Neurodevelopmental impairment, dopamine sensitisation, and social adversity in schizophrenia

During the early 1980s, the view emerged that schizophrenia had a significant developmental component (1-4), and subsequently the 'neurodevelopmental hypothesis' became the dominant model in schizophrenia research. However, while few now doubt that neurodevelopmental disturbance plays an important role in schizophrenia, it is increasingly clear that the original simple 'doomed from the womb' model offers only a partial explanation for the disorder. Consequently, attention has recently turned to a) whether the process of aberrant development is an exclusively biological one or whether there is also a psychosocial component; and b) what causes the onset of psychosis. There is increasing evidence that dopamine sensitisation provides the link between early developmental deviance and the vulnerability of the preschizophrenic individual to certain stressors in the social environment.

EVIDENCE FOR DEVELOPMENTAL IMPAIRMENT

Risk-increasing effect of early environmental events

A major impetus to the formulation of the neurodevelopmental hypothesis was the evidence that pregnancy and delivery complications (collectively termed obstetric complications or OCs) increase risk of schizophrenia (2). Two meta-analyses summarised much of the early data (5,6), and confirmed an association between OCs and schizophrenia with an odds ratio of approximately 2. There were many criticisms of the methodology in the studies included in these meta-analyses, but the last decade has seen the publication of a series of registerbased longitudinal studies of a higher quality. A third meta-analysis, this time of these registerbased studies, confirms that exposure to OCs is a modest risk factor for schizophrenia (7). The mechanism underlying the link between OCs and schizophrenia remains uncertain, although one possibility is foetal hypoxia. According to Cannon et al (8), the odds of schizophrenia increase linearly with an increasing number of hypoxic-ischaemic complications, possibly mediated by the excitotoxic effects of hypoxia on the foetal/neonatal brain.

A slight increase in risk for schizophrenia exists among individuals born in late winter/ early spring (9). While some studies suggest this seasonal effect could be secondary to exposure to influenza in utero during winter, other research fails to find such a link (10,11). Other potential, but not yet replicated, early hazards are intrauterine rubella (12), as well as maternal malnutrition (13) and maternal diabetes mellitus (14). Finally, Rantakallio et al (15) demonstrated that the window of opportunity for risk increasing insults is wider than was previously thought, as those exposed to childhood viral central nervous system infections were five times more likely to develop schizophrenia than those not exposed.

Neuroimaging evidence of impaired development

It is well known that there are volumetric differences in certain brain structures between schizophrenic patients and controls. Enlargement of the lateral and third ventricles is the most prominent finding - a meta-analysis of 58 magnetic resonance imaging (MRI) studies reported a 26% enlargement of ventricular volume in schizophrenic subjects (16). Subtle volume reductions are frequently reported in the medial temporal lobe, superior temporal gyrus, prefrontal cortex, cingulate gyrus, thalamus and insula (16,17).

Several lines of evidence suggest that at least some of the structural changes are likely to result from impairment of normal neurodevelopment. Firstly, many are detectable in patients at or near the onset of psychosis, including enlargement of the lateral and third ventricles and reduced volume of cortical grey matter, the thalamus, the left hippocampus/amygdala and the left posterior superior temporal gyrus (18- 20). Thus these abnormalities cannot be wholly secondary to disease progression or treatment.

Secondly, some of the brain abnormalities are also found in the unaffected relatives of patients. McDonald et al (21) examined a large number of families either multiply or singly affected with schizophrenia and found that there was substantial enlargement of lateral and third ventricular volume in those relatives at high likelihood of carrying susceptibility genes for schizophrenia, whereas those relatives at lower likelihood of carrying such genes had similar ventricular volume to the control sample. Enlarged third ventricular volume in unaffected siblings was also found by Staal et al (22), and other structural abnormalities identified in unaffected adult relatives or in younger high risk subjects include reduced volume of cortical grey matter (23), the amygdala-hippocampal complex and the thalamus (24,25). The existence of morphological deviations in unaffected relatives indicate that such abnormalities are not restricted to the pathological process of psychosis but are a manifestation of familial risk factors, the most likely candidates being genes influencing neurodevelopment.

