Association of the ZFPM2 Gene with Antipsychotic Induced Parkinsonism in Schizophrenia Patients

Lior Greenbaum,1 Robert C. Smith,2,3 Mordechai Lorberboym,4,6 Anna Alkelai,1 Polina Zozulinsky,1 Tzuri Lifshytz,1 Yoav Kohn,1 Ruth Djaldetti,5,6 Bernard Lerer. 1

1Biological Psychiatry Laboratory, Department of Psychiatry, Hadassah – Hebrew University Medical Center, Jerusalem, Israel;

2Department of Psychiatry, New York University Medical School, New York, NY, USA;

3Nathan Kline Institute for Psychiatric Research, Orangeburg, New York;

4Department of Nuclear Medicine, Edith Wolfson Medical Center, Holon, Israel;

5Department of Neurology, Rabin Medical Center, Petah Tiqwa, Israel;

6Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Tel Aviv; Israel.

Corresponding Author:

Prof. Bernard Lerer, Biological Psychiatry Laboratory, Department of Psychiatry, Hadassah – Hebrew University Medical Center, Ein Karem, Jerusalem 91120

Israel.

Tel: 972-2-6777185; Fax: 972-2-6439294; Email:

Conflict of interest: BL, LG and AA have a pending patent application relating to the role of ZFPM2 in antipsychotic-induced parkinsonism. All other authors declare that they have no conflicts of interest.

Keywords: Schizophrenia, antipsychotic induced parkinsonism, ZFPM2, FOG2.

Abstract

Rationale: Antipsychotic induced parkinsonism (AIP) is a severe adverse affect of antipsychotic drug treatment. Recently, our group performed a genome-wide association study (GWAS) for AIP severity, and identified several potential AIP risk variants.

Objectives: To validate our original AIP-GWAS susceptibility variants, and to understand their possible function.

Methods:We conducted a validation study of 15 SNPs in an independent sample of 178 US schizophrenia patients treated for at least a month with typical or atypical antipsychotics. Then, a sample of 49 Jewish Israeli Parkinson's disease (PD) patients with available neuroimaging ([123I]-FP-CIT-SPECT) data was analyzed, to study association of confirmed AIP SNPs with level of dopaminergic deficits in the putamen. Finally, association of confirmed AIP related SNPs with gene expression was examined in-silico in human lymphoblasts.

Results: Using logistic regression and controlling for possible confounders, we found nominal association of the intronic SNP, rs12678719, in the Zinc Finger Protein Multitype 2 (ZFPM2) gene with AIP (62 affected/116 unaffected), in the whole sample (p=0.009; P=5.97x10-5 in the GWAS), and in the African American sub-sample (N=111; p=0.002). The same rs12678719-G AIP susceptibility allele was associated with lower levels of dopaminergic neuron related ligand binding in the contralateral putamen of PD patients (p=0.026). and with differences in ZFPM2 gene expression in lymphoblastoid cells (p=0.002)

Conclusions: Our preliminary findings support association of the ZFPM2 SNP, rs12678719, with AIP. At the functional level, this variant is associated with deficits in the nigrostriatal pathway in PD patients that may be related to latent subclinical deficits among AIP-prone individuals with schizophrenia. Further validation studies in additional populations are required.

Introduction

Antipsychotic induced parkinsonism (AIP) is an acute side effect of antipsychotic drug treatment and is the most common manifestation of antipsychotic-induced extrapyramidal symptoms (EPS) (Rochon et al, 2005; Tenback et al, 2006). AIP prevalence data vary widely among studies, ranging from 15% to more than 50% of antipsychotic treated patients (Hansen et al, 1997; Hirose, 2006). According to data from the 1960s which relate to first generation antipsychotics (FGA), 50% of AIP-affected individuals manifest parkinsonian symptoms within the first month of drug administration and 90% during the first 72 days (Ayd et al, 1960). Other researchers observed that the majority of patients develop AIP within 20 days (Freyhan et al, 1959), while Medinar et al (1962) reported AIP within the first week of treatment. Improvement and recovery of AIP symptoms within two months was reported in two thirds of patients (Stephen et al, 1984), while in the remaining cases, parkinsonian features were found to persist or even worsen (Tinazzi et al, 2009) However, AIP is also observed as a late onset manifestation (Lerner et al, 2007). Clinically, AIP is similar to idiopathic Parkinson disease (PD), and characterized by bradykinesia, tremor, rigidity and stooped posture (Hansen et al, 1997; Hirose, 2006; Haddad et al, 2008).

