Evidence that the COMTVal158Met polymorphism moderates sensitivity to stress in psychosis: An experience-sampling study†
Please cite this article as follows: van Winkel R, Henquet C, Rosa A, Papiol S, Faňanás L, De Hert M, Peuskens J, van Os J, Myin-Germeys I. 2007. Evidence That the COMTVal158Met Polymorphism Moderates Sensitivity to Stress in Psychosis: An Experience-Sampling Study. Am J Med Genet Part B 147B:10–17.
Abstract
Gene–environment interactions involving the catechol-O-methyltransferase Val158Met polymorphism (COMTVal158Met) have been implicated in the causation of psychosis. Evidence from general population studies suggests that Met/Met subjects are sensitive to stress, a trait associated with psychosis. We hypothesized that the Met allele would moderate the effects of stress on negative affect (NA) in controls, and on NA and psychosis in patients with a psychotic disorder. Thirty-one patients with a psychotic disorder and comorbid cannabis misuse and 25 healthy cannabis users were studied with the experience sampling method (ESM), a structured diary technique assessing current context and emotional and psychotic experiences in daily life. A significant interaction between COMTVal158Met genotype and ESM stress in the model of NA was found for patients (interaction χ2 = 7.4, P = 0.02), but not for controls (interaction χ2 = 3.8, P = 0.15). In the model of ESM psychosis, a significant interaction between COMTVal158Met genotype and ESM stress was also apparent (interaction χ2 = 11.6, P < 0.01), with Met/Met patients showing the largest increase in psychotic experiences as well as NA in reaction to ESM stress. The findings suggest that the COMTVal158Met polymorphism moderates affective and psychotic responses to stress in patients with psychosis, providing evidence for gene–environment interaction mechanisms in the formation of psychotic symptoms. © 2007 Wiley-Liss, Inc.
INTRODUCTION
The catechol-O-methyltransferase gene (COMT) has received much attention as a candidate gene in recent psychosis research, because of its function in inactivating catecholamines at postsynaptic sites in the human brain. COMT contains a functional polymorphism that results in a change from Valine (Val) to Methionine (Met) (COMTVal158Met). The amino acid change affects the function of the enzyme: individuals with the Val/Val genotype have a 40% higher COMT enzyme activity in the brain than individuals with the Met/Met genotype [Chen et al., 2004a]. However, the results from association studies attempting to examine the COMTVal158Met polymorphism as a specific risk gene for schizophrenia have shown great inconsistency. Earlier studies found an association of schizophrenia with the Met allele [Ohmori et al., 1998; Kotler et al., 1999; Park et al., 2002], whereas later evidence favored an association with the Val allele [Glatt et al., 2003; Kremer et al., 2003; Wonodi et al., 2003; Chen et al., 2004b; Sanders et al., 2005]. Two recent meta-analyses found minimal or no evidence for an association between the COMTVal158Met polymorphism and schizophrenia [Fan et al., 2005; Munafo et al., 2005], comparable to a recent study in two large samples [Williams et al., 2005].
Recent studies have focused on gene–environment interactions involving COMTVal158Met rather than genotypic main effects. These studies focused on the Val allele, providing evidence of synergism between the Val allele and exposure to cannabis in the causation of psychosis; carriers of the Val allele were most likely to exhibit psychotic symptoms and to develop schizophreniform disorder years after the initial exposure to cannabis in an epidemiological study [Caspi et al., 2005]. Val carriers were also found to display more psychotic experiences in reaction to cannabis use in an experimental challenge study [Henquet et al., 2006] and an experience-sampling study [Henquet et al., submitted], consistent with interaction, although these findings were conditional on prior evidence of psychometric psychosis liability in both studies.
Therefore, rather than a main effect, the influence of COMTVal158Met on behavior may be better understood in terms of gene–environment interactions. Interestingly, a number of studies in general population samples suggested that carriers of the Met allele may be differentially affected by the environment, by showing larger sensitivity to stress. Oswald et al. [2004], for example, found the Met allele to be associated with a greater stress hormone response to a physical stressor in 46 healthy controls. A study of 35 healthy subjects by Smolka et al. [2005] showed greater brain activation in the limbic region to unpleasant stimuli (but not pleasant stimuli) in carriers of the Met allele, as measured by fMRI. These findings were extended by the work of Drabant et al. [2006] in 100 healthy control subjects, who found a greater limbic reactivity in Met carriers using an emotional face-processing task. Furthermore, the Met allele was associated with excessive hippocampal and prefrontal cortex activity. In this study, Met/Met subjects also showed greater amygdala-orbitofrontal cortex connectivity that in turn was associated with low novelty seeking. In the context of provocative environments (e.g. stress), this could lead to an increased susceptibility for negative mood states [Drabant et al., 2006].
