INTRODUCTION

Thirty years after the description of a direct linear correlation between dopamine (DA) D2 receptor-binding affinities and antipsychotic drug potencies for ameliorating psychosis (Seeman et al, 1976; Creese et al, 1976), this association remains an important cornerstone of current hypotheses of both the etiology and treatment of psychotic disorders (Emilien et al, 1999; Seeman, 2002; Seeman and Tallerico, 1998). However, over the past quarter century significant advances have been made in expanding our understanding of receptor-binding affinities and clinically effective drug dosages for a steadily expanding list of antipsychotic medications. These changes may influence the relationships identified in earlier analyses.

First, Civelli et al, in cloning the DA D2 receptor (Bunzow et al, 1988) provided the foundation for identifying a family of three receptor subtypes, termed DA D2, D3 (Sokoloff et al, 1990), and D4 receptors (Van Tol et al, 1991). Together, these receptor subtypes constituted the receptor-binding affinities measured in earlier studies (Seeman et al, 1976; Creese et al, 1976). Each of these individual receptor subtypes has, in turn, been cloned, and binding data for antipsychotic drugs to these individual cloned DA receptor subtypes and their splice variants, which was not available in 1976, can now be consulted. Variability in antipsychotic medication binding to individual D2 family receptor subtypes (ie D2, D3, and D4) could help to elucidate the contribution of individual DA receptor subtypes in psychosis.

Secondly, many second-generation antipsychotic medications, which were not available in the 1970s, can now be similarly studied. Only one atypical antipsychotic medication, clozapine, had been released at the time of the original analyses in 1976. Further adding to the rationale for re-evaluating the relationship between DA receptor binding and antipsychotic effect is the degree to which D2 family (Kapur and Remington, 2001) compared with serotonergic (5-HT) receptor binding (Meltzer, 1999) may contribute to the antipsychotic properties of atypical antipsychotic medications.

Third, commonly prescribed doses of antipsychotic medications have changed dramatically over the past 25 years (Baldessarini et al, 1993; Vuckovic et al, 1990). Although there are few dose-finding studies adequately powered to clearly identify minimally effective antipsychotic drug dose (Zimbroff et al, 1997), currently available data (Geddes et al, 2000; Bollini et al, 1994), as well as revisions in clinical practice, have led to a marked reduction in the average prescribed doses of most typical D2 receptor antagonists since 1976.

Fourth, cloning and pharmacological characterization of multiple serotonin receptor subtypes, including the 5-HT1A (Fargin et al, 1988), 5-HT2A (Pritchett et al, 1988), and 5-HT2C (Julius et al, 1988) serotonin receptor, provide an opportunity to examine the relationship between antipsychotic binding affinities to these serotonin receptor subtypes and clinically effective antipsychotic drug dosages, an analysis not possible in 1976.

Finally, several frequently prescribed antipsychotic medications, including loxapine and perphenazine, were omitted from the original Seeman analysis, whereas an antidepressant medication (trazodone) was included (Seeman et al, 1976). It is therefore of interest to re-examine the correlation between average antipsychotic drug dosages used in the treatment of psychosis and DA and serotonin receptor subtype binding, including all commonly prescribed antipsychotic medications.

METHODS

In order to minimize variability in assay conditions that may lead to differences in Ki value for a given receptor (Strange, 2001), drug affinity Ki values determined by the NIMH Psychoactive Drug Screening Program (Roth et al, 2004) were used in our analysis. Ki values selected for analysis were those listed as NIMH Psychoactive Drug Screening Program assay certified data, determined from assays using the cloned human receptors with drugs of interest as test ligands. For Ki values for which PDSP certified assay data were not listed, the average Ki value from assay data compiled on the PDSP web site (Roth et al, 2004) using the cloned human receptor with drug of interest as the test ligand was utilized. Ki values from cloned human receptor for three drug/receptor combinations not listed in the PDSP database were identified from published literature. Ki values used in our analysis, with data source, are listed in Tables 1 and 2. As noted, all of the binding data analyzed in our study has been previously reported by other investigators.

