René S. Kahn and Kenneth L. Davis

The last 10 years have witnessed far-reaching changes in the understanding of dopamine (DA) and its possible role in the pathogenesis of schizophrenia. Although the original hypothesis that has so stimulated the study of DA in schizophrenia has proven to be untenable, a role for DA in schizophrenia appears even more likely than it did 10 years ago.

The original DA hypothesis of schizophrenia postulated that schizophrenia was characterized by increased DA function (1). This hypothesis was based primarily on the correlation between the ability of neuroleptics to displace DA antagonists in vitro and their clinical potency (2). However, in the last decade it has become increasingly evident that this hypothesis was in need of revision. One of the principal reasons driving the demand for reconceptualizing the original DA hypothesis was the appreciation that some core symptoms of schizophrenia are negative symptoms and cognitive deficits. These symptoms, though amenable to some extent to neuroleptic treatment, are far less responsive to treatment with DA antagonists than are psychotic, positive symptoms. This, in turn, suggests that some of the core symptoms of schizophrenia may be unrelated to increased DA activity. Additionally, knowledge about the DA system has expanded considerably over the last decade and, combined with the above questions, has stimulated further studies into DA and schizophrenia, leading to an increased and refined understanding of its role in that illness. (See Mesocorticolimbic Dopaminergic Neurons: Functional and Regulatory Roles and Dopamine Receptors: Clinical Correlates , for related discussion.)

The discovery of multiple DA receptors serves as a good illustration of the progress made over the last decade in understanding the DA system. Ten years ago, D2 and D1 were the only DA receptors known, but now D3, D4, and D5 receptors have also been identified. D1 receptors are coupled to adenylate cyclase, have a low binding affinity to [3H]spiperone, and are found predominantly in the cortex of humans (3). D5 receptors resemble D1 (4), but they have a higher affinity for DA than do D1 receptors (5). D2 receptors are negatively coupled to adenylate cyclase, display high binding affinity to [3H]spiperone (6), and are most prominent in the striatal and limbic structures in humans; and their presence, if at all, in the human cortex, is limited (3). The D2 receptor has also been cloned and two D2 isoforms, labeled D2a and D2b, have been identified (7). The D3 receptor has been cloned and is primarily present in the nucleus accumbens with very low levels in the caudate and putamen (8). It also exists in two isoforms (9). [In one study, no linkage was found in four Icelandic pedigrees between schizophrenia and the D3 receptor gene (10).] It bears no resemblance to either the D1 or the D2 system (11). Finally, D4 receptors have been identified displaying a higher affinity for the atypical neuroleptic, clozapine (12). The identification of these various DA receptors has important implications. The high affinity of clozapine to the DA4 receptor, for instance, raises the issue of whether atypical neuroleptics are effective by blocking D4 receptors more effectively than they block D2 receptors. Indeed, it has been argued that blockade of D4 receptors is related to the efficacy of neuroleptics, whereas blockade of D2 receptors is related to their extrapyramidal side-effect profile (13). The anatomical localization of D3 receptors to limbic regions has intriguing possibilities for the development of antipsychotic compounds.

Another discovery of importance in understanding the role of dopaminergic transmission in schizophrenia has been the elucidation of an interaction between cortical and striatal DA systems: An inhibitory regulation of cortical DA systems on striatal DA neurons has been found. When DA neurons are lesioned in the prefrontal cortex (PFC) in rats, increased levels of DA and its metabolites as well as increased D2 receptor binding sites and D2 receptor responsivity are found in striatum (14, 15, 16, 17, 18). Conversely, injection of the DA agonist, apomorphine, in the PFC of rats reduced the DA metabolites, homovallic acid (HVA) and dihydroxyphenylacetic acid (DOPAC), by about 20% in the striatum (19).

In a modification to the model proposed by Pycock et al. (14), Deutch (20) proposed that the effect of DA depletion in the PFC on striatal DA activity is particularly revealed after the animal has been stressed. Specifically, when animals were stressed, larger increases in striatal DA activity were found in animals whose mesocortical DA neurons had been lesioned than in animals with intact PFC DA systems (20). This suggests that the sensitivity of striatal (mesolimbic) DA neurons to physiological (i.e., stress) challenge is enhanced when DA function in the PFC is decreased. These studies therefore indicate that decreasing prefrontal cortical DA activity increases striatal DA turnover, D2 receptor sensitivity, and D2 receptor function, whereas increased DA function in the PFC decreases striatal DA activity particularly in response to stress. Thus, it appears that DA systems in the PFC display an inhibitory modulatory effect on subcortical, striatal DA systems. Decreased activity in the PFC may render the subject particularly sensitive to stress-induced increases in subcortical DA activity.