The pathophysiology of AIP is unclear; however, nigrostriatal (NS) pathway dopamine D2 receptor occupancy by antipsychotics is directly related to it (Casey, 2004). All clinically effective antipsychotics are D2 receptor blockers (Miyamoto et al, 2005). D2 receptor occupancy of more than 80%, as produced by typical antipsychotics, significantly increases the risk of AIP (Hirose, 2006). Well documented clinical and demographic risk factors for AIP are antipsychotic type (FGA vs. second generation antipsychotics [SGA]), high doses of antipsychotics, older age and female gender (Ebadi and Srinivasan, 1995; Caligiuri et al, 2000). The substantial heterogeneity in AIP prevalence may stem from inter-study differences in medication regimens, patient demographic background data and variable phenotype definitions (Hassin-Baer et al, 2001).

In addition to the epidemiological risk factors, genetic factors may contribute to inter-individual differences in AIP susceptibility (Basile et al, 2002; Arranz et al, 2007; Greenbaum et al, 2007, 2009; Kasten et al, 2011). Research in this field is limited, since many studies have tended to regard different forms of drug induced movement disorders as a single clinical entity (e.g. Alhadithy et al, 2008). Studies that focus specifically on AIP are rare and have mostly concentrated on functional variants within selected candidate genes such as dopamine and serotonin receptors (DRD2, DRD3, HTR2A, HTR2C) (Gunes et al 2007; Alhadithy et al, 2008). Our group has reported association of the regulator of G protein signaling-2 (RGS2) gene with AIP in independent Jewish Israeli and American samples of schizophrenia patients (Greenbaum et al, 2007; Greenbaum et al, 2009) but findings in other populations are inconsistent (Al Hadithy et al, 2009; Higa et al, 2010).

Recently, we published the first genome-wide, pharmacogenomic association study (GWAS) for AIP severity (Alkelai et al, 2009). This study employed a secondary analysis of phenotype and genotype data from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study (Lieberman et al, 2005). CATIE was a large, multi-center, double blind study that compared the clinical effectiveness and adverse effects of five different antipsychotics drugs. The study took place in the US between January 2001 and December 2004 (Lieberman et al, 2005). 397 American schizophrenia patients who participated in the CATIE GWAS project (Sullivan et al, 2008) were included in our AIP analysis. They had been randomized to treatment with antipsychotic monotherapy (perphenazine, olanzapine, quetiapine, risperidone and ziprasidone) during phase 1 of the study. Participation time ranged from two weeks (minimum) to 18 months (maximum). Patients were assessed for AIP with a modified Simpson Angus scale (SAS) at baseline, after one and three months, and then every three months up to 18 months. For statistical analysis, patients were dichotomized as cases (average Simpson Angus (SAS) global score >0.3, N=199) or controls (average SAS global score zero for the entire period of study, N=198) (Alkelai et al, 2009). Using logistic regression and controlling for possible confounders, we identified several single nucleotide polymorphism (SNP) associations with AIP severity (Supp. Table 1).

Although none of our associated SNPs reached GWAS significance level (probably due to the relatively small sample size), we decided to perform a validation study of the top results in an independent sample. This strategy assists in identifying true-positive associations in GWAS samples. In addition, we raise a novel hypothesis, that AIP related SNPs may also influence nigrostriatal pathway integrity. To obtain functional insights and provide proof of concept, we studied possible effects of AIP risk variants on level of nigrostriatal degeneration (evaluated by DAT SPECT imaging) among patients with early, idiopathic Parkinson’s disease (PD), in order to link genetic variants to pre-synaptic dopaminergic changes. Finally, in silico genotype-expression analysis was undertaken in human lymphoblasts to show how genetic variation may affect gene expression.

Methods

AIP Association Study.

Sample description and clinical methods: The sample is essentially the same as reported by us in a previous study (Greenbaum et al, 2009) although slightly smaller in terms of participant number (DNAs of 6 patients were unavailable). The sample and study design are described in detail in a previous publication (Smith et al, 2005). In brief, this was a cross-sectional study of patients with schizophrenia or schizoaffective disorder diagnosed according to DSM-IV criteria who were hospitalized at one of three tertiary care public hospitals in the United States and had been treated with a single antipsychotic agent (clozapine, olanzapine, risperidone or a typical antipsychotic) for at least a month. Patients gave written informed consent for participation in the study after the purpose and procedures were explained. The protocol and consent forms were approved by the Internal Review Board of each institution. Recruitment was consecutive and sampling procedures were continued until there were approximately 50 patients in each of the 4 groups. Clinical state was evaluated by Positive and Negative symptoms scale (PANSS). AIP was evaluated by the Simpson Angus Scale (SAS) (Simpson and Angus, 1970). The scale was administered on two separate occasions, separated by at least a week, by the same clinician. The mean score of the two SAS assessments was used for data analysis. Data on evaluation of tardive dyskinesia and akathisia were also collected but were not analyzed in the current context.