Thirdly, some of the morphological abnormalities identified in adult schizophrenic patients suggest pathophysiologic processes affecting the early developing brain. These include uncommon developmental brain lesions such as aqueduct stenosis, arachnoid and septal cysts, and agenesis of the corpus callosum, which occur with excess frequency in schizophrenia (26). Normal structural cerebral asymmetries such as those of the planum temporale and fronto-occipital petalias also develop during foetal life and some studies have reported loss or reversal of such asymmetries in schizophrenia (27). Similarly, gyrification of the prefrontal cortex achieves stability soon after birth and Vogeley et al (28) have reported right frontal cortical hypergyrification in schizophrenia.

Fourthly, a series of correlations have been reported between early environmental risk factors for schizophrenia and structural deviations in the adult brain. These include early reports (2) of an association between increased ventricle-to-brain ratio on computed tomography (CT) scans and exposure to OCs, although other studies did not find a relationship between OCs and lateral ventricular enlargement (reviewed by McGrath and Murray [10]). Recent evidence points to an interaction between genetic risk for schizophrenia and hypoxic birth events upon ventricular volume, since patients were more likely to demonstrate ventricular enlargement and reduced cortical grey matter volume if they had also experienced OCs than were controls (21,29). Other structural abnormalities such as reduced hippocampal volume have also been linked to OCs (30,31).

Childhood evidence of neurodevelopmental impairment

Children who go on to develop schizophrenia tend to display early neurological and cognitive problems. As early as 1977, Fish (32) pointed out that the increased prevalence of neurological signs in multiple sensorimotor systems in the offspring of schizophrenics was consistent with an 'inherited neurointegrative deficit'. High-risk studies concur that 25-50% of children born to mothers with schizophrenia have developmental abnormalities, especially poor motor co-ordination in early childhood, and attention and information processing deficits later (33).

Since the majority of schizophrenic subjects do not have an affected parent, Jones et al (34) examined the 1946 British Birth Cohort Study which followed up 4746 children for 43 years. The 30 who developed schizophrenia, as a group, had delayed milestones, more speech problems, and lower educational test scores. Cannon et al (35) compared elementary school records of 400 children who later developed schizophrenia and 400 healthy controls in individuals born in Helsinki. Poor performance in sports and handicrafts, which may indicate motor co-ordination deficits, were risk factors for schizophrenia.

How specific are such predictors for schizophrenia? In the British National Child Development Study, Done et al (36) showed that those who developed schizophrenia had significantly more neurocognitive problems than controls at the age of 7 years, especially if they were males. Children who went on to develop affective psychoses did not differ from controls, while pre-neurotic children, especially if female, manifested poorer social adjustment than controls at age 11. In the Dunedin follow-up study, in which 761 children were regularly studied till the age of 26 years, those who went on to develop anxiety disorders, mania and schizophrenia shared a range of emotional and conduct difficulties. However, only those who later developed schizophreniform psychosis showed neurocognitive or neuromotor impairment. Thus, one can consider the preschizophrenic child as exhibiting a general liability to adult psychiatric disorder compounded by a more specific neurodevelopmental impairment (37).

Cognitive impairment has been repeatedly reported in those destined to develop schizophrenia (34). David et al (38) investigated a cohort of nearly 50,000 18-year-old males who were conscripted into the Swedish army in 1969-1970. There was a highly significant linear association between low IQ scores and the subsequent development of schizophrenia, with risk gradually increasing as IQ fell at all levels of intellectual ability. Similarly, Davidson et al (39) examined assessments of nearly 10,000 16 and 17 year old boys entering the Israeli army, and again found a linear relationship between IQ and risk of schizophrenia. Such findings raise the question of whether the association between low IQ and schizophrenia is directly causal, with cognitive impairment compromising information processing and leading to false beliefs and perceptions.

Is there a social developmental component?

All the children in the Dunedin study discussed above were given a structured interview by a child psychiatrist at age 11 years. This included four questions regarding quasipsychotic ideas (Have other people ever read your mind? Have you ever had messages sent just to you through TV or radio? Have you ever thought that people are following you or spying on you? Have you heard voices other people can't hear?). Interestingly, children who reported having such ideas had a 16-fold increased risk of schizophreniform disorder at age 26 years (40). Furthermore, having these quasi-psychotic ideas was strongly predicted by motor and language deficits.

