Oxidative stress in tardive dyskinesia: Genetic association study and meta-analysis of NADPH quinine oxidoreductase 1 (NQO1) and Superoxide dismutase 2 (SOD2, MnSOD) genes

Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

Contents lists available at ScienceDirect

Progress in Neuro-Psychopharmacology & Biological Psychiatry j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p n p

Oxidative stress in tardive dyskinesia: Genetic association study and meta-analysis of NADPH quinine oxidoreductase 1 (NQO1) and Superoxide dismutase 2 (SOD2, MnSOD) genes Clement C. Zai a,1, Arun K. Tiwari a,1, Vincenzo Basile a, Vincenzo de Luca a, Daniel J. Müller a, Aristotle N. Voineskos a, Gary Remington a, Herbert Y. Meltzer b, Jeffrey A. Lieberman c, Steven G. Potkin d, James L. Kennedy a,⁎ a

Neurogenetics Section, Centre for Addiction and Mental Health, 250 College Street, Toronto, Ontario, Canada M5T 1R8 Psychiatric Hospital at Vanderbilt University, 1601 23rd Avenue South, Suite 306, Nashville, Tennessee 37212, USA c New York State Psychiatric Institute, Columbia University Medical Centre, New York City, New York, USA d Brain Imaging Center, Irvine Hall, Room 166, University of California, Irvine, Irvine, California 92697-3960, USA b

a r t i c l e

i n f o

Article history: Received 6 July 2009 Received in revised form 16 September 2009 Accepted 16 September 2009 Available online 22 September 2009 Keywords: Manganese Superoxide Dismutase (MnSOD SOD2) NADPH quinine oxidoreductase 1 (NQO1) Oxidative stress Pharmacogenetics Schizophrenia Tardive dyskinesia (TD)

a b s t r a c t Introduction: Tardive dyskinesia (TD) is a potentially irreversible side effect of antipsychotic medication treatment that occurs in approximately 25% of chronically treated schizophrenia patients. Oxidative stress has been one of the proposed mechanisms influencing TD risk. Pae et al. (2004) originally reported a significant association between TD and the NADPH quinine oxidoreductase 1 (NQO1) gene Pro187Ser (C609T, rs1800566) polymorphism in Korean schizophrenia patients; however, subsequent studies have not consistently replicated these findings. Similarly, Hori et al. (2000) reported an association between TD and the Manganese superoxide dismutase SOD2 (MnSOD) gene Ala9Val (rs4880) polymorphism in a Japanese sample, but most research groups failed to replicate their positive findings. Aims: We investigated the role of the NQO1 polymorphism Pro187Ser and SOD2 (Ala9Val) in a group of wellcharacterized schizophrenia patients (N = 223) assessed for TD. We also performed a meta-analysis of all the previously published TD studies, including data from our sample, on these polymorphisms, Pro187Ser (N = 5 studies) and Ala9Val (N = 9 studies). Results: We did not observe a significant association of the Pro187Ser or Ala9Val polymorphism with TD occurrence or AIMS scores in our Caucasian and African American samples when analyzed independently. Meta-analysis did not reveal a significant association of the Pro187Ser/Ala9Val alleles or genotypes with TD occurrence. Conclusions: Neither the NQO1 Pro187Ser nor the SOD2 Ala9Val appear to play a major role in TD risk, although additional polymorphisms should be tested before the role of NQO1 and SOD2 in TD can be completely excluded. © 2009 Elsevier Inc. All rights reserved.

1. Introduction Tardive Dyskinesia (TD) in conjunction with antipsychotic use was first described by Faurbye et al. (1964), with “tardive” emphasizing delayed onset of involuntary movements. This is reflected in the current research criteria detailed in DSM-IV (APA 2000), which requires at least three months of antipsychotic exposure in nongeriatric samples. TD is a potentially irreversible movement side effect

Abbreviations: (MnSOD, SOD2), Manganese Superoxide Dismutase; (NQO1), NADPH quinine oxidoreductase 1; (TD), Tardive dyskinesia. ⁎ Corresponding author. Centre for Addiction and Mental Health, 250 College Street, Room 30, Toronto, Ontario, Canada M5T 1R8. Tel.: +1 416 979 4987; fax: +1 416 979 4666. E-mail address: [email protected] (J.L. Kennedy). 1 Contributed equally to the work. 0278-5846/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.pnpbp.2009.09.020

that occurs in approximately 25% of patients exposed to chronic typical antipsychotic treatment (Margolese et al., 2005), and is characterized by involuntary athetoid (slow), choreiform (fast), and/or rhythmic (stereotypic) movements affecting mostly orofacial muscles, with more severe cases involving the trunk and limbs. These movements can be both debilitating and a source of social stigmatization, adversely affecting medication adherence and outcome. Unfortunately, clear predictors of who is likely to develop TD have, to date, remained elusive. The familial nature of TD suggests a strong genetic basis, while factors such as age, female gender, diagnosis of affective disorder and organic brain dysfunction have also been identified as risk factors (reviewed in Müller et al., 2004). Various factors have been implicated in TD's pathophysiology, including oxidative stress (Lohr et al., 2003). The brain is vulnerable to oxidative stress for several reasons. First, it uses a great deal of energy,

