Brain-derived neurotrophic factor (BDNF) and ...

Psychiatry Research 226 (2015) 1–13

Contents lists available at ScienceDirect

Psychiatry Research journal homepage: www.elsevier.com/locate/psychres

Review article

Brain-derived neurotrophic factor (BDNF) and neurocognitive deficits in people with schizophrenia: A meta-analysis Anthony O. Ahmed a,b,n, Andrew M. Mantini b, Daniel J. Fridberg c, Peter F. Buckley b a

Department of Psychiatry, Weill Cornell Medical College, 21 Bloomingdale Road, White Plains, NY 10605, USA Department of Psychiatry and Health Behavior, Georgia Regents University, 997 Saint Sebastian Way, Augusta, GA 30912, USA c Department of Psychiatry, University of Chicago, 5841 S. Maryland Avenue, Chicago, IL 60637, USA b

art ic l e i nf o

a b s t r a c t

Article history: Received 16 March 2014 Received in revised form 15 December 2014 Accepted 19 December 2014 Available online 28 January 2015

Studies suggest that the BDNF Val66Met (rs6265) polymorphism is associated with the incidence of schizophrenia and neurocognitive functioning. These associations appear to be however somewhat mixed. We conducted two separate meta-analyses to investigate (1) the association between the Val66Met polymorphism and neurocognition in people with schizophrenia and (2) the association between peripheral expression of BDNF and neurocognitive phenotypes. For the first aim, we identified 12 studies and 67 comparisons of Met allele carriers and Val homozygotes. These comparisons included 1890 people with schizophrenia (men¼ 1465, women¼553), of whom 972 were Met allele carriers and 918 were Val homozygotes. For the second aim, we identified five studies and 25 correlations of peripheral BDNF and neurocognitive scores. The meta-analysis for the second aim included 414 people with schizophrenia (men¼ 292, women¼ 170). First, we found non-significant difference between the genotype groups on most neurocognitive domains. Second, correlations between peