Psychiatric Genetics 2011, 21: a Division of Psychological Medicine, Institute of Psychiatry, King s College

Transcription

1 98 Original article COMT (Val 158/108 Met) genotype moderates the impact of antipsychotic medication on verbal IQ in twins with schizophrenia Irene Rebollo-Mesa a,b, Marco Picchioni a,c, Madiha Shaikh a, Elvira Bramon a, Robin Murray a and Timothea Toulopoulou a Objectives In this study, we aimed to assess the moderating effects of the catechol-o-methyl transferase (COMT) (Val 158/108 Met) genotype on antipsychotic medication-induced changes in the cognitive performance of patients with chronic schizophrenia. Methods The sample consisted of 85 monozygotic and 53 dizygotic twin pairs, of varying concordance for schizophrenia, and healthy control twins. Cognitive ability was measured using the Wechsler Adult Intelligence Scale-third edition. We used structural equation modelling to estimate main and interaction effects of the COMT status and antipsychotic medication dose on verbal intelligence quotient (VIQ) and performance intelligence quotient scores. Results There was no evidence of a main or interaction effect of the COMT status or chlorpromazine equivalent dose on the performance intelligence quotient. There were no main effects of COMT or chlorpromazine equivalent dose on VIQ; however, there was evidence of a statistically significant interaction (P < 0.01) between the COMT and chlorpromazine equivalents on VIQ. The VIQ performance of val/val individuals was significantly lower with increasing antipsychotic medication dose, up to 12 intelligence quotient points lower than met carriers treated with medication. In the absence of medication, the three genotypes did not significantly differ, whereas at the highest doses (1500), the val/val homozygotes and Met carriers differed by more than one standard deviation. Conclusion Our results show that the verbal abilities of val homozygotes of the COMT gene are cognitively impaired by higher doses of antipsychotic medication. This association is reversed in Met carriers. These data are consistent with an earlier study that found evidence of moderating effects of antipsychotic medication on N-back and verbal fluency tasks. Psychiatr Genet 21: c 2011 Wolters Kluwer Health Lippincott Williams & Wilkins. Psychiatric Genetics 2011, 21: Keywords: cognition, catechol-o-methyl transferase, endophenotypes, intelligence quotient, schizophrenia a Division of Psychological Medicine, Institute of Psychiatry, King s College London, b MRC Centre for Transplantation, King s College London, London and c St Andrew s Academic, Institute of Psychiatry, Northampton, UK Correspondence to Irene Rebollo-Mesa, PhD, MRC Centre for Transplantation, King s College London, 5th Floor, Tower Wing, Guy s Hospital, Great Maze Pond, London SE19RT, UK Tel: ; fax: ; irene.rebollo_mesa@kcl.ac.uk Received 14 May 2010 Revised 5 November 2010 Accepted 15 November 2010 Introduction Neurocognitive dysfunction is a core feature of schizophrenia (Forbes et al., 2009). It is unlikely to be a unitary phenomenon and its causes are complex, although genes are likely to contribute. For example, there is some evidence supporting a specific association between cognitive performance in psychosis and the catechol-o-methyl transferase (COMT) (Val 158/108 Met) genotype (Egan et al., 2001; Goldberg et al., 2003; Barnett et al., 2007, 2009). However, the studies are inconsistent, failing to replicate across different cognitive domains (Goldberg et al., 2003; Dickerson et al., 2007; Mata et al., 2008; Barnett et al., 2008). The most consistent results are seen for working memory and executive processing, whereas the outcomes for intelligence quotient (IQ), verbal fluency and processing speed are less robust. Evidence has also emerged suggesting that the COMT genotype can modulate an individual s sensitivity to aetiologically relevant environmental stressors in schizophrenia. For example, Caspi et al. (2005) have shown an increased vulnerability to develop schizophreniform disorder in val allele carriers, but not in Met homozygotes, after adolescent cannabis consumption. Similarly, Van Winkel et al. (2008) reported that Met/Met homozygotes were prone to develop greater negative effect and more psychotic symptoms after stressful events, than those with the val allele, suggesting that the relationship between COMT genotype and psychosis is complex. Antipsychotic medication can be thought of as an environmental factor, and in particular, one that patients can differ in their responsiveness, in part driven by their COMT (Val 158/108 Met) polymorphism status. To date, three pharmacogenetic studies have been carried out that support a moderating effect of COMT status on the cognitive impact of antipsychotic medication. In the first study, Bertolino et al. (2004) found that Met/Met c 2011 Wolters Kluwer Health Lippincott Williams & Wilkins DOI: /YPG.0b013e a7

2 COMT, antipsychotic and IQ Rebollo-Mesa et al. 99 individuals improved significantly more in N-back performance than other genotypes after 8 weeks of olanzapine treatment. Weickert et al. (2004) replicated Bertolino s results on the N-back task but not on the Wisconsin Card Sorting Test, nor for IQ or verbal fluency performance. Subsequently, Woodward et al. (2007), in a larger study, followed patients who were treated with clozapine for 6 months. They found that verbal fluency performance improved in Met carriers but not in val homozygotes. The interaction pattern did not replicate in any other cognitive tasks, that is, attention or working memory. The three pharmacological studies mentioned above offer the methodological advantages of being able to recruit well defined and relatively homogeneous patient groups, with defined and restricted prestudy pharmacological histories. However, by virtue of exactly the same considerations, they are universally small in sample size and limited in terms of follow-up time. Thus, although they offer excellent internal validity, the generalizability of their results to the wider patient population with schizophrenia is necessarily questionable. In a more naturalistic observational study of chronic patients with psychotic disorders, Mata et al. (2006) reported a positive association between the number of COMT val alleles and cognitive deterioration after antipsychotic medication treatment. In particular, homozygotes for the val allele, taking second-generation antipsychotic medications, deteriorated cognitively more than twice as much as the Met homozygotes. To summarize, the four published pharmacogenetic studies suggest a modulating role for the Val 158/108 Met genotype on the effects of antipsychotic medication on the cognitive performance of patients with schizophrenia. There are experimental inconsistencies among the studies, principally with regard to the specific cognitive abilities affected, and to the nature or direction of any effects, that is, whether they represent cognitive improvement or deterioration. In this study, as part of the Maudsley Twin Study of Schizophrenia, we chose to explore the moderating effects of the COMT (Val 158/108 Met) genotype on the cognitive performance of patients in a naturalistic setting. Our design offers a number of experimental advantages. (1) This study includes a large national sample of patients and co-twins with schizophrenia. (2) The recruitment strategy and inclusion criteria meant that each patient was in a stable phase of their illness. (3) We robustly assessed the antipsychotic dosing schedules. (4) The sample included both affected patients and unaffected participants, ensuring an accurate estimation of the main genotype effects on the phenotype, before considering medication effects. (5) Using twins allowed us to control the possible population stratification effects. (6) We explored the specificity of any cognitive effect by examining multiple subscales of the Wechsler Adult Intelligence Scale-third edition (WAIS-III), including verbal IQ (VIQ) and performance IQ (PIQ), rather than simply the full-scale IQ score. On the basis of earlier studies, we predicted that we would find the main effect of the COMT (Val 158/108 Met) genotype on cognition, more so, for verbal tasks that involve executive functioning (Barnett et al., 2009), with an advantage for Met carriers. In addition, we expected the genotype differences in cognitive performance to be exaggerated by the doses of antipsychotic medication. Given the similarity of this study to that of Mata et al. (2006) study, we expected cognitive performance to be more impaired at higher medication doses. Materials and methods Participants Twin probands with schizophrenia were recruited nationally throughout the United Kingdom, referred by their treating psychiatrist. In the United Kingdom, the National Health Service is a comprehensive national treatment service funded centrally and free at the point of delivery for all aspects of health care. By virtue of its financial and organizational structure, it is hugely inclusive and as a consequence, treats most patients with schizophrenia with care, making it a highly representative recruitment source. Control twins were recruited from the Institute of Psychiatry Volunteer Twin Register. Exclusion criteria for all the groups were age under 18 years, a history of a neurological disorder or a systemic illness with known neurological complications, a history of significant head injury associated with loss of consciousness for more than 1 min and a current harmful or aetiologically relevant substance use or dependence (defined as within the last 12 months). The study was approved by the UK Multicenter Research Ethics Committee and all the participants gave written informed consent before participating. Clinical assessment We established Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) diagnoses using either the Schedule for Affective Disorders and Schizophrenia Lifetime Version (Spiltzer and Endicott, 1978), supplemented by additional information or the Structured Clinical Interview for DSM-IV (First et al., 1997). Psychotic symptoms in the probands, in the month before testing, were assessed using the Scales for the Assessment of Positive Symptoms (SAPS) (Andreasen, 1984b) and Scales for the Assessment of Negative Symptoms (SANS) (Andreasen, 1984a). Zygosity was determined by the assessment of 12 highly polymorphic microsatellite markers and a standardized twin likeness questionnaire. Medication status was recorded at the time of assessment, and age at first contact with psychiatric services was ascertained to serve as a proxy index of age at the

3 100 Psychiatric Genetics 2011, Vol 21 No 2 onset of illness. We converted the doses of prescribed medication at assessment into chlorpromazine equivalents (CPZEQ) according to an algorithm based on each drug, D2-receptor blockade and pharmacokinetic properties using the best available evidence (Rey et al., 1989; Woods, 2003; American Psychiatric Association Practice Guidelines, 2004; Bezchlinbnyk-Butler and Jeffries, 2007; Bazire, 2007; Taylor et al., 2009). In concordant pairs, both members fulfilled the criteria for DSM-IV schizophrenia or schizoaffective disorder. In discordant pairs, one member was diagnosed with schizophrenia, whereas their co-twin was free of any psychotic illness. In control pairs, both members were free of personal or family history of psychosis or a schizophrenia spectrum disorder. The probability that any of the discordant pairs would become concordant for schizophrenia in the future was low, given that a mean standard deviation of (10.19) years had elapsed since the onset of illness in the patients. Neurocognitive assessments Participants completed the full UK version of the WAIS-III (Wechsler, 1997). This consists of 14 subtests, each designed to engage a distinct cognitive domain. In addition to the full scale IQ, the subtests of the WAIS-III can be reclassified providing indices of four more refined cognitive domains: (i) working memory, comprising the composite score of arithmetic, digit span and letternumber sequencing subtests; (ii) processing speed, the composite score of digit symbol and symbol search subtests; (iii) perceptual organization, the composite score of picture completion, block design and matrix reasoning subtests and (iv) verbal comprehension, the composite score of vocabulary, similarities and information subtests (Deary, 2001). Genotyping The COMT Val 158/108 Met polymorphism was genotyped using a TaqMan SNP Genotyping Assay (Applied Biosystems, California, USA; Assay ID C_ _50, Lot Number ). Statistical analyses Generalized estimating equations We used generalized estimating equations (GEE), in SPSS 15.0 (Chicago, Illinois, USA), to compare the three genotype groups and the affected (patients with schizophrenia) and unaffected individuals (participants), with respect to their demographical characteristics. We also used GEE to study the relationship between severity of symptoms (SAPS and SANS), antipsychotic dose (CPZEQ) and the WAIS-III VIQ and PIQ. The GEE procedure uses a robust covariance estimator ensuring an accurate estimate of standard errors for nonindependent data. Structural equation modelling To study the effects of genotype and medication status on cognitive scores, while controlling for the effects of covariates, we used structural equation modelling, within Van den Oord and Snieder s (2002) general framework. They proposed that structural equation modelling could include measured genotypes in statistical models to allow the complex interplay between measured environmental factors and measured genotypes to be studied. This model has been described and published in detail by Van den Oord and Snieder s (2002), we will limit our model description to the specifics of this study. Figure 1 shows the path diagram of the model used to analyse the four domains of the WAIS-III. Fig. 1 Twin 1 Twin 2 Diag 1 Age 1 Sex 1 Diag 2 Age 2 Sex 2 G1 1 G1 2 CPZEQ 1 VIQ 1 PIQ 1 VIQ 2 PIQ 2 CPZEQ 2 G2 1 G VCI 1 WMI 1 POI 1 PSI 1 VCI 2 WMI 2 POI 2 PSI 2 Path diagram of full model for a twin pair. Diag, Diagnosis; CPZEQ, chlorpromazine equivalents; PIQ, performance intelligence quotient; POI, perceptual organization index; PSI, processing speed index; VCI, verbal comprehension index; VIQ, verbal intelligence quotient; WMI, working memory index.