Thus, there exists a certain continuity of psychotic symptoms from childhood to adulthood, and the origin of delusions is often more than a decade before psychosis is formally diagnosed. These findings are important in that they demonstrate that quasi-psychotic ideas in childhood are on the causal pathway to frank psychosis and are not simply epiphenomena of an early neural lesion. While it is clear that neuropsychological deficits facilitate the development of such early ideas, the Dunedin findings open the question of whether non-biological factors can do likewise.

Rearing factors

In the British 1946 cohort, those 4-year-old children rated as having a poor mother-child relationship had a 6- fold increase in risk for schizophrenia later on in life (34). Similar findings were reported in the Dunedin study. Of course, this does not tell us whether poor mothering was a causal risk factor, or whether the pre-schizophrenic child was so deviant as to be unable to form a close bond with the mother. Also those adopted-away children of schizophrenic mothers who are reared in adverse circumstances have a higher risk than those brought up in loving homes by stable adoptive parents (41). Furthermore, Mirsky et al (42) noted that children with known genetic risk for schizophrenia were more likely to develop the disorder if they lived in a kibbutz, rather than a family home.

Being brought up in a city

Several studies have indicated that being born or brought up in a city increases the risk for schizophrenia. For example, Mortensen et al (43) showed in a Danish national sample that the relative risk of schizophrenia associated with urban birth was 2.4, and that the larger the town of birth, the greater the risk. van Os and his colleagues (44) have shown that it is not only frank psychosis but also minor quasi-psychotic ideas which are more common in urban dwellers. Pedersen and Mortensen (45) have gone on to demonstrate that there is a dose response relationship between urban rearing and risk of schizophrenia, such that the longer an individual lived in a city as a child the higher the risk. A further risk-increasing factor was move of household particularly during adolescence; the greater the number of moves the higher the risk. The authors raise the question of whether individuals with shy schizoid personalities may have particular difficulties in making new friends following such moves and become increasingly isolated.

Social isolation

The Swedish conscript study discussed before found that young men who felt they were more sensitive than their peers, had fewer than two close friends and did not have a girl friend, had an increased risk of later developing the disorder (46). Once again this raises the question of whether these characteristics are an expression of a schizoid or schizotypal personality or whether they are in themselves independent risk factors. Until proven otherwise, it is wise to consider that both may be true, i.e. individuals with a schizoid or schizotypal personality may be less able to make social relationships and then the social isolation itself may cause them to become increasingly deviant. van Os et al (47) found that people who were single had a slightly higher risk of developing psychosis if they lived in a neighbourhood with fewer single people, compared to a neighbourhood with many other single people. The authors suggested that single status might give rise to perceived (or actual) social isolation if most other people are living with a partner. The question of whether social isolation may increase the risk of schizophrenia (or rather whether a close relationship may be protective) is also raised by Jablensky et al (48), who showed that marriage had a protective effect for males, and that this was not simply a consequence of better-adjusted males being able to marry.

Migration

As far back as the 1930s, Odegaard (49) noted that Norwegian migrants to the USA were at increased risk of schizophrenia, while as recently as 1999, Mortensen et al (43) reported that children born in Greenland to Danish mothers had a relative risk of 3.7 for schizophrenia. The most striking findings have come from the UK, where numerous studies have reported an increased incidence of schizophrenia among African-Caribbean people (50). Hutchinson et al (51) found that morbid risks for schizophrenia were similar for parents and siblings of white and first generation African-Caribbean patients. However, morbid risk for siblings of second generation African- Caribbean psychotic probands was approximately 7 times higher than that for their white counterparts. This study, which almost exactly replicates the work of Sugarman and Craufurd (52), suggests an environmental agent that is operating on this population in the UK but not in the Caribbean. Again social isolation could be relevant; Boydell et al (53) found that the incidence of psychosis in the ethnic minority population in South London (mainly black) is much higher in those areas where few of the minority population live compared with those areas which have a substantial minority population.

WHAT CAUSES THE ONSET OF PSYCHOSIS?

Maturational brain changes?

