Dopamine D₂ Receptor Occupancy and Cognition in Schizophrenia: Analysis of the CATIE Data

Abstract

Introduction: Antipsychotic drugs exert antipsychotic effects by blocking dopamine D₂ receptors in the treatment of schizophrenia. However, effects of D₂ receptor blockade on neurocognitive function still remain to be elucidated. The objective of this analysis was to evaluate impacts of estimated dopamine D₂ receptor occupancy with antipsychotic drugs on several domains of neurocognitive function in patients with schizophrenia in the Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) trial. Methods: The dataset from the CATIE trial was used in the present analysis. Data were extracted from 410 subjects who were treated with risperidone, olanzapine, or ziprasidone, received assessments for neurocognitive functions (verbal memory, vigilance, processing speed, reasoning, and working memory) and psychopathology, and provided plasma samples for the measurement of plasma antipsychotic concentrations. D₂ receptor occupancy levels on the day of neurocognitive assessment were estimated from plasma antipsychotic concentrations, using population pharmacokinetic analysis and our recently developed model. A multivariate general linear model was used to examine effects of clinical and demographic characteristics, including estimated D₂ occupancy levels, on neurocognitive functions. Results: D₂ occupancy levels showed significant associations with the vigilance and the summary scores. Neurocognitive functions, including vigilance, were especially impaired in subjects who showed D₂ receptor occupancy level of >77%. Discussion: These findings suggest a nonlinear relationship between prescribed antipsychotic doses and overall neurocognitive function and vigilance. This study shows that D₂ occupancy above approximately 80% not only increases the risk for extrapyramidal side effects as consistently reported in the literature but also increases the risk for cognitive impairment.

Introduction

Neurocognitive impairment is considered to be a core feature in schizophrenia and has a strong correlation with real-world functioning of patients.¹ While antipsychotic drugs play a principal role in the treatment of schizophrenia, high doses of antipsychotic drugs have been associated with negative consequences on global neurocognitive function² as well as specific cognitive domains, including processing speed,³ visual memory, delayed recall, performance IQ, and executive function.⁴ Antipsychotic drugs, including atypical antipsychotics, have been related with mixed results in terms of effects on the neurocognitive impairment due to this illness^5,6; however, many of these studies involved methodological shortcomings, including small sample sizes and insufficient neurocognitive measures.^7–9 Therefore, effects of antipsychotic drugs on neurocognitive function still remain to be elucidated.¹⁰

Accumulated evidence has shown that the dopaminergic system in the central nervous system is profoundly associated with cognition.¹¹ For example, the availability of dopamine D₂ receptors has been reported to have a significant impact on neurocognitive function, including attention and executive function, in healthy subjects.¹¹ Similarly, the blockade of dopamine D₂ receptors by risperidone has a negative correlation with attention in patients with schizophrenia.¹² Furthermore, neurocognitive performance, including verbal fluency, spatial span, planning, and sequence generation, was found to be positively correlated with both dopamine D₁ and D₂ receptor binding levels, but mainly with D₂ binding levels in patients with Huntington’s disease.¹³ Animal studies have also endorsed the pivotal role of the dopaminergic system in neurocognitive function; in mutant mice, the absence of D₂ receptors has been demonstrated to impair performance in spatial working memory and perceptual discrimination.^14,15

We have recently reported that striatal dopamine D₂ receptor occupancy by antipsychotic drugs, including risperidone, olanzapine, and ziprasidone, can be reliably estimated from plasma concentrations of these drugs.¹⁶ In addition, recent advances in nonlinear mixed-effects population pharmacokinetic methods have made it possible to predict individual pharmacokinetic parameters for antipsychotic drugs, including peak and trough plasma concentrations, using 2 or more sparsely collected blood samples in a real-world setting.¹⁷ By combining these models, the dopamine D₂ receptor occupancy levels at peak and trough can be reliably estimated using the measurement of antipsychotic plasma concentrations at 2 separate time points.¹⁸