Others suggest an additional link between (diminished) DA activity in the PFC and stress-sensitive changes in subcortical DA activity. It has been hypothesized that the release of DA in subcortical sites is under the control of two independent mechanisms: phasic and tonic DA release (21). Phasic DA release appears associated with behavioral stimuli (stress, for instance), whereas the degree of tonic DA activity determines the magnitude of the phasic response to environmental stimuli. Decreased prefrontal DA activity in schizophrenic patients is hypothesized to reduce tonic DA release, leading to compensatory increases in (for instance) receptor sensitivity, resulting in exaggerated responses to phasic release of DA in response to stress (21).

In summary, findings suggesting a regulatory effect for PFC DA systems on subcortical DA function have changed the focus from solely subcortical DA systems to the interaction between the subcortical and cortical DA systems as one of the primary regions of interest in schizophrenia. These findings have far-reaching and important implications for the role of DA in schizophrenia: Not only do they suggest the usefulness of increasing DA activity (in the PFC), but even more importantly, increasing DA function in the PFC may be used as an intervention to prevent (stress-induced) increases in subcortical DA activity (i.e., psychosis) and thus may be considered as maintenance treatment in schizophrenia (see also Schizophrenia and Glutamate).

Understanding of the DA system in humans has also been enhanced by the development of peripheral measures that reflect central DA function. Measurement of the DA metabolite, homovanillic acid (HVA), in plasma has proven to be such a tool, appearing particularly useful when DA function is putatively manipulated, as during administration of neuroleptics. The HVA found in plasma is produced primarily by brain DA neurons and peripheral noradrenergic (NA) neurons. Secondary sources of HVA are peripheral DA and brain NA neurons. Animal and human studies suggest that brain DA turnover can be reflected by plasma HVA (pHVA) concentrations (22, 23). Although the precise proportion of pHVA deriving from brain HVA has not been fully elucidated (24), measurement of this DA metabolite in plasma of schizophrenic patients appears to be a valid method to investigate DA in this disorder provided certain conditions are met.

For example, highly consistent findings have been produced when HVA is measured in plasma prior to and during neuroleptic treatment in schizophrenic patients. All studies found chronic neuroleptic treatment to lower pHVA, and all found this decrement to relate to treatment outcome (Table 1). Moreover, six out of eight studies found higher pretreatment pHVA concentrations to be related to good neuroleptic treatment response (Table 2). When HVA is measured during the steady state, however, less consistent results have been generated: pHVA differentiates patients from controls only in some studies, and results of studies trying to link pHVA concentrations to specific schizophrenic symptoms, or even to severity of illness, have been inconsistent (Table 3).

Thus, while the results when HVA is used as an indication of baseline DA function are conflicting, when HVA is used as an index of change in DA function the results are quite consistent. This may be the result of the relatively large changes in HVA production when DA activity is manipulated as compared to the steady state. For instance, when striatal HVA is reduced (after administration of apomorphine) by about a third, pHVA decreases by about 25% in rodents (22); when HVA increases fourfold (after administration of haloperidol) in striatum, it almost doubles in plasma of rats (25). A single administration of haloperidol roughly doubles pHVA concentrations in human subjects (26). Thus, the changes induced by perturbation of DA function lead to large changes in both central and peripheral HVA concentrations. Possibly, when DA function is manipulated, the changes that occur are profound enough to be detected in metabolite concentrations in plasma. In contrast, when steady-state DA function is assessed, DA metabolite concentrations may be much more prone to multiple confounding factors (27).

The distribution and development of the DA system in primates have recently been elucidated. This may indirectly help to understand the role of DA in schizophrenia (see ref. 28). In contrast to those in rodents, DA fibers in adult monkeys are widespread in every cortical region, although they are most pronounced in layers I and III (28). Neonatal monkeys, however, display a more uneven distribution of DA fibers—that is, fewer DA axons in layers III–VI. Axonal growth in these layers takes place during the first 2–3 months of life, so that in early adult life the distribution of DA fibers resembles that in adult monkeys. Interestingly, in monkeys that have been socially isolated since birth, the neuronal growth in layers III–VI appears to have been stunted (29), suggesting that early age is critical for the normal development of DA neuronal networks.