These studies suggest that Met carriers may have greater reactivity to situations that provoke negative affective states. This is in keeping with findings from the general population that the Met allele is associated with a propensity to anxiety [Mathew et al., 1980; Enoch et al., 2003; Olsson et al., 2005; Stein et al., 2005], reduced extraversion [Reuter and Hennig, 2005; Stein et al., 2005], and reduced novelty seeking [Benjamin et al., 2000; Tsai et al., 2004; Reuter and Hennig, 2005].
The findings connecting the Met allele to stress-responsivity are relevant to schizophrenia, as increased stress-sensitivity was found to be present not only in patients diagnosed with schizophrenia, but also in their relatives in an experience sampling study [Myin-Germeys et al., 2001]. In addition, experience sampling studies also found that increased stress-sensitivity in schizophrenia is associated with increases in both affective and psychotic responses [Myin-Germeys et al., 2001, 2005c] in both patients and their first degree relatives.
It can thus be hypothesized that the Val and Met allele are both, albeit differentially, implied in the development of psychopathology through divergent interactions with specific environmental stimuli. Whereas the Val allele may interact with cannabis in shaping the risk for psychotic disorder as shown previously [Caspi et al., 2005; Henquet et al., 2006; Henquet et al., submitted], the Met allele may be specifically associated with stress-sensitivity in terms of affective and psychotic responses to small stressors in the flow of daily life.
The experience sampling method (ESM) [Myin-Germeys et al., 2001], a self-assessment technique that is used to assess context, thoughts, affect, and symptoms in the flow of daily life, can be used to investigate these hypothesized gene–environment interactions in a momentary, ‘real-world’ design. Using this method, the current study aimed to evaluate the relationship between sensitivity to stress and the COMTVal158Met polymorphism in a sample of 31 patients with a clinical diagnosis of psychotic disorder and 25 non-psychotic controls. The aim of the current study was to investigate the hypothesis that the Met allele would moderate affective responses to stressful events in daily life both in patients and control subjects. In addition, since stress also elicited psychotic reactions in patients, it was hypothesized that the Met allele would also moderate the effects of stress on psychotic symptoms in patients with a psychotic disorder but not in the control group.
Of note, this same sample has been previously shown to have an interaction between the Val allele and cannabis use in their effect on psychotic symptoms [Henquet et al., submitted].
MATERIALS AND METHODS
Sample
Thirty-one cannabis users with a clinical diagnosis of psychotic disorder according to DSM-IV criteria and 25 non-psychotic cannabis users were recruited through inpatient and outpatient mental health service facilities in South-Limburg, The Netherlands, or were recruited from the general population as described previously [Henquet et al., 2006; Henquet et al., submitted]. The diagnosis of patients was assessed according to the Research Diagnostic Criteria (RDC) using the OPCRIT computer program [McGuffin et al., 1991]. Patients had diagnoses of schizophrenia (n = 6), schizoaffective disorder (n = 23), and psychosis not otherwise specified (n = 2).
Subjects attended a briefing session during which a complete description of the study was provided and written informed consent was obtained. In addition, past and current psychiatric illness and past and current substance use were assessed. Given the focus of the study by Henquet et al. on the effects of cannabis use on psychosis according to genotype [Henquet et al. submitted], the following exclusion criteria were applied in order to reduce possible confounding by the use of other substances, and to minimize health risks: alcohol use in excess of 5 U per day; use of illicit drugs (other than cannabis) during the 6 consecutive days of the study; respiratory, cardiovascular, or neurological disease. Pregnant women were also excluded. A positive family history for psychosis was a further exclusion criterion for the control group. The study was carried out in accordance with the World Medical Association's Declaration of Helsinki [Edinburgh modification, 2000] and was approved by the local Medical Ethics Committee.
COMTVal158Met Genotyping
Buccal mucosa was obtained from all the individuals by brushing of a large surface inside both cheeks by means of a cotton swab. DNA was extracted using a BuccalAmp DNA Extraction Kit. COMTVal158Met genotype was assayed by polymerase chain reaction (PCR) and enzymatic digestion with NlaIII followed by Acrylamide gel electrophoresis as described by Daniels et al., 1996.
