The classical DA hypothesis, formulated over 40 years ago, proposed that schizophrenia is associated with hyperactivity of dopaminergic neurotransmission (1,2). This hypothesis was essentially based on the observation that all effective antipsychotic drugs provided at least some degree of D₂ receptor blockade (3,4), an observation that is still true today. As D₂ receptor blockade is most effective against positive symptoms (delusions and hallucinations), the DA hyperactivity model appeared to be most relevant to the pathophysiology of these symptoms. This idea was further supported by the fact that sustained exposure to DA agonists such as amphetamine can induce a psychotic state characterized by some features of schizophrenic positive symptomatology (emergence of paranoid delusions and hallucinations in the context of a clear sensorium) (5,6). These pharmacological effects suggest, but do not establish, a dysregulation of DA systems in schizophrenia.

On the other hand, negative and cognitive symptoms are generally resistant to treatment by antipsychotic drugs. Functional brain imaging studies have suggested that these symptoms are associated with prefrontal cortex (PFC) dysfunction (7). Studies in nonhuman primates have demonstrated that deficits in DA transmission in the PFC produce cognitive impairments reminiscent of those observed in schizophrenic patients (8), suggesting that a deficit in DA transmission in the PFC may be implicated in the cognitive impairments associated with schizophrenia (9,10).

Thus, a contemporary view of the role of DA in schizophrenia is that subcortical mesolimbic DA projections may be hyperactive (resulting in positive symptoms) and that the mesocortical DA projections to the PFC may be hypoactive (resulting in negative symptoms and cognitive impairment). Furthermore, these two abnormalities may be related, as the cortical DA system generally exerts an inhibitory action on subcortical DA systems (11,12). The advent in the early 1980s of techniques based on PET and SPECT to measure indices of DA activity in the living human brain opened the possibility of direct investigation of these hypotheses.

Abnormalities of 5-HT transporters (SERT), 5-HT_2A receptors, and, more consistently, 5-HT_1A receptors have been described in postmortem studies in schizophrenia (see references in Abi-Dargham and Krystal [88]). However, these postmortem results were generally not confirmed by molecular imaging studies.

The concentration of SERT in the midbrain measured by [¹²³I]-β-CIT is unaltered in patients with schizophrenia (62). This observation was confirmed and extended to thalamus and striatum using the more selective radiotracer [¹¹C]-DASB (89). Yet, SERT imaging agents currently available are not valid to study SERT in areas of relatively low SERT density, such as the PFC, where SERT has been reported to be reduced in three of four postmortem studies in schizophrenia (88).

Decreased 5-HT_2A receptors have been reported in the PFC in four of eight postmortem studies (see references in Abi-Dargham and Krystal [88]). Three PET studies in drug-naive or drug-free patients with schizophrenia reported normal cortical 5-HT_2A receptor binding (90,91,92), while one study reported a significant decrease in PFC 5-HT_2A binding in a small group (n = 6) of drug-naive schizophrenic patients (93).

The most consistent abnormality of 5-HT parameters reported in postmortem studies in schizophrenia is an increase in the density of 5-HT_1A receptors in the PFC, reported in seven of eight studies (88). Several groups evaluated the binding of this receptor in vivo with PET and [¹¹C]-WAY100907 in schizophrenia and reported inconsistent results. One study reported increased 5-HT_1A binding in the temporal lobe of patients with schizophrenia (94), another study reports decreased binding in the amygdala (95), and a third study failed to detect any changes in regional 5-HT_1A availability (96). These inconsistencies indicate that the alterations of 5-HT_1A receptors demonstrated in postmortem studies in schizophrenia could not be reliably detected with PET and might not be related to the pathophysiology of the illness.

A robust body of findings suggests a deficiency of GABAergic function in the PFC in schizophrenia (for reviews, see Lewis [97] and Benes [98]). In vivo evaluation of GABAergic systems in schizophrenia has so far been limited to evaluation of benzodiazepine receptor densities with SPECT and [¹²³I]-iomazenil, and three of three studies comparing patients with schizophrenia and controls reported no significant regional differences (99,100,101). Although significant correlations between symptom clusters and regional benzodiazepine densities have been observed in some studies, these relationships have not been replicated. Thus, taken together, these studies are consistent with an absence of marked abnormalities of benzodiazepine receptor concentration in the cortex and patients with schizophrenia. Alterations of GABAergic systems in schizophrenia might not involve benzodiazepine receptors (102), or be restricted to certain cortical layers or classes of GABAergic cells that are beyond the resolution of current radionuclide-based imaging techniques.