Multiple studies using computerized tomography (CT) have found evidence of enlarged ventricles in schizophrenic patients as compared to healthy controls (for a review see ref. 30). These studies provide only nonspecific evidence of diminished brain tissue. With the advent of magnetic resonance imaging (MRI), evidence of more localized abnormalities have materialized. Decreased volume of the frontal and temporal cortex have been found as well as decreased volume of the hippocampus, although the findings have not been consistent (see ref. 30). Differences in results may depend on imaging techniques (resolution of MRI scanners) including the slice thickness of the images. The potential significance of these findings can best be viewed in relation to functional imaging studies.

Decreased function of the frontal lobes has been repeatedly demonstrated with both measurements of cerebral blood flow as measured by single photon emission computerized tomography (SPECT) and positron emission tomography (PET) (for a review see ref. 31). In a cognitive task linked to frontal lobe function, the Wisconsin Card Sort Task (WCST), schizophrenic patients failed to show an increase in cerebral blood flow to the same degree as normal controls (32). Facility at this task has been associated with the dorsolateral prefrontal lobe. Similarly, schizophrenic patients showed decreased blood flow and activation of the left mesial frontal cortex on performing the "Tower of London" task (31). This lack of activation and decreased blood flow was similar in drug-naive and medicated patients, but occurred only in patients with high negative symptoms scores (31). Indeed, negative symptomatology has been associated with prefrontal hypometabolism (33). Furthermore, decreased frontal blood flow is not related to medication effects (34). Hence, frontal hypofunction seems a key feature of schizophrenia, particularly to patients with prominent negative or deficit symptoms. However, a critical question is whether the findings of decreased volume and function of the prefrontal cortex in schizophrenia have any relationship to the role of DA in schizophrenia. Obviously, atrophy of the frontal cortex could affect various neurotransmitter systems. Similarly, decreased function of the PFC may be the result of hypofunction of multiple neurotransmitters. However, several lines of evidence suggest that decreased function of the PFC may be related to decreased activity of mesocortical DA neurons.

Indirect evidence has suggested that cortical hypofunctionality is associated with diminished cortical DA activity. For example, a strong positive correlation was found between the ability to activate the PFC (on the Wisconsin Card Sort Test) and CSF HVA concentrations (32). Indeed, cognitive deficits attributed to activity of the frontal cortex, such as WCST performance, were associated with lowered CSF HVA concentrations, suggesting a relationship between decreased DA function and impaired frontally mediated cognitive function (35). Moreover, blood flow in the prefrontal cortex increases in schizophrenic patients after administration of the DA agonists amphetamine (36) and apomorphine (37), suggesting that the hypofrontality found in schizophrenic patients can be redressed by increasing DA activity in the PFC. The increase in prefrontal blood flow after amphetamine also correlated significantly with improved performance on the WCST (36), indicating that increasing DA activity improves a cognitive deficit linked to diminished prefrontal cortical activity.

If negative symptoms were related to decreased function of the mesocortical DA system, one would expect treatment with DA agonists to improve negative symptoms of schizophrenia. Various studies have attempted to improve schizophrenic symptoms by increasing DA activity. Most have failed to find clinically meaningful effects (see ref. 38). However, recently the DA reuptake inhibitor, mazindole (2 mg/day), improved negative symptoms as compared to placebo (39). In that study, mazindole or placebo were added to neuroleptic treatment after patients had been stabilized on neuroleptic for 4 weeks. However, well-controlled large studies are needed to explore the efficacy of increasing DA activity in the negative symptoms of schizophrenia, although the data reviewed here certainly encourage such an approach.

In summary, evidence suggests that the negative symptoms and some of the cognitive deficits of schizophrenia may be related to decreased PFC function which, in turn, based on indirect evidence, may be associated with decreased mesocortical DA activity.

Increased DA activity of the subcortical, striatal DA neurons has been the basis of the original DA hypothesis. Although unlikely to be the only, or even the main, dopaminergic abnormality in schizophrenia, some evidence does suggest that increased striatal or mesolimbic DA activity is related to some schizophrenic symptoms. Increased activity in those areas is suggested by anatomical and functional imaging studies and more indirectly by measurement of pHVA.