For the purpose of elucidating the relationship between neurocognitive function and estimated dopamine D₂ receptor blockade by antipsychotics, the Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) trial provides an ideal dataset in light of its unprecedented large sample size, comprehensive neurocognitive assessments, and assessment of plasma antipsychotic concentrations with which population pharmacokinetic models have already been developed for risperidone, olanzapine, and ziprasidone.^19–21 The objective of this report was to evaluate impacts of estimated dopamine D₂ receptor occupancy with risperidone, olanzapine, and ziprasidone on several domains of neurocognitive function in patients with schizophrenia in the CATIE trial. Our working hypothesis was that the relations between cognitive functions and D₂ occupancy would be U-shaped. That is, cognition remains impaired at low D₂ occupancy, and medium D₂ occupancy may improve cognition; however, high occupancy of D₂ receptors impairs cognition. This hypothesis came from the following observations in the literature: (1) a moderate amount of antipsychotic drugs generally improves cognitive functions,⁵ (2) an excessive blockade of dopamine D₂ receptors by antipsychotics is associated with worsening in cognitive functions,¹² and (3) an insufficient blockade of dopamine D₂ receptors by antipsychotics does not fully exert its therapeutic effect.²²

Methods

Study Design

The CATIE trial was funded by the National Institute of Mental Health to compare the effectiveness of atypical antipsychotics and a single conventional antipsychotic medication in patients with schizophrenia; the details of the study were reported elsewhere.²³ Briefly, the study was performed between January 2001 and December 2004 at 57 clinical sites in the United States. One thousand four hundred and ninety-three patients between ages 18 and 65 years with a diagnosis of schizophrenia on the basis of the Structured Clinical Interview of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition participated in the CATIE trial. Patients were initially randomized to olanzapine (7.5–30 mg/day), risperidone (1.5–6.0 mg/day), ziprasidone (40–160 mg/day), quetiapine (200–800 mg/day), or perphenazine (8–32 mg/day) under double-blind conditions and received treatments for up to 18 months or until treatment was discontinued for any reason (phase 1).

Data used in the present analysis were derived from subjects who were receiving risperidone, olanzapine, or ziprasidone, completed assessments for psychopathology and neurocognitive function at months 1 and 2, respectively, and provided plasma samples for the assessment of plasma antipsychotic concentrations. These 3 drugs were included in the present study because the nonlinear mixed-effect models were already established for them using the data from the CATIE studies.^19–21 All participants gave written informed consent to participate in the protocols approved by the local institutional review boards.

Assessments for Cognition, Psychopathology, and Extrapyramidal Symptoms

Neurocognitive assessment was performed at month 2. The neurocognitive tests were chosen by a group of advisors based upon the following considerations; sensitivity to neurocognitive impairment in schizophrenia, relation to functional outcome, potential sensitivity to treatment, and practicality for various antipsychotic clinical trials for schizophrenia.^7,24,25 The following 5 neurocognitive domain scores were calculated from 9 neurocognitive test summary scores and standardized to create z scores for each domain in the CATIE trial: verbal memory, vigilance, processing speed, reasoning, and working memory. The verbal memory domain score was calculated from the Hopkins Verbal Learning Test, which assesses verbal learning and memory. The vigilance domain score was calculated from the Continuous Performance Test, which assesses attention. The processing speed domain score was obtained from category instances, the Grooved Pegboard, and the Revised Wechsler Adult Intelligence Scale Digit Symbol Test, which represents processing speed. The reasoning domain score was derived from the Wisconsin Card Sorting Test and the Revised Wechsler Intelligence Scale for Children Mazes. The working memory domain score was calculated from the Letter-number test of auditory working memory and a computerized test of visuospatial working memory. A neurocognitive summary score was calculated by creating a z score of the average of the 5 standardized domain scores. The following information was also collected: age, sex, years of education, and concomitant mediations. The Positive and Negative Syndrome Scale (PANSS) and Simpson-Angus Scale (SAS) were also conducted at month 1.

Population Pharmacokinetic Analysis

Subjects who participated in the CATIE trial provided plasma samples for the measurement of concentrations of risperidone plus 9-hydroxyrisperidone (active moiety), olanzapine, or ziprasidone at more than one time points. Using these samples, plasma antipsychotic concentrations at peak and trough that corresponded to the dose given on the day of cognitive assessment were calculated for each individual using the established population pharmacokinetic models and extracting the Empirical Bayes Estimates for the pharmacokinetic parameters from each of these individuals.^26,27 These parameters were utilized to calculate the expected peak and trough concentrations for each of risperidone, 9-hydroxyrisperidone, ziprasidone, and olanzapine for the dosage regimen on the day of the neurocognitive assessments. Thus, the model-predicted values of the plasma concentrations were used to calculate the peak and trough dopamine D₂ receptor occupancy levels for each individual on the day of neurocognitive assessments. The precision and reliability of this estimation has recently been confirmed in our population pharmacokinetic study.²⁸ The nonlinear mixed-effect models for olanzapine, risperidone, and ziprasidone were previously established using the CATIE data.^19–21 These original studies used to establish the population pharmacokinetic models comprised 1236 risperidone and 9-hydroxyrisperidone concentrations from 490 subjects, 1527 olanzapine concentrations from 523 subjects, and 568 ziprasidone concentrations from 233 subjects, respectively. All 3 compounds were adequately described using a one-compartment linear model with first-order absorption. The previously established models utilized exponentiated or log-normal interindividual variability on each pharmacokinetic parameters, a mixture distribution to assign the tri-modal distribution of clearance as CYP 2D6 genotype was not available for risperidone, an age effect on clearance of the 9-hydroxyrisperidone moiety, and sex, race, and age effects on olanzapine disposition.

