Introduction and history The dopamine (DA) hypothesis is the oldest and most established of the schizophrenia hypotheses. It has evolved from clinical observations, and received empirical validation from antipsychotic treatment and more direct testing from imaging studies. Although clearly not sufficient to explain the complexity of this disorder, it offers a direct relationship to symptoms and to their treatment. We will review here the history of this hypothesis, followed by the evidence from multiple lines of research supporting the presence of alterations of dopamine in schizophrenia. Furthermore, we will provide updates from studies that have been published since the original review was posted on this site nearly six years ago.
The first formulation of the DA hypothesis of schizophrenia proposed that hyperactivity of DA transmission was responsible for the disorder (Rossum, 1966). This was based on the early observations that dopamine receptors are activated by psychostimulants, that non-reserpine neuroleptics are dopamine antagonists, and that dopamine plays an important role in the extrapyramidal motor system. Much of the evidence behind these concepts derived from the seminal work of Arvid Carlsson characterizing the presence of dopamine in the brain and the effects of neuroleptics on monoaminergic indices (Carlsson and Lindqvist, 1963).
The classical dopamine hypothesis received further support from the correlation between clinical doses of antipsychotic drugs and their potency to block DA D2 receptors (Creese et al., 1976; Seeman and Lee, 1975), and from the psychosis-inducing ("psychotogenic") effects of DA-enhancing drugs (for review, see Angrist and van Kammen, 1984; Lieberman et al., 1987a). Given the predominant localization of DA terminals and D2 receptors in subcortical regions such as the striatum and the nucleus accumbens, the classical DA hypothesis of schizophrenia focused on subcortical regions.
Over the years, the increasing awareness of the importance of enduring negative symptoms (flattening of affect, apathy, poverty of speech, anhedonia, and social withdrawal) and cognitive symptoms (deficits in attention, working memory, and executive functions) in this illness and of their resistance to D2 receptor antagonism led to a reformulation of the classical DA hypothesis. Functional brain imaging studies suggested that these symptoms might arise from altered prefrontal cortex (PFC) functions (for reviews, see Knable and Weinberger, 1997). A wealth of preclinical studies emerged documenting the importance of prefrontal DA transmission at D1 receptors (the main DA receptor in the neocortex) for optimal PFC performance (for review, see Goldman-Rakic et al., 2000). Together, these observations led to the hypothesis that a deficit in DA transmission at D1 receptors in the PFC might be implicated in the cognitive impairments and negative symptoms of schizophrenia (Davis et al., 1991; Weinberger, 1987), while the excess DA transmission may be related only to the core or "positive" symptoms (hallucinations, delusions).
As a result, an imbalance in DA with hyperactive subcortical mesolimbic projections (resulting in hyperstimulation of D2 receptors and positive symptoms) and hypoactive mesocortical DA projections to the PFC (resulting in hypostimulation of D1 receptors, negative symptoms, and cognitive impairment) became the predominant hypothesis. In addition, a relationship between these two was suggested by the initial observation of Pycock and colleagues (1980) (for review, see Tzschentke, 2001). An abundant literature suggests that prefrontal DA activity exerts an inhibitory influence on subcortical DA activity (Deutch et al., 1990; Karreman and Moghaddam, 1996; Kolachana et al., 1995; Wilkinson, 1997). Based on these observations, Weinberger (Weinberger, 1987) proposed that both arms of the DA imbalance model might be related, inasmuch as a deficiency in mesocortical DA function might translate into disinhibition of mesolimbic DA activity. We will review here the pharmacological, postmortem, and imaging data that implicate DA alterations in schizophrenia.
1. Pharmacological evidence
1.1. Propsychotic pharmacological effects The psychotogenic effect of amphetamine and other DA-enhancing drugs, such as methylphenidate and L-DOPA, is a cornerstone of the classical DA hypothesis of schizophrenia. Two sets of observations are relevant to this issue.
1.1.1. Repeated exposure to high doses of psychostimulants in non-schizophrenic subjects gradually induces paranoid psychosis First mentioned in 1938 (Young and Scoville, 1938), amphetamine-induced psychosis was recognized as a possible consequence of chronic amphetamine use upon the publication of a 42-case monograph by Connell (1958).