Table 1 Antipsychotic Medication Dopamine Receptor Ki Values
Table 2 Antipsychotic Medication Serotonin Receptor Ki Values

Average daily antipsychotic drug dose was determined from data in randomized, controlled clinical trials wherever possible (Leucht et al, 1999), supplemented by the consensus clinical experience of three psychiatrists (NMR, PEK, SMS) who regularly prescribe these medications. The drug dosage ranges determined by consensus clinical experience are similar in each case to recommended dosage ranges for published drug reference guides, including ePocrates Rx (ePocrates, San Carlos, California); the Physicians' Desk Reference (Kaplan et al, 1994; Bernstein, 1988). The midpoint of the dose range was used in subsequent calculations. Values for antipsychotic drug dose were established before any data analysis, blind to specific Ki values, and are listed in Tables 1 and 2.

Data Analysis

Antipsychotic doses and binding affinities were log-transformed before analysis. Linear regression was used to estimate the association between these quantities. A binary variable indicating antipsychotic class (typical or atypical) was also entered into the model and allowed to interact with (log) dose, so that a different regression equation could be estimated for each antipsychotic class. A test of whether the interaction term was statistically different from zero was then used to determine if separate regression equations were indicated for each antipsychotic class. A statistically significant interaction would allow rejecting the possibility that the relationship between binding and effective dosage was the same for both antipsychotic classes. It should be noted, however, that the statistical power of these tests to detect cases where class-specific relationships exist may be low, because of the limited number of data points. Consequently, failure to detect a statistical difference between the best-fitting regressions for each antipsychotic class should not be taken as proof that such a difference does not exist.

Data were analyzed using separate univariate analyses for each receptor subtype. It was not feasible to analyze data using multivariate methods because the limited number of drugs for which Ki values were available for all receptor subtypes examined did not allow for a meaningful statistical analysis.

The linear correlation coefficient (r) is reported as a standardized measure of strength of association for the regressions within each class and, in cases where equality of the regressions could not be statistically rejected, for an additional regression based on combining data from both classes.

RESULTS

The correlation between average clinically effective antipsychotic dose and binding affinity to the cloned human D2 receptor is illustrated in Table 3 and Figure 1. Clinically effective dose and binding affinity to D2 DA receptor were directly correlated for typical antipsychotic medications (r=0.54, p=0.046), but not for second-generation antipsychotic medications (r=0.41, p=0.311). In testing for a difference between the relationships for first- and second-generation drugs, an equality of regression parameters could not be statistically rejected (interaction p=0.72). Combining typical and atypical medications into a single analysis resulted in a linear correlation with r=0.48, p=0.023.

Table 3 Correlation Between Clinically Effective Antipsychotic Dose and Receptor Binding Affinity
Figure 1
figure 1

Clinically effective antipsychotic dose vs binding affinity to cloned human DA D2 receptor for (a) typical and (b) atypical antipsychotic medications.

The relationship between average clinically effective dose and binding affinity to the cloned human D3 receptor is also shown in Table 3. There are no clear correlations between these variables for typical (r=0.33, p=0.292) or atypical antipsychotic medications (r=0.42, p=0.287) analyzed separately. Combining typical and atypical medications into a single analysis yielded a similar degree of correlation (r=0.37, p=0.100).

The relationship between average clinically effective antipsychotic dose and binding affinity to the cloned human D4 receptor is shown in Table 3. For typical antipsychotic drugs, there was no correlation between these two measures (r=0.24, p=0.390). In contrast, for atypical antipsychotics the two measures were more moderately correlated (r=0.59, p=0.123). Equality of regression parameters could not be statistically rejected, however, in testing for a difference between the relationship for first- and second-generation drugs (interaction p=0.30). Combining typical and atypical medications into a single analysis did not demonstrate a significant correlation between the two variables (r=0.37, p=0.088).

The relationship between average clinically effective antipsychotic dose and binding affinity to the cloned human 5-HT1A receptor is shown in Table 3. There were no detectable direct relationships between clinically effective antipsychotic dose and receptor-binding affinity for typical (r=0.14, p=0.684) or atypical antipsychotic medications (r=−0.14, p=0.791).