Estimation of Dopamine D₂ Receptor Occupancy

By using the predicted plasma concentrations of antipsychotics at peak and trough on the day of cognitive assessment, corresponding dopamine D₂ receptor occupancy levels were estimated, using our recently developed model.¹⁶ Briefly, dopamine D₂ receptor occupancy levels were estimated by incorporating the predicted plasma concentration of risperidone active moiety, olanzapine, or ziprasidone into the following one-site binding model: occupancy (%) = a × (plasma level/[plasma level + ED₅₀]), where a is the maximum receptor occupancy attributable to the antipsychotic drug, and ED₅₀ is the estimated plasma concentration of the antipsychotic drug associated with 50% of receptor occupancy, which was obtained in the systematic review and pooled analysis (Risperidone active moiety: a = 88.0%, ED₅₀ = 4.9 ng/ml; olanzapine: a = 90.7%, ED₅₀ = 7.1 ng/ml; and ziprasidone: a = 88.2%, ED₅₀ = 32.9 ng/ml).¹⁶ Mean values of those peak and trough dopamine D₂ receptor occupancy levels were obtained for further analyses.

Statistical Analysis

Statistical Analyses Were Carried Out Using SPSS Version 19.0 (SPSS Inc., Chicago). To test the hypothesis, subjects were divided into an equal number of 4 groups (ie, 102 or 103) based on the predicted dopamine D₂ receptor occupancy on the day of cognitive testing (low D₂ occupancy group: 15.5–62.7%, n = 102; slightly low D₂ occupancy group: 62.7–71.8%, n = 102; slightly high D₂ occupancy group: 71.9–77.2%, n = 103; and high D₂ occupancy group: 77.2–85.8%, n = 103). A multivariate general linear model was used to examine effects of antipsychotic drugs (ie, risperidone, olanzapine, or ziprasidone), dopamine D₂ receptor occupancy levels (ie, those 4 groups), age, education years, PANSS total score, SAS mean score, and the use of anticholinergics on 5 neurocognitive domain and summary scores. In addition, to exclude a possibility of potential interaction between age and estimated dopamine D₂ receptor occupancy, we performed additional analysis, using the data from subjects aged less than 50; another multivariate general linear model was used to examine effects of the above demographic and clinical characteristics other than age on 5 neurocognitive domain and summary scores. Variables of interest were compared among the 4 dopamine D₂ receptor occupancy groups, using a one-way ANOVA for parametric data and chi-square test for categorical variables. When appropriate, we also examined group differences with pairwise comparisons using Turkey-Kramer HSD (honestly significant difference). A P value of <.05 was considered statistically significant (2-tailed).

Results

Subject Characteristics

Four hundred and ten subjects who provided plasma samples of risperidone plus 9-hydroxyrisperidone, olanzapine, or ziprasidone and received neurocognitive assessments at month 2 and the PANSS and SAS at month 1 were identified. Demographic and clinical characteristics of these subjects were summarized in table 1. Mean ± SD daily doses of risperidone, olanzapine, and ziprasidone on the day of neurocognitive assessments were 3.9 ± 1.3 mg, 19.7 ± 7.0 mg, and 100.5 ± 57.9 mg, respectively. Mean ± SD PANSS total and SAS mean scores were 69.9 ± 18.3 and 0.19 ± 0.29, respectively. Characteristics of groups stratified by dopamine D₂ receptor occupancy levels are detailed in table 2. No significant difference was found in any of the total, positive, negative, or general psychopathology scale PANSS score.

Table 1.

