generalisation (126), deficits in working (parvocellular division) (127) and episodic memory (magnocellular division) (128), impairment in executive function (129), and retarded eye-blink conditioning (130), while limbic seizures are regulated by the interplay between GABAergic and glutamatergic transmission in the mediodorsal thalamic nucleus (131); furthermore, the mediodorsal thalamic nucleus is involved in Korsakoff's amnesia (132), psychosis associated with temporal lobe epilepsy (133) and limbic motor seizures (134). Taken together, these data support the involvement of the medio-dorsal thalamus in psychotic behaviour.
D3 receptor and D2 receptor densities in human thalamic nuclei are low-to-moderate and do not differ significantly, in contrast with other brain areas, where D2 receptors generally prevail; mRNA for these receptors are moderately expressed in neurones of the mediodorsal and anteroventral thalamic nuclei, i.e., the two nuclei which post-mortem studies suggest they are involved in the pathophysiology of schizophrenia; it is noteworthy that in the mammillothalamic tract, D3 receptors show the highest concentration for a dopamine receptor of the D2 group, which is compatible with this receptor being expressed at an axonal level, whereas the D2 receptor is only synaptic (135), suggesting that the two receptors are regulated in a different way. However, despite the function of dopaminergic receptors in the thalamus is still unknown, and in particular, it is not known whether they directly influence the glutamatergic thalamocortical projection, it has been suggested that the thalamus is one of the major sites for antipsychotic drug action, as such drugs increase c-fos expression in the thalamus (136,137). This increases is mainly apparent in the midline thalamic paraventricular (138,139), centromedian and rhomboid nuclei and the nucleus reuniens (139), but also the mediodorsal thalamic nucleus (140). The ventromedial prefrontal cortex is another site where atypical antipsychotics induce c-fos (141), but this effect might not be directly elicited through dopaminergic receptor blockade (142). Interestingly, disinhibition of glutamatergic mediodorsal thalamic neurones effected through modulation of the dopaminergic system, increased c-fos in the prefrontal cortex of rats (143). The circuitry involved could be the one depicted in Figure 5. In fact, GABAergic interneurones in the rostrodorsal thalamic reticular nucleus inhibit the firing of glutamatergic neurones to the prefrontal cortex, reducing dopamine release in this area (144,145); such dopamine derives from the pigmented parabrachial nucleus of the A10 area (146) and is released on dendrites of the chandelier GABAergic interneurones that control glutamatergic pyramidal neurones, which in turn feed back to the ventral tegmental area (147). It is possible that schizophrenia is associated with failure of this pathway.
Since the early eighties, a glutamatergic hypothesis of schizophrenia has been worked-out on the basis of reduced cerebrospinal levels of glutamate in schizophrenic patients (148) and various, albeit non-uniform, abnormalities of metabotropic (149,150) and ionotropic (15-155) Glu receptors (AMPA/kainate), NMDA receptors (156-158) and their associated sites, such as sigma sites (159) (but not phencyclidine sites (159) ), and glycine (160) sites in schizophrenic brains, reduced glutamate synaptosomal release (161) and altered excitatory amino acid re-uptake (150). Furthermore, neuroleptic enhance NMDA-mediated glutamatergic transmission (162), and sigma ligands (163,164) as well as glycine analogues or agonists (165-169) have been used (with limited success) in add-on to treat schizophrenic positive and negative symptoms. The glutamatergic deficit hypothesis (170) is integrated with the dopamine/5-HT2A imbalance (171-174), the GABAergic predominance (175) and the dopaminergic hypotheses of schizophrenia (176) and with the phencyclidine model (177). The thalamus, given its rich glutamatergic content and innervation is a good candidate for finding possible alterations of glutamatergic function, but as yet, only two studies investigated glutamatergic receptors in this area. The one concerning metabotropic receptors failed to identify any significant difference in schizophrenic individuals (71), the other focused on ionotropic receptors and showed significant reductions of many receptor subtypes and sites (72); these reductions regarded mainly the anterior, the centromedial and the mediodorsal nuclei, but also the anterior nucleus, without assessing whether the anteroventral subdivision was more or less involved. Although the investigators speculated that these areas are those connected with the limbic system, it could also be the case that in the mediodorsal nucleus glutamatergic receptors are reduced on neurones receiving input from the prefrontal cortex. In the thalamus, the metabotropic receptors are more distributed outside the nuclei that were found to be involved in schizophrenia (178), so it is not surprising that they were found not to differ significantly between schizophrenics and controls. Ionotropic receptors on the other hand, both NMDA and non-NMDA, are co-localised in reticular and ventroposterior nuclei both on corticothalamic and lemniscal synapses and on GABAergic dendrites (179), hence the finding of reduced content in schizophrenia fits with the GABAergic hyperactivity/glutamate hypoactivity hypothesis. The differential distribution of AMPA preferring subunits (180) is consistent with mediodorsal and centromedial thalamic GluR1 as playing a major role betwixt this group of receptors in the pathophysiology of schizophrenia.