C.C. Zai et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

and the brain has large amounts of polyunsaturated fatty acids that are substrates for lipid peroxidation cascades. Further, certain brain regions, the basal ganglia in particular, contain large amounts of transition metals, some of which are involved in the formation of hydroxyl radicals via superoxide dismutase (SOD). The basal ganglia are also rich in dopamine, a neurotransmitter that can auto-oxidize to produce dopamine quinone free radicals, or be metabolized by monoamine oxidase (MAO) to produce hydrogen peroxide. Acute haloperidol treatment increases the oxidized-to-reduced glutathione ratio in murine brain (Cohen and Spina, 1989). Furthermore, long-term haloperidol administration reduces rat brain SOD and catalase, with a corresponding increase in membrane lipid oxidation products (Pillai et al., 2007). These observations suggest that haloperidol increases oxidative stress. Sagara (1998) have also reported increased levels of oxidized glutathione in neuronal cultures, finding that oxidative stress originated from the mitochondria and not dopamine metabolism. In accord with this line of thinking, Vitamin E and melatonin, both with antioxidant activity, have been tested in TD. A meta-analysis of the effect of Vitamin E on TD found 28.3% of Vitamin-E treated patients compared to 4.6% of placebo-treated patients had their AIMS scores (a scale that measures involuntary movements) decrease by at least one third (Barak et al., 1998); however, the results were preliminary and need to be confirmed with larger, long-term randomized prospective trials (Soares and McGrath, 2001). A double-blind placebo controlled trial involving melatonin reported an average decrease in AIMS scores of 2.45 points, compared to 0.77 points in the placebo group (Shamir et al., 2001). Finally, antioxidants such as melatonin, quercetin and AD4 have been reported to reduce haloperidol-induced vacuous chewing moments in rats, an accepted animal model for TD (Adler et al., 1999; Naidu et al., 2003, Sadan et al., 2005). Based on this above evidence, several genes involved in the maintenance of cellular redox balance have been investigated, namely neuronal nitric oxide synthase (nNOS; NOS1), endothelial NOS (eNOS; NOS3), glutathione S-transferase (GST; GSTMI, GSTTI, GSTPI), superoxide dismutase 2 (MnSOD, SOD2) and NAD(P)H: quinone oxidoreductase 1 (NQO1). The nitric oxide synthase NOS1 gene has not been found to be associated with TD (Shinkai et al., 2002; Wang et al., 2004; Thelma et al., 2007). The NOS3 (Liou et al., 2006) and glutathione S-transferase GSTM1 (de Leon et al., 2005) positive findings require replication, while the glutathione peroxidase (GPX1; Shinkai et al., 2006), GSTT1 (de Leon et al., 2005; Thelma et al., 2007), GSTP1 (Shinkai et al., 2005; Thelma et al., 2007), and APOE (Halford et al., 2006) genes need to be examined more thoroughly. The SOD2 gene (MIM# 147460; 6q25.3) codes for the Manganese Superoxide Dismutase (MnSOD), an important mitochondrial enzyme involved in detoxification of superoxide radical to hydrogen peroxide. Any perturbation in this activity can leave the superoxide and/or hydrogen peroxide unattended which can be cytotoxic. Increased levels of lipid peroxidation in both CSF and plasma are significantly correlated with severity of TD (Lohr et al., 1990; Peet et al., 1993; Brown et al., 1998), while reduced SOD activity in erythrocytes as well as CSF has been reported in schizophrenia patients with TD compared to those without TD (Yamada et al., 1997; Tsai et al. 1998). Zhang et al. (2007) have reported decreased levels of SOD, glutathione peroxidase and catalase along with increased levels of malondialdehyde in the plasma of TD patients. The SOD2 gene contains a functional polymorphism Ala9Val (rs4880) in the mitochondrial targeting sequence (MTS), of which the Val allele causes a conformational change in the MTS that misdirects intracellular trafficking of the protein (ShimodaMatsubayashi et al., 1996; Rosenblum et al., 1996). Genetic studies have suggested that the SOD2 Ala allele may be protective against TD (Hori et al., 2000; Galecki et al., 2006), though neither Zhang et al. (2002b) nor Thelma et al. (2007) replicated this finding. A recent meta-analysis, which included 4 studies, found a significant overall protective effect of Val9 against TD (Bakker et al., 2008).

51

NAD(P)H:quinine oxidoreductase 1 (NQO1; MIM# 125860) is a 35.5 kb gene consisting of 6 exons and is located on chromosome 16q22.1. It codes for a phase II detoxifying homodimeric flavoprotein that catalyzes two-electron reduction of quinines to hydroquinones. It is also involved in the detoxification of superoxide radicals to hydrogen peroxide. Its expression is induced by high free radical concentrations under conditions of oxidative stress (Siegel et al., 2004). NQO1, in the presence of electron donors NADPH/GSH and transition metals copper or iron, produces hydroxyl radicals, a more toxic reactive oxygen species. NQO1 is expressed in the substantia nigra and carries a common functional polymorphism, Pro187Ser (C609T, rs1800566), in exon 6. Individuals carrying variant Ser/Ser genotype (TT) express just trace amount of NQO1 protein and exhibit only 2–4% of the quinone reductase activity compared to the wildtype Pro/Pro genotype (CC; Traver et al., 1997; Siegel et al., 1999). Individuals heterozygous for this polymorphism exhibit an approximately two-fold decrease in NQO1 activity (Misra et al., 2000). Hori and coworkers (2006) could not replicate the association between TD and the NQO1 Ser allele that was first reported by Pae et al. (2004), though Liou et al. (2005) found higher average AIMS scores in Ser/Ser homozygotes than in Pro/Pro homozygotes and Pro/Ser heterozygotes, which is in agreement with the Pae et al. paper. Thelma et al. (2007) reported an opposite trend, with patients carrying the Ser/ Ser genotype (TT) having the lowest average AIMS scores compared to patients carrying the Pro/Pro and Pro/Ser genotypes. The present study represents an investigation of the role of NQO1 Pro187Ser (C609T) and the SOD2 Ala9Val polymorphisms in a group of well-characterized schizophrenia patients evaluated for TD, with Caucasian and African American patients analyzed separately. We also present a meta-analysis of all the previous studies that have included these two polymorphisms and discuss our observations in light of these findings. 2. Methods 2.1. Subjects The sample for this study is largely the same as that reported in Zai et al. (2007). Subjects were recruited from four clinical sites in North America: Center for Addiction and Mental Health in Toronto, Ontario (Gary Remington (GR), N = 92); Case Western Reserve University in Cleveland, Ohio (Herbert Y Meltzer (HYM), N = 69); Hillside Hospital in Glen Oaks, New York (Jeffrey A Lieberman (JAL), N = 50); University of California at Irvine, California (Steven G Potkin (SGP), N = 12). In total, 223 schizophrenia patients were studied. All patients had undergone at least one year of treatment with typical antipsychotics. More specifically, half of the sample (HYM, JAL, SGP) was collected in the context of a clinical trial where patients were treated with typical antipsychotics and assessed for TD before administration of clozapine. The majority of patients in the remaining sample (GR) were on atypical antipsychotics at the time of TD assessment; however, they had all been previously treated with typical antipsychotics for at least one year. The presence of TD was assessed using the Abnormal Involuntary Movement Scale (AIMS) or the modified Hillside Simpson Dyskinesia Scale (HSDS) in the case of 50 patients (JAL) using the Schooler and Kane criteria; that is, a patient has TD if he/she has mild dyskinesia in at least two body items or moderate dyskinesia in at least one body item (Guy, 1976; Schooler and Kane, 1982; Basile et al., 1999). 2.2. Genotyping Genomic DNA was purified from whole blood samples using the non-enzymatic method previously described (Lahiri and Nurnburger, 1991). Genotyping was done after the subjects had completed the follow-up, and all laboratory staff was blind to AIMS scores. The non-synonymous polymorphism NQO1 Pro187Ser (also known as