Demographic and Clinical Characteristics of 410 Subjects

Characteristics	Values
Age, years, mean ± SD (range)	40.9 ± 10.3 (18–62)
Male, n (%)	303 (73.9)
Ethnicity, n (%)
White	259 (63.2)
Others	151 (36.8)
Duration of education, years, mean ± SD (range)	12.3 ± 2.0 (3–21)
Duration of treatment, years, mean ± SD (range)	16.5 ± 10.9 (0–56)
Marital status
Married, n (%)	39 (9.5)
Previously married, n (%)	130 (31.7)
Never married, n (%)	241 (58.8)
Employment status
Unemployed, n (%)	349 (85.1)
Antipsychotics
Risperidone, n (%)	150 (36.6)
Olanzapine, n (%)	184 (44.9)
Ziprasidone, n (%)	76 (18.5)
Use of anticholinergics, n (%)	69 (16.8)
PANSS total score, mean ± SD (range)	69.9 ± 18.3 (32–131)
SAS mean score, mean ± SD (range)	0.19 ± 0.29 (0–1.83)

Note: PANSS, Positive and Negative Syndrome Scale; SAS, Simpson-Angus Scale.

Table 2.

Characteristics of Subjects Stratified by Dopamine D₂ Occupancy

Characteristics	Low (n = 102)	Slightly Low (n = 102)	Slightly High (n = 103)	High (n = 103)
D2 occupancy, %, range (mean ± SD)	15.5–62.7 (49.0 ± 11.4)	62.7–71.8 (67.6 ± 2.6)	71.9–77.2 (74.4 ± 1.6)	77.2–85.8 (80.1 ± 1.9)
Age, years, mean ± SD (range)a	39.2 ± 10.6 (20–62)	39.8 ± 10.0 (20–62)	41.5 ± 9.4 (21–59)	43.2 ± 10.9 (18–62)
Male, n (%)	78 (76.5)	77 (75.5)	79 (76.7)	69 (67.0)
Duration of education, years, mean ± SD (range)	12.2 ± 2.4 (3–21)	12.2 ± 1.8 (6–18)	12.4 ± 1.8 (7–21)	12.3 ± 1.9 (3–18)
Duration of treatment, years, mean ± SD (range)	14.8 ± 10.0 (0–40)	16.0 ± 11.2 (0–39)	17.9 ± 11.7 (0–56)	17.3 ± 10.4 (0–42)
Antipsychoticsb
Risperidone, n (%)	16 (15.7)	42 (41.2)	53 (51.5)	39 (37.9)
Olanzapine, n (%)	20 (19.6)	50 (49.0)	50 (48.5)	64 (62.1)
Ziprasidone, n (%)	66 (64.7)	10 (9.8)	0 (0)	0 (0)
Use of anticholinergics, n (%)	14 (13.7)	16 (15.7)	17 (16.5)	22 (21.4)
PANSS total score, mean ± SD (range)	68.0 ± 18.8 (32–108)	68.7 ± 17.3 (32–113)	71.7 ± 18.3 (34–131)	71.1 ± 18.6 (33–120)
PANSS positive score, mean ± SD (range)	16.1 ± 5.9 (7–33)	16.4 ± 5.4 (7–31)	16.8 ± 5.5 (7–35)	17.1 ± 5.6 (7–31)
PANSS negative score, mean ± SD (range)	18.4 ± 6.7 (7–36)	18.2 ± 6.0 (8–35)	19.7 ± 6.3 (7–38)	19.7 ± 6.8 (7–38)
PANSS general score, mean ± SD (range)	33.6 ± 9.7 (16–56)	34.0 ± 9.2 (16–57)	35.1 ± 9.4 (18–58)	34.3 ± 9.3 (16–62)
SAS mean score, mean ± SD (range)c	0.16 ± 0.26 (0–1.5)	0.13 ± 0.19 (0–0.8)	0.25 ± 0.34 (0–1.83)	0.22 ± 0.31 (0–1.5)

Note: Abbreviations are explained in the first footnote to table 1. No statistically significant differences were found in any of the other variables.

^a F _{3, 406} = 3.15, P = .03 by the one-way ANOVA; no further statically significant difference was found by the Turkey-Kramer HSD (honestly significant difference).

^bχ² ₆ = 200.8, P < .001.

^c F _{3, 406} = 3.58, P = .01 by the one-way ANOVA; no further statically significant difference was found by the Turkey-Kramer HSD.