In the early 1970s, several studies experimentally induced amphetamine psychosis in non-schizophrenic amphetamine abusers in order to better document the clinical pattern of this syndrome (Angrist and van Kammen, 1984; Gershon, 1970; Bell, 1973; Griffith et al., 1968). These experiments formally established that sustained psychostimulant exposure can produce paranoid psychosis in non-schizophrenic individuals in the context of a clear sensorium (intact sensory and perceptual abilities). Since these studies were performed before the conceptualization of the symptoms of schizophrenia into positive and negative (Crow, 1980), they did not formally assess negative symptoms and only included anecdotal reports of emotional blunting, withdrawal, or alogia, suggesting that sustained and excessive stimulation of DA systems does not predominantly induce what is now defined as the "negative" symptoms of schizophrenia.
Ellinwood (1967; 1973) described amphetamine-induced psychosis as a continuum that evolves from stimulation of interpretative mental activities to enhancement of perceptual acuity, and reversal and projection onto others (persecution), leading to paranoia and ideas of references. The "enhancement of sensitive acuity" develops into hallucinations, initially auditory, then visual and tactile. The sensorium remains clear until toxic delirium is reached. Thought disorders manifest towards the end of the continuum near the toxic stage.
Another important property of psychostimulants is their ability to induce reverse tolerance or "sensitization" (Kalivas et al., 1993; Robinson and Becker, 1986). Long-term sensitization to psychostimulants is a process whereby repeated exposure to these drugs results in an enhanced response upon subsequent exposures, a process that may be relevant to the pathophysiology of schizophrenia (Laruelle, 2000b; Lieberman et al., 1997). Subjects who abused psychostimulants and experienced stimulant-induced psychotic episodes are reported to remain vulnerable to low doses of psychostimulants (Connell, 1958; Ellinwood et al., 1973; Sato et al., 1983). In these subjects, exposure to psychostimulants at doses that do not normally produce psychotic symptoms can trigger a recurrence of these symptoms. The similarity between these patients and the patients with schizophrenia in terms of vulnerability to the psychotogenic effects of psychostimulants has led to the theory that schizophrenia might be associated with an "endogenous" sensitization process (Glenthoj and Hemmingsen, 1997; Laruelle, 2000b; Lieberman et al., 1990).
Considerable research efforts have been devoted to the identification of neuronal substrates involved in sensitization. Several studies have shown that sensitization is associated with increased stimulant-induced DA release in the axonal terminal fields (for references, see Laruelle, 2000b). One brain imaging study has suggested that, in humans, sensitization to the effects of amphetamines involves increased amphetamine-induced DA release (Boileau et al., 2003). Imaging studies reviewed below show that patients with schizophrenia display an enhanced amphetamine-induced DA release, indirectly supporting the notion of an endogenous sensitization process of the subcortical DA system in schizophrenia.
1.1.2. Low doses of psychostimulants that are not psychotogenic in healthy subjects can induce or worsen psychotic symptoms in patients with schizophrenia A number of studies, reviewed by Lieberman et al. (1987b), showed that patients with schizophrenia, as a group, display increased sensitivity to the psychotogenic effects of acute psychostimulant administration. In other terms, some, but not all, patients with schizophrenia present emergence or worsening of psychotic symptoms after acute exposure to psychostimulants at doses that do not induce psychosis in healthy subjects. The psychotic response appears to be state dependent. First, patients who responded with a psychotic reaction to a psychostimulant challenge during an acute episode failed to show such a response when they were in remission. Second, the propensity to present a psychotic reaction to a psychostimulant challenge is predictive of relapse upon antipsychotic discontinuation. Thus, the clinical response to stimulants might "reveal" an active phase of the illness that is not readily identifiable by the clinical symptomatology in the absence of psychostimulant administration.
1.2. Therapeutic antipsychotic pharmacological effects. D2 receptor antagonism or functional antagonism is the only pharmacological property shared by all antipsychotic drugs Since the discovery of the antipsychotic properties of chlorpromazine (Delay et al., 1952) in 1952, antipsychotic medications have fundamentally altered the course and the prognosis of schizophrenia by reducing severity of symptoms and preventing relapse. D2 receptor antagonism is fundamental to their beneficial effects, as evidenced by the following observations:
1.2.1. All antipsychotics bind to D2 receptors D2 receptor occupancy by antipsychotic drugs has been confirmed by a large number of imaging studies (reviewed in Talbot and Laruelle, 2002). Two studies performed with low doses of relatively selective D2 receptor antagonists (haloperidol and raclopride) suggest that a minimum of 50 percent occupancy is required to observe a rapid clinical response (Kapur et al., 2000; Nordstrom et al., 1993). Imaging studies have repeatedly confirmed the existence of a striatal D2 receptor occupancy threshold (about 80 percent) above which extrapyramidal symptoms (EPS) are likely to occur (Farde et al., 1992). Together, these data suggest the existence of a therapeutic window between 50 and 80 percent striatal D2 receptor occupancy. Within this window, the relationship between occupancy and response is unclear, presumably because of the variability in endogenous DA (Frankle et al., 2004).