The overall sample for the current study (clinical ratings and DNA available) consisted of 178 patients of whom 111 were African-American (AA) and 67 Caucasians (40 of Hispanic and 27 of European origin). Further demographic and clinical data, including distribution among the antipsychotic treatment groups, antipsychotic dose at time of SAS evaluation (adjusted to chlorpromazine units) and total PANSS score, are given in Table 1.

SNP selection and genotyping: 15 SNPs which were associated with AIP in our previous study (Alkelai et al, 2009) with a P value 0.0001 were selected for the current study. SNP genotyping was performed with the Sequenom MassARRAY system, at the Washington University Human Genetics Division Genotyping Core, St. Louis, USA. Quality control measures were implemented.

Phenotype definition: A dichotomized AIP severity phenotype was used, based on the average of the two SAS mean scores (SASms) rated for a particular patient during his research participation period. Since a SASms threshold of 0.3 for parkinsonism is commonly accepted (Jano et al, 2005) and was also used in our previous AIP-GWAS (Alkelai et al, 2009), cases (AIP+) were defined as individuals whose average SASms was 0.3 and above, and controls (AIP-) as patients with average SASms less than 0.3.

For an additional analysis of extreme distribution of the phenotype, we defined controls as patients whose SASms was zero (absence of any parkinsonian features on two measurements), while cases were the same as described above (AIP+). This is a much more rigorous definition for controls, identical to the control definition in the AIP GWAS (Alkelai et al, 2009). However, only a small number of patients (41) met this extreme control criterion (Table 1).

Data analysis: To study association of the genotyped SNPs with AIP severity (dichotomized definition of the phenotype) we used logistic regression (additive model). Due to the mixed ethnicity of the US sample and its possible influence on allele frequencies, self reported ancestry (African Americans, White, Hispanic) was included as a covariate. In addition, based on clinical considerations (see Introduction), we identified 5 potential covariates to be checked for inclusion in our regression model: gender; age; antipsychotic type; antipsychotic dose at time of AIP assessment (standardized to chlorpromazine units) and total PANSS score (indication for disease severity). To select covariates, we checked for association of these variables with the dependent variable (AIP+/AIP-) using t test, Mann-Whitney or Chi-Square tests. Only significant variables (p<0.05) in the univariate analysis (conducted with SPSS Inc., Chicago, IL, USA) were included in the regression model. The final analysis, including Hardy-Weinberg equilibrium for the studied SNPs, was performed using PLINK software (Purcell et al, 2007). Hardy-Weinberg equilibrium for the studied SNPs was calculated with Haploview, version 4.1. The same analysis was implemented for the African-Americans subsample (without ancestry covariate).

Level of statistical significance required: Applying Bonferroni correction for multiple testing would have required a significance level of approximately 0.003. However, as in our previous study (Greenbaum et al, 2010), and due to the low power of this sample, we followed the criteria of Sullivan (2007) and Van der Oord et al (2008) concerning the appropriate required significance level. According to these criteria, an uncorrected standard P value of <0.05 may be used in a replication trial, if association is for the same SNP and phenotype and the direction of effect is the same as in the original report (see discussion section).

DAT SPECT Neuroimaging

Patient selection: The study sample (see Table 2) consisted of 49 Jewish patients with early stage (defined as maximum of 3 years disease duration) idiopathic PD attending the Movement Disorders Unit of the Rabin Medical Center (Petah Tiqwa, Israel), who underwent [123I]-FP-CIT SPECT (a ligand which binds to dopamine transporter) imaging. Clinical diagnosis of PD was based on the criteria of the United Kingdom PD Society Brain Bank (Hughes et al, 1992). Patients who manifested secondary parkinsonism were excluded. Some patients were being treated with anti PD agents when undergoing the imaging. The study was approved by the local Ethical Committee of Rabin Medical Center and all patients signed an informed consent form.

SPECT imaging: All patients received potassium iodide orally to block thyroid uptake of free radioactive iodide. A dose of 185 MBq [123I]-FP-CIT was injected intravenously and imaging was performed 3 h later. The SPECT study was performed using a dual-head -camera (Helix; Elscint) equipped with a low-energy, high-resolution collimator. A 20% window was centered on the 159-keV photopeak of 123I. One hundred twenty frames of 15s each were acquired into a 128 x 128 image matrix using a circular rotation mode. Transaxial, coronal, and sagittal slices 1 pixel thick were reconstructed using a third-order Metz filter set to 12-mm full width at half maximum. Attenuation was corrected with a constant linear attenuation coefficient of 0.11 cm–1.