Association Between D₂ Receptor Occupancy and Neurocognitive Function

Results of the multivariate general linear model are detailed in table 3. While age and education years were found to have significant effects on all of the subscale scores, dopamine D₂ receptor occupancy levels also showed significant association with the vigilance and the summary scores. Mean values of neurocognitive scores, including the vigilance and the summary scores, and dopamine D₂ receptor occupancy levels demonstrated a nonlinear relationship, where there seemed to be a cliff-fall-off at higher occupancies (table 4 and figure 1). We therefore conducted additional analyses to try to find the threshold. Subjects were divided into the 2 groups based on consecutive cutoff points in increments of 1% between 70% and 85% in the dopamine D₂ receptor occupancy. Then, other general linear models that included the same clinical and demographic variables as the original model were built to examine which threshold(s) resulted in the presence of significant effects of dopamine D₂ receptor occupancy on vigilance and summary scores. The results revealed that dichotomization of subjects based on 77%, 78%, or 79% and 77%, 78%, 79%, 80%, or 81% were associated with the presence of statistically significant effects of dopamine D₂ receptor occupancy on vigilance and summary scores, respectively. Another significant finding was an association between the processing speed score and the PANSS total score (table 3).

Table 3.

Relationship Between Neurocognitive Scores and Subjects’ Characteristics (n = 410)

Characteristics	Verbal Memory		Vigilance		Processing Speed		Reasoning		Working Memory		Summary Score
	F df1,df2	P Value	F df1,df2	P Value	F df1,df2	P Value	F df1,df2	P Value	F df1,df2	P Value	F df1,df2	P Value
D2 occupancy level	0.583,399	.63	5.523,399	.002	1.813,399	.14	1.273,399	.28	1.373,399	.25	3.023,399	.03
Age	8.301,399	.004	11.551,399	.001	51.331,399	<.001	63.911,399	<.001	31.081,399	<.001	51.281,399	<.001
Education year	43.991,399	<.001	16.191,399	<.001	31.751,399	<.001	28.361,399	<.001	28.221,399	<.001	53.261,399	<.001
PANSS	2.681,399	.10	0.601,399	.44	5.111,399	.02	0.441,399	.51	1.631,399	.20	3.211,399	.07
SAS	0.521,399	.47	0.171,399	.68	0.781,399	.38	0.071,399	.79	0.471,399	.49	0.321,399	.57
Antipsychotic medication	1.472,399	.23	0.952,399	.39	1.572,399	.21	1.692,399	.19	2.432,399	.09	1.702,399	.19
Use of anticholinergics	0.411,399	.52	1.421,399	.23	2.951,399	.09	3.271,399	.07	3.061,399	.08	3.581,399	.06

Note: Abbreviations are explained in the first footnote to table 1. Statistics for these general linear models are as follows: verbal memory: F _{10, 399} = 5.97, P < .001, R ² = .13; vigilance: F _{10, 399} = 4.10, P < .001, R ² = .10; processing speed: F _{10, 399} = 7.03, P < .001, R ² = .20; reasoning: F _{10, 399} = 6.99, P < .001, R ² = .19; working memory: F _{10, 399} = 4.64, P < .001, R ² = .13; and summary score: F _{10, 399} = 8.22, P < .001, R ² = .22. Statistically significant effects with P value of <.05 were described in bold.

Table 4.

Neurocognitive Scores Stratified by Dopamine D₂ Occupancy

D2 Occupancy Level	D2 Occupancy Range (%)	z Score (Mean ± SD)
		Verbal Memory	Vigilance	Processing Speed	Reasoning	Working Memory	Summary Score
Low (n = 102)	15.5–62.7	0.084 ± 1.077	0.174 ± 0.964	0.159 ± 0.958	0.311 ± 0.866a	0.293 ± 0.787	0.275 ± 0.920
Slightly low (n = 102)	62.7–71.8	0.188 ± 1.015	0.306 ± 0.992b	0.223 ± 0.889	0.259 ± 0.899	0.229 ± 0.880	0.323 ± 0.937
Slightly high (n = 103)	71.9–77.2	0.180 ± 0.865	0.385 ± 0.951c	0.167 ± 0.913	0.285 ± 0.853	0.309 ± 0.736	0.354 ± 0.839d
High (n = 103)	77.2–85.8	0.025 ± 0.990	−0.110 ± 0.934	−0.085 ± 0.903	−0.013 ± 0.949	0.095 ± 1.017	−0.008 ± 0.951

Note: Significant differences were found in vigilance score, reasoning score, and summary score (F _{3, 406} = 5.21, P = .002; F _{3, 406} = 2.90, P = .04; F _{3, 406} = 3.36, P = .02, respectively) by the one-way ANOVA.