Experience Sampling Method (ESM)
The ESM has been described in previous reports [Myin-Germeys et al., 2001, 2005c]. It is a self-assessment technique that is used to assess current context, thoughts, mood and symptoms in the flow of daily life. Subjects are provided a wrist-watch and a set of booklets (one booklet for each day). At random moments, the watch emits a “beep,” after which subjects are asked to report their current context, thoughts, appraisals and mood. In the current study, the watch emitted a beep 12 times a day, during 6 consecutive days. Beeps occurred at random, between 7.30 am and 0.30 pm. Self-assessments were rated on a 7-point Likert scale.
The ESM procedure was explained to the subjects during an initial briefing session, and a practice form was filled in to confirm that subjects were able to understand the 7-point Likert scale format. Subjects were asked to complete their reports immediately after the beep and to record the time at which they completed their report, in order to minimize memory distortions. For reasons of reliability and validity as described in detail before [Delespaul, 1995], all reports completed more than 15 min after the beep and/or subjects with fewer than 24 reports were excluded from the analysis. In line with earlier reports [Myin-Germeys et al., 2005a,c], recruited patients were clinically remitted. This strategy was chosen because: (i) the completing and adequacy of the reports was assumed to be better in clinically remitted patients and (ii) clinically remitted patients were assumed to be less likely to quit the study because of paranoid ideas interfering with the research protocol.
Measures
As described in previous work, stress reactivity was conceptualized as affective and psychotic reactivity to daily life events and minor disturbances in daily life [Myin-Germeys et al., 2001, 2002, 2005a,c; Henquet et al., submitted]. Measures of stress, affective response, and psychotic symptoms were derived from the experience sampling reports as described below.
Assessment of stress
Stress in the model of ESM (hereafter, ‘ESM stress’) was conceptualized as the subjective appraised stressfulness of distinct events in the natural flow of daily life. Subjects were asked to report the most important event that happened between the current and the previous report, and to rate this event on a 7-point scale (−3, very unpleasant; 0, neutral; 3, very pleasant). ESM stress was linearly defined with most stress being related to the ‘unpleasant’ side of the continuum and least stress being associated with the ‘pleasant’ side of the continuum. In order to allow high scores to reflect high ESM Stress, responses were recoded (−3, very pleasant; 0, neutral; 3, very unpleasant).
Assessment of psychosis
Psychotic experiences were assessed with nine ESM items rated on 7-point Likert scales (1, not at all; 7, very). To allow self-reported assessments, these nine items included aspects of the mental state that subjects are aware of and that are directly associated with psychotic experiences such as delusions and hallucinations. These nine items were: “auditory hallucinations,” “visual hallucinations,” “preoccupation,” “suspicion,” “feeling unreal,” “feeling controlled,” “difficulty of expressing thoughts,” “racing thoughts,” and “fear of losing control”. As previous reports have shown that subjects with hallucinatory experiences can distinguish between hearing real voices and auditory hallucinations [Delespaul, 1995; Escher et al., 2002], the presence of hallucinations was asked directly (“Do you hear voices?,” “Do you see things that others cannot see?”). Guided by a previous paper addressing the presence of psychotic symptoms using ESM [Myin-Germeys et al., 2005c], the presence of delusions was assessed with the use of items that assess aspects of mental states that are directly related to delusional ideation in psychosis. These were: “preoccupation,” “suspicion,” “feeling unreal,” and “feeling controlled.” Finally, three items were added that are related to psychosis in general. These were: “I have difficulty expressing my thoughts,” “My thoughts are racing,” and “I am afraid to lose control.” The sum of these nine items formed the variable ‘ ESM psychosis’ (Cronbach's α = 0.77). A sumscore was preferred over other approaches (e.g. using the highest symptom score as measure for psychosis) because the ESM psychosis measure had to allow to pick up subtle fluctuations in psychotic symptomatology reactive to small, daily life stressors; a score that ranges from 9 to 63 (as in a sumscore) is more likely to allow this than a score that only ranges from 1 to 7 (as in the highest psychotic symptom score). Furthermore, the use of a sumscore was chosen because of its consistency with previous ESM studies that examined psychotic reactivity to daily life stressors [Myin-Germeys et al., 2005a,b].
Factor analysis of these nine psychosis items yielded two factors with eight values greater than 1. The first factor loaded high on the items that assessed hallucinations and consisted of the mean of the items “auditory hallucinations” and “visual hallucinations” (“ESM hallucinations;” Cronbach's α = 0.70). The second factor loaded high on items that were intended to assess delusions or thought disorder, and consisted of the mean of the items “preoccupation,” “suspicion,” “feeling controlled,” “racing thoughts,” “feeling unreal,” “fear of losing control,” and “difficulty expressing thoughts” (“ESM delusions;” Cronbach's α = 0.76).