The thalamus contains many perikarya with many interacting neurotransmitters, such as acetylcholine, glutamate, aspartate, gamma-aminobutyric acid (GABA), enkephalin and noradrenaline, but the latter is confined in the paraventricula, ventral posterior and lateral geniculate nuclei and in the pulvinar (181). Receptors for many neurotransmitters and neuromodulators are also found in the thalamus. Opioid receptors belonging to all three major opioid subtypes, m, d, and k, are present all over the thalamus (182). These receptors are thought to be involved in pain and somatosensory integration and have not been specifically investigated in schizophrenic thalami. Angiotensin II thalamic sites were localised in the anterodorsal, laterodorsal, posterior, lateral geniculate, and reticular nuclei, as well as in the zona incerta (183); these sites also were not investigated in schizophrenic thalami. Among neuropeptides for which an involvement in the pathophysiology of schizophrenia has been proposed, terminals or receptors for somatostatin (184,185), cholecystokinin (185-189), neuropeptide tyrosine (NPY) (190-194,185) (a Y1-5, but also a sigma receptor ligand (195-197), mainly sigma1 (198,199), which regulates glutamatergic (199,200) and dopaminergic transmission (198,199,201,202) ), substance P (187,203), neurotensin (188,204) and vasoactive intestinal polypeptide (VIP) (188,189) have been found in the thalamus; cholecystokinin (205), substance P (185,206) and all the other peptides were found to be reduced post-mortem in the thalamus of schizophrenic patients (207). Studies are lacking on distinct thalamic nuclear distribution of these peptides in schizophrenia. These peptides frequently co-exist in the same neurones and are co-transmitted with classical neurotransmitters; they all affect neuronal transmission mediated through classical neurotransmitters, so it is possible that they play a role in the pathophysiology of schizophrenia.
Other receptors for classical neurotransmitters in the thalamus include cholinergic muscarinic (208,209), which are widely diffused in the mediodorsal nucleus, on neurones projecting to the cingulate cortex (208), as well as in the parvocellular division of the anteroventral nucleus (209); some of these receptors are M2 presynaptic heteroreceptors localised on terminals in the anterior and laterodorsal nuclei not synthesised in local neurones (204). Also present are GABAergic bezodiazepine-sensitive, neurosteroid-modulated GABAA receptors (210). GABABreceptors, which are most dense in the habenular nucleus, are also found in the reticular, ventrobasal and geniculate thalamic nuclei, but are powerfully reduced in adult age in the mediodorsal nucleus (211), as are a1A and a1B-adrenoceptors (212), and b2-adrenergic presynaptic heteroreceptors (204), which mediate increases in thalamic regional blood flow (213). 5-HT2A/2C receptors in the submedian nucleus (214) are related with the nociceptive circuitry. Low levels of 5-HT2A receptors (215) which do not undergo variations with age (216) are expressed in the whole area as visualised by PET. It appears that at least some of these receptors are localised on glutamatergic thalamocortical axons and that they are carried to cortical terminals, where they enhance as presynaptic heteroreceptors glutamatergic transmission in the cortex (217) (Fig. 5). Abundant 5-HT2C receptors are found in the laterodorsal and lateral geniculate nuclei (218), and high levels of 5-HT7 receptors are displayed in the entire thalamus (219-,221), comprised its intralaminal nuclei (222), but their function at this level is still obscure. On the contrary, 5-HT6 receptors are absent from the human thalamus (223). 5-HT1Db and 5-HT1F receptors are also present in most thalamic nuclei (224) and presynaptic 5-HT1B/1D heteroreceptors (204), which are transiently expressed on thalamocortical neurones at high levels during corticothalamic migration (225) and affect their growth (226) and organisation (227) overlap with the 5-HT1F/1Db population. In spite of the fact that neuroleptic/antipsychotic treatment affects most of these receptors to various extents, have not been investigated as to their possible changes in the thalamus of schizophrenic patients. It is conceivable that the search for changes in schizophrenia of these receptors should await the development of pharmacological tools that better discriminate them, as well as their more precise localisation in the human.