1.2.2. Affinity at D2 receptors is both necessary and sufficient for the antipsychotic effect The introduction of the second-generation antipsychotic (SGA) drugs since the early 1990s has not fundamentally altered the prominence of D2 receptor antagonism in the current treatment of schizophrenia. Most SGAs potently interact with other receptors, such as the serotonin 5HT2A receptors, but the possibility of achieving an "atypical" profile with a pure D2 receptor antagonist such as amisulpride indicates that binding at serotonin receptors is not necessary.
On the other hand, imaging studies have generally reported lower occupancies of striatal D2 receptors at therapeutic doses of SGAs compared to first-generation antipsychotic drugs (FGAs). This seems to be especially true for amisulpride, clozapine, and quetiapine, which provide 50-60 percent D2 receptor occupancy at clinically effective doses (for review and references, see Abi-Dargham and Laruelle, 2005). In contrast, studies with FGAs often reported occupancies exceeding 75 percent. Thus, a parsimonious hypothesis to account for the widespread use of SGAs is that, in general, clinical results are obtained in the presence of moderate occupancies (50-75 percent), and that, for a variety of reasons, SGAs tend to maintain lower occupancies than do FGAs, thus avoiding side effects associated with very high occupancies. The alternate hypothesis is that the D2 receptor occupancy required for therapeutic effects is lower in SGAs than in FGAs. Should the alternate hypothesis be true, the mechanisms responsible for the gain in the occupancy/efficacy relationship of SGAs remain to be fully elucidated. It should be noted that this alternative hypothesis is not supported by the recent evidence from the CATIE trial showing equivalent efficacy between both classes of drugs (Stroup, McEvoy et al., 2003).
A potentially important synergistic effect of 5HT2A and D2 receptor antagonism is to increase prefrontal DA, an effect not observed with selective D2 or 5HT2A receptor antagonists administered alone (Gessa et al., 2000; Ichikawa et al., 2001; Melis et al., 1999; Pehek and Yamamoto, 1994; Youngren et al., 1999). This effect might be mediated by the stimulation of 5HT1A receptors: it is blocked by 5HT1A antagonists and is also observed following the combination of 5HT1A receptor agonism and D2 receptor antagonism (Ichikawa et al., 2001; Rollema et al., 2000). As discussed below, decreased prefrontal DA function contributes to the cognitive deficits present in patients with schizophrenia, and it is possible that an increase in prefrontal DA induced by SGAs might mediate some of the modest cognitive improvements induced by these drugs (Keefe et al., 1999). Yet, it is unclear whether or not this increase in prefrontal DA, documented as an acute response in animal studies, is sustained during the course of treatment in patients with schizophrenia.
2. Postmortem studies The discovery of the antipsychotic effect of D2 receptor blockade spurred numerous postmortem studies of dopaminergic parameters in schizophrenia.
Tissue DA and metabolites. Some studies of the tissue content of DA and its metabolites (e.g., homovanillic acid) have reported higher DA tissue levels in samples from patients with schizophrenia in subcortical regions such as caudate (Owen et al., 1978), nucleus accumbens (Mackay et al., 1982), or amygdala (Reynolds, 1983). However, consistent and reproducible abnormalities are lacking (for review, see Davis et al., 1991; Reynolds, 1989).
D3 receptors. A significant increase in D3 receptor number in the ventral striatum (VST) samples from patients with schizophrenia who were off neuroleptics at the time of death has been reported in one study (Gurevich et al., 1997). In contrast, in patients who had been treated with neuroleptics up to the time of death, D3 receptor levels did not differ significantly from those of controls (Gurevich et al., 1997). These data were interpreted as indicating that antipsychotics downregulate the D3 receptor in schizophrenic patients who otherwise have a higher density of this receptor in the VST. The D3 receptor gene expression is under the control of the neurotrophin BDNF, which is synthesized either in the VTA or the prefrontal cortex and released in the VST, where it maintains the expression of the D3 receptor (Guillin et al., 2001). Studies have shown increases (Takahashi et al., 2000) and decreases (Hashimoto et al., 2005; Weickert et al., 2003) in BDNF levels in the brains of patients with schizophrenia, and a consistent story with BDNF in schizophrenia has not yet emerged.