Assessment of affect
The affective states reported after each beep were assessed with nine mood-related items, that were rated on 7-point Likert scales (1, not at all; 7, very). In line with previous reports [Myin-Germeys et al., 2001, 2002], affective states were expressed as “positive affect (PA)” and “negative affect” (NA). The items “happy,” “cheerful,” “relaxed,” and “satisfied” formed the PA scale, (Cronbach's α = 0.85); the items “down,” “guilty,” “insecure,” “lonely,” and “anxious” formed the NA scale, (Cronbach's α = 0.80). As there was no a priori hypothesis regarding PA, only the results with regard to NA are shown.
Statistical Analyses
ESM data are hierarchically structured: multiple observations (level 1) are nested within subjects (level 2). Since observations from the same subject are more similar than observations from other subjects, conventional regression techniques are not suited for analyzing ESM data. Therefore, multilevel linear random regression was used, hereby taking into account the variance components at two levels (ESM-beep level and subject level) [Goldstein, 1987].
As it is also necessary to correct for autocorrelation of data in a time-series like the ESM data, data were analyzed with the XTREGAR module in Stata [StataCorp, 2005], that takes the possibility of autocorrelation into account. The β is the fixed regression coefficient, and can be interpreted in the same fashion as the parameters of unilevel regression coefficients. In order to test the hypothesis that COMTVal158Met genotype moderates the psychotic response to daily life stress, multilevel linear random regression analyses were conducted with ESM psychosis, ESM delusions, and ESM hallucinations as the respective dependent variables, COMTVal158Met genotype (0 = Val/Val; 1 = Val/Met; 2 = Met/Met) and ESM stress as independent variables, as well as their interaction term. Similarly, in order to test the hypothesis that COMTVal158Met genotype moderates the affective response to daily life stress, a multilevel linear random regression was performed with NA as the respective dependent variable, and COMTVal158Met genotype and ESM stress as independent variables, as well as their interaction term. As the current study was performed in a sample of cannabis users [Henquet et al., submitted], ESM cannabis use was included in the analyses as a time-dependent covariate. Antipsychotic use was included in the analyses as a time-independent covariate. Furthermore, use of sedative medication related to perceived stress as indicated by the subject (“ESM stress”), was included in the analyses as a time-dependent potential confounder. Both benzodiazepines and first-generation, sedative antipsychotics that were not prescribed on a daily basis were defined as “sedative medication.”
In order to estimate effect sizes for each genotype separately, the appropriate linear combinations were calculated from the interaction terms using the STATA LINCOM routine, after fitting the stress × genotype interaction term with genotype as two dummy variables, using Val/Val as reference category. Main effects and interactions were assessed by Wald test [Clayton and Hills, 1993]. As there was no a priori hypothesis regarding the type of psychotic symptoms related to the hypothesized gene–stress interaction, the results were corrected for multiple testing using the Simes' correction, which in the case of correlated outcomes represents an improvement over the Bonferroni procedure [Simes, 1986]. According to the Simes' procedure, observed P-values greater than Simes' P-values are considered statistically non-significant. Simes' P-values are shown for P-values below 0.05. Since, it was hypothesized that the increased responses to stress associated with Met would differ between patients and controls, stratified analyses were conducted.
RESULTS
Subjects
All subjects had used cannabis in the past 12 months with a frequency of use of once a day or more in 84% of subjects. No illicit drugs other than cannabis were used during the study. In the patient group, 7 patients were taking typical antipsychotics, 12 were taking atypical antipsychotics, and 12 were not taking antipsychotic medication. There were no differences between patients and controls with regard to frequency of cannabis use, age at onset of cannabis use, gender or living situation, but patients were significantly older and had less education (Table I). Patients had significantly more ESM symptomatology, but did not experience more ESM stress. The number of completed ESM reports did not significantly differ between patients and controls (Table I). Of all ESM moments, cannabis use was reported in 40.6%. The use of sedative medication was not confined to patients, but patients were much more likely to use sedative medication (t = −4.0, P < 0.001).