Only very recently have imaging studies been able to focus on volumetric measurement of the subcortical structures with the availability of high-resolution MRI scanners with section thickness of 3 mm. Increased volume of the left caudate nucleus has been described in a study comparing 44 schizophrenic patients with 29 healthy controls (40). This effect may be medication-related, because it was not found in neuroleptic-naive patients but, instead, appeared only after patients had been receiving neuroleptic treatment.

In vivo measurement of D2 receptor affinity in humans, using PET, has provided conflicting results. An increase in D2 receptor numbers in striatum of 10 neuroleptic-naive schizophrenic patients has been reported, using [11C]methylspiperone as a D2 ligand (41). In contrast, D2 receptor density was not different in 15 (42) and 18 (43) similarly drug-naive schizophrenic patients as compared to normal controls when studied with [11C]raclopride. Similarly, when [76Br]bromospiperone was used to compare D2 receptor density in 12 schizophrenic patients (who were either drug-naive or at least 1 year drug-free) with 12 controls, no group differences in D2 receptor density were found (44). Interestingly, the more acutely ill patients had higher D2 receptor density in the striatum than did the more chronically ill patients and higher than the control subjects, suggesting that DA2 receptor density may be state-dependent. Part of these conflicting data may be due to the ligand used. For instance, methylspiperone, but not raclopride, binds potently to 5HT2 receptors. Moreover, the methods with which PET data were analyzed varied across studies. In addition, as the study using [76Br]bromospiperone suggests, differences in patient population may partly explain the different D2 receptor densities found in schizophrenic patients. Finally, the ligands used occupied different populations of DA receptors, and they may therefore point toward an increase in number in only the receptors occupied by methylspiperone but not raclopride.

The relationship between pHVA concentrations and neuroleptic treatment response suggests an association between the effects of neuroleptics on DA activity and treatment outcome (Table 1). Neuroleptics initially increase (45) and subsequently decrease pHVA concentrations (49!popup(ch113, 50, 51, 52, 53). Both the initial increase and the subsequent decrease by neuroleptics are associated with clinical response. Interestingly, increased pretreatment pHVA concentrations (49, 50, 52, 53), 54 but also see refs. 45) and 55) are predictive of good treatment response to neuroleptics. Conversely, clinical decompensation after discontinuation of neuroleptic is associated with increases in pHVA levels (56, 57, 58). Thus, pHVA studies suggest that neuroleptics initially increase and subsequently decrease DA activity. This is consistent with studies in rodents where, in the nigrostriatal (A9) and mesolimbic (A10) DA systems, a single dose of a neuroleptic increases DA neuron firing (59) while chronic (3–4 weeks) neuroleptic administration decreases DA neuron firing in A9 and A10 below pretreatment levels. Interestingly, atypical neuroleptics—that is, antipsychotics that do not induce extrapyramidal side effects, such as, for instance, clozapine—are anatomically more selective in their effect on DA neuronal firing than typical neuroleptics in that they decrease DA activity in A-10 only (59). On the basis of these data, it has been proposed that decreased activity in A9 is responsible for induction of extrapyramidal side effects, while in A10 it leads to the antipsychotic effects of neuroleptics (59).

The effects of clozapine on pHVA are less clear-cut than those of typical neuroleptics. Clozapine treatment decreased pHVA concentrations with larger decrements associated with good treatment response (60). However, in another study the effect of clozapine on pHVA was less robust, although treatment responders tended to show a decrement in pHVA while nonresponders did not (57). A complicating factor in examining clozapine's effect on pHVA concentrations is the fact that, unlike typical neuroleptics such as haloperidol (Davidson, unpublished results) and fluphenazine (60), it increases plasma norepinephrine (NE) concentrations. Because about one-third of NE is metabolized into HVA in the peripheral nervous system (24), the clozapine-induced increase in plasma NE (pNE), may partially overshadow a possible lowering effect of clozapine on pHVA. Consequently, measurement of pHVA as a reflection of clozapine's effect on (central) DA turnover may be compromised by its concomitant opposite effect on NE metabolism. Therefore, a relationship between symptom improvement on clozapine and its effects on pHVA could be obscured by this potent effect of clozapine on pNE.