Tyrosine hydroxylase (TH) immunolabeling. A postmortem study in patients with schizophrenia showed decreased TH labeled axons in layers 3 and 6 of the entorhinal cortex (EC) and in layer 6 of the PFC, a finding suggesting that schizophrenia might be associated with a deficit in DA transmission in the EC and PFC (Akil et al., 2000; Akil et al., 1999). This finding was clearly unrelated to premortem neuroleptic exposure. Benes et al. (1997) observed no significant changes in TH positive varicosities in the DLPFC. In the anterior cingulate region (layer 2), these authors observed a significant shift in the distribution of TH varicosities from large neurons to small neurons.
In conclusion, postmortem measurements of indices of DA transmission generated a number of consistent observations in the striatum:
1. The binding of radioligands to D2-like receptors in the striatum of patients with schizophrenia is increased, but the magnitude of this increase varies with the type of radioligands used, and it is difficult to exclude the contribution of premortem antipsychotic exposure in this set of findings.
2. Striatal DAT and D1 receptor density are unaffected in schizophrenia. Several interesting observations such as an increase in D3 receptors in the ventral striatum and alteration in TH immunolabeling in several cortical regions do not appear to be consequences of premortem neuroleptic exposure, but these findings have yet to be independently confirmed.
3. Imaging studies
3.1. Striatal DA function The development of PET and SPECT imaging techniques in the late 1980s made possible, for the first time, the examination of DA function in vivo in patients with schizophrenia never exposed to antipsychotic drugs.
Striatal D2 and D1 receptors. Striatal D2 receptor density in schizophrenia has been extensively studied with PET and SPECT imaging. In a major meta-analysis (Weinberger and Laruelle, 2001), 17 imaging studies comparing D2 receptor parameters in patients with schizophrenia were analyzed (including a total of 245 patients and 231 control subjects) (Abi-Dargham et al., 1998; Abi-Dargham et al., 2000; Blin et al., 1989; Breier et al., 1997; Crawley et al., 1986; Hietala et al., 1994; Knable et al., 1997; Laruelle et al., 1996; Martinot et al., 1990; Martinot et al., 1991; Pilowsky et al., 1994; Wong et al., 1986a). Updated with a more recent study (Yang et al., 2004), this meta-analysis revealed a small (12 percent) but significant elevation of striatal D2 receptors in untreated patients with schizophrenia. No clinical correlates of increased D2 receptor binding parameters could be identified. Studies performed with butyrophenones (n = 7) show an effect size of 0.96 +/- 1.05, significantly larger than the effect size observed with other ligands (benzamides and lisuride, n = 11, 0.19 +/- 0.25, p = 0.02). This difference might be due to differences in vulnerability of the binding of these tracers to endogenous DA, and elevation of endogenous DA in schizophrenia (Seeman, 1988; Seeman et al., 1989). A more recent meta-analysis confirmed these observations, showing that the elevation of D2/3 receptors in schizophrenia is of small magnitude, with a Cohen's d effect size of 0.26, is most predominant in previously treated patients, and varies with the imaging methods (Howes, Kambeitz et al., 2012).
One study showed that D2 receptor levels are increased in healthy monozygotic twins compared to dizygotic twins of patients with schizophrenia, and led to the conclusion that the caudate dopamine D2 receptor upregulation may be related to genetic risk for schizophrenia (Hirvonen et al., 2005). As for D1 receptors, imaging studies have consistently failed to detect abnormalities of D1 receptor availability in the striatum of patients with schizophrenia (Abi-Dargham et al., 2002; Karlsson et al., 2002; Okubo et al., 1997).
Striatal amphetamine-induced DA release. The decrease in [11C]raclopride and [123I]IBZM binding in vivo following acute amphetamine challenge has been well validated as a measure of the change in D2 receptor stimulation by DA due to amphetamine-induced DA release (Breier et al., 1997; Laruelle et al., 1997b; Villemagne et al., 1999).
Three studies (Abi-Dargham et al., 1998; Laruelle et al., 1996; Breier et al., 1997) have shown that the amphetamine-induced decrease in [11C]raclopride or [123I]IBZM binding, an index of the magnitude of dopamine release, is significantly greater in untreated patients with schizophrenia compared to well-matched controls. The clinical significance of this dysregulation (Laruelle et al., 1999) is summarized as follows: the increase in dopamine response is related to the transient induction or worsening of positive symptoms; it is observed in both first-episode/drug-naive patients and previously treated patients; it is larger in patients experiencing an episode of illness exacerbation than in patients in remission at the time of the scan; and it does not appear to be a nonspecific effect of stress, as higher self-reports of anxiety before the experiments were not associated with larger effect of amphetamine on [123I]IBZM binding. Furthermore, nonpsychotic subjects with unipolar depression, who reported levels of anxiety similar to the schizophrenic patients at the time of the scan, showed normal amphetamine-induced displacement of [123I]IBZM (Parsey et al., 2001).