| Patients | Controls | t, P | |
|---|---|---|---|
| Age (SD) | 36.2 (9.6) | 26.1 (6.6) | t = −4.3, P < 0.001 |
| Gender (male/female) | 23/8 | 22/3 | t = −1.3, P = 0.20 |
| Education (n = 58) | t = 2.6, P = 0.01 | ||
| Elementary school | 5 | 3 | |
| Secondary school | 21 | 11 | |
| Higher education | 2 | 11 | |
| Living situation (n = 61) | t = −0.6, P = 0.57 | ||
| Alone | 15 | 12 | |
| With partner | 6 | 5 | |
| With parents or family | 5 | 8 | |
| Protected housing | 5 | 0 | |
| Frequency of cannabis usea | t = −0.15, P = 0.88 | ||
| One to two times per week | 2 | 0 | |
| Three to four times per week | 2 | 5 | |
| One time per day | 5 | 3 | |
| More than 1 time per day | 22 | 17 | |
| Age of onset cannabis use | 18.3 (6.9) | 16.0 (1.8) | t = −1.6, P = 0.11 |
| Number of ESM reports | 41.9 (1.9) | 44.2 (2.1) | t = 0.8, P = 0.43 |
| ESM psychosis | 17.5 (7.6) | 12.9 (4.3) | t = −2.7, P = 0.01 |
| ESM delusions | 2.2 (1.0) | 1.6 (0.6) | t = −2.5, P = 0.02 |
| ESM hallucinations | 1.3 (0.7) | 1.0 (0.1) | t = −2.5, P = 0.02 |
| ESM stress | −1.3 (1.8) | −1.5 (1.6) | t = −1.1, P = 0.28 |
| ESM positive affect | 4.5 (1.3) | 5.1 (1.2) | t = 2.1, P = 0.04 |
| ESM negative affect | 1.7 (1.0) | 1.3 (0.7) | t = −2.3, P = 0.03 |
- a Over the last year.
COMTVal158Met Genotype
The COMTVal158Met genotype distribution was in Hardy–Weinberg equilibrium for both patients and controls, with the Met/Met genotype being present in 14.3% of subjects (n = 8; 5 patients (16.1%); 3 controls (12.0%)), Val/Met in 55.4% (n = 31; 18 patients (58.1%); 13 controls (52.0%)), and Val/Val in 30.4% (n = 17; 8 patients (25.8%); 9 controls (36.0%)). Genotype was not associated with sex, ESM cannabis use, ESM psychosis, ESM hallucinations, ESM delusions, or ESM stress, but there was a significant main effect of genotype on NA (Table II).
| Val/Val | Val/Met | Met/Met | P | |
|---|---|---|---|---|
| Age (SD) | 32.2 (8.4) | 31.3 (10.4) | 27.7 (8.2) | 0.26 |
| Sex (M/F) | 16/3 | 24/8 | 8/2 | 0.67 |
| Number of ESM reports (SD) | 42.7 (8.5) | 47.1 (10.8) | 44.5 (8.2) | 0.45 |
| ESM cannabis usea | 48.8% | 36.7% | 39.7% | 0.25 |
| ESM psychosis (SD) | 14.9 (6.3) | 15.6 (7.0) | 14.8 (6.2) | 0.79 |
| ESM delusions (SD) | 1.8 (0.9) | 1.9 (0.9) | 1.8 (0.8) | 0.95 |
| ESM hallucinations (SD) | 1.1 (0.4) | 1.1 (0.6) | 1.3 (0.6) | 0.23 |
| ESM stress (SD) | −1.5 (1.6) | −1.4 (1.8) | −1.1 (1.7) | 0.21 |
| ESM negative affect (SD) | 1.3 (0.5) | 1.6 (0.9) | 1.8 (1.0) | 0.03 |
- a Percentage of ESM moments where cannabis use is reported.
Models for NA in Patients and Controls
The main effect of genotype on NA was significant in controls (β = 0.27, 95% CI 0.03–0.51, P = 0.03), but failed to reach significance in patients, although the relationship was directionally similar (β = 0.37, 95% CI −0.02 to 0.76, P = 0.07), with subjects of the Met/Met genotype having the highest NA scores in both groups (Table III). ESM stress had a significant main effect on NA in the total group (β = 0.03, 95% CI 0.02–0.04, P < 0.001). Stratified analyses revealed that this effect was also significant for patients, but just failed to reach significance for controls (Table III).
| Model | Patients | Controls | ||
|---|---|---|---|---|
| Effect size (β, P) (SE) | Model fit (Wald χ2; P) | Effect size (β, P) (SE) | Model fit (Wald χ2; P) | |
| ESM stress only | 0.04; P < 0.001 (0.01) | 25.3; P < 0.001 | 0.01; P = 0.07 (0.01) | 9.1; P = 0.06 |
| Genotype only | 7.3; P = 0.30 | 13.7; P = 0.02 | ||
| Val/Val | 0.00a | 0.00a | ||
| Val/Met | 0.55; P = 0.07 (0.30) | 0.07; P = 0.66 (0.16) | ||
| Met/Met | 0.68; P = 0.09 (0.40) | 0.69; P < 0.01 (0.24) | ||
| Genotype + ESM stress | 26.6; P < 0.001 | 17.6; P < 0.01 | ||
| Val/Val | 0.05; P < 0.001 (0.01) | 0.01; P = 0.32 (0.01) | ||
| Val/Met | 0.57; P = 0.05 (0.29) | 0.07; P = 0.65 (0.15) | ||
| Met/Met | 0.74; P = 0.05 (0.39) | 0.73; P = 0.001 (0.22) | ||
| Genotype × ESM stress | 35.1; P < 0.001 | 21.2; P < 0.01 | ||
| Val/Val | 0.05; P = 0.13 (0.03) | 0.01; P = 0.49 (0.02) | ||
| Val/Met | 0.03; P < 0.01 (0.01) | 0.00; P = 0.92 (0.01) | ||
| Met/Met | 0.12; P < 0.001 (0.03) | 0.08; P = 0.04 (0.04) | ||
- In the composite models, only genotype effect sizes are displayed.
- a Val/Val is the reference category.
- * The Wald χ2's and their statistical significance can be used to compare the fit of the models. The reported effect sizes are relative effect sizes as compared to the Val/Val reference category that can only be interpreted within a given model and not across models. Models were adjusted for the following confounders: cannabis use, use of sedative medication, and antipsychotic use.
A significant interaction effect was found for COMTVal158Met genotype and ESM stress in the model of NA for patients (interaction χ2 = 7.4, P = 0.02) but not for controls (interaction χ2 = 3.8, P = 0.15), indicating that COMTVal158Met genotype moderated the affective response to ESM stress in patients only. The Met/Met patients reported a larger increase in NA in reaction to ESM stress than the patients of the other genotypes (Table III). Although the interaction between COMTVal158Met genotype and ESM stress in the model of NA was only statistically significant for patients, the fit of the model improved in both groups after the addition of the interaction term (Table III).
Models for ESM Psychosis in Patients and Controls
In the patient group, ESM stress was significantly associated with the ESM psychosis measure (β = 0.18, 95% CI 0.05–0.32, P < 0.01; Table IV). ESM stress was not associated with the ESM psychosis measure in controls (Table IV). In the model of ESM psychosis, a significant interaction was apparent for patients between COMTVal158Met genotype and ESM stress (interaction χ2 = 11.6, P < 0.01; <Simes' P of 0.016), indicating that in patients, the psychotic response to ESM stress was moderated by COMTVal158Met genotype. The Met/Met patients showed a significantly greater increase in psychotic experiences in reaction to ESM stress than the patients of the Val/Met or Val/Val genotype (Table IV). Similar results were also found for ESM delusions (interaction χ2 = 12.4, P < 0.01; <Simes' P of 0.033), but not for ESM hallucinations (interaction χ2 = 4.9, P = 0.08). As expected, there was no interaction between COMTVal158Met genotype and ESM stress in the model of ESM psychosis in controls (interaction χ2 = 3.3, P = 0.20).
| Model | Patients | Controls | ||
|---|---|---|---|---|
| Effect size (β, P) (SE) | Model fit (Wald χ2; P) | Effect size (β, P) (SE) | Model fit (Wald χ2; P) | |
| ESM stress only | 0.18; P = 0 < 0.01 (0.07) | 10.5; P = 0.06 | −0.03; P = 0.52 (0.05) | 2.8; P = 0.59 |
| Genotype only | 1.5; P = 0.96 | 4.5; P = 0.47 | ||
| Val/Val | 0.00a | 0.00a | ||
| Val/Met | 1.83; P = 0.47 (2.51) | −0.45; P = 0.74 (1.36) | ||
| Met/Met | 2.87; P = 0.39 (3.34) | −2.26; P = 0.28 (2.10) | ||
| Genotype + ESM stress | 5.1; P = 0.65 | 6.6; P = 0.36 | ||
| Val/Val | 0.15; P = 0.07 (0.08) | −0.12; P = 0.05 (0.06) | ||
| Val/Met | 1.99; P = 0.44 (2.57) | −0.56; P = 0.68 (1.36) | ||
| Met/Met | 3.11; P = 0.37 (3.45) | −2.41; P = 0.25 (2.08) | ||
| Genotype × ESM stress | 16.7; P = 0.05 | 10.1; P = 0.26 | ||
| Val/Val | 0.25; P = 0.30 (0.24) | −0.11, P = 0.27 (0.10) | ||
| Val/Met | 0.01; P = 0.95 (0.09) | −0.08, P = 0.32 (0.08) | ||
| Met/Met | 0.77; P < 0.001 (0.21) | −0.56, P = 0.03 (0.25) | ||
- In the composite models, only genotype effect sizes are displayed.