4 COMT, antipsychotic and IQ Rebollo-Mesa et al. 101 Measurement model: We specified a factor model with two latent factors, VIQ and PIQ, after the theoretical cognitive structureofthewais-iii.thus,wedefinedviqandpiq as variations that are common to the two domains that represent each, excluding measurement error and task specificity. This measurement model is estimated for both members of the twin pair simultaneously, and all the parameters in the model are constrained to be equal across members of a pair. Given that each factor has only two indicators, the factor loadings on the observed variables are fixed to one, whereas the factor variance contains the common variance between the two. The two latent factors, VIQ and PIQ, are allowed to correlate. To account for the dependency between their observations, all the variables in the model are correlated freely between the members of the twin pair. Regression model: Subsequently, a regression model was specified on the means model of the two latent factors, VIQ and PIQ. The predictors shown in grey in Fig. 1 represent covariates whose effects we want to control, for example, affected (schizophrenia) status, age and sex. The predictors shown in black represent those independent variables whose effects we want to study, namely COMT genotype (G1 and G2) and CPZEQ. From those, the dashed paths represent the main effects of genotype and CPZEQ on the latent factors, whereas the continuous paths represent the effects of the interaction between them. According to this model, the model for the mean, for example, VIQ is equal to: MeanVIQ ¼ a þ b c1 AfS þ b c2 Age þ b c3 Gender þ a 1 G1 þ a 2 G2 þ b cpz CPZEQ þ b gxeð1þ CPZEQ G1 þ b gxeð2þ CPZEQ G2 þ e where a is the intercept, and b c1 and b c3 are the effects of the covariates. G1 and G2 are dummy variables of the measured genotype coded to represent homozygous and heterozygous effects, respectively; a 1 is the additive effect of the genotype, a 2 is the dominant effect and b cpz is the main effect of CPZEQ. b gxe(1) and b gxe(2) are the interaction effects between CPZEQ and the additive and dominant effects of the gene, respectively. Finally, e is the residual variance, not explained by any of the other terms. We estimated the same model for both VIQ and PIQ. We tested the statistical significance of each of these effects by fixing them to zero and comparing the goodness of fit of the constrained model with the less restrictive model. Mx (Mx, Richmond, Virginia, USA) (a software package; Neale et al., 2006) was used to estimate the models with full information maximum likelihood estimation for the raw data. Nested models were compared with a likelihood ratio test, which has a w 2 distribution with the number of degrees of freedom equal to the difference in the degrees of freedom between the models being compared. It is recognized in genetic association studies that unnoticed subgroups within the sample might themselves differ in allele frequencies and prevalence of the disease and can lead to an increase of both false positives and false negatives. The twin design in this study allowed us to address this possible effect of population stratification by incorporating the effects because of family clustering in the regression equation. We classified the twin pairs as a function of their genotype combinations (F1, F2, F3y). If the genotypic effects were an artefact of population stratification, they should be corrected under this model. For a more detailed description of this strategy see Van den Oord and Snieder (2002). Results Table 1 describes the demographic characteristics of the sample by the genotype and the affected (for schizophrenia) status. The total sample consisted of 85 monozygotic (MZ) twin pairs [53 concordant unaffected (controls), 19 concordant affected and 13 discordant] and 53 dizygotic (DZ) pairs (39 concordant unaffected and 14 discordant). The twin correlations for VIQ were for MZ twins and for DZ twins; for PIQ the correlations were for MZ twins and for DZ twins, consistent with a strong additive genetic effect on both types of IQ. Approximately 62.4% of the patients Table 1 Description of the sample Unaffected Affected val/val val/met Met/Met val/val val/met Met/Met Disease status P value Genotype P value N Percentage of male * Percentage of Caucasian Mean age Mean CPZEQ Mean years of education Mean age of onset CPZEQ, chlorpromazine equivalence. Disease status and genotype *P values correspond to the Wald w 2 statistic from generalized estimating equations, using disease status and genotype as predictors of each descriptive variable.