52

C.C. Zai et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

rs1800566, C609T) was genotyped using Assays-on-Demand (Assay ID C___2091255_30, Applied Biosystems Inc (ABI), Foster City, CA). 10 μL Polymerase Chain Reactions (PCRs) on 20 ng genomic DNA were performed using Assays-on-Demand with the following conditions: 95 °C 10 min, followed by 50 cycles of 92 °C 15 s, 60 °C 1 min. Determination of alleles was performed using the ABI model 7500 machine with the Allelic Discrimination module in the sequence detection software v1.3.1.21. The non-synonymous polymorphism SOD2 Ala9Val (also known as rs4880) was genotyped using restriction fragment length polymorphism. First, a 172-bp fragment of the human SOD2 mitochondrial targeting sequence was amplified by PCR using the primers: 5′-GGC TGT GCT TTC TCG TCT TC-3′ and 5′-GGT GAC GTT CAG GTT GTT CA-3′. The PCR products were digested with the restriction endonuclease BsaWI and submitted to electrophoresis on 3% agarose gels. The Ala9 allele was shown as one ~85-bp band on the agarose gel indicating two DNA fragments of 85 bp and 87 bp, while the Val9 allele was shown as an undigested 172-bp fragment. 2.3. Statistics Statistical analyses of individual polymorphisms and haplotypes in TD and AIMS scores were carried out as previously described (Zai et al., 2007). Statistical analysis was conducted using SPSS version 15.0. Genotype frequency distribution was tested for fitness to Hardy– Weinberg equilibrium. The association of genotype frequencies with age and AIMS scores were assessed using ANOVA, and where the variances of AIMS scores among genotypes differed significantly with the Levene's Test for Homogeneity of Variances, AIMS scores were tested with the Kruskal–Wallis test. Gender differences in genotype frequencies were assessed using the χ2 test. The differences in allele and genotype frequencies between patients with and without TD were analyzed by χ2 test. For contingency tables with at least one expected cell count of less than five, two-tailed Fisher's Exact Tests were performed (URL: http://home.clara.net/sisa/fiveby2.htm). 2.4. Meta-analysis The PubMed database of National Centre for Biological Information (NCBI) was searched using the key terms “Tardive Dsykinesia” and “Superoxide dismutase 2” or SOD2 or Manganese superoxide dismutase or MnSOD or NAD(P)H:quinine oxidoreductase 1 or NQO1. Furthermore, articles cited in the papers retrieved were also screened for any relevant information. Data was extracted from these publications. In case the information was not available in the publication then it was obtained from the corresponding author. We analyzed the data using the STATA Release 8 statistical software package (StataCorp. Stata

Statistical Software: Release 8. College Station, TX: StataCorp LP). The odds ratios and confidence intervals of NQO1 Pro187Ser and SOD2 Ala9Val for TD from the individual studies were calculated using the “metan” command, with the pooled odds ratio and standard error calculated under the random effects and fixed effects model. The possible effects of ethnicity, age, and sex ratio on heterogeneity amongst the studies were assessed by meta-regression analysis using the “metareg” command. Publication bias was estimated using the “metabias” command using the Begg's and Egger's tests. 3. Results The sample in the present study consisted of 193 caucasians with a mean age of 37.7 +/− 10.1 years (66% males). Of these, 76 were diagnosed with TD, and a total of 162 patients had AIMS scores available for quantitative analysis. The African American sample consisted of 31 patients with a mean age of 32.5 +/− 10.9 (70% male). A total of 11 patients were diagnosed to be suffering from TD (Tables 1 and 2). Due to the small sample size, genotype and haplotype analyses will not yield conclusive information. Thus, African Americans were used only in the allele frequency association tests. 3.1. Association study of NQO1 Pro187Ser and SOD2 Ala9Val with TD The NQO1 Pro187Ser and the SOD2 Ala9Val genotypes were in Hardy–Weinberg Equilibrium for both of our Caucasian and African American samples (p > 0.1). The allelic and genotypic distributions were not significantly different between the patients with and without TD (Table 1). The average total AIMS scores were not significantly different among the genotype groups (Table 1). We performed logistic regression on TD status with age and sex as covariates. The results were similarly non-significant. 3.2. Meta-analysis of NQO1 Pro187Ser (C609T) in TD We carried out a computer search on the National Library of Medicine's PubMed online search engine database for all papers published up to August 2008 using the search terms “tardive dyskinesia” and “NAD(P)H:quinine oxidoreductase 1” or “NQO1”. In all, four genetic association studies were found reporting on TD and NQO1. The number of patients with and without TD and genotypes for NQO1 Pro187Ser was available for all four studies (Pae et al., 2004; Liou et al., 2005; Hori et al., 2006; Thelma et al., 2007). All subjects were selected based on their diagnoses of schizophrenia or schizoaffective disorder, according to DSM-III-R or DSM-IV, using case records with or without patient interviews. The presence of TD was assessed using the AIMS, or the modified Hillside

Table 1 Results from statistical analysis of demographics (sex, age) as well as total AIMS scores among NQO1 Pro187Ser and SOD2 Ala9Val genotypes, and TD diagnoses with genotypes and alleles of the NQO1 Pro187Ser and SOD2 Ala9Val polymorphism. Polymorphism

Genotypes

N(M/F)a

Agea (years)

Total AIMS scoresa

TDa (Yes/No)

Alleles

TDa (Yes/No)

TDb (Yes/No)

NQO1 rs1800566 C609T, Pro187Ser

Pro/Pro Pro/Ser Ser/Ser

123 (78/45) 60 (43/17) 5 (3/2) 0.551⁎ 51 (33/18) 87 (59/28) 47 (30/17) 0.876

37.50 +/− 9.64 38.75 +/− 10.98 37.20 +/− 13.85 0.732 37.73 +/− 9.38 37.43 +/− 10.97 38.75 +/− 9.28 0.763