Analysis of SPECT data: For analysis of striatal [123I]-FP-CIT binding, the two transaxial slices representing the most intense striatal binding were summed and subjected to qualitative analysis of tracer activity in the striatal regions. For quantitative analysis of tracer uptake, regions of interest (ROIs) were constructed manually with the help of a brain atlas in areas corresponding to the right and left putamen, caudate, and overall striatum. For background evaluation, ROI's were also drawn bilaterally in areas corresponding to the medial occipital lobe. For each ROI, mean counts were measured, and specific [123I]-FP-CIT uptake was calculated according to the following formula: specific [123I]-FP-CIT uptake = (mean activity in ROI – mean activity in occipital cortex) / mean activity in occipital cortex. SPECT analysis was performed by an expert in nuclear medicine.

Data analysis: To study the association of the ZFPM2 intronic SNP, rs12678719, with nigrostriatal pathway degeneration level, contralateral (CL) putamen uptake was selected as the outcome variable because it shows the most severe reduction of ligand uptake in early PD (Mozley et al, 2000; Winogrodzka et al, 2003; Marek et al, 2001).

For analysis of striatal [123I]-FP-CIT binding, the two transaxial slices representing the most intense striatal binding were summed and subjected to quantitative analysis of tracer activity in the CL Putamen region. Mean counts were measured, and specific [123I]-FP-CIT uptake was calculated according to the following formula: specific [123I]-FP-CIT uptake = (mean activity in Cl. putamen – mean activity in occipital cortex) / mean activity in occipital cortex. SPECT analysis was performed by an expert in nuclear medicine. This variable was distributed normally in our sample (Kolmogorov-Smirnov test).

We checked for significant difference in allele frequency of rs12678719 between the Ashkenazi and non-Ashkenazi groups, and no statistically difference was detected (Chi square=0.43, p=0.51) Age, gender, duration of PD and treatment with anti-parkinsonian agents may influence contralateral putamen uptake in early PD. We analyzed the association of these variables with contralateral putamen uptake; significant variables (p<0.05) were included the regression model. Finally, stepwise linear regression was used to evaluate the effect of ZFPM2 SNP rs12678719 together with relevant covariates, on contralateral putamen uptake (SPSS,Version 15).

nalysis

To study association of rs12678719 with ZFPM2 gene expression, we extracted genotype and ZFPM2 expression data in lymphoblastoid cell lines deposited in Genevar ( (Yang et al, 2010). The cells are derived from 4 population groups: European (Utah residents with Northern and Western European ancestry- CEU), Yoruba, Chinese and Japanese. Data analysis was performed by linear regression, under the additive model (allele dose effect). Further details regarding RNA preparation, gene expression quantification and normalization of raw data is given by Stranger et al (2007).

Results

AIP association study

The 15 selected SNPs (P<1x10-4 in the original AIP-GWAS) were successfully genotyped in the US AIP sample. One SNP (rs7174597) had a minor allele frequency <5%, and was therefore excluded. None of the remaining SNPs showed deviation from HWE. In the univariate analysis, we detected significant association of total PANSS score with AIP (extreme phenotype) in both the overall and African-American subsamples. This variable, in addition to ethnicity, was included in the regression model as a covariate. There was no significant association of age, sex and antipsychotic type or dose with AIP (wide and extreme phenotype definitions); thus they were not included in the regression model. Out of 14 analyzed SNPs, nominal association of the ZFPM2 gene intronic SNP, rs12678719, with AIP (P=5.97 x10-5 in the GWAS) was demonstrated. Controlling for ethnicity and PANSS total score, the 'G' allele of this SNP was found to be a susceptibility allele (same direction as in the original report), when comparing AIP affected patients (AIP+, SASms>0.3) (N=62) to those who do not have AIP (AIP-, SASms<0.3) (N=116) (p=0.009; OR=1.93) (Table 3). When applying the same extreme phenotype definition used in the original GWAS, identifying controls as individuals with SASms of zero (N=41) and cases as individuals with SASms>0.3 (N=62), the nominal association is still observed (p=0.017; OR=2.19) despite the decreased power. Since African-Americans represent the major ethnic group in this US sample, we studied the association of rs12678719 with AIP severity among AA separately. Association of the 'G' allele with AIP susceptibility was demonstrated, comparing AIP+ (N=38) to AIP- (N=73) (p=0.002; OR=2.85). Data regarding the consistency of ''G' risk allele over representation in cases versus controls in subgroups of this sample and our former CATIE AIP GWAS are presented in Table 3C. None of the other SNPs studied reached statistical significance (Supp. Table 1).