A major problem for the original neurodevelopmental model was to explain why damage, presumed to be present since foetal or neonatal life, does not cause psychosis until decades later. The solution proposed was that the crucial lesion(s) could lie silent until the operation of brain maturational processes in adolescence expose neuronal circuits that are underdeveloped and are not functional in childhood. The developing brain has a large excess of neurones and axons, which thin out during early development, thus serving to eliminate early errors of connection and to strengthen those that are useful. The relevance of pruning of cortical neurons during adolescence for schizophrenia was introduced by Feinberg (1), who proposed there may be a fault in progressive synapse elimination. Murray et al (54) postulated immature circuitry laid bare by synaptic pruning, a process which continues until after puberty. Early injury could interact with such processes, resulting in patterns of dysconnectivity.

However, if maturational change is considered necessary to interact with an earlier abnormality, might maturational changes on their own be sufficient to initiate schizophrenia? Keshavan et al (55) used magnetic resonance spectroscopy to show that people with schizophrenia show a phosphomonoesterase pattern suggestive of failure of new synapse production and excessive synaptic reduction. This process is postulated to lead to a loss of synaptic connectivity below a critical level (56). Postmortem studies of schizophrenia consistently show reductions in neuronal size, dendritic spine density and length, synaptic proteins and synaptic gene expression (57). These post-mortem findings are suggestive of pathology of synapses and their connectivity and are compatible with theories which implicate abnormalities of pruning in adolescence.

Structural brain changes which reflect the above processes of late brain maturation can be seen in late childhood and adolescence. These include age-related grey matter volume reductions, greatest in the frontal and parietal convexities (58). Such changes appear to be particularly prominent in childhood-onset cases of schizophrenia, where increasing ventricular enlargement and progressive reduction of cerebral volume, grey matter and temporal lobe structures have been reported (59,60).

Dopamine sensitisation

The neurodevelopmental hypothesis helps us to understand the development of the adolescent at risk for schizophrenia. However, this does not adequately explain what converts an odd, socially isolated adolescent with some deficits in cognition and strange ideas, into a frankly psychotic individual. The answer may lie in the impact of developmental impairment on the mesolimbic dopamine system. As is well known, the dopamine hypothesis of schizophrenia derives from the evidence that all antipsychotics block dopamine D2 receptors, whereas direct or indirect dopamine agonists elicit positive symptoms of schizophrenia. This longstanding theory has recently been bolstered by evidence which directly implicates dopamine dysregulation in the pathogenesis of the positive psychotic symptoms (61). For example, a series of single photon emission computerised tomography (SPECT) and positron emission tomography (PET) studies have demonstrated that patients with schizophrenia release excessive amounts of dopamine in response to an amphetamine challenge and that there is a clear relationship between degree of this release and psychotic symptoms (62,63). Kapur (61) has gone so far as to regard acute psychosis as "a disorder of dopamine-induced aberrant salience."

It is postulated (61,62,64) that the dopamine dysregulation arises from the development of sensitisation, the process whereby repeated exposure to a drug induces not tolerance but rather reversed tolerance with progressively increased neurochemical and behavioural responses. Thus, Kapur (61) states that "somewhere in their lateteens (the usual age for onset of psychosis in schizophrenia) patients develop an abnormality of the dopamine system such that there is an exaggerated release of dopamine, out of synchrony with the stimuli that usually induce them. This state does not lead to any physical feelings, but, leads to the assignment of inappropriate salience and motivational significance to external and internal stimuli."

The relevance of early developmental impairment for dopamine sensitisation is demonstrated by studies which show that animals subject to a range of perinatal lesions develop dopamine systems which are particularly prone to such dysregulation when they mature. For example, Lipska et al (65) reported that hippocampal lesions to neonatal rats remain relatively silent until adult life, when the animals develop hyper-responsiveness to amphetamine and also to stress. Lipska's model (66) mimics many of the abnormalities which are found in schizophrenia, but of course it results from a gross artificial lesion. Boksa and her colleagues (67,68) have shown that a more physiological perinatal stress - anoxia during caesarean section - induces altered mesocorticolimbic dopamine transmission in the rat together with increased behavioural responses to amphetamines and to stress. There is also evidence that perinatal injury in humans can similarly induce subcortical dopaminergic overactivity. Thus, Kapucu et al (69) used SPECT to examine 20 infants with hypoxicischaemic brain damage, and noted that neostriatal dopamine D2 binding decreased as the severity of the injury increased. A similar end-result might arise from abnormal cortical pruning. Thus, Laruelle and Abi- Dargham (64) postulate that "during late adolescence, the failure of cortical development in schizophrenia might limit the capacity of the brain to modulate stress-related activity of the mesolimbic DA neurons. This failure of normal homeostatic and buffering mechanisms results in a process of endogenous sensitization". They postulate that the increased dopamine activity triggers neuroplastic adaption downstream from the mesolimbic dopamine synapse, and that eventually the neuroplastic changes become independent of dopamine so that positive symptoms circuits become 'hard wired' and the patient becomes treatment resistant (64).