- a Val/Val is the reference category.
- * The Wald χ2's and their statistical significance can be used to compare the fit of the models. The reported effect sizes are relative effect sizes as compared to the Val/Val reference category that can only be interpreted within a given model and not across models. Models were adjusted for the following confounders: cannabis use, use of sedative medication, and antipsychotic use.
DISCUSSION
Findings
The current study aimed to test two main hypotheses. First, it was hypothesized that the Met allele would moderate the effects of stress on NA in healthy controls as well as in patients with a psychotic disorder. This hypothesis was only partly supported, as a significant interaction effect for COMTVal158Met genotype and ESM stress in the model of NA was found for patients but not for controls. Nevertheless, the fit of the model improved in both groups after the addition of the interaction term, suggesting that a lack of power may explain the lack of significant interaction in the controls group. As hypothesized, the Met/Met patients reported a larger increase in NA in reaction to ESM stress than the patients of the other genotypes.
Second, it was hypothesized that the Met allele would moderate the effects of stress on psychotic symptoms in patients only. In the model of ESM psychosis, a significant interaction between COMTVal158Met genotype and ESM stress was indeed apparent for patients, indicating that the psychotic response to ESM stress was also moderated by COMTVal158Met genotype in patients. Again, the Met/Met patients showed a significantly greater increase in psychotic experiences in reaction to ESM stress than the patients of the Val/Met or Val/Val genotype. Similar results were also found for ESM delusions, but not for ESM hallucinations, probably because of a relative scarcity of hallucinations in the patient group (Table I). As hypothesized, there was no evidence for an interaction between COMTVal158Met genotype and ESM stress in the model of ESM psychosis in the controls group.
ESM stress had a significant main effect on NA and on ESM psychosis in patients but not in controls. These findings suggest that patients with psychosis react more strongly to stress, showing an increase in psychotic and affective symptoms. Interestingly, the model containing ESM stress, genotype and their interaction term predicted the symptomatic response clearly better than the main model only containing ESM stress, indicating the importance of taking gene–environment interaction into account in explaining transitions in momentary symptomatology in psychosis. The observation that the interaction between ESM stress and COMTVal158Met was only present in patients is in line with findings regarding the interaction between the COMTVal158Met Val allele and cannabis use, where the cannabis × genotype interaction was dependent on a pre-existing liability to psychosis [Henquet et al., 2006; Henquet et al., submitted]. Given these findings, it can be hypothesized that the Met allele can be a causal factor not only for depression and bipolar disorder as shown by Mandelli et al., 2006, but also for psychosis, specifically in subjects with a pre-existing liability to psychosis.
Whereas the effect of genotype on NA in patients was only apparent in interaction with ESM stress, a significant main effect of genotype on NA was found in controls. Since the effect of ESM stress on measures of symptomatology was only shown in patients, it is unclear whether the effect of genotype on NA in the controls group was independent of ESM stress, or alternatively, whether the perceived stress for controls during the ESM registration was insufficient to provoke and detect these kind of gene–environment interactions. The observation that the fit and statistical significance of the model containing the interaction term was clearly better than the fit of the main effect model (Table III) could be in line with the latter possibility.
COMTVal158Met Polymorphism and Emotional Processing
These findings support the hypothesized role of the COMTVal158Met polymorphism in sensitivity to stress in the patients with a psychotic disorder and its possible contribution to symptomatology in the general population, and are in line with previous findings connecting the Met allele with sensitivity to stress. The findings support the involvement of the COMTVal158Met polymorphism in the emotional processing of environmental stimuli, with Met carriers experiencing more difficulties. The tonic-phasic dopamine model may provide an underlying biological rationale for COMTVal158Met interactions with the environment. According to this model, the enzyme containing Val favors reduction in tonic dopamine but increases in phasic dopamine. A milieu characterized by predominance of phasic over tonic dopamine may give rise to reduced cognitive stability, but increased cognitive flexibility. For the enzyme containing Met, the opposite would hold [Nolan et al., 2004; Rosa et al., 2004; Blasi et al., 2005; Bruder et al., 2005; Stefanis et al., 2005]. It has been suggested that cognitive stability in Met carriers would be adaptive in circumstances where holding information is demanded (e.g. working memory), but dysfunctional in circumstances where rapid adjustments to changing external stimuli are required [Bilder et al., 2004; Tunbridge et al., 2006]. The lack of an allele dosage effect in the current study, meaning that the Val/Met individuals clustered with Val/Val individuals rather than being intermediate between homozygote groups, is interesting in this regard, and supports the idea that excessive cognitive stability, that is only present in subjects with two Met alleles, could be problematic in these circumstances. The current study is the first to demonstrate the relevance of these emotional processing difficulties in Met carriers for psychopathological symptoms such as psychosis and NA. It extends the findings of Mandelli et al., 2006, who showed that the Met allele was associated with depression following adverse life events in a large sample of patients with major depression and bipolar disorder.