5 102 Psychiatric Genetics 2011, Vol 21 No 2 were taking second-generation antipsychotics, whereas the remaining 37.6% were prescribed first-generation antipsychotics. The genotype frequencies were in Hardy Weinberg equilibrium in the entire sample [w 2 (1) = , P =0.792],andinbothpatients[twin1: w 2 (1) = , P = 0.490; twin 2: w 2 (1) = , P = 0.850] and controls [twin 1: w 2 (1) = 0.001, P = 0.974; twin 2: w 2 (1) = , P = 0.867]. The genotypes did not differ on any of the demographic characteristics, sex, ethnicity, age, CPZEQ, education or age at onset. However, there was an excess of male patients in the schizophrenia group. Sex was included as a covariate in subsequent analysis. Using GEE, we found a statistically significant positive association between CZPEQ and positive symptoms (SAPS, P < 0.01), but neither SAPS, SANS nor CPZEQ were significantly associated with VIQ or PIQ (all P values > 0.05). Table 2 shows the results of the model fitting procedure, in which we tested the statistical significance of each of the regression coefficients. There were no effects of sex on either VIQ or PIQ. There was a statistically significant effect of age on PIQ, with a positive coefficient indicating slightly better performance in the older patients. Illness status had a large and significant effect on both VIQ and PIQ, with a difference of almost 11 VIQ and 17 PIQ points between patients and participants. There was neither any evidence of a significant interaction between COMT genotype and CPZEQ nor any main effect of COMT genotype and CPZEQ on PIQ. There was, however, a significant interaction between CPZEQ and the COMT genotype on VIQ. The confidence intervals (CIs) of the final parameter estimates show that both the additive and dominant components were necessary to define the shape of the interaction. Figure 2 shows the parameter estimates included in the final model. Those effects that were not significant in the model fitting procedure were removed. The main effects of genotype and CPZEQ were left in the final model, because of the presence of the interaction. However, the CIs of the final parameter estimates show that neither differed significantly from zero (a 1 = 1.446, 95% CI = ; a 2 = 3.179, 95% CI = ; b cpz = 0.003, 95% CI = ). The latent factors representing VIQ and PIQ were highly correlated as expected. Figure 3 shows the pattern implied by the interaction terms estimated in the model for the two scales of VIQ. First, the dot plots for unmedicated patients show a significantly better performance in the unaffected participants compared with patients. The lines show how patients performance differs with genotype, as a function of the CPZEQ dose they were taking. The results suggest that larger doses of antipsychotic medication are associated with a greater impairment of verbal abilities in val allele homozygotes, whereas Met carriers tend to experience a slight improvement in performance. Further analyses using GEE showed that the interaction between genotype and CPZEQ differs significantly Fig. 2 CPZEQ G1 G Diag VIQ Age Sex PIQ VCI WMI POI PSI Path diagram of final model. Diag, Diagnosis; CPZEQ, chlorpromazine equivalents; PIQ, performance intelligence quotient; POI, perceptual organization index; PSI, processing speed index; VCI, verbal comprehension index; VIQ, verbal intelligence quotient. Table 2 Testing the statistical significance of effects through model fitting Model -2LL D.f. Versus Change in w 2 Change in d.f. P Final parameter estimate 95% CI 1 Full Sex effects = NI 3 Age effects VIQ = NI 4 Age effects PIQ = (0.078, 0.336) 5 Disease status effects VIQ = ( 15.83, 6.15) 6 Disease status effects PIQ = ( 20.61, 13.11) 7 COMT by CPZEQ on PIQ = NI 8 COMT by CPZEQ on VIQ = g1 = g2 = ( 0.009, 0.001) (0.003, 0.014) 9 CPZEQ on PIQ = NI 10 COMT G1 on PIQ = NI 11 COMT G2 on VIQ = NI CI, confidence interval; COMT, catechol-o-methyl transferase; CPZEQ, chlorpromazine equivalence; d.f., degrees of freedom; IQ, intelligence quotient; NI, not included in final model; PIQ, performance IQ; VIQ, verbal IQ.

6 COMT, antipsychotic and IQ Rebollo-Mesa et al. 103 Fig. 3 Verbal comprehension Working memory Verbal comprehension index Patient val/val Patient val/met Patient met/met Control val/val Control val/met Control met/met Working memory index Patient val/val Patient val/met Patient met/met Control val/val Control val/met Control met/met Chlorpromazine equivalents Interaction between the val/met catechol-o-methyl transferase genotype and antipsychotic medication for verbal intelligence quotient indexes from the Wechsler Adult Intelligence Scale-third edition. between the val/val and the Met/Met genotypes (P = 0.023), but not between the Met/Met and Val/Met genotypes (P = 0.329). Discussion The results of this study do not support a main effect of the COMT (Val 158/108 Met) genotype on either VIQ or PIQ when measured using the WAIS-III. This interpretation is consistent with earlier studies that found significant main effects of COMT on working memory and executive functioning, but no association with more general cognitive indices such as the Repeated Battery for the Assessment of Neuropsychological Status (Dickerson et al., 2007) or the WAIS-R (Egan et al., 2001; Goldberg et al., 2003). A recent review (Lewandowski, 2007) speculated that more specific cognitive measures might show the greatest association with the COMT genotype, in contrast to general cognitive measures, in which compensatory strategies could come into play. Interestingly, our study confirmed a modulatory role for COMT (Val 158/108 Met) polymorphism on the effects of antipsychotic medication on VIQ in patients with schizophrenia, but not PIQ. The VIQ domain of the WAIS-III is composed of working memory and verbal comprehension indices. These tap specific cognitive abilities overlaping those reported in earlier studies by Bertolino et al. (2004), Woodward et al. (2007) and Weickert et al. (2004). It is worth highlighting that we also found evidence of lower cognitive performance in val homozygotes, similar to the results obtained by Mata et al. (2006), rather than the earlier studies that seemed to detect evidence of better performance in val carriers. This discrepancy could be seen in the context of the earlier studies failure to take antipsychotic dose into account. We have shown that this determines the degree with which cognitive performance is negatively affected in val/val carriers. Furthermore, there may be a temporally dynamic component to these effects. Woodward et al. (2007) and Bertolino et al. (2004) found that patients experienced an initial short-term improvement in cognitive performance through amelioration of the acute psychotic symptoms, distress and concentration, before a secondary, delayed but more profound, and sustained change in prefrontal brain function. We suggest that a longer follow-up of those same patients, subjected to larger doses or the cumulative effects of antipsychotic medication, might have shown decay in their verbal/ memory performance, which is consistent with the findings of Mata et al. (2006) and our own study. Our finding that individuals cognitive response to antipsychotic medication was determined in part by their genotype is consistent with other researchers, who suggest that Met carriers respond better, from a symptom perspective, to antipsychotic treatment (Arranz and de Leon, 2007). Furthermore, our results are compatible with the study by Mattay et al. (2003), in which patients with the val/val genotype experienced enhanced cortical efficiency during a working memory task after amphetamine stimulation, whereas Met/Met patients cortical efficiency deteriorated, but only under high-working memory load. Similarly, Barnett et al. (2009) recently reported a diplotype within the COMT allele, which shows a quadratic relationship with working memory. The genotypes associated with intermediate levels of dopamine availability were associated with optimal performance. Overall, these studies support the hypothesis that dopamine influences prefrontal cortex function in a U-shaped dose-dependent curve. Val/val individuals