Studies examining a relationship between steady-state pHVA and schizophrenic symptoms have been less consistent than studies examining the effect of neuroleptic treatment on pHVA (Table 3). Four studies have found a positive correlation between pHVA levels and clinical severity (27, 46, 61, 62), while three studies did not (47, 53, 63). The most likely explanation for the different results across studies is the number of pHVA samples taken as a basis for the correlational studies. The studies employing more than one sampling of pHVA found significant positive correlations between pHVA and severity of symptoms, whereas studies using one single measurement of pHVA did not. The studies producing significant correlations between pHVA and severity of schizophrenic symptoms averaged two (27), three (46), four (61), or thirteen (62) pHVA samples, whereas the studies that produced negative findings assessed pHVA only once (47, 53, 63). Repeated pHVA measurements in the same individual therefore appears to increase the signal/noise ratio for pHVA by reducing the intra-individual variance in pHVA concentrations (see also Schizophrenia and Glutamate).

Although HVA and DA concentrations in postmortem brains of schizophrenic patients consistently show patient–control differences, the localization of these differences are not consistent. Increased HVA concentrations in schizophrenic patients have been found in caudate and nucleus accumbens (64) and cortex as compared to normal brains. The difference in caudate was attributable to prior medication history, while the finding in accumbens only applied in the medication-free patients. Similarly, although DA was found to be increased in nucleus accumbens in schizophrenic patients compared to controls (65), another study found increased DA in the caudate of schizophrenic patients, but not in nucleus accumbens (66). Finally, increased DA has been found in the amygdala of schizophrenic patients, mostly in the left hemisphere (67). These inconsistencies may be due to differences in medication status of the patients studied, varying analytical and statistical methods used, and, finally, genuine variability in the location of DA abnormalities in schizophrenia.

Receptor affinity studies have found increases in D2, but not D1, receptors in the striatum of schizophrenics (68, 69, 70, 71, 72; see Table 4). Although these results could have been a result of prior medication use, most studies show that those patients who were neuroleptic-free for at least 1 year prior to study or were drug-naive still have increased striatal D2 receptors. Moreover, a bimodal distribution of D2 receptor numbers in brains of schizophrenic patients indicates that neuroleptics do not uniformly increase D2 receptor numbers (68). That neuroleptic treatment alone cannot explain the increased D2 receptor affinity in schizophrenia is also suggested by postmortem studies in other patient groups treated with neuroleptics: Patients with Alzheimer's disease and Huntington's disease who had been treated with neuroleptics prior to death showed increases in striatal DA receptors of only 25% as compared to controls, whereas schizophrenics had greater than 100% increases (68). Thus, the available data indicate that D2 (but not D1) receptor density is increased in schizophrenia, and that this finding cannot be accounted for by medication history alone.

D4 receptors have also been reported to be elevated in postmortem schizophrenic brain in subcortical regions (73). Because a selective D4 ligand was not used in this study, subtraction of two different ligands was used to infer the D4 receptor number. The differences found between schizophrenic and controls was quite robust, but awaits confirmation (see all Cytochrome P450 Enzymes and Psychopharmacology).

Increased striatal DA activity has not been demonstrated directly in schizophrenia. Postmortem and in vivo receptor binding studies provide some, but not consistent, evidence that striatal DA function is increased, while studies examining pHVA prior to and after neuroleptic treatment only provide an indirect suggestion that modulatory DA activity in schizophrenia can alter symptomatology. pHVA appears to be a useful indicator of central DA activity, and studies examining pHVA justify the following conclusions: (a) Increased DA turnover is related to good response to neuroleptic treatment, and (b) neuroleptic treatment decreases DA turnover, and this effect is related to treatment response.

An increasing number of MRI studies indicate abnormalities in the temporal lobes (more pronounced on the left side) in schizophrenic patients. Decreases of 10% in total temporal lobe volume (74) or 20% of temporal lobe gray matter have been found, present at first episode (75). Interestingly, the abnormalities of the temporal cortex in schizophrenia appears to be associated with specific positive symptoms, such as auditory hallucinations (76) and thought disorder (77). Additional, indirect evidence that the temporal lobes are associated with (positive) schizophrenic symptoms is the discovery that stimulation of the superior temporal gyrus (left and right) elicits auditory experiences (78) and that psychotic symptoms in temporal lobe epilepsy patients appear related to anatomical abnormalities in the medial temporal lobe (established at postmortem examination) (79). Although speculative, since increased D2 receptor binding has been found in the temporal cortex of brains of schizophrenic patients (80), the abnormalities found in the temporal cortices of schizophrenic patients and its association with some of the schizophrenic symptoms may be related to dysfunctional DA systems in those areas. These findings are particularly provocative in light of the fact that hippocampal lesions to rat pups produces subcortical hyperdopaminergia and an enhanced stress response at adulthood (81).