Drug abuse

In animal studies, repeated exposure to drugs such as amphetamines and cocaine induces dopamine sensitisation; in humans this can precipitate psychosis (61,64). The development of persecutory delusions, bizarre delusions as well as olfactory and auditory hallucinations and a variety of thought disorders is well documented in experimental studies (70). Repeated exposure to cannabis and its constituent tetrahydrocannabinol also induces alterations in dopamine transmission and sensitisation to amphetamine in rats (71) and dopamine release in humans (72). It is also known to trigger brief psychotic episodes, and to exacerbate pre-existing psychotic symptoms (73). Several studies show that its repeated use is a risk factor for psychosis. In the best known, Andreasson et al (74), who examined 45,570 Swedish 18-year conscripts, found that the relative risk of developing schizophrenia over the next 15 years was 6.0 for heavy users of cannabis compared to non-users (at time of conscription). The findings of the Swedish Army study have been supported by other studies currently in press which are likely to put the issue beyond doubt (75,76). The latter group, who examined the Dunedin cohort, found that cannabis consumption at age 15 was a risk factor for later psychosis even when one took into account childhood psychotic symptoms.

Do individuals with certain genotypes selectively expose themselves to drugs with psychotogenic effects to which they are particularly vulnerable? In the Swedish conscript study, over half of those who admitted heavy cannabis use at age 18 already had a psychiatric diagnosis (74). Similarly, McGuire et al (77) showed that individuals with cannabis-associated psychosis had an increased genetic risk of schizophrenia. In a large study of methamphetamine abusers, Chen et al (78) noted that such individuals were more likely to have a particular variant of the D4 receptor gene than controls, and further that the 164 abusers who developed a psychosis had a stronger genetic loading and more schizoid/schizotypal personality traits than those abusers who never went psychotic. Thus, it may be that some individuals abuse drugs because they are genetically predisposed to have psychiatric difficulties and, among those individuals who abuse drugs, it is those who have such a genetic predisposition who are particularly likely to develop psychosis.

Social adversity

In animals there is ample evidence that dopamine sensitisation secondary to developmental insult is associated with an exaggerated response to social stress (67). The same process has been postulated for humans (64) and could explain why an adverse social environment (e.g. social isolation) could have a particularly detrimental effect on the developmentally compromised child. Furthermore, prospective studies have found an association between life events and onset of psychosis (79); stressful life events in the three weeks preceding onset or relapse seem particularly important. One can envisage that such events could produce excessive dopaminergic response in the pre-psychotic individual and that, as dopamine sensitisation becomes increasingly established, smaller and smaller stresses could induce the abnormal response.

IS THERE EVIDENCE OF DEGENERATION?

The neurodevelopmental hypothesis fails to adequately explain the malignant course shown by some patients over time. Wyatt (80) suggested that the likelihood of experiencing deterioration is correlated with the duration and number of periods of active psychosis. Associations between the duration of untreated psychosis and likelihood of developing a more 'malignant' illness initially seemed to support this (81). However, Verdoux et al (82) found that when premorbid characteristics such as childhood function, family history and negative symptoms were taken into account, there was little independent evidence for an adverse effect of duration of untreated psychosis.

It is difficult to attribute schizophrenia with its onset late in life (over 45 years) to neurodevelopmental damage. There is some evidence that these late onset cases have less genetic loading for schizophrenia or exposure to early environmental insults. For this reason, Murray et al (83) suggested that these cases might be secondary to neurodegeneration.