These findings have generally been interpreted as reflecting a greater magnitude of synaptic DA following amphetamine in the schizophrenic group compared to controls. Another interpretation of these observations would be that schizophrenia is associated with increased affinity of D2 receptors for DA. A study using [11C]PHNO to assess the high-affinity sites of the D2/3 receptors showed no differences between patients and controls, arguing that affinity for dopamine may not be different (Graff-Guerrero, Mizrahi et al., 2009). However, this interpretation was questioned by P. Seeman, who raised the possibility that D2/3 high-affinity states may be masked by higher levels of synaptic dopamine and may be predominantly occupied under baseline conditions, suggesting that this study would be most informative under conditions of dopamine depletion.
DAT transporters. Three imaging studies have confirmed the in-vitro observation of normal striatal DAT density in schizophrenia (Laakso et al., 2000; Laruelle et al., 2000a). In addition, no association between amphetamine-induced DA release and DAT density was found (Laruelle et al., 2000b), suggesting that the increased presynaptic output revealed by the studies reviewed above is not due to higher terminal density.
Vesicular monoamine transporter. Using the radiotracer [11C]DTBZ, Taylor and colleagues (2000) were not able to show any difference in vesicular monoamine transporter binding potential in patients with schizophrenia compared to healthy subjects.
Baseline occupancy of striatal D2 receptors by DA. In rodents, acute depletion of synaptic DA is associated with an acute increase in the in-vivo binding of [11C]raclopride or [123I]IBZM to D2 receptors (for review, see Laruelle, 2000a). The increased binding is observed in vivo but not in vitro, indicating that it is not due to receptor upregulation (Laruelle et al., 1997a), but to removal of endogenous DA and unmasking of D2 receptors previously occupied by DA. A similar acute DA depletion technique paired with D2 receptor imaging in humans using αMPT has been developed to assess the degree of occupancy of D2 receptors by DA (Laruelle et al., 1997a). In schizophrenia, there was a higher occupancy of D2 receptors by DA in patients experiencing an episode of illness exacerbation compared to healthy controls (Abi-Dargham et al., 2000). Again assuming normal affinity of D2 receptors for DA, the data are consistent with higher DA synaptic levels in patients with schizophrenia. Higher synaptic DA levels in patients with schizophrenia were predictive of good therapeutic response of these symptoms following six weeks of treatment with atypical antipsychotic medications. This observation has now been replicated with PET and [11C]raclopride in our lab. We observed that the increase at the level of the striatum observed initially is essentially accounted for by an increase in dopamine transmission at the level of the associative striatum and, in particular, the precommissural or rostral caudate (preDCA), rather than the limbic or sensorimotor striatum (Kegeles, Abi-Dargham et al., 2010). The preDCA is the area of the striatum that receives most of the corticostriatal projections from the DLPFC (Alexander et al., 1986; Ferry et al., 2000; Hoover and Strick, 1993; Joel and Weiner, 2000; Parent and Hazrati, 1995), the neocortical area most implicated in the pathophysiology of schizophrenia. The information is processed in the preDCA and sent back to the DLPFC via the globus pallidus (pars interna)/substantia nigra and ventral anterior thalamic nuclei. Thus, while subcortical DA dysregulation has historically been conceptualized as a possible consequence of DLPFC dysfunction, these findings suggest that, in addition, alterations of subcortical DA transmission in the preDCA, might, in turn, negatively impact DLPFC function. Furthermore, the rostral caudate is the only area of the striatum that receives input from both the DLPFC and the limbic cortical regions, thus integrating across functional domains (Haber, Kim et al., 2006). This integration may be biased in the presence of excess dopamine, and may lead to aberrant salience, a construct believed to relate to psychosis (Kapur, 2003).