6.36 +/− 7.79 5.60 +/− 6.85 4.00 +/− 5.83 0.692 5.57 +/− 6.72 5.79 +/− 7.65

123 (47/76) 60 (25/35) 5 (1/4) 0.729⁎ 51 (21/30) 91 (31/60) 48 (24/24) 0.186

Pro Ser

119/187 27/43

20/31 2/7

Ala Val

0.961 73/120 79/108

0.464⁎ 9/13 13/23

0.379

0.715

p-value SOD2 rs4880 Ala9Val p-value

Ala/Ala Ala/Val Val/Val

0.441

Abbreviations: TD, tardive dyskinesia; AIMS, abnormal involuntary movement scale; N; number of patients; M/F, males/females. ⁎Expected cell count less than 5. p-value for Fisher's Exact Test shown. a Caucasian patients. b African American patients.

C.C. Zai et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

53

Table 2 Demographics of studies included in the present allelic meta-analysis. NQO1 studies

Ethnicity

Age

M/Fa

Pae et al. (2004) Liou et al. (2005) Hori et al. (2006) Thelma et al. (2007) Present study Present study P(metareg) P(metareg) (excluding Pae'04 study)

East Asian East Asian East Asian N. Indian Caucasian Afr. Amer. 0.952 0.624

45.4 +/− 9.59 46.7 +/− 9.29 53.2 +/− 11.5 32.28 +/− 10.9 37.7 +/− 10.1 32.5 +/− 10.9 0.622 0.723

73/34 172/108 115/107 147/125 129/64 21/9 0.157 0.670

SOD2 studies

Ethnicity

Age

M/Fa

Hori et al. (2000) Zhang et al. (2002) Akyol et al. (2005) Thelma et al. (2007) Pae et al. (2007) Hitzeroth et al. (2007) Galecki et al. (2006) Kang et al. (2008) Present study Present study P(metareg) (excluding Akyol'05 study) P(metareg) (excluding Akyol'05, Hitzeroth'07, Pae'07, and Galecki'06 studies)

East Asian East Asian Caucasian N. Indian East Asian Afr. Amer. Caucasian East Asian Caucasian Afr. Amer. 0.365 0.670

55.61 +/− 9.09 55.34 +/− 8.53 37.68 +/− 10.82 31.98 +/− 10.87 44.69 +/− 9.50 34.14 +/− 10.39 55.55 +/− 9.98 45.22 +/− 9.55 37.7 +/− 10.1 32.5 +/− 10.9 0.160 0.755

97/95 99/0 94/59 164/135 178/84 206/52 75/47 110/99 129/64 21/9 0.790 0.493

MAF (Ser) 0.39 0.46 0.36 0.29 0.19 0.15 0.542 0.883 MAF (Ala) 0.12 0.16 0.44 0.54 0.09 0.42 0.26 0.10 0.51 0.38 0.685 0.487

Abbreviations: M/F, males/females; MAF, minor allele frequency. a Percentage male as variable for the meta-regression analysis.

Simpson Dyskinesia Scale (HSDS) in the case of 50 patients. TD ratings were performed once on each patient for the majority of the studies. In all, 1071 schizophrenia patients were genotyped for NQO1 Pro187Ser. Of those, 396 were positive for the diagnosis of TD. Demographic information for each of the studies included in the metaanalysis is shown in Table 2. In all studies, the sample sets were in Hardy–Weinberg equilibrium (p > 0.05). The allelic as well as genotypic analysis, using the random effect model, did not show a significant association with TD occurrence [Table 3a: (Odds Ratio) ORSer = 1.05, 95% confidence interval (95% CI): 0.78–1.41, p = 0.76; Table 3b: ORPro/Ser & Ser/Ser = 0.97, 95% CI: 0.66– 1.41, p = 0.87]. There was weak evidence for heterogeneity among the six studies (Allelic test, p = 0.08; Genotypic test p = 0.12). Since Pae et al. (2004) reported the first and most significant finding and showed an opposite trend from all other studies, we excluded the sample and reanalyzed the data. The allelic as well as genotypic results remained non-significant [Table 3a: ORSer = 0.906, 95% CI: 0.74–1.12; p = 0.36; Table 3b: ORPro/Ser & Ser/Ser = 0.82, 95% CI: 0.61–1.08, p = 0.16]. Furthermore, heterogeneity was not observed after the removal of the Pae et al. (2004) sample (Allelic test, p = 0.97; Genotypic test p = 0.70). Ethnicity, gender ratio, and age did not have any influence on the results observed for Pro187Ser (P > 0.1; Table 2). Publication bias was not significant with [p(Begg) = 0.13; p(Egger) = 0.44)], and without [(p(Begg) = 1; p(Egger) = 0.71)] the Pae et al. (2004) sample. Similar lack of effect was observed when the fixed effect model was

used to conduct the meta-analysis, excluding the Pae et al. (2004) paper (data not shown). 3.3. Meta-analysis of SOD2 Ala9Val in TD We carried out a computer search on the National Library of Medicine's PubMed online search engine database for all papers published up to August 2008 using the search terms “tardive dyskinesia” and “Superoxide dismutase 2” or “SOD2” or “Manganese superoxide dismutase” or “MnSOD”. In all, eight genetic association studies were found reporting on TD and SOD2 (MnSOD). The number of patients with and without TD and genotypes for SOD2 Ala9Val was available for all eight studies (Hori et al., 2000; Zhang et al., 2002; Akyol et al., 2005; Thelma et al., 2007; Pae et al., 2007; Hitzeroth et al., 2007; Galecki et al., 2006; Kang et al., 2008). All subjects were selected based on their diagnoses of schizophrenia or schizoaffective disorder, according to DSM-III-R or DSM-IV, using case records with or without patient interviews. The presence of TD was assessed using the AIMS, or the modified Hillside Simpson Dyskinesia Scale (HSDS) in the case of 50 patients from our study as described above, except for the study by Hitzeroth (2007). TD ratings were performed once for each patient in the majority of the studies. In all, 1771 schizophrenia patients were genotyped for SOD2 Ala9Val, and 484 were positive for the diagnosis of TD. Demographic information for each of the studies included in the meta-analysis is shown in Table 2.