The relationship between average clinically effective antipsychotic dose and binding affinity to the cloned human 5-HT2A receptor is shown in Table 3 and Figure 2. There is no direct relationship between clinically effective antipsychotic dose and receptor-binding affinity for typical antipsychotic medications (r<−0.10, p=0.812). In contrast, for atypical antipsychotic medications binding affinity and clinically effective dose were nonsignificantly correlated, with a moderate effect size (r=0.62, p=0.133). In testing for a difference between these two relationships, the equality of these regression parameters could not be rejected (interaction p=0.14). Combining typical and atypical medications into a single analysis eliminated the correlation between the two measures (r=0.17, p=0.514).

Figure 2
figure 2

Clinically effective antipsychotic dose vs binding affinity to cloned human serotonin 5-HT2A receptor for (a) typical and (b) atypical antipsychotic medications.

The relationship between average clinically effective dose and binding affinity to the cloned human 5-HT2C receptor is shown in Table 3 and Figure 3. For typical antipsychotic drugs, binding affinity and clinically effective dose were surprisingly negatively correlated (r=−0.68, p=0.021, Figure 3a). For atypical antipsychotic medications, the direction of the correlation was opposite (Figure 3b), and there was no significant direct relationship between clinically effective antipsychotic dose and binding affinity (r=0.47, p=0.285). In testing for a difference between these two relationships, the regression parameters were significantly different (interaction p=0.020). As would be expected, combining typical and atypical medications into a single analysis eliminated any correlation between clinically effective dose and 5-HT2C receptor binding (r=−0.20, p=0.40).

Figure 3
figure 3

Clinically effective antipsychotic dose vs binding affinity to cloned human serotonin 5-HT2C receptor for (a) typical and (b) atypical antipsychotic medications.

In order to evaluate possible interactions between receptor subtypes playing a role in mechanism of antipsychotic efficacy, we analyzed correlations between log (average dose) and log (ratio of binding affinities) for combinations of individual receptor subtypes, as shown in Table 4. As illustrated in Figure 4a, there is a modest negative correlation between average clinically effective dose and ratio of binding affinities for 5-HT2A/D2 receptors for typical antipsychotic medication (r=−0.52, p=0.082), however, surprisingly for atypical antipsychotics, there was no detectable relationship between dose and 5-HT2A/D2 binding affinity ratio (r=−0.08, p=0.869, Figure 4b).

Table 4 Correlation Between Clinically Effective Antipsychotic Dose and Receptor Binding Affinity Ratios
Figure 4
figure 4

Clinically effective antipsychotic dose vs ratio of binding affinity to cloned human serotonin 5-HT2A/DA D2 receptor for (a) typical and (b) atypical antipsychotic medications.

As shown in Figure 5a, typical antipsychotic medication dose and 5-HT2C/D2 receptor-binding affinity ratios were strongly and inversely correlated (r=−0.81, p=0.003). In contrast, there was no detectable relationship between dose and 5-HT2C/D2 receptor-binding affinity ratio for second-generation antipsychotic medications (r=−0.30, p=0.507, Figure 5c).

Figure 5
figure 5

Clinically effective antipsychotic dose vs ratio of binding affinity to cloned human serotonin 5-HT2C/DA D2 receptor for (a) typical and (b) atypical antipsychotic medications.

A similar analysis of 5-HT2A/D3 receptor-binding affinity ratios did not identify correlations between these values and clinically effective dosages of typical (r=−0.40, p=0.216) or atypical (r<0.01, p=0.997), or pooled first- and second-generation antipsychotic medications, as illustrated in Table 4.

In contrast, 5-HT2C/D3 receptor-binding affinity ratios were correlated with clinically effective dosages of typical antipsychotic medications (r=−0.67, p=0.023, Figure 6a). 5-HT2C/D4 receptor-binding affinity ratios were similarly correlated with clinically effective dosages of typical antipsychotic medications (r=−0.70, p=0.016, Figure 7a). A comparable relationship, however, was not observed for atypical antipsychotic medications, as 5-HT2C/D3 (Figure 6b) and 5-HT2C/D4 (Figure 7b) receptor-binding affinity ratios were not correlated with clinically effective dosages of atypical antipsychotic medications.