7 104 Psychiatric Genetics 2011, Vol 21 No 2 express low levels of signal at baseline, whereas Met/Met individuals depart from optimal levels near the peak of prefrontal cortical function (Mattay et al., 2003). This theory could explain why antipsychotic medication can have a detrimental effect on prefrontal cortical function in val/val individuals, pushing them even further from the optimal levels of dopamine signalling. This study was limited by its relatively small sample size and the apparent, although nonsignificant, overrepresentation of noncaucasians in val homozygous patients (see Table 1). To exclude an artefact of admixture, we repeated the model fitting analyses only in Caucasians. The results replicated those of the entire sample, with significant disease effects on PIQ and VIQ, and a significant interaction between COMT and CPZEQ, in the same direction and magnitude (P < 0.001, b g1xcpzeq = 0.006, b g2 = 0.009) (further details available from the author on request). Our failure to find the main effect of COMT on IQ, as in Barnett et al s study (2008) (d = 0.06) may be due to lack of statistical power or may be by being masked by the inclusion of the interaction term. This study has shown that val homozygotes for the COMT (Val 158/108 Met) polymorphism experience significant impairment of VIQ ability on exposure to antipsychotic medication, whereas Met carriers can experience a degree of improvement. This interpretation is consistent with earlier studies that found similar moderating effects on N-back and verbal fluency performance. Our study is the largest to date, using well-characterized clinical samples, and whose results can be generalized to the general chronic schizophrenic population. An observational longitudinal study that followed large samples of first-episode patients, with different degrees of symptomatology and under different medication regimens, would provide valuable new insights to the neurocognitive effects of antipsychotic medication. Measuring additional polymorphisms within the COMT gene will provide further insight into the biological mechanisms behind the interaction. Acknowledgements This work was supported by the European Community s Sixth Framework Programme through a Marie Curie Training Network called the European Twin Study Network on Schizophrenia (EUTwinsS), and NARSAD (through a Young Investigator Award to Dr Timothea Toulopoulou). Timothea Toulopoulou, Irene Rebollo Mesa and Marco Picchioni had full access to all the data in the study and to take responsibility for the integrity of the data and the accuracy of the data analysis. Professor Murray received honoraria for lectures from Astra-Zeneca, Janssen, Lilly and BMS. Dr Picchioni has received travel awards from Pfizer, Janssen-Cilag, and Eli Lily. Drs Toulopoulou, Rebollo Mesa, Bramon and Miss Shaikh report no competing interests. References American Psychiatric Association Practice Guidelines (2004). Practice Guidelines for the Treatment of Schizophrenia. Washington DC: American Psychiatric Press Inc. Andreasen NC (1984a). Scale for the assessment of negative symptoms. Iowa City, USA: University of Iowa Press. Andreasen NC (1984b). Scale for the assessment of positive symptoms. Iowa City, USA: University of Iowa Press. Arranz MJ, De Leon J (2007). Pharmacogenetics and pharmacogenomics of schizophrenia: a review of last decade of research. Mol Psychiatry 12: Barnett JH, Jones PB, Robbins TW, Muller U (2007). Effects of the catechol-omethyltransferase Val158Met polymorphism on executive function: a metaanalysis of the Wisconsin Card Sort Test in schizophrenia and healthy controls. Mol Psychiatry 12: Barnett JH, Scoriels L, Munafo MR (2008). Meta-analysis of the cognitive effects of the catechol-o-methyltransferase gene Val158/108Met polymorphism. Biol Psychiatry 64: Barnett JH, Heron J, Goldman D, Jones PB, Xu K (2009). Effects of catechol-omethyltransferase on normal variation in the cognitive function of children. Am J Psychiatry 166: Bazire S (2007). Psychotropic drug directory. Salisbury: Fivepin Publishing. Bertolino A, Caforio G, Blasi G, De Candia M, Latorre V, Petruzzella V, et al. (2004). Interaction of COMT (Val(108/158)Met) genotype and olanzapine treatment on prefrontal cortical function in patients with schizophrenia. Am J Psychiatry 161: Bezchlinbnyk-Butler KZ, Jeffries J (2007). Clinical handbook of psychotropic drugs. 