Striatal DOPA decarboxylase activity. Eight studies have reported rates of DOPA decarboxylase, which converts the precursor L-DOPA to dopamine, in patients with schizophrenia, using [18F]DOPA or [11C]DOPA. Six of these reported increased accumulation of DOPA in the striatum of patients with schizophrenia (Dao-Castellana et al., 1997; Elkashef et al., 2000; Hietala et al., 1999; Hietala et al., 1995; Lindstrom et al., 1999; McGowan et al., 2004; Meyer-Lindenberg et al., 2002; Reith et al., 1994a); one reported no change (Dao-Castellana et al., 1997); and one study reported reduced [18F]DOPA striatal uptake (Elkashef et al., 2000). Three studies involved first-episode schizophrenia, and all three showed an increase of DOPA in the striatum (Hietala et al., 1999; Hietala et al., 1995; Lindstrom et al., 1999). Interestingly, one study observed a relationship between poor prefrontal activation during the Wisconsin Card Sorting task and elevated [18F]DOPA accumulation in the striatum, suggesting a link between alteration of DLPFC and increased striatal DA activity in schizophrenia (Meyer-Lindenberg et al., 2002). In rats, as in anesthetized pigs, increases in DOPA decarboxylase activity have been reported following acute treatment with dopamine antagonists in vitro and in vivo (Cho et al., 1997; Danielsen et al., 2001; Zhu et al., 1993). Conversely, acute treatment with the dopamine agonist apomorphine decreases 11C-DOPA influx in monkeys (Torstenson et al., 1998). Evidence for such effects in humans, however, is extremely limited. Thus, in the only comprehensive study to date, Grunder et al. recently reported a decrease in [18F]DOPA uptake in nine patients with schizophrenia following subchronic treatment with haloperidol (Grunder et al., 2003), suggesting that chronic neuroleptic administration will tend to decrease DOPA decarboxylase activity and, hence, dopamine synthesis. Interestingly, acute administration of antipsychotics increases DA neuron firing, whereas chronic administration decreases the number of spontaneously active DA neurons in the rat substantia nigra (Grace, 1991), suggesting that the different effects of antipsychotics on DOPA decarboxylase activity in the living brain could reflect such phenomena.
New studies performed with [18F]DOPA and PET in subjects at risk for schizophrenia have significantly advanced our knowledge regarding the timing and functional significance of increased presynaptic dopamine synthesis rate in schizophrenia. In an elegant series of studies, Oliver Howes and his colleagues have shown that [18F]DOPA accumulation is increased in the rostral caudate in subjects at risk, relates to attenuated positive symptoms and deficits on a verbal fluency task (Howes, Montgomery et al., 2009), predicts conversion to schizophrenia at three-year follow-up (Howes, Bose et al., 2011), and spreads out to the sensorimotor striatum on follow-up (Howes, Bose et al., 2011).
3.2. Prefrontal DA function and schizophrenia Indirect evidence supports the hypothesis that a deficit in prefrontal DA function might contribute to prefrontal impairment in schizophrenia. Preclinical studies have documented the importance of prefrontal DA function for cognition (for review, see Goldman-Rakic, 1994; Goldman-Rakic et al., 2000). This important role has been confirmed in humans by the repeated observation that carriers of the high-activity allele of catechol-O-methyltransferase (COMT), an enzyme involved in DA metabolism, display lower performance in various cognitive tasks compared to carriers of the allele that induces higher concentration of DA in PFC (for review, see Goldberg and Weinberger, 2004). Clinical studies have suggested a relationship between low cerebrospinal fluid homovanillic acid, a measure reflecting low DA activity in the prefrontal cortex, and poor performance at tasks involving white matter in schizophrenia (Kahn et al., 1994; Weinberger et al., 1988). Administration of DA agonists might have beneficial effects on the pattern of prefrontal activation measured with PET during these tasks (Daniel et al., 1991; Dolan et al., 1995). While these observations are consistent with the hypothesis of a hypodopaminergic state in the PFC of patients with schizophrenia, they do not constitute direct evidence.