Figure 6
figure 6

Clinically effective antipsychotic dose vs ratio of binding affinity to cloned human serotonin 5-HT2C/DA D3 receptor for (a) typical and (b) atypical antipsychotic medications.

Figure 7
figure 7

Clinically effective antipsychotic dose vs ratio of binding affinity to cloned human serotonin 5-HT2C/DA D4 receptor for (a) typical and (b) atypical antipsychotic medications.

Analysis of DA and serotonin receptor-binding affinities singly and in simple combinations did not identify correlations between receptor subtype binding and clinical efficacy for atypical antipsychotic medications. We therefore further evaluated the relationship between receptor binding and efficacy using a more comprehensive set of binding affinity ratios. Although there is not a universal consensus on this point, it has previously been suggested that the antipsychotic effect of atypical antipsychotic medications results from a balance of inhibition at serotonin 5-HT2A, 5-HT2C, and DA D2 receptors (Meltzer, 1989; Meltzer, 1995; Leysen et al, 1993; Huttunen, 1995), coupled with simultaneous agonist effects at serotonin 5-HT1A receptors (Meltzer, 1999; Millan, 2000; Protais et al, 1994). In order to identify therapeutic benefit resulting from the interaction between simultaneous effects at these receptor subtypes, we determined the relationship between clinically effective antipsychotic medication dose and ratios incorporating the binding affinities for each of these receptor systems. As shown in Table 4 and Figure 8 (lower panel), atypical antipsychotic medication dose and D2 (5-HT2A/5-HT1A) binding affinity ratio are highly correlated (r=0.80, p=0.031). Similarly, D2 (5-HT2C/5-HT1A) binding affinity ratio and atypical antipsychotic medication dose are also highly correlated (r=0.78, p=0.038, Figure 9, lower panel). In contrast, neither binding affinity ratio is significantly correlated with clinically effective dose for typical antipsychotic medications (Figures 8 and 9, upper panels). Removing the 5-HT1A receptor-binding affinity term from the equation by correlating antipsychotic medication dose with (5-HT2A × D2) or (5-HT2C × D2) binding affinity ratio lessens the resulting degree of correlation (Table 4). Similarly, the receptor-binding relationships can be modified, so that 5-HT1A and D2 receptor binding no longer have functionally opposite roles, and D2 binding no longer has a functionally similar action as 5-HT2A and 5-HT2C binding, by inverting the serotonin receptor affinity terms (Table 4, lower right two columns). This modification completely eliminates the correlation between binding affinity ratio and drug dosage for atypical antipsychotic medications. Typical antipsychotic drug dosage, in contrast, is significantly correlated with the resulting binding affinity ratio D2 (5-HT1A/5-HT2C) (r=0.75, p=0.013). Comparing this result to the 5-HT2C/D2 binding affinity vs typical antipsychotic drug dosage correlation described above (Table 4) suggests that the D2/5-HT2C term contributes the majority of influence to this relationship.

Figure 8
figure 8

Clinically effective antipsychotic dose vs ratio of binding affinity to cloned human D2 (5-HT2A/5-HT1A) receptor for (a) typical and (b) atypical antipsychotic medications.

Figure 9
figure 9

Clinically effective antipsychotic dose vs ratio of binding affinity to cloned human D2 (5-HT2C/5-HT1A) receptor for (a) typical and (b) atypical antipsychotic medications.