^a P =.049 by the Turkey-Kramer HSD (honestly significant difference), vs high D₂ group.

^b P = .012 by the Turkey-Kramer HSD, vs high D₂ group.

^c P = .002 by the Turkey-Kramer HSD, vs high D₂ group.

^d P = .025 by the Turkey-Kramer HSD, vs high D₂ group.

Fig. 1.

Neurocognitive domain scores and estimated dopamine D₂ receptor occupancy. Pearson’s correlation analysis did not find any statistically significant linear correlation between neurocognitive scores and dopamine D₂ occupancy. Note that trend lines are shown only for descriptive purposes; refer to table 4 for detailed comparison to test the nonlinear relationship between neurocognitive scores and dopamine D₂ occupancy.

Since the mean differences across the D₂ groups may account for other factors that were entered into the model, especially age, another model was generated, using the data from 327 subjects aged less than 50 (see online s upplementary table 1) Estimated dopamine D₂ receptor occupancy levels showed significant association with the vigilance, processing speed, working memory, and the summary scores. Again, a nonlinear relationship was found between mean values of these neurocognitive scores and dopamine D₂ receptor occupancy levels (see online supplementary table 2).

Discussion

To our knowledge, this is the largest study to investigate neurocognitive function in relation to dopamine D₂ receptor occupancy with antipsychotic drugs in patients with schizophrenia. The results demonstrated a nonlinear relationship between prescribed antipsychotic doses and overall neurocognitive function and vigilance; they were especially impaired in subjects who showed D₂ receptor occupancy level of >77%. Thus, our hypothesis regarding the association of very high D₂ receptor occupancy with impaired cognitive functions was supported, while there was no evidence that supported the relationship between very low dopamine D₂ occupancy and cognitive dysfunction.

The association between a high dopamine D₂ receptor occupancy of >77% by antipsychotic drugs and impaired neurocognitive function was observed in this study, which is in line with the findings in the literature. Patients with schizophrenia who were treated with high doses of antipsychotic drugs (ie, 1134 ± 840 mg/day of chlorpromazine equivalents [mean ± SD]) showed significantly poorer performance than those with standard doses (ie, 473 ± 268 mg/day of chlorpromazine equivalents [mean ± SD]) on visual memory, delayed recall, performance IQ, and executive function.⁴ Similarly, one clinical positron emission tomography (PET) study demonstrated that the attentional deficits were observed above 74% blockade of dopamine D₂ receptor by risperidone in patients with schizophrenia.¹² In animal experiments, the 5-choice serial reaction time task (5CSRTT) provides substantial validity as a direct measure of attention and bears a good analogy to the continuous performance test²⁹ that was adopted in the CATIE trial to measure vigilance. The neurochemical lesion of nucleus accumbens septi (NAS) induced by intracerebral infusions of neurotoxin 6-hydroxydopamine (6-OHDA) in rats produced an 87% depletion of dopamine in the NAS, which attenuated both speed and impulsivity of responding on the 5CSRTT.³⁰ In human cortex, long-term potentiation (LTP) of synaptic efficacy is considered as a fundamental mechanism of learning and memory. A single oral dose of dopamine antagonist, haloperidol, depressed significantly the paired associative stimulation–induced LTP-like plasticity at the systems level of human cortex in 8 healthy subjects.³¹ Thus, excessive blockade of dopamine D₂ receptor by antipsychotic drugs or relative paucity of dopamine may have detrimental effects on neurocognitive function in patients with schizophrenia.

On the other hand, dopamine D₂ receptor blockade with antipsychotic drugs has also been shown to improve cognitive functions. Keefe et al⁵ conducted a meta-analysis of 15 studies and found that atypical antipsychotic drugs improved attention, executive function, working memory, visuospatial analysis, verbal fluency, and digit symbol substitution in patients with schizophrenia. Similarly, in a systemic review of 20 previous reports that examined changes in neurocognitive function followed by the treatment with atypical antipsychotic drugs in patients with schizophrenia, significant improvements in overall neurocognitive function were observed⁷; especially, effects for domains related with vigilance were consistently large. Combined with the observation that an excessive blockade of dopamine D₂ receptor by antipsychotic drugs was associated with impaired neurocognitive function as described above, these findings may suggest that a moderate degree of dopamine D₂ receptor blockade may provide amelioration in neurocognitive dysfunction in patients with schizophrenia.