The tendency of schizophrenic patients to abuse nicotine is a widely known phenomenon (228) that has tentatively been explained as an attempt at self-medication (228-230) as the side effects of classical neuroleptics include unpleasant cognitive side effects and nicotine was shown to reduce them (228), and since nicotine is able to lessen negative symptoms of schizophrenia associated with abnormal sensory gating (230). Receptor abnormalities of the cholinergic transmission in schizophrenia involve both the muscarinic (231,232) (particularly the M1 subtype (233,234) )and the nicotinic (235,236) subtypes. In particular, the nicotinic receptors are heterogeneous and involve one or two subunits combined as homo- or hetero-oligomeric proteins, with specific distribution patterns, concerning area and type or portion of neurone. Subunit type and factors related to the relative participation of subunits in the construction of the receptor and the number and stoichiometric proportion of subunits involved confer each nicotinic receptor its unique pharmacological and action profile; for example, sensitivity to the agonists imidacloprid, epibatidine, nicotine and the natural endogenous ligand, acetylcholine, is determined mostly by the alpha subunit (237), with the alpha1 and alpha3 subunits determining preference for imidacloprid (238), while alpha2 and alpha4 subunits shift the preference towards nicotine. Sensitivity to the competitive antagonist, alpha-conotoxin-MII is mediated by specific domains of the beta2 subunit (239); furthermore, the relative receptor composition influences affinity for the prototypic cholinergic muscarinic antagonist, atropine, determining whether the drug will act as a non-competitive nicotinic antagonist on a channel site or as a competitive nicotinic agonist (240). Usually, a pentameric structure prevails, composed of five alpha units of the same subunit or of two of the same alpha and three of the same beta subunits, but this is not always the rule (241). These receptors may be further subdivided into alpha-bungarotoxin- and methyllycaconitine-sensitive receptors, which bear the alpha7 subunit, and high-affinity sites, which are composed of various alpha subunits (alpha1 to alpha4) and a beta subunit, mainly beta2, but also beta4.
Alpha7 nicotine receptors mediate the release of cortical glutamate from thalamocortical output and enhance glutamate-stimulated mesoprefrontal dopamine output together with alpha4beta2 high affinity receptors (242,243), while somatodendritic alpha4beta2 nicotine receptors in the ventral tegmental area inhibit stress-induced dopaminergic firing (243) (Fig. 5). In the hippocampus and cortex, alpha7 and alpha4beta2 receptors activate transmitter release of both pyramidal neurones and GABAergic interneurones (244), with the result that activation is modulated by GABAergic inhibition. The release of dopamine in the striatum is positively regulated by nicotinic alpha3beta2 or alpha4beta2 (245,246). Whereas at this level, the most predominant nicotine subtype alpha4beta2 is located on somatodendritic sites, in the human cortex is located on axonal preterminal segments and on presynaptic boutons of interneurones (247). GABAergic neurones in the hippocampus are also enhanced by alpha7 and alpha4beta2 nicotine receptors (248) and this enhancement results in inhibition of CA1 pyramidal neurones (249). In the ventral tegmental area, the release of dopamine is also positively regulated by nicotine through a nitric oxide-dependent stimulation of glutamatergic terminals (250).