Extrastriatal D1 receptors. The main parameter of extrastriatal DA transmission that is currently quantifiable using noninvasive in-vivo studies is D1 receptor availability. Three PET studies of prefrontal D1 receptor availability in patients with schizophrenia have been published. Two studies were performed with [11C]SCH 23390. The first reported decreased [11C]SCH 23390 binding potential in the PFC (Okubo et al., 1997), and the other reported no change (Karlsson et al., 2002). One study was performed with [11C]NNC 112 (Abi-Dargham et al., 2002), and reported increased [11C]NNC 112 binding potential in the DLPFC, and no change in other regions of the prefrontal cortex such as the medial prefrontal cortex (MPFC) or the orbitofrontal cortex. In patients with schizophrenia, increased [11C]NNC 112 binding in the DLPFC was predictive of poor performance on a working memory task (Abi-Dargham et al., 2002). Many potential factors, including patient heterogeneity and differences in the boundaries of the sampled regions, might account for these discrepancies. However, the severity of deficits at tasks involving working memory was reported to be associated with both decreased PFC [11C]SCH 23390 binding potential in one study (Okubo et al., 1997) and increased PFC [11C]NNC 112 binding potential in another one (Abi-Dargham et al., 2002), suggesting that both alterations might reflect a common underlying deficit. Interestingly, a recent study showed an effect of genetic loading on cortical D1 levels as well as an effect of previous antipsychotic treatment (Hirvonen et al., 2006). A downregulation of D1 receptors by D2 antagonists had been previously shown in the cortex of nonhuman primates (Lidow et al., 1997). Finally, in another study (Abi-Dargham, Xu et al., 2011) using [11C]NNC 112 in 12 drug-naive (DN), 13 drug-free (DF) patients with schizophrenia, and 40 healthy control subjects (n = 24 per comparison group) matched for age, gender, ethnicity, parental socioeconomic status, and cigarette smoking, we measured the binding potential BPP, corrected for partial volume effects due to significant differences in volumes of regions among groups. This measure was significantly higher in DN versus controls in cortical regions. No such increases were found in the DF versus controls comparison. Furthermore, in the DF group, DF interval correlated positively with cortical BPP. This study suggested that upregulation of D1 receptors in schizophrenia is related to the illness itself, and may be corrected and normalized by chronic antipsychotic treatment, in agreement with the observation in nonhuman primates (Lidow et al., 1997).
Because of the prevalent view that schizophrenia is associated with a deficit in prefrontal DA activity, the impact of acute and subchronic DA depletion on the in-vivo binding of [11C]SCH 23390 and [11C]NNC 112 is highly relevant to the interpretation of these data (Guo et al., 2001). Acute DA depletion does not affect the in-vivo binding of [11C]NNC 112, but results in decreased in-vivo binding of [3H]SCH 23390, a paradoxical response that might be related to DA depletion-induced translocation of D1 receptors from the cytoplasmic to cell surface compartment (Dumartin et al., 2000; Laruelle, 2000a; Scott et al., 2002). In contrast, chronic DA depletion is associated with increased in vivo [11C]NNC 112 binding, presumably reflecting a compensatory upregulation of D1 receptors. Interestingly, chronic DA depletion did not result in enhanced in-vivo binding of [3H]SCH 23390, possibly as a result of opposite effects of receptor upregulation and externalization on the binding of these tracers. A study performed in the same sample of six patients (Kosaka, Takahashi et al., 2010) using both radiotracers found decreases with both radiotracers compared to controls in the striatum and in the cortex. One concern is the very small sample size and the unexpected finding of a decrease in the striatum, which has not been observed previously in vivo. A replication in a larger set would be needed to confirm these results.
One additional complication in the interpretation of all these studies is the lack of selectivity of both radiotracers for D1 versus 5HT2A in the cortex in monkeys (Ekelund et al., 2006) and in humans (Slifstein, Kegeles et al., 2007). These findings flag the need for better tracers going forward with investigations of cortical dopaminergic transmission in schizophrenia.
In summary, studies of D1 receptors in the cortex have yielded inconsistent results, and have suggested an effect of antipsychotic treatment. Furthermore, because available radiotracers are not selective for cortical D1, more selective tracers are needed for future investigations and more direct examination of dopamine tone in the cortex.
Extrastriatal D2 receptors. The recent availability of high-affinity D2 radiotracers allowed the study of D2 receptors in low-density regions such as the substantia nigra, thalamus, and temporal cortex in patients with schizophrenia compared to controls. A first study found decreases in temporal cortex in both hemispheres in a very small group of patients compared to controls (seven patients and seven controls) (Tuppurainen et al., 2003). A similarly small study found low thalamic binding; however, this was later confirmed in a larger sample of drug-naive patients (Talvik et al., 2006; Talvik et al., 2003). Suhara et al. (2002) found decreased DA D2 receptors in the anterior cingulate cortex and thalamic subregions in patients with schizophrenia, while Glenthoj found a significant correlation between frontal D2 receptor and positive symptoms in male schizophrenic patients (Glenthoj et al., 2006). A more recent and larger study (21 patients and 22 controls) did not confirm these observations but found binding potential values larger in postcommissural caudate, thalamus, and lower in uncus in patients compared to controls. Loss of D2/3 receptors with age was also found in striatal and extrastriatal regions, and was greater in neocortex (Kegeles, Slifstein et al., 2010). Overall, these results show that a consistent pattern of abnormalities in D2/3 in extrastriatal regions in schizophrenia has not yet emerged.