The efficacy and tolerability of all available dopamine antagonist antipsychotics have been linked to their binding to dopamine D₂ receptors.³² For most antipsychotic drugs, PET studies have suggested the presence of a therapeutic window of striatal dopamine D₂ receptor occupancy (65%–80%) in younger patients,^33,34 with extrapyramidal side effects more likely at more than 80% dopamine D₂ receptor occupancy.^34,35 Our recent pooled analysis also supports the presence of the therapeutic window in young adults with schizophrenia.²² Interestingly, the results of this study may also endorse the upper limit of this established therapeutic window of dopamine D₂ receptor occupancy in terms of neurocognitive function. If the observations in the present study are confirmed in future studies with a specific focus on the causal relationship between dopamine D₂ receptor blockade with antipsychotics and cognitive function, the therapeutic window could also be used to predict the therapeutic dose range of antipsychotic drugs to maximize therapeutic effects and minimize detrimental effects of antipsychotic drugs from a perspective of neurocognitive function.

The impairment of processing speed has been consistently observed in patients with schizophrenia.^36,37 A recent meta-analysis has shown that patients with schizophrenia presented the most profound impairment on a digit symbol coding test that measures processing speed among various common neuropsychological measures.³⁸ A reduced processing speed is known to be observed in patients with schizophrenia prior to the onset of the illness and is associated with clinical and functional outcomes.^36,37 These data suggest that the decline of processing speed represents an important behavioral marker of the pathophysiology of schizophrenia. The significant association between the processing speed and the PANSS total score that we observed in this study is compatible with these findings.

In the present study, dopamine D₂ receptor occupancy was not measured, using brain imaging techniques, but estimated with our recently developed model; the prediction performance of the model has been shown to be reliable.¹⁶ Furthermore, in theory, dopamine D₂ receptor occupancy is not determined by plasma drug concentrations but by free and unbound concentrations. We therefore compared the estimated dopamine D₂ receptor occupancy levels between the model that we used in our study¹⁶ and the method using the following formula: f = C/(K + C), where f is the fraction of D₂ receptors occupied, where C is free concentration in the plasma samples, and where K is the drug dissociation constant at D₂. This additional analysis was not performed for risperidone because of the following reason. It was possible to estimate dopamine D₂ receptor occupancy for either of risperidone or 9-hydroxyrisperidone; however, it was unclear how to estimate the combined effects of risperidone plus 9-hydroxyrisperidone in terms of dopamine D₂ receptor occupancy. In addition, the percentage of free and unbound concentration of ziprasidone has been reported in the literature; however, it has consistently been reported as “greater than 99%.”³⁹ This lack of an exact value also discouraged us from including ziprasidone in the additional analysis. Thus, this additional analysis was performed, using the data from subjects on olanzapine (K = 2.3 ng/ml⁴⁰; percentage of free and unbound drug concentration = 7%⁴¹). Although the values obtained from those 2 models were found to be closely correlated (Pearson’s r = .98 and P < .0001), the values obtained from the formula were generally lower by 10%–20% than those from the Uchida prediction model. Because we did not actually measure dopamine D₂ receptor occupancy levels in the present study, it was impossible to compare the precision of prediction performance between those 2 methods. Therefore, we decided to perform an additional analysis, using the dataset of measured plasma drug concentrations and corresponding dopamine D₂ receptor occupancy levels that were used when the Uchida prediction model was developed. We estimated dopamine D₂ receptor occupancy levels (n = 42), using the formula, and compared the results with those obtained from the prediction model by Uchida et al. As shown in online supplementary figure 1, the values from the model by Uchida et al look more comparable to the measured values than those from the formula; in fact, the mean prediction error (%) and squared mean prediction error (%)⁴² also corroborate this finding (−0.1 [95% CI: −1.2–1.2] vs 23.1 [95% CI: 16.8–29.4]; 4.6 [95% CI: 3.5–5.8] vs 24.5 [95% CI: 19.3–29.7], respectively). However, this finding does not always mean that the prediction model by Uchida et al that was used in this study is superior to the formula that has a robust theoretical basis. In theory, again, dopamine D₂ receptor occupancy is dependent upon free and unbound drug concentrations. This modeling issue clearly warrants further investigations. Moreover, the region of interest in 97% of the PET data used for the development of this prediction model by Uchida et al. was striatum. Although a potential difference in D₂ receptor blockade by antipsychotics between striatal and extrastriatal regions could be attributable to the methodology used,⁴³ the findings in the present study need to be replicated in future investigations, using radiotracers that can assess extrastriatal dopamine receptors. In addition, dopamine D₂ receptor occupancy levels were calculated with an unconstrained one-site occupancy model in this study. Therefore, for example, 77% D₂ occupancy corresponds to 81%, 79%, and 83% that were estimated with a maximum constrained occupancy of 100% for risperidone, olanzapine, and ziprasidone, respectively.¹⁶ Even though the differences are small, it is important to remember that the estimated D₂ receptor occupancy levels are not absolute values.