The negative or deficit symptoms—that is, decreased social interaction, apathy and avolition—are considered to be core symptoms of schizophrenia. Indeed, Bleuler proposed that deficit state symptoms represent pathognomonic signs of schizophrenia and are at the root of the poor social and work function that characterize people with chronic schizophrenia. Primate studies suggest that insufficient frontal cortical functioning is responsible for poor social skills: Monkeys with frontal lobe ablations not only have an inability to suppress irrelevant stimuli, poor concentration, and impaired delayed response testing, but also exhibit the poor social function that is reminiscent of deficit state symptoms which characterize schizophrenia (82).

Only a handful of studies have directly attempted to link decreased activity of the PFC in schizophrenia with negative symptoms. Decreased activation of the PFC as measured by SPECT was only found in schizophrenic patients with predominantly negative symptoms (31). Furthermore, negative symptoms were associated with decreased frontal blood flow as assessed by PET (33). Although preliminary, these data do suggest a link between negative symptoms and impaired cortical function in schizophrenia.

There are data indicating that frontal lobe dysfunction can be associated with psychotic symptoms. Evidence of frontal lobe damage leading to abnormal behaviors strikingly similar to some of the more persistent symptoms observed in schizophrenia can be found in anecdotal and case series describing (a) patients with frontal lobe injury and (b) primates with frontal lobe ablations (e.g., see ref. 83). Although there is great individual variation in the severity and constancy of the symptoms that emerge in patients even with severely damaged frontal lobes, some of these bear a remarkable resemblance to the deficit state symptoms in schizophrenia. For example, orbitofrontal and anteromedial lesions can produce flattened affect.

That negative symptoms are associated with decreased DA function (in the mesocortical DA system) is suggested indirectly. pHVA concentrations levels were lower in chronic, treatment refractory schizophrenic patients than in normal subjects (62). Treatment with the DA reuptake blocker, mazindole (35), or with the DA agonist, SKF393939 (84), appears to ameliorate negative symptoms in some schizophrenic patients. Although indirect, these data imply that decreased DA activity can modulate negative symptoms in schizophrenia.

Schizophrenic patients perform poorly on cognitive tests that are thought to depend on activation of the PFC, such as the WCST (e.g., see ref. 32) and the "Tower of London" (31). Animal studies suggest that some of these cognitive deficits may be due to decreased mesocortical DA activity: (a) Surgical ablation of the PFC or selective destruction of mesocortical DA neurons in monkeys impaired performance of the spatial delayed-response task, a test thought to depend on activation of the frontal cortical areas in monkeys (85); (b) iontophoretically applied DA in area 46 [corresponding to the dorsolateral aspects of the PFC (DLPFC) in humans] improved performance in the delayed-response task in monkeys (86); and (c) administration of D1 antagonists dose-dependently produced deficits in performance during the delayed response task, while the selective D2 antagonist raclopride did not (86). Because the terminals of the mesocortical DA system consists of the D1 (and likely D5) receptor subtype (87), these findings suggest that the mesocortical DA system is important for memory and retrieval functions in high-order primates, and by inference in humans as well. These data are consistent with the notion that the decreased cognitive performance on frontally mediated tasks in schizophrenia may be the result of decreased activity of the mesocortical (D1/5) system. Indeed, single-dose administration of DA agonists, such as apomorphine and amphetamine, ameliorate cognitive performance on frontally mediated tasks (36). Studies examining the effect of selective D1/5 agonists on cognitive function in schizophrenia have yet to be conducted.

Andreasen et al. (31) and Wolkin et al. (33) demonstrated that these cognitive deficits occur predominantly in negative-symptom schizophrenics. By inference, the cognitive deficits and negative symptoms in schizophrenia may both be related to decreased mesocortical DA function.

The persistent symptoms of schizophrenia appear to be the deficit state symptoms rather than the positive symptoms and appear to be related to decreased DA function in the cortex rather than being related to increased DA activity in the subcortical regions. Thus, it is not surprising that these symptoms are resistant to treatment with DA antagonists. Indeed, one would expect these symptoms to be amenable to treatment with DA agonists with cortical selectivity. Because mesocortical DA neurons are primarily of the D1 and D5 type, it can be hypothesized that selective D1 or D5 agonists would be particularly helpful for these symptoms. Moreover, consistent with the finding by Jaskiw et al. (88) that increasing prefrontal cortical DA activity reduces striatal DA activity, D1 or D5 agonists would be expected to decrease the hypothesized increased DA activity in subcortical DA neurons and thus be useful (in combination with traditional D2 antagonists) in the treatment of acute psychoses as well. Preliminary data from treatment of nonresponsive patients treated with mazindole or SKF39393 are consistent with this notion (39, 84).