Occupancy of extrastriatal D2/3 receptors by antipsychotics has also been examined with various results which are beyond the scope of this review; however, two studies (Agid et al., 2006; Kegeles, Slifstein et al., 2008) found no relationship to treatment response of psychotic symptoms. These findings suggest that extrastriatal D2 may not play an important role in the therapeutics of schizophrenia. Overall, this is an emerging field, and more research is needed to understand the role of extrastriatal D2 transmission and dopamine in schizophrenia.
4. Convergence of DA hypothesis with other major hypotheses in schizophrenia An additional factor supporting the evidence for DA dysregulation in schizophrenia is its plausibility in the overall context of other transmitter systems that may be altered in schizophrenia, in particular, with the NMDA hypofunction hypothesis. More specifically, imaging studies have shown that NMDA hypofunction can lead to DA alterations similar to those observed in schizophrenia, namely, subcortical DA excess and cortical D1 alterations (Abi-Dargham and Moore, 2003; Kegeles et al., 2000; Laruelle et al., 2003; Narendran et al., 2005). More recently, NMDA antagonists have been shown to engender some of the alterations in GABA neurons in the DLPFC that have been described in schizophrenia (Lewis and Gonzalez-Burgos, 2006). Such convergence suggests that the main neurochemical dysregulations described in schizophrenia are not mutually exclusive. Glutamatergic and GABAergic dysfunction in schizophrenia could be linked, and could lead to or be associated with an inefficient control of cortical input onto subcortical striatal dopamine, as well as inefficient cortico-cortical connectivity and function. DA dysregulation may be the endpoint of a cascade of upstream events, an endpoint that is most directly associated with symptoms of the illness and their treatment. This is further supported by the studies in the rodent MAM model of schizophrenia, showing that hippocampal glutamatergic drive through the ventral pallidum could lead to a dysregulation of firing patterns of midbrain dopamine neurons (Grace, 2012). More recently, data have emerged suggesting that dopamine dysfunction, if present during development, may induce pathology of the remaining circuitry. D2 overexpression in transgenic mice leads to abnormal, irreversible cortical dysfunctionm as well as reversible social and motivational deficits (Kellendonk, Simpson et al., 2006; Drew, Simpson et al., 2007; Bach, Simpson et al., 2008; Ward, Kellendonk et al., 2009). This important finding has suggested a radical shift in thinking about dopamine dysfunction as not just an endpoint, but a participating early event that contributes to the overall pathology.
Figure 1: Striatal dopaminergic synapse in schizophrenia. Evidence for excess dopamine transmission derives from pre- and postsynaptic studies. Excess dopamine transmission may impair glutamatergic NMDA transmission by a D2-mediated decrease of presynaptic glutamate release and an imbalance of D1/D2 opposing effects onto NMDA receptors. Image courtesy of Anissa Abi-Dargham (Click image for larger view.)
Conclusion Imaging studies have conclusively shown that schizophrenia is associated with hyperactivity of subcortical transmission at D2 receptors (see Figure 1). These results are consistent with the known mode of action of current antipsychotic treatment (D2 receptor blockade) and with the psychotogenic effects of sustained stimulation of DA function by psychostimulants. Most of the evidence points to a presynaptic dopamine dysregulation: more synthesis, and more release, while postsynaptic dysregulation remains unclear. In addition, the relationship between these two is also unclear: which is primary and which is secondary? One recent study has suggested a functional supersensitivity of postsynaptic D2 receptors, in the absence of increased expression, in a population of drug-abusing patients with schizophrenia (Thompson, Urban et al., 2012). Thus pre- and postsynaptic factors may both be at play. In addition, the data suggest that the DA hyperactivity of subcortical systems is episodic in nature, and accounts directly for aspects of positive symptomatology. On the other hand, imaging and postmortem studies have suggested that hypodopaminergia in the DLPFC contributes to the pathophysiology of cognitive symptoms endured by patients with schizophrenia, although more evidence for and better characterization of these relationships is needed.
Remaining unanswered questions relate to extrastriatal dopamine transmission, the cellular characterization of excess subcortical dopamine transmission, and definitive studies of cortical dopaminergic transmission. These answers will come from more selective tracers and more direct ways of measuring cortical dopaminergic transmission, including agonists for the D1 receptor and tracers sensitive to acute fluctuations of dopamine tone in the cortex.