Our focus on the dopaminergic system is not intended to insist that neurocognitive function in schizophrenia is solely related to effects in the dopaminergic system. Clearly, this relationship is far more complex, and we certainly do not exclude the involvement of other systems. For example, manipulations of the central serotoninergic system can produce specific changes in cognitive functioning.^29,44,45 We therefore estimated 5-HT₂ receptor occupancy levels, using the ED₅₀ values reported in the literature.^46,47 5-HT₂ receptors were almost saturated in a majority of the subjects; in fact, 71.5% of the subjects showed >90% occupancy and 92.4% showed >80% occupancy, which may suggest a limited impact of the serotonergic system on cognition in this study. However, potential confounding effects through this system clearly warrant further investigations. In addition, anticholinergic effects have been reported to impair cognitive function both globally⁴⁸ as well as in specific domains, including memory⁴⁹ and executive functioning.^50,51 The association between anticholinergic activity and cognitive performance are also strongly supported by the studies that measured serum anticholinergic activity.^52,53 Overall, anticholinergic burden due to all prescribed medications was not evaluated in the present study although the use of antiparkinsonian anticholinergic drugs was taken into consideration, which has to be acknowledged in light of the exposure-dependent detrimental effects of those medications on cognition.⁴⁸

Several other limitations qualify our conclusions. First, mean values of predicted peak and trough dopamine D₂ receptor occupancy levels on the day of cognitive assessment were used in this analysis; however, they did not always represent the levels at the time of neurocognitive assessments. Second, subjects were divided into 4 groups according to their estimated dopamine D₂ receptor occupancy levels in order to test the hypothesis; however, this classification could be considered arbitrary, In our hypothesis, we expected that effects of dopamine D₂ receptor occupancy on cognitive functions would not be simply linear but inverted U-shaped. To test this hypothesis while taking other clinical and demographic variables into consideration, we decided to divide subjects to 4 groups based on their dopamine D₂ receptor occupancy levels and then examined the effects of dopamine D₂ receptor occupancy on cognition, using a multivariate general linear model. Still, it would have been ideal to handle dopamine D₂ receptor occupancy as a continuous variable from a statistical perspective. Third, extrapyramidal symptoms are also expected to affect cognition; we therefore included the SAS mean score in the model but failed to find any statistically significant effect on cognition in this study. However, this does not always exclude any possibility of potential effects of extrapyramidal symptoms on cognition. Similarly, sedative effects of antipsychotic drugs could also affect cognition, which was not taken into consideration in the present study. These limitations clearly emphasize the need for prospective studies with more comprehensive assessments. Finally, it would have been ideal to include perphenazine that demonstrated comparable clinical effects to newer antipsychotic drugs although any model for the prediction of dopamine D₂ receptor occupancy for this drug is not available.

In conclusion, the degree of dopamine D₂ receptor occupancy levels estimated from plasma concentrations of antipsychotic drugs were associated with overall neurocognitive function and vigilance in patients with schizophrenia. This study shows that D₂ occupancy above 80% not only increases the risk for extrapyramidal side effects as consistently reported in the literature but also increases the risk for cognitive impairment.

Supplementary Material

Supplementary material is available at http://schizophreniabulletin.oxfordjournals.org.

Funding

Data used in the preparation of this article were obtained from the limited access datasets distributed from the National Institutes of Health (NIH)-supported “Clinical Antipsychotic Trials of Intervention Effectiveness in Schizophrenia” (CATIE-Sz). This is a multisite clinical trial of persons with schizophrenia comparing the effectiveness of randomly assigned medication treatment. The study was supported by National Institute of Mental Health (NIMH) Contract #N01MH90001 to the University of North Carolina at Chapel Hill. The ClinicalTrials.gov identifier is NCT00014001. The version of the dataset used was 1.0. This study was also supported by grant R01MH064173 from the NIMH and was ancillary to CATIE, N01MH90001, from the NIMH. The present analysis was also funded by Grant-in-Aid for Young Scientists-B from the Ministry of Education, Culture, Sports, Science, and Technology (HU); this organization had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.