DISCUSSION

Here, we present data evaluating the relationship between binding affinity to several catecholamine receptor subtypes and drug dosage for antipsychotic efficacy. Our analysis is similar in concept to prior studies demonstrating a linear correlation between antipsychotic drug dose and D2-family DA receptor-binding affinity (Seeman et al, 1976; Creese et al, 1976). Our goal was to evaluate additional DA and serotonin receptor subtypes, which had not been identified at the time of the earlier analyses, in order to determine whether affinity to individual receptor subtypes could be correlated with antipsychotic potency of these medications. Although we expected to identify common DA receptor subtypes mediating antipsychotic efficacy for both typical and atypical antipsychotic medications, our analysis instead identified surprising differences in serotonergic mechanisms mediating antipsychotic efficacy for typical vs atypical medications. The major findings identified by analysis of this data are discussed below.

Typical Antipsychotic Medications

In agreement with earlier studies, we determined that antipsychotic drug dosage for typical antipsychotic medications is directly correlated with binding to D2 DA receptors, however, the strength of this correlation was less robust than anticipated. Our data suggest that this may be related in part to interactions between typical antipsychotic medications and serotonin 5-HT2C receptors. The observation that serotonin 5-HT2C receptor affinity is negatively correlated with antipsychotic drug dosage for typical antipsychotic medications was an unexpected outcome of our data analysis. Additionally, 5-HT2C and D2 receptor-binding affinities of typical antipsychotic medications interact such that the ratio of serotonin 5-HT2C/D2 receptor-binding affinity more accurately predicts dosage needed for antipsychotic effect than do 5-HT2C or D2 binding affinities independently. Thus, increasing serotonin 5-HT2C receptor antagonist affinity lowers antipsychotic potency at any given level of D2 blockade, suggesting that signaling through 5-HT2C receptors interacts with and improves antipsychotic effects achieved via D2 receptor blockade. In contrast, the correlation between serotonin 5-HT2C receptor affinity and clinically effective antipsychotic drug dose for atypical antipsychotic medications differs from the correlation for typical medications (p=0.02), is in the opposite direction, and the degree of correlation is less pronounced (Figure 3). Although a potential role for 5-HT2C receptor antagonism in the therapeutic effect of atypical antipsychotic medications has previously been discussed (Meltzer et al, 2003; Meltzer, 1995; Wood et al, 2006), it has also been suggested that 5-HT2C agonism could be therapeutic (Meltzer, 1999; Marquis et al, 2007) based on a wide range of preclinical measures demonstrating that serotonin 5-HT2C receptor stimulation inhibits the mesolimbic DA system (Alex et al, 2005; Pozzi et al, 2002; Di Giovanni et al, 1999; De Deurwaerdere and Spampinato, 1999; Millan et al, 1998; Di Matteo et al, 1999; Di Matteo et al, 2002). These findings are consistent with our observation, and suggest a potential mechanism for 5-HT2C receptor blockade to worsen psychotic symptoms. Human data supporting the concept that 5-HT2C blockade lowers the antipsychotic potency of first-generation antipsychotic medications has not been previously elucidated to our knowledge, however.