Table 3a Results from meta-analysis of NQO1 Pro187Ser alleles in tardive dyskinesia. Study

TD(+) Pro

TD(+) Ser

TD(−) Pro

TD(−) Ser

OR (Ser)

95% LCI

95% UCI

%Weight

Pae et al. (2004) Liou et al. (2005) Hori et al. (2006) Thelma et al. (2007) Present study Present study Overallas Overall (excluding Pae'04 study)b

43 170 60 113 119 19

45 144 32 39 27 3

86 132 222 274 187 32

39 118 130 118 43 6

2.308 0.948 0.911 0.801 0.987 0.842 1.047 0.906

1.313 0.679 0.563 0.525 0.579 0.188 0.780 0.735

4.055 1.322 1.473 1.224 1.682 3.765 1.405 1.117

15.62 24.90 18.52 20.83 16.62 3.497 100 100

Abbreviations: TD, tardive dyskinesia; OR, odds ratio, LCI, lower confidence interval; UCI, upper confidence interval. a Heterogeneity χ2 = 9.78 (df = 5) p = 0.082; Test of OR = 1: z = 0.30 p = 0.762. b Heterogeneity χ2 = 0.50 (df = 4) p = 0.973; Test of OR = 1: z = 0.92 p = 0.357.

54

C.C. Zai et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

Table 3b Results from meta-analysis of NQO1 Pro187Ser genotypes in tardive dyskinesia. Study

TD(+) Pro/Pro

TD(+) Pro/Ser and Ser/Ser

TD(−) Pro/Pro

TD(−) Pro/Ser and Ser/Ser

OR (Pro/Ser and Ser/Ser)

95% LCI

95% UCI

%Weight

Pae et al. (2004) Liou et al. (2005) Hori et al. (2006) Thelma et al. (2007) Present study Present study Overalla Overall (excluding Pae'04 study)b

12 50 19 44 47 8

32 107 27 32 26 3

31 30 74 95 76 13

32 95 102 101 39 6

2.583 0.676 1.031 0.684 1.078 0.813 0.969 0.815

1.130 0.398 0.533 0.401 0.583 0.157 0.665 0.613

5.907 1.148 1.992 1.168 1.994 4.197 1.410 1.083

13.62 22.30 17.97 22.14 19.33 4.630 100 100

Abbreviations: TD, tardive dyskinesia; OR, odds ratio, LCI, lower confidence interval; UCI, upper confidence interval. a Heterogeneity χ2 = 8.86 (df = 5) p = 0.115; Test of OR = 1: z = 0.17 p = 0.868. b Heterogeneity χ2 = 2.17 (df = 4) p = 0.704; Test of OR = 1: z = 1.41 p = 0.159.

In all studies except for Akyol et al. (2005), the sample sets were in Hardy–Weinberg equilibrium (p > 0.05). We removed the Akyol et al. (2005) study from our meta-analysis. The allelic as well as genotypic analysis, using the random effect model, did not show a significant association with TD occurrence [Table 4a: ORAla = 0.83, 95% CI: 0.54– 1.29, p = 0.42]; Table 4b: ORAla/Ala & Ala/Val = 0.72, 95% CI: 0.43–1.21, p = 0.21]. There was evidence for heterogeneity among the nine samples (Allelic test, p < 0.001; Genotypic test p < 0.001). We removed the Hitzeroth et al. (2007) study because TD assessment in this study was different from the others. The results did not differ significantly, and significant heterogeneity remained. We then excluded the two samples with the largest and smallest Odds Ratios (Galecki et al., 2006; Pae et al.,

2007) and reanalyzed the data. The allelic as well as genotypic results remained non-significant [Table 4a: ORAla = 0.90, 95% CI: 0.66–1.23; p = 0.53; Table 4b: ORAla/Ala & Ala/Val = 0.79, 95% CI: 0.52–1.19, p = 0.26]. Heterogeneity was not significant after the removal of the Galecki et al. (2006) and Pae et al. (2007) samples (Allelic test, p = 0.18; Genotypic test p = 0.19). Ethnicity, sex ratio, and age did not have a significant influence on the results observed for Ala9Val (P > 0.05; Table 2). Publication bias was not significant with [p(Begg) = 0.60; p(Egger) = 0.94)], and without [(p(Begg) = 0.71; p(Egger) = 0.99)] the Hitzeroth et al. (2007), Galecki et al. (2006), and Pae et al. (2007) samples. Similar lack of effect was observed when the fixed effect model was used to conduct the allelic and genotypic meta-analyses that excluded the Akyol

Table 4a Results from meta-analysis of SOD2 Ala9Val alleles in tardive dyskinesia. Study

TD(+) Ala

TD(+) Val

TD(−) Ala

TD(−) Val

OR (Ala)

95% LCI

95% UCI

%Weight

Hori et al. (2000) Zhang et al. (2002) Akyol et al. (2005) Thelma et al. (2007) Pae et al. (2007) Hitzeroth et al. (2007) Galecki et al. (2006) Kang et al. (2008) Present study Present study Overall (excluding Akyol'05 study)a Overall (excluding Pae'04, Akyol'05, Hitzeroth'07, and Galecki'07 studies)b

3 12 16 93 12 30 12 20 73 9

75 72 30 83 76 30 102 146 79 13

42 21 118 228 37 165 52 21 120 13

264 97 142 194 399 243 78 231 108 23

0.251 0.770 0.642 0.953 1.703 1.473 0.176 1.507 0.832 1.225 0.83 0.90

0.076 0.356 0.334 0.670 0.849 0.855 0.088 0.789 0.551 0.412 0.54 0.66

0.834 1.666 1.234 1.356 3.414 2.536 0.353 2.876 1.254 3.638 1.29 1.23

7.2 10.5 0 14.1 11.2 12.6 11.2 11.6 13.7 7.9 100 100

Abbreviations: TD, tardive dyskinesia; OR, odds ratio, LCI, lower confidence interval; UCI, upper confidence interval. a Heterogeneity χ2 = 35.42 (df = 8) p < 0.001; Test of OR = 1: z = 0.81 p = 0.418. b Heterogeneity χ2 = 7.54 (df = 6) p = 0.183; Test of OR = 1: z = 0.63 p = 0.528.

Table 4b Results from met a-analysis of NQO1 Pro187Ser genotypes in tardive dyskinesia. Study

TD(+) Ala/Ala and Ala/Val

TD(+) Val/Val

TD(−) Ala/Ala and Ala/Val

TD(−) Val/Val

OR (Ala/Ala and Ala/Val)

95% LCI

95% UCI

%Weight

Hori et al. (2000) Zhang et al. (2002) Akyol et al. (2005) Thelma et al. (2007) Pae et al. (2007) Hitzeroth et al. (2007) Galecki et al. (2006) Kang et al. (2008) Present study Present study Overall (excluding Akyol'05 study)a Overall (excluding Pae'04, Akyol'05, Hitzeroth'07, and Galecki'06 studies)b

3 12 14 67 12 21 11 17 52 7

36 30 9 21 32 9 46 66 24 4

39 21 106 161 37 143 40 20 90 11

114 38 24 50 181 61 25 106 24 7

0.244 0.724 0.352 0.991 1.834 0.995 0.149 1.37 0.578 1.114 0.72 0.79

0.071 0.308 0.137 0.553 0.865 0.431 0.065 0.667 0.298 0.236 0.43 0.52

0.836 1.703 0.908 1.777 3.891 2.297 0.341 2.793 1.119 5.255 1.21 1.19

8.4 11.2 0 13.5 12.1 11.4 11.5 12.4 12.9 6.6 100 100

Abbreviations: TD, tardive dyskinesia; OR, odds ratio, LCI, lower confidence interval; UCI, upper confidence interval. a Heterogeneity χ2 = 28.27 (df = 8) p < 0.001; Test of OR = 1: z = 1.24 p = 0.214. b Heterogeneity χ2 = 7.44 (df = 6) p = 0.19; Test of OR = 1: z = 1.13 p = 0.259.

C.C. Zai et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

et al. (2005), Hitzeroth et al. (2007), Galecki et al. (2006), and Pae et al. (2007) papers (data not shown).

55

of studies identifying genetic factors in TD genesis have been conducted although consensus loci are lacking, highlighting the need for ongoing research in this area.

4. Discussion Cellular redox imbalance has been attributed to the development of several neurodegenerative diseases such as Alzheimer's, Parkinson's, and Huntington's disease. The brain is especially prone to oxidative stress due to its high oxygen consumption, abundance of polyunsaturated fatty acids, transition metals, and relatively low levels of antioxidant enzymes (Calabrese et al. 2000). The importance of the antioxidant system in TD is supported by the alterations in levels of antioxidant enzymes and lipid peroxidation in TD patients as well as in animal models. In this study, we investigated the possible role of the NQO1 Pro187Ser (C609T) and SOD2 Ala9Val polymorphisms in TD as candidates risk genes for TD based on their role in detoxification of superoxide radical, a putative mechanism for development of TD. NQO1 has been investigated for association with TD, and Pae et al. (2004) reported a significant association with both the occurrence and severity of TD. However, none of the studies with patients of similar (East Asians: Liou et al., 2005; Hori et al., 2006) or different ethnic groups (North Indians; Thelma et al., 2007) replicated the effect reported by Pae et al. (2004). We also failed to replicate the association of NQO1 Pro187Ser in our TD sample. To further investigate if this polymorphism has any significance in TD pathogenesis, we conducted a meta-analysis of all the samples reported to date. The results were not significant, suggesting that the NQO1 polymorphism may not play a major role in the etiopathogenesis of TD. It is possible that the sample sizes in the meta-analysis, as well as the genetic association study we conducted, were insufficient to detect small genetic effect sizes commonly observed in complex traits; additional studies with larger sample sets are required to categorically rule out the role of NQO1 gene in TD development. The previous association studies for SOD2 (MnSOD) have also yielded conflicting results. Hori et al. (2000) and Galecki et al. (2006) reported significant associations between the Ala9Val polymorphism and TD. However, other research groups have not replicated these observations (Zhang et al., 2002; Akyol et al., 2005; Thelma et al., 2007; Pae et al., 2007; Hitzeroth et al., 2007; Kang et al., 2008). It is important to note that the Val allele frequency is higher in the East Asian population than in other ethnicities although neither minor allele frequencies nor ethnicities contributed to the findings in our metaanalyses. Nonetheless, we conducted a stratified meta-analysis of SOD2 Ala9Val for East Asian and Caucasian samples separately but did not observe a significant contribution of this polymorphism to TD. Recently Bakker et al. (2008) conducted a meta-analysis of the Ala9Val polymorphism and reported a protective effect for Ala/Val [OR = 0.37 (0.17–0.79]; p = 0.009) and for the Val/Val and Ala/Val genotypes– [OR = 0.49 (0.24–1.00), p = 0.047]. We did not replicate their findings despite having a larger sample size (n = 1771). Based on the observations made in this study, we conclude that the Pro187Ser polymorphism in NQO1 and the Ala9Val polymorphism in SOD2 have a limited role in the development of TD. However, the role of the NQO1 and SOD2 genes in TD cannot be dismissed yet as other polymorphisms within and surrounding these genes have not been systematically tested. In addition, heterogeneity due to mixed ethnicities and treatment history may have obscured the results of the meta-analysis. Additional sufficiently powered studies in different ethnicities may bring up NQO1 and SOD2 as a risk factor for TD in certain populations and not others. Environmental factors, such as diet, are likely to differ among the geographical locations and may interact with genetic predispositions to influence susceptibility to oxidative stress in individual populations. In this regard, learning more about environmental influences in TD susceptibility will help limit confounding factors in future genetic studies of TD. To date a number

Acknowledgements We acknowledge our funding sources: the Canadian Institute for Health Research (CIHR) MOP79525, CIHR fellowship to AKT, the Bebensee Foundation, the C. R. Younger Foundation, the Prentiss Foundation, and the Ritter Foundation; CIHR, the Bebensee Foundation, the C. R. Younger Foundation, the Prentiss Foundation, and the Ritter Foundation had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. We also thank the participants in our study.