The thalamus receives cholinergic innervation not only from the major brain contributor of cholinergic fibres, Meynert's nucleus basalis (251-253), bust most importantly also from mesopontine nuclei (254) and in particular from the lateral tegmental, lateral parabrachial and pedunculopontine nuclei (255), which project to the reticular nucleus (255), but also to the midline and intralaminar groups of nuclei (256), where they overlap with thalamostriatal neurones (257), and to the intermediodorsal, mediodorsal, paratenial, posteromedian, ventromedian, ventrolateral and rhomboid thalamic nuclei (257), as well as to the anteroventral nucleus and to the zona incerta (258), where they innervate also other types of neurones projecting to the cortex (259) and to other brain structures. The major output of the laterodorsal tegmental nucleus is to the anteroventral thalamus, whereas, the pedunculopontine nucleus innervates both striatal structures and the thalamus, in a uniform manner (260). It is worth mentioning that the pedunculopontine nucleus (Ch5) and the laterolodorsal tegmental (Ch6) nuclei, which are under inhibitory muscarinic autoreceptor modulation (261) and postsynaptic serotonergic modulation mediated by 5-HT1A receptors (262) from the dorsal raphé (263), also project to the A9 (substantia nigra, pars compacta) (264) and A10 areas (ventral tegmentum) (265) and to the nucleus accumbens (266) (Fig. 6) and are believed to be involved in the pathophysiology of schizophrenia in some (267-269), but not all (270) patient populations.
Binding for nicotine is highest in the anteroventral nucleus and mediodorsal nuclei of the thalamus, while the labelling was lower in the posterolateral nucleus and posterolateroventral nucleus, as revealed by autoradiography (271). Nicotinic receptor in situ hybridisation signals in humans are strong in the dorsomedial, lateral posterior, ventroposteromedial and reticular nuclei of the thalamus for the alpha3 subunit, and in the the pulvinar and ventroposterolateral nuclei of the thalamus for the alpha7 subunit (272), but low in the mouse (273); the alpha7 signal is not affected by age, as in other brain regions (274). High cortical levels of alpha7 receptors are found on rat thalamocortical neurones (275,276), where they were found to enhance, conjointly with alpha4beta2 receptors, glutamate release and AMPA-mediated transmission (277). Alpha4beta2 receptors in the thalamus are under a negative allosteric modulation by neurosteroids (278). Generally, high affinity sites are denser than alpha-bungarotoxin sites in the human thalamus, except for the reticular nucleus (279), where high levels are found both in terms of alpha7 mRNA hybridisation signal and alpha-bungarotoxin binding (280). Both beta2 and beta4 signals are detected in the primate thalamus, but the latter is more abundant than the former, differently from the cortex (281). The alpha4beta2 subunit combination was found also to be fairly represented in primate thalamic nuclei in vivo, as assessed by single photon emission computerised tomography (282). Taken together, these data suggest that both high-affinity and alpha-bungarotoxin-sensitive nicotinic receptors (mainly alpha7, but possibly also alpha8 or alpha 9) may positively modulate thalamocortical output when localised on glutamatergic neuronal bodies, dendrites or axonal segments and terminals, and they also regulate it negatively, when localised on inhibitory GABAergic interneurones, as they always tend to activate neural transmission. It appears that, in the cortex, nicotinic receptors are exclusively localised on thalamocortical terminals, whereas muscarinic receptors are found on cortical neuronal dendrites (283).
Specific nicotine receptor abnormalities have been found in the brain of schizophrenic patients. The cloned human alpha-7 gene was found to be partially duplicated proximal to the full-length gene and this duplication was found both in brain and peripheral blood cells of healthy, but not in a sub-population of schizophrenic patients (284). These alpha bungarotoxin-sensitive sites were found to be reduced in the hippocampus (285,286) and in the frontal, but not parietal cortex (287). It appears that nicotine receptors rapidly desensitise following exposure to nicotine in schizophrenic patients (288) and that this is followed by P50 auditory evoked potential alterations, thus resulting in sensory gating deficit in the same patients (289). The high affinity, [3H]cytisine-labelled alpha4-beta2 nicotine receptors, which are up-regulated by nicotine and tobacco smoking (290,291), were also found to be reduced in number in the basal ganglia of schizophrenic patients (292). The nicotine-sensitive receptors (alpha4 or 3/beta2 or 4), which are sensitised with nicotine use and revert this sensitisation after smoking cessation, were lower in the thalamus of schizophrenic smokers than control smokers; moreover, the normally found correlation between receptor binding and tobacco use was lost in schizophrenic individuals (293). The alpha-bungarotoxin-sensitive receptors (predominantly alpha7) were found to be reduced in schizophrenic thalami by about 25%, a proportion which is half the one observed in Lewy body dementia, but still significant (70). While these studies confirm the importance of not only area-specific abnormalities in the aetiopathogenesis of schizophrenia, but also of a failing circuitry, they add a cholinergic component to this circuitry, the importance of which was largely underestimated until recently.
In recent years, there has been a shift of interest towards molecules expressed transiently during development in schizophrenia research and abnormal persistence of Wnt-1 in schizophrenic hippocampal CA4 area (294), as well as consensual reduction in beta-catenin was found in the same structure, with reductions of both beta- and gamma-catenin also in the CA3 area (295). Another early developmental gene polymorphism that was found to be potentially linked to schizophrenia is Pax-6, for which evidence was obtained for a genotypic association of the high-activity variant with paranoid schizophrenia (296). These molecules provide for transcriptional regulation in the developing and adult human brain and may be affected by a neurodevelopmental defect. Reelin is a glycoprotein secreted into extracellular matrix by Retzius-Cajal cells during development and by GABAergic interneurones in the adult cortex and cerebellum, binding on integrin receptors at the level of dendritic spines at post-synaptic density of glutamatergic pyramidal neurones (297) and inducing disabled-1 gene product (DAB1), which is a cytosolic adaptor that mediates the action of reelin. This molecule was found to be significantly reduced in the prefrontal cortex and cerebellum of psychotic patients (298), schizophrenic (298,299) or bipolar (298). Reelin was found during development in the thalamus and in other sites of the circuitry that could be involved in schizophrenia (300). A possible abnormal persistence of this molecule has not been sought to date in schizophrenic thalamus, a region where the GABAergic/glutamatergic is prominent as in the cortex; should it be found in future studies, it will further add to the neurodevelopmental theory of schizophrenia. Other transient molecules providing axonal guide during corticogenesis, such as netrin-1, which provides for interplay between attraction and repulsion in thalamocortical neuronal guidance (301) and is involved in plasticity (302), are good candidates for research in thalamic abnormalities in schizophrenia. The neuronal defective migration theory does not explain why lesions so early established do not show-up for several years. An explanation could be that the defect in migration creates a vulnerable structure that fails upon increased demands occurring at some critical points of life, but how this is mediated is not known, and why individuals with similar alterations (i.e., patient relatives) do not fail cannot be explained by current MRI data on the thalamus. It would be interesting to obtain some post-mortem material from patient relatives to compare the latter with the patients and with healthy controls.
The evidence of thalamic involvement in schizophrenia is sound, and points at a reduction of thalamic volume and function in this disease (303). Some discrepancies encountered in various studies, where the thalamus was found to be increased or hyperperfused, might be explained admitting that the primary defect is in the frontal cortex, which gives feedback innervation to both ipsilateral and contralateral thalamus; in fact, brain tumours that destroy frontal cortex result in reduction of thalamic volume ipsilaterally and an increase contralaterally (304). The thalamus is a structure that underwent considerable enlargement during human adaptation (305). While the anterior nuclei might be regarded as a primitive group (306), it is noteworthy that the anteroventral nucleus sends excitatory aminoacidergic (307) fibres to (309-315,308) and receives from (316,317) the cingulate cortex, according to a neurogenetic gradient (i.e., older thalamic cells synapse with younger cortical cells and vice versa) (318). Abnormalities within the cingulate cortex were reported in schizophrenia and they regard principally the anterior cingulate (319-323) (which is involved conjointly with the anteroventral nucleus in discriminative learning and conditioning processes (324-329) that upregulate muscarinic receptors in both parvicellular anteroventral thalamic nucleus and rostral cingulate cortex (330) ), whereas anteroventral thalamic innervation of the cingulate cortex is mainly directed to the posterior cingulate (313). However, it should be noted that the anterior cingulate receives innervation from the posterior cingulate (309), a fact that could explain the conjoint involvement of anteroventral and anterior cingulate cortex in avoidance behaviour (321-324). It appears that behaviour acquisition relies in cingulate activity and that anteroventral thalamic activity is involved in its retention and maintenance (324). Whereas total cingulate volume is only marginally decreased in schizophrenia, as assessed through MRI (331), schizophrenic patients display a decreased N-acetyl aspartate signal in the anterior cingulate with respect to controls, which correlated with age, disease duration, and classical neuroleptic medication in one study (332), but not in another (333). Dopaminergic (334), GABAergic (49,50,335,336) and excitatory amino acidergic (337) abnormalities, as well as abnormalities in synaptic proteins (338,339) and neural cell adhesion molecule (339), were all shown in schizophrenic cingulate. Decreased metabolic rate with respect to controls was shown with PET in schizophrenic anterior cingulate both at rest (340) and during cognitive task performance (341); the latter is reversed through the administration of apomorphine (342). A probably compensatory increase in metabolic rate was observed in the posterior cingulate cortex in schizophrenic patients (340) (Fig. 7). Combining these data, we may hypothesise that a neurodevelopmental migration defect results in ectopic interneuronal localisation in hippocampus (343,344) or prefrontal cortex (107), which in turn is reflected in a dopaminergic innervation shift from pyramidal to non-pyramidal neurones in layer II cingulate cortex (345,346); this miswiring is explicated by dopaminergic terminals towards parvalbumin positive GABAergic chandelier interneurones at the expense of layer II or V/VI glutamatergic or aspartatergic pyramidal neurone innervation (336,347,348). This could be translated into both reduced cingulate cortical activation (341) and dopaminergic hypersensitivity (342,349,350) of "denervated" (351) or erroneously not innervated synapses, with decreased feedback to the anteroventral thalamus and consequent positive feedback hyperactivity of the posterior cingulate (340) (Fig. 7). It should be underlined that the cingulate is a key region in controlling prefrontal-temporal cortex integration and that this modulation fails in schizophrenia (352).
The size of the phylogenetically more advanced mediodorsal nucleus parallels that of the prefrontal cortex in all species (306), hence its major involvement in processes such as attention, memory formation, rather than retrieval (353), and thought, which are impaired in schizophrenia. With the expansion of the cerebral cortex, the thalamus had to connect with a far wider area, so this structure in the human, by being involved in particular emotional and cognitive processes, achieved a high degree of complexity. Hence, since it works as an integrative gateway to the cortex and at crossroads with the rest of the brain, it had to develop into an organiser of brain activity. The particular connectivity the human thalamus achieved as a result of human adaptation to its particular environment determined its involvement in brain functions, such as human language and thought processes, but also consciousness (354) and construct of self, according to Metzinger's self-model (355), in particular, agency and spatial perspectivity, which are impaired in schizophrenia giving rise to symptoms like hallucinations, derealisation and depersonalisation (356). Hence, the disorganisation of this connectivity results in disorganisation of such processes, so characteristic of schizophrenia, which cannot be adequately simulated by any animal model, due to the fact that thought processes and language of animal species are difficult to understand completely by humans.