The neuroanatomical mechanism(s) underlying this finding may be related to the tonic inhibitory control exerted by serotonin operating through 5-HT2C receptors over limbic dopaminergic pathways (De Deurwaerdere and Spampinato, 1999). Serotonergic cell bodies originating in the raphe nucleus project diffusely to targets throughout the brain, and strong 5-HT2C receptor expression has been observed in nucleus accumbens and ventral striatum, with modest expression in prefrontal cortex (Lopez-Gimenez et al, 2001; Eberle-Wang et al, 1997). In prefrontal cortex, 5-HT2C receptors are co-expressed with DA D4 receptors (Vysokanov et al, 1998). Within the substantia nigra/ventral tegmentum, 5-HT2C receptors are expressed on inhibitory GABA-ergic interneurons (Eberle-Wang et al, 1997). 5-HT2C receptor stimulation inhibits reward system-related behaviors including cocaine-induced hyperlocomotion (Filip and Cunningham, 2003; Grottick et al, 2000). 5-HT2C receptor blockade increases both dopaminergic cell firing and DA release in nucleus accumbens and frontal cortex (Alex et al, 2005; Di Matteo et al, 1999; Millan et al, 1998). 5-HT2C receptor blockade could oppose the actions of D2 DA receptor blockade either through direct effects on second messenger systems in neurons co-expressing DA D2 and 5-HT2C receptors, or indirectly through a systems effect on components of limbic neurotransmission. In addition to antagonist effects through inhibition of basal serotonin tone, constitutive activity of 5-HT2C receptor isoforms have also been previously described (Niswender et al, 1999; Westphal et al, 1995), and this constitutive activity participates in the tonic inhibition of mesolimbic DA function (De Deurwaerdere et al, 2004). Antipsychotic medications could therefore also function as inverse agonists through second messenger pathways of 5-HT2C isoforms (Navailles et al, 2006; Rauser et al, 2001). Although previous studies of the 5-HT2C inverse agonist properties of antipsychotic medications have identified the potential role for inverse agonism in the mechanism of action of antipsychotic efficacy (Navailles et al, 2006), our data suggest an alternative possibility that 5-HT2C inverse agonism may also directly oppose acute antipsychotic efficacy. Our analysis supports the concept that 5-HT2C agonists, in contrast, may have therapeutic potential as adjunctive medications to improve antipsychotic efficacy for patients receiving typical antipsychotic medication, and suggests that these medications could have applications for treatment refractory psychosis. The potential therapeutic benefit of 5-HT2C stimulation in inhibiting psychotic symptoms through inhibition of meso-accumbens DA function must be balanced with the potential for worsening of cognitive and negative symptoms through decreased mesocortical DA function.

5-HT2C/D3 and to a lesser extent 5-HT2C/D4 binding affinity ratios were also correlated with clinically effective antipsychotic medication dose for typical antipsychotic medications. Our data therefore suggest the likelihood of an interaction between binding at serotonin 5-HT2C and DA D2, D3, and D4 receptors in the mechanism of action of typical antipsychotic medications.

Our analysis identifies a modest correlation between antipsychotic drug dosage and the ratio of serotonin 5-HT2A/D2 receptor affinity for typical antipsychotic medications.

Atypical Antipsychotic Medications

Therapeutic efficacy for atypical antipsychotic medications has been suggested to result from a balance of inhibition at DA D2, serotonin 5-HT2A, and 5-HT2C receptors (Meltzer, 1989, 1995; Leysen et al, 1993; Huttunen, 1995), whereas serotonin 5-HT1A receptor stimulation appears to contribute to antipsychotic efficacy in rat models (Protais et al, 1994; Meltzer et al, 2003; Millan, 2000). Consistent with these observations, clinically effective dosages of atypical antipsychotic medication are highly correlated with the ratios of D2 (5-HT2A/5-HT1A) and D2 (5-HT2C/5-HT1A) receptor-binding affinities. Thus, our analysis suggests that the therapeutic effects of atypical antipsychotic medications are determined by interactions among three different domains: (1) increasing D2 DA receptor-binding affinity enhances antipsychotic potency. (2) Increasing 5-HT2C and 5-HT2A receptor-binding affinities also facilitate antipsychotic efficacy. (3) Increasing 5-HT1A receptor-binding affinity, in contrast, reduces antipsychotic efficacy.

We are not aware of other studies demonstrating that serotonin 5-HT2C receptor blockade has opposite effects in typical and atypical antipsychotic medications. It has previously been suggested, however, that both 5-HT2C antagonism (Meltzer et al, 2003; Meltzer, 1995; Wood et al, 2006) and 5-HT2C receptor stimulation (Meltzer, 1999; Marquis et al, 2007) could facilitate antipsychotic activity. Our data support the view that this seemingly paradoxical finding may result from the relatively higher 5-HT2A receptor blockade in atypical vs typical medications. Thus, simultaneous 5-HT2A and 5-HT2C receptor blockade may be more effective in mediating antipsychotic effects than blockade of either receptor separately (Meltzer et al, 2003).