17th ed. Toronto, Canada: Hogrefe and Huber. Caspi A, Moffitt TE, Cannon M, McClay J, Murray R, Harrington H, et al. (2005). Moderation of the effect of adolescent-onset cannabis use on adult psychosis by a functional polymorphism in the catechol-o-methyltransferase gene: longitudinal evidence of a gene X environment interaction. Biol Psychiatry 57: Deary IJ (2001). Human intelligence differences: a recent history. Trends Cogn Sci 5: Dickerson FB, Boronow JJ, Stallings C, Origoni AE, Sullens A, Yolken RH (2007). The catechol-o-methyltransferase Val158Met polymorphism is not associated with broad-based cognitive functioning in schizophrenia. Schizophr Res 96: Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, et al. (2001). Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci U S A 98: First MB, Spitzer RL, Gibbon M, Williams JBM (1997). Structured clinical interview for DSM-IV axis I disorders (SCID). New York, USA: New York State Psychiatric Institute. Forbes NF, Carrick LA, McIntosh AM, Lawrie SM (2009). Working memory in schizophrenia: a meta-analysis. Psychol Med 39: Goldberg TE, Egan MF, Gscheidle T, Coppola R, Weickert T, Kolachana BS, et al. (2003). Executive subprocesses in working memory: relationship to catechol- O-methyltransferase Val158Met genotype and schizophrenia. Arch Gen Psychiatry 60: Lewandowski KE (2007). Relationship of catechol-o-methyltransferase to schizophrenia and its correlates: evidence for associations and complex interactions. Harv Rev Psychiatry 15: Mata I, Arranz MJ, Staddon S, Lopez-Ilundain JM, Tabares-Seisdedos R, Murray RM (2006). The high-activity Val allele of the catechol-o-methyltransferase gene predicts greater cognitive deterioration in patients with psychosis. Psychiatr Genet 16: Mata I, Perez-Iglesias R, Pelayo-Teran JM, Rodriguez-Sanchez JM, Gonzalez-Blanch C, Carrasco-Marin E, et al. (2008). Lack of influence of COMT Val158Met genotype on cognition in first-episode non-affective psychosis. Schizophr Res 102: MattayVS,GoldbergTE,FeraF,HaririAR,TessitoreA,EganMF,et al. (2003). Catechol O-methyltransferase val158-met genotype and individual variation in the brain response to amphetamine. Proc Natl Acad Sci U S A 100: Neale M, Boker SM, Xie G, Maes HHM (2006). Mx: statistical modeling, VCU Box Richmond, VA 23298: Department of Psychiatry. Rey MJ, Schulz P, Costa C, Dick P, Tissot R (1989). Guidelines for the dosage of neuroleptics. I: chlorpromazine equivalents of orally administered neuroleptics. Int Clin Psychopharmacol 4:

8 COMT, antipsychotic and IQ Rebollo-Mesa et al. 105 Spiltzer RL, Endicott J (1978). Schedule for affective disorders and schizophrenialifetime version. New York, USA: New York State Psychiatric Institute. Taylor D, Paton C, Kapur S (2009). The Maudsley prescribing guidelines. 10th ed. London, UK: Martin Dunitz. Van den Oord EJCG, Snieder H (2002). Including measured genotypes in statistical models to study the interplay of multiple factors affecting complex traits. Behav Genet 32:1 22. Van Winkel R, Henquet C, Rosa A, Papiol S, Fananas L, De Hert M et al. (2008). Evidence that the COMT(Val158Met) polymorphism moderates sensitivity to stress in psychosis: an experience-sampling study. Am J Med Gen Part B147B: Wechsler D (1997). Wechsler adult intelligence scale, third edition: administration and scoring manual. London, England: Psychological Corp. Weickert TW, Goldberg TE, Mishara A, Apud JA, Kolachana BS, Egan MF, Weinberger DR (2004). Catechol-O-methyltransferase val108/158met genotype predicts working memory response to antipsychotic medications. Biol Psychiatry 56: Woods SW (2003). Chlorpromazine equivalent doses for the newer atypical antipsychotics. J Clin Psychiatry 64: Woodward ND, Jayathilake K, Meltzer HY (2007). COMT val108/158met genotype, cognitive function, and cognitive improvement with clozapine in schizophrenia. Schizophr Res 90:86 96.