Perhaps the most widespread use of neuroreceptor imaging in schizophrenia over the past two decades has been the assessment of
receptor occupancy achieved by typical and atypical antipsychotic drugs, a topic that has been reviewed elsewhere (103,104). The main focus has been on D₂ receptor occupancy, but 5-HT_2A and D₁ receptors have also been studied. Studies have repeatedly confirmed the existence of a threshold of occupancy of striatal D₂ receptors (about 80%) above which extrapyramidal side effects (EPS) are likely to occur (105). In general, studies have failed to observe a relationship between the degree of D₂ receptor occupancy and clinical response (106,107). However, most studies were performed at doses achieving more than 50% occupancy. Two studies performed with low doses of relatively selective D₂ receptor antagonists (haloperidol and raclopride) suggested that 50% to 60% occupancy was required to observe a rapid clinical response (108,109). Clozapine, at clinically therapeutic doses, has been found to achieve only 40% to 60% D₂ receptor occupancy (105,107,110), which, in conjunction with its anticholinergic properties, may account for its low risk for EPS. Occupancy of 5-HT_2A receptors by “5-HT_2A/D₂ balanced antagonists” such as risperidone does not confer protection against EPS, since the threshold of D₂ receptor occupancy associated with EPS is not markedly different between these drugs and drugs devoid of 5-HT_2A antagonism (111,112,113,114). Studies with quetiapine suggest that, at least with this agent, transient high occupancy of D₂ receptors might be sufficient to elicit a clinical response (115,116).

An interesting question relates to putative differences in the degree of occupancy achieved by antipsychotic drugs in striatal and extrastriatal areas. Pilowsky et al. (117) initially reported lower occupancy of striatal D₂ receptors compared to temporal cortex D₂ receptors in seven patients treated with the atypical antipsychotic drug clozapine, using the high affinity SPECT ligand [¹²³I]-epidipride. In contrast, typical antipsychotics were reported to achieve similar occupancy in striatal and extrastriatal areas, as measured with [¹¹C]-FLB457 (118) or [¹²³I]-epidipride (119). It should be noted, however, that these very high affinity ligands do not allow accurate determination of D₂ receptor availability in the striatum (120). Conversely, [¹⁸F]-fallypride enables accurate determination of D₂ receptor availability in both striatal and extrastriatal areas (121). Occupancy studies using [¹⁸F]-fallypride confirmed that clozapine and quetiapine, but not olanzepine or haloperidol, achieved higher D₂ receptor occupancy in temporal compared to striatal regions (122,123,124). Occupancy studies performed with [⁷⁶Br]-FLB457 also reported higher occupancies in cortex compared to striatum for a number of antipsychotic drugs, including typical antipsychotic drugs (125). Conversely, a study combining [¹¹C]-FLB457 imaging for extrastriatal D₂ receptor receptors and [¹¹C]-raclopride imaging for striatal D₂ receptors suggested similar occupancy of D₂ receptors in both regions for both typical and atypical antipsychotic drugs (126). Thus, at this point in time, there is a strong suggestion that many, if not most, antipsychotic drugs achieve higher occupancies in temporal cortex compared to striatum, although this phenomenon has not been universally observed. Factors underlying this difference remain to be elucidated.

Finally, it is important to point out that the most robust preclinical evidence relative to the site of therapeutic effect of antipsychotic drugs points toward the ventral striatum (65,127), while the imaging studies reviewed above contrasted striatal versus mesotemporal D₂ receptor binding. Furthermore, D₂ receptor occupancy levels in striatum have been shown to be more predictive of therapeutic response than in temporal cortex (128). Thus, the observation that, in a restricted dose range, D₂ receptor occupancy by antipsychotic drugs is higher in temporal cortex than in striatum does not necessarily imply that the temporal cortex is the therapeutic site of actions of these agents.

Another unresolved question is the discrepancy in the value of D₂ receptor occupancy obtained with [¹¹C]-raclopride versus [¹¹C]-NMSP. The haloperidol plasma concentration associated with 50% inhibition of [¹¹C]-NMSP binding (3 to 5 mg/mL) (129) is ten times higher than that associated with 50% inhibition of [¹¹C]-raclopride binding (0.32 ng/mL) (130). Quetiapine, at a dose of 750 mg, decreased [¹¹C]-raclopride specific binding by 51%, but failed to affect [¹¹C]-NMSP specific binding (131). These observations contribute to the debate regarding differences between benzamide and butyrophenone binding to D₂ receptors.