Our finding that a simple linear correlation between D2 receptor binding and clinically effective drug dosage is not apparent for atypical antipsychotic drugs was an unexpected outcome. Previous studies have determined that atypical antipsychotic medications can be distinguished from typical antipsychotic drugs based on the ratio of 5-HT2/D2-binding affinities (Meltzer et al, 1989a, 1989b). Thus, we had anticipated a more direct relationship between D2 binding and antipsychotic efficacy for second-generation medications, and an interaction between serotonin 5-HT2A and D2 effects. Instead, we observed a modest correlation between atypical antipsychotic drug dosage and serotonin 5-HT2A receptor binding, and a similarly modest correlation between DA D4 receptor-binding affinity and atypical antipsychotic drug dosages. It has previously been suggested that a subset of atypical antipsychotic medications derive their efficacy in part from selective effects at D4 DA receptors (Seeman et al, 1997). Our data do not provide evidence supporting the concept that a simple ratio of binding to 5-HT2A and D2 receptors accounts for a significant proportion of atypical antipsychotic medication efficacy, although this ratio does appears to distinguish between atypical and typical medication classes (Meltzer et al, 1989a, 1989b). Addition of more members of the atypical class to the relatively small number of drugs available for our analysis might help to further clarify this issue.

Limitations

Our analysis is limited to antipsychotic medication effects on positive psychotic symptoms, and does not address efficacy for negative symptoms or cognition, which may be more important in terms of long-term functional outcome. Importantly, the strength of correlations between receptor binding and antipsychotic efficacy identified in our analysis are restricted by a wide range of limiting factors. Medication differences in absorption; metabolism; protein binding; and the presence of pharmacologically active metabolites all serve to weaken the observed correlations. Additionally, the antipsychotic medication dose prescribed to patients may be determined in part by side effects, and might therefore not accurately reflect the ‘ideal’ efficacy dose. The paucity of adequately powered clinical trials to determine optimal dose for antipsychotic medications further limits the accuracy of medication dosages employed in our analysis. Also, the binding data used in the current analysis, measuring ligand binding to cloned human receptors expressed in cell culture systems, may be distinct from binding to limbic neurotransmitter receptor populations in vivo. Differences in receptor phosphorylation, glycosylation, and/or dimerization to hetero-oligomers (Nimchinsky et al, 1997; Scarselli et al, 2001; Zawarynski et al, 1998; Lee et al, 2000) between in vivo and cell culture systems lacking post-translational machinery could potentially alter receptor-binding affinity. Additionally, atypical antipsychotic medications as a group tend to have more rapid dissociation rates from DA receptors than typical antipsychotics (Kapur and Seeman, 2001), an effect that might further complicate the relationship between receptor affinity and clinically effective drug dose. And finally, this approach is inherently limited by the complexities of brain circuitry in which DA and serotonin receptors may function as a ‘brake’ in one brain region, and simultaneously as an ‘accelerator’ in a different brain region. For example, blockade of D2 DA autoreceptors in cell body regions of the ventral tegmentum increases both synthesis and release of DA, which could worsen psychotic symptoms, whereas blockade of postsynaptic D2 receptors in limbic terminal regions would likely have an opposite behavioral effect. Thus, the dysfunction of schizophrenia, resulting from a complex interaction of multiple receptor and neurotransmitter systems (Carlsson et al, 1999), does not lend itself ideally to an analysis of isolated receptor systems.

In summary, we present data demonstrating correlations between clinical efficacies of antipsychotic medications and binding affinities to D2, D3, D4, 5-HT1A, 5-HT2A, and 5-HT2C receptor subtypes. Given the numerous limitations inherent in this approach (listed above), the strength of correlations described in this analysis suggest that the DA and serotonin receptor subtypes analyzed provide the preponderance of antipsychotic effect of these medications. The specific mechanism(s) underlying this clinical effect, however, remains obscure. The ‘disconnect’ between the pharmacokinetics of receptor blockade and the extended time lag until clinical benefit suggests antipsychotic efficacy, while initiated through binding to neurotransmitter receptor target(s), is likely the result of a downstream cascade of changes in gene transcription and translation. Studies identifying the specific targets of altered gene transcription resulting from these drug-neurotransmitter receptor interactions would therefore have high likelihood of improving specificity and efficacy of antipsychotic medications.