COMTVal158Met Polymorphism and Psychosis
The current findings also add evidence to the theoretical framework that links dopamine to the development of psychosis through aberrant salience [Kapur, 2003, 2004], in the sense that a part of the “dopamine system” (the COMTVal158Met polymorphism, that regulates breakdown of intrasynaptic dopamine) was shown to moderate the psychotic response to daily life stressors. These daily life stressors were shown to interact with COMTVal158Met genotype to increase momentary delusional ideas, which indeed suggests some kind of attribution of aberrant salience to these daily life stressors and, given the interaction with COMTVal158Met genotype, also suggests a role for dopamine in the development of this aberrant salience. The current results also suggest that the attribution of aberrant salience is not “context-independent,” as hypothesized by Kapur and co-workers, but rather that the more stressful these daily life stressors are perceived, the greater the chance that they are being interpreted in a “context-inappropriate,” delusional way. Interestingly, COMT has its effect primarily in the prefrontal cortex, whereas aberrant salience is considered to occur through excess striatal dopamine. Future research will have to investigate this issue more specifically.
Strengths and Caveats
First, a major strength of the current study is the use of the ESM for the assessment of a gene–environment interaction between the COMTVal158Met polymorphism and the responsivity to stress. This method allows gathering insight in the course of symptoms and the influence of genes and environmental stimuli in a momentary, “real-world” design. The use of the ESM facilitated the test of the hypothesis that both the COMTVal158Met Val and Met allele differentially moderated the psychotic response to different specific environmental stimuli. Although, this finding needs to be replicated, the confirmation of this hypothesis by the current data clearly provides greater insight in the role of the COMTVal158Met polymorphism in the occurrence and course of psychopathological symptoms. In future research, the ESM could also provide an attractive paradigm to study the influence of several other candidate genes, including gene–gene and gene–environment interactions.
A second strength of the current study is that it was specifically performed in a sample in which it was previously shown that the Val allele moderated the psychotic response to cannabis [Henquet et al., submitted]. In this study on the same sample as that of the current study, subjects with the Val/Val genotype, but not those with one or more Met alleles, showed an increase in hallucinations after cannabis exposure, that was conditional on prior evidence of psychometric psychosis liability. These findings allowed the test of the hypothesis that within an identical sample, both the COMTVal158Met Val and Met allele moderate differential vulnerabilities in interaction with specific environmental factors.
However, apart from being a strength of the current study by allowing to directly test this hypothesis, the approach of testing COMTVal158Met moderation of stress reactivity in severe cannabis smokers also represents the main limitation of the current study, as it obviously limits the generalizibility of our findings. Although, limited data are available, cannabis use has been suggested to result in increased synaptic dopamine activity, at least as a short-term effect [Voruganti et al., 2001]. Furthermore, chronic cannabis use was found to induce a prefrontal hypodopaminergic state [Eldreth et al., 2004] and desensitization of CB1 receptors [Sim-Selley, 2003]. Therefore, a replication of our findings in a sample of subjects abstinent of any drug use is needed.
A second limitation is the cross-sectional nature of the analyses, which makes it impossible to establish causal relationships. Therefore, it is impossible to determine whether stress measures influenced affect or psychotic symptoms, or whether affect and psychotic symptoms influenced the subjective appraisal of stress. However, either explanation has clinical relevance. A third limitation is the unequal use of sedative medication in patients and controls, which could have biased the results, as could the use of antipsychotic medication in patients. However, the therapeutic effect of these drugs is intended to be stress-reducing and antipsychotic. This implies that if the current results are indeed biased by the unequal intake of these drugs in patients and controls, the bias is most likely to reduce the results towards the null rather than to create spurious results.
In conclusion, the current results provide evidence for a gene–environment interaction between COMTVal158Met genotype and stressful events in the flow of daily life in patients with psychosis and comorbid cannabis abuse, with Met/Met patients showing the largest increase in psychotic experiences as well as NA in reaction to ESM stress. The findings of this study need replication in samples abstinent of any drug use in order to confirm these gene–stress interactions.
Acknowledgements
Inez Myin-Germeys was supported by a 2006 NARSAD Young Investigator Award and by the Dutch Medical Council (VENI-grant).