The question of whether a neurodegenerative process also plays a part in earlier onset schizophrenia has been reopened by studies claiming progression of morphological changes after the onset of psychosis. DeLisi et al (84) performed a prospective follow-up of first episode cases of schizophrenia and age-matched controls for a minimum of four years and compared the rate of change over time in the size of certain brain structures. There were differences in the rate of change in the overall volumes of left and right hemispheres and right cerebellum and in the area of the isthmus of the corpus callosum, interpreted by the authors as evidence for a subtle active brain process in the first few years of a schizophrenic illness. Other studies of first episode patients have also found apparent progression of brain structural changes, such as volume reduction of the frontal lobes (85). Wood et al (86) report a reduction of whole brain volume in both first episode and chronic patients at an average 2.3-years follow-up. Pantelis et al (87), who looked at a small group of high-risk subjects, claim that those who subsequently developed psychosis showed left-sided reductions in parahippocampal, fusiform and orbitofrontal cortex. They concluded that this process may start in the prodrome before frank psychotic symptoms are expressed.

Reports of progression of structural abnormalities are more common in, but not confined to, first episode and early onset cases. Mathalon et al (88) reported that patients with chronic schizophrenia had accentuated loss of fronto-temporal grey matter and enlargement of ventricular and sulcal cerebrospinal fluid spaces at an average 4-year follow-up. Saijo et al (89) reported progressive lateral ventricular volume enlargement in a small sample over a follow-up period of up to 10 years. Several studies report that progression is associated with a more severe clinical course and poorer outcome (85,88,90).

However, not all follow-up studies of first episode patients report progressive changes. DeGreef et al (91) found no change in cortical or ventricular volume over a 1-2 year period, and James et al (92) report no progression during late adolescence of ventricular enlargement in adolescent onset patients after an average follow-up period of 2.7 years. Keshavan et al (93) found that reduced volume of the left superior temporal gyrus in a small sample of first episode patients actually showed evidence of reversal at one-year follow-up. Moreover, as Weinberger and McClure (94) point out, there are numerous methodological difficulties associated with longitudinal quantitative neuroimaging studies. Such difficulties include different resolution of imaging techniques, small sample sizes and inadequate matching of controls and subjects, as well as the uncertain impact of prescribed medication and substance misuse. Furthermore, there is no support whatsoever from neuropathology studies that the findings of volume decrements in follow-up MRI studies reflect neurodegeneration (see review by Harrison [57]). Finally, if there are such changes in the brain, as suggested by some MRI studies, in the first five years following onset of psychosis, how is it that at the same time neuropsychological studies show that cognitive function is either static or improving? Any plausible degenerative theory must explain this curious paradox.

It is worth noting that many studies reporting progression of structural changes have been performed on younger patients in their first episode of illness or with onset in childhood or adolescence, when the brain is still at a crucial stage of development. Thus, one explanation for the confusion over whether or not there is progression of brain changes after onset of psychosis is that those changes noticed are in fact changes associated with late brain development, as suggested by Rapoport et al (60). It may be that maturation accentuates the trajectory of neurodevelopmental deviance in already compromised brains, without the need to invoke an additional neurodegenerative process.

CONCLUSION

It is indisputable that there is a developmental component to schizophrenia. Dysplastic neural development could account not only for some of the clinical characteristics of the illness but also for the loss of neuropil seen in postmortem studies and the dysfunctional circuitry seen in functional imaging studies (95). However, the original developmental hypothesis requires to be modified in two important ways. Firstly, to encompass the evidence that neuropsychological and social deviance in childhood are not simply epiphenomena of an early lesion but are part of the causal pathway to psychosis; thus, some children show quasi-psychotic ideas many years before the onset of illness. Secondly, the developmental hypothesis needs to incorporate the evidence implicating dopamine sensitisation in the initiation of psychosis; developmental insults such as hypoxia can establish a vulnerability to dopamine sensitisation which can then be later exacerbated by drug abuse and social stress. The sensitised dopamine state provides a mechanism through which social adversity can have an effect and indeed, there is increasingly robust evidence that there exist social factors of aetiological significance such as isolation and migration. As yet we cannot say whether it is necessary to invoke additional neurodegenerative processes; the evidence remains too uncertain. The alternative view is that the deterioration shown by some schizophrenic patients results from the psychological and social consequences of the psychotic state itself, and leads to a downward spiral of increasing social alienation, increasing dopamine dysregulation, and increasingly distorted belief systems.