To treat both positive and negative symptoms of schizophrenia, a balance between increasing DA activity at D1/5 receptors and decreasing it at D2, D3, or D4 receptors may be needed. Studies examining such combination treatments, using selective D1 agonists and D2, D3, or D4 antagonists, have yet to be conducted, but promise to be scientifically and possibly practically fruitful.

It would be overly simplistic to hypothesize that the pathophysiology of schizophrenia is only dopaminergic in nature. Several authors have suggested that it may be abnormalities in the interaction between monoaminergic systems in general and between serotonin (5HT) and DA systems in particular (38), rather than abnormalities in any one system, that is relevant to the pathophysiology of schizophrenia. Indeed, it is particularly difficult to discuss DA without mentioning its interactions with 5HT. Both neurotransmitter systems are highly intertwined, anatomically and functionally, with 5HT having an inhibitory modulation on DA function (89). Moreover, human studies consistently find high correlations in CSF between the DA and 5HT metabolites HVA and 5-hydroxyindolic acid (5HIAA), respectively (see ref. 90). This appears to be the result of a functional interaction, with 5HIAA "controlling" HVA (90), and is not due to a shared transport mechanism (91).

Blockade of 5HT receptors diminishes extrapyramidal side effects induced by DA antagonists. Indeed, ritanserin, a selective 5HT1c/2 antagonist, significantly reduced extrapyramidal side effects when added to neuroleptic treatment in schizophrenic patients in several placebo-controlled, double-blind studies (92). Thus the addition of 5HT2 antagonism to DA2 receptor blockade may lead to decreased extrapyramidal side-effect potential of DA receptor blockade. Although more speculative, it has also been suggested that blockade of 5HT2 (38) or 5HT1c (89) receptors mediates, in part, the superior clinical efficacy of clozapine.

Another interesting relationship is the one between 5HT3 systems and DA function. For instance, 5HT3 antagonists fail to alter basal DA activity, but they reverse the increase in DA release that results from behavioral and biological stressors (93, 94). This may have important implications for the treatment of schizophrenia and schizophrenia spectrum disorders. If 5HT3 antagonists prevent stress-induced increases in DA activity, these drugs would be particularly useful in the prevention of relapse in schizophrenic patients and may also have a role in patients that are prone to display psychotic decompensations, such as borderline personality disorders.

The explosion in knowledge concerning DA in general, and its possible role in modulating the symptoms of schizophrenia in particular, offers rich ground for drug development and for further elucidating the biology of schizophrenia and its symptomatology. The following seem to be particularly exciting directions.

1. The development of drugs with selectivity for frontal cortical regions could be a viable approach to the treatment of the negative or deficit symptoms of schizophrenia. Dopamine D1 or D5 receptors would be most appropriate targets.

2. The development of specific D4 antagonists will be an important test of the centrality of this DA receptor subtype in alleviating the positive symptoms of schizophrenia, and it will further our understanding of the relatively unique properties of clozapine.

3. The role of the corticostriatal glutamatergic pathway and its likely role in mediating the reciprocal relationship between cortical and subcortical dopaminergic activity needs to become a target for investigation both in antemortem and postmortem protocols. With the generation of new antibodies for the glutamatergic receptors, the latter may be a particularly worthwhile pursuit.

4. The importance of stress in precipitating subcortical hyperdopaminergia following lesions to the cortex has obvious implications for understanding the initiation of schizophrenic symptoms. Studies in schizophrenic patients that attempt to rigorously document stressful events in a longitudinal context, and correlate them with changes in dopaminergic parameters as well as with symptom fluctuation, would be particularly informative.

5. Some link must be sought between the morphometric abnormalities that have been found in postmortem examination of schizophrenic tissue and the bidirectionality of dopaminergic systems.

With the inevitable conduct of the above investigations, real advances in testing the validity of current conceptualizations regarding DA and schizophrenia will finally be made.

This work was supported, in part, by the Schizophrenia Biological Research Center Grant from the Veterans Administration (4175-020) and by RO1-MH3792206.

published 2000