References Adler LA, Rotrosen J, Edson R, Lavori P, Lohr J, Hitzemann R, et al. Vitamin E treatment for tardive dyskinesia. Veterans Affairs Cooperative Study#394 Study Group. Arch Gen Psychiatry 1999;56:836–41. Akyol O, Yanik M, Elyas H, Namli M, Canatan H, Akin H, Yuce H, Yilmaz HR, Tutkun H, Sogut S, Herken H, Ozyurt H, Savas HA, Zoroglu SS. Association between Ala-9Val polymorphism of Mn-SOD gene and schizophrenia. Prog Neuropsychopharmacol Biol Psychiatry 2005;29(1):123–31. APA (American Psychiatric Association). Diagnostic and Statistical Manual of Mental Disorders (4th edn – text revision). Washington, DC: American Psychiatric Association, 2000. Bakker PR, van Harten PN, van Os J. Antipsychotic-induced tardive dyskinesia and polymorphic variations in COMT, DRD2, CYP1A2 and MnSOD genes: a metaanalysis of pharmacogenetic interactions. Mol Psychiatry. 2008;13(5):544–56. Barak Y, Swartz M, Shamir E, Stein D, Weizman A. Vitamine E (a-Tocopherol) in the treatment of tardive dyskinesia. Ann Clin Psychiatry. 1998;10(3):101–5. Basile VS, Masellis M, Badri F, Paterson AD, Meltzer HY, Lieberman JA, Potkin SG, Macciardi F, Kennedy JL. Association of the MscI polymorphism of the dopamine D3 receptor gene with tardive dyskinesia in schizophrenia. Neuropsychopharmacol. 1999;21:17–27. Brown K, Reid A, White T, Henderson T, Hukin S, Johnstone C, et al. Vitamin E, lipids, and lipid peroxidation products in tardive dyskinesia. Biol Psychiatry. 1998;43(12):863–7 Jun 15. Calabrese V, Bates TE, Stella AM. NO synthase and NO-dependent signal pathways in brain aging and neurodegenerative disorders: the role of oxidant/antioxidant balance. Neurochem Res 2000;25:1315–41. Cohen G, Spina MB. Deprenyl suppresses the oxidant stress associated with increased dopamine turnover. Ann Neurol. 1989;26:689–90. De Leon J, Susce MT, Pan R-M, Koch WH, Wedlund PJ. Polymorphic variations in GSTM1, GSTT1, PgP, CYP2D6, CYP3A5, and Dopamine D2 and D3 Receptors and their association with tardive dyskinesia in severe mental illness. J Clin Psychopharmacol 2005;25:448–56. Faurbye A, Rasch PJ, Rasch PB, et al. Neurological syndromes in pharmacotherapy of psychosis. Acta Psychiatr Scand 1964;40:10–27. Galecki P, Pietras T, Szemraj J. Manganese superoxide dismutase gene (MnSOD) polymorphism in schizophrenics with tardive dyskinesia from central Poland. Psychiatr Pol 2006;40(5):937–48. Guy W. ECDEU Assessment Manual for PsychopharmacologyRevised edn. . Washington, DC: Department of Health, Education and Welfare; 1976. Halford J, Mazeika G, Slifer S, Speer M, Saunders AM, Strittmatter WJ, Morgenlander JC. APOE2 allele increased in tardive dyskinesia. Mov Disord 2006;21(4):540–75. Hitzeroth A, Niehaus DJ, Koen L, Botes WC, Deleuze JF, Warnich L. Association between the MnSOD Ala-9Val polymorphism and development of schizophrenia and abnormal involuntary movements in the Xhosa population. Prog Neuropsychopharmacol Biol Psychiatry 2007;31(3):664–72. Hori H, Ohmori O, Shinkai T, Kojima H, Okano C, Suzuki T, et al. Manganese superoxide dismutase gene polymorphism and schizophrenia: Relation to tardive dyskinesia. Neuropsychopharmacol 2000;23:170–7. Hori H, Shinkai T, Matsumoto C, Ohmori O, Nakamura J. No association between a functional NAD(P)H: quinone oxidoreductase gene polymorphism (Pro187Ser) and tardive dyskinesia. Neuromol Med 2006;8(3):375–80. Kang SG, Choi JE, An H, Park YM, Lee HJ, Han C, Kim YK, Kim SH, Cho SN, Joe SH, Jung IK, Kim L, Lee MS. Manganese superoxide dismutase gene Ala-9Val polymorphism might be related to the severity of abnormal involuntary movements in Korean schizophrenic patients. Prog Neuropsychopharmacol Biol Psychiatry 2008;32 (8):1844–7. Lahiri DK, Nurnburger Jr JI. A rapid non-enzymatic method for the preparation of HMV DNA from blood for RFLP analysis. Nuc Acids Res 1991;19:5444. Liou Y-J, Wang Y-C, Lin C-C, Bai Y-M, Lai I-C, Liao D-L, Chen J-Y. Association analysis of NAD(P)H:quinone oxidoreductase (NQO1) Pro187Ser genetic polymorphism and tardive dyskinesia in patients with schizophrenia in Taiwan. Int J Neuropsychopharmacol 2005;8:483–6.

56

C.C. Zai et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 34 (2010) 50–56

Liou YJ, Lai IC, Lin MW, Bai YM, Lin CC, Liao DL, et al. Haplotype analysis of endothelial nitric oxide synthase (NOS3) genetic variants and tardive dyskinesia in patients with schizophrenia. Pharmacogenet Genomics 2006;16(3):151–7. Lohr JB, Kuczenski R, Niculescu AB. Oxidative mechanisms and tardive dyskinesia. CNS Drugs 2003;17:47–62. Lohr JB, Kuczenski R, Bracha HS, Moir M, Jeste DV. Increased indices of free radical activity in the cerebrospinal fluid of patients with tardive dyskinesia. Biol Psychiatry 1990;28:535–9. Margolese HC, Chouinard G, Kolivakis TT, Beauclair L, Miller R, Annable L. Tardive dyskinesia in the era of typical and atypical antipsychotics. Part 2: Incidence and management strategies in patients with schizophrenia. Can J Psychiatry 2005;50 (11):703–14. Misra V, Grondin A, Klamut HJ, Rauth AM. Assessment of the relationship between genotypic status of a DT-diaphorase point mutation and enzymatic activity. Br J Cancer. 2000;83:998-1002. Müller DJ, Shinkai T, De Luca V, Kennedy JL. Clinical implications of pharmacogenomics for tardive dyskinesia. Pharmacogenomics J. 2004;4(2):77–87. Naidu PS, Singh A, Kulkarni SK. Reversal of haloperidol-induced orofacial dyskinesia by quercetin, a bioflavonoid. Psychopharmacology (Berl) 2003;167:418–23. Pae C, Yu H, Kim J, Lee C, Lee S, Jun T, et al. Quinone oxidoreductase (NQO1) gene polymorphism (609C/T) may be associated with tardive dyskinesia, but not with the development of schizophrenia. Int J Neuropsychopharmacol 2004;7:495–500. Pae CU, Kim TS, Patkar AA, Kim JJ, Lee CU, Lee SJ, Jun TY, Lee C, Paik IH. Manganese superoxide dismutase (MnSOD: Ala-9Val) gene polymorphism may not be associated with schizophrenia and tardive dyskinesia. Psychiatry Res 2007;153 (1):77–81. Peet M, Laugharne J, Rangarajan N, Reynolds GP. Tardive dyskinesia, lipid peroxidation, and sustained amelioration with vitamin E treatment. Int Clin Psychopharmacol. 1993;8(3):151–3 Fall. Pillai A, Parikh V, Terry Jr AV, Mahadik SP. Long-term antipsychotic treatments and crossover studies in rats: differential effects of typical and atypical agents on the expression of antioxidant enzymes and membrane lipid peroxidation in rat brain. J Psychiatr Res. 2007;41:372–86. Rosenblum JS, Gilula NB, Lerner RA. On signal sequence polymorphisms and diseases of distribution. Proc Natl Acad Sci U S A 1996;93:4471–3. Sadan O, Bahat-Stromza M, Gilgun-Sherki Y, Atlas D, Melamed E, Offen D. A novel braintargeted antioxidant (AD4) attenuates haloperidol-induced abnormal movement in rats: implications for tardive dyskinesia. Clin Neuropharmacol. 2005;28:285–8. Sagara Y. Induction of reactive oxygen species in neurons by haloperidol. J Neurochem. 1998;71:1002–12. Schooler NR, Kane JM. Research diagnoses for tardive dyskinesia. Arch Gen Psychiatry 1982;39(4):486–7. Shamir E, Barak Y, Shalman I, Laudon M, Zisapel N, Tarrasch R, Elizur A, et al. Melatonin treatment for tardive dyskinesia. A double-blind, placebo-controlled, crossover study. Arch Gen Psychiatry 2001;58:1049–52.

Shimoda-Matsubayashi S, Matsumine H, Kobayashi T, Nakagawara-Hattori Y, Shimizu Y, Mizuno Y. Structural dimorphism in the mitochondrial targeting sequence in the human manganese superoxide dismutase gene. Biochem Biophys Res Comm 1996;226:561–5. Shinkai T, Ohmori O, Hori H, Nakamura J. Allelic association of the neuronal nitric oxide synthase (NOS1) gene with schizophrenia. Mol Psychiatry 2002;7:560–3. Shinkai T, De Luca V, Hwang R, Matsumoto C, Hori H, Ohmori O, Remington G, et al. Association study between a functional glutathione S-transferase (GSTP1) gene polymorphism (Ile105Val) and tardive dyskinesia. Neurosci Lett 2005;388(2):116–20. Shinkai T, Müller DJ, De Luca V, Shaikh S, Matsumoto C, Hwang R, et al. Genetic association analysis of the glutathione peroxidase (GPX1) gene polymorphism (Pro197Leu) with tardive dyskinesia. Psychiatry Res 2006;141:123–8. Siegel D, Gustafson DL, Dehn DL, Han JY, Boonchoong P, Berliner LJ, Ross D. NAD(P)H: quinine oxidoreductase 1: role as a superoxide scavenger. Mol Pharmacol 2004;65:1238–47. Siegel D, McGuinness SM, Winski SL, Ross D. Genotype-phenotype relationships in studies of a polymorphism in NAD(P)H:quinone oxidoreductase 1. Pharmacogenetics. 1999;9:113–21. Soares KV, McGrath JJ. Vitamin E for neuroleptic-induced tardive dyskinesia. Cochrane Database Syst Rev 2001;4:CD000209. Thelma BK, Tiwari AK, Deshpande SN, Lerer B, Nimgaonkar VL. Genetic susceptibility to Tardive Dyskinesia in chronic schizophrenia subjects: role of oxidative stress pathway genes. Schizophr Res 2007;92(1–3):278–9. Traver RD, Siegel D, Beall HD, Phillips RM, Gibson NW, Franklin WA, Ross D. Characterization of a polymorphism in NAD(P)H: quinone oxidoreductase (DTdiaphorase). Br J Cancer 1997;75:69–75. Tsai G, Goff DC, Chang RW, Flood J, Baer L, Coyle JT. Markers of glutamatergic neurotransmission and oxidative stress associated with tardive dyskinesia. Am J Psychiatry 1998;155:1207–13. Wang Y-C, Liou Y-J, Liao D-L, Bai Y-M, Lin C-C, Yu S-C, et al. Association analysis of a neural nitric oxide synthase gene polymorphism and antipsychotics-induced tardive dyskinesia in Chinese schizophrenic patients. J Neural Transm 2004;111:623–9. Yamada K, Kanba S, Anamizu S, Ohnishi K, Ashikari I, Yagi G, et al. Low superoxide dismutase activity in schizophrenic patients with tardive dyskinesia. Psychol Med 1997;27:1223–5. Zai CC, Hwang RW, De Luca V, Müller DJ, King N, Zai GC, et al. Association study of tardive dyskinesia and twelve DRD2 polymorphisms in schizophrenia patients. Int J Neuropsychopharmacol. 2007;10(5):639–51. Zhang XY, Tan YL, Zhou DF, Cao LY, Wu GY, Haile CN, et al. Disrupted antioxidant enzyme activity and elevated lipid peroxidation products in schizophrenic patients with tardive dyskinesia. J Clin Psychiatry 2007;68(5):754–60 May. Zhang ZJ, Zhang XB, Hou G, Sha WW, Reynolds GP. The increased activity of plasma manganese superoxide dismutase in tardive dyskinesia in untreated to the Ala9Val polymorphism. J Psychiatry Res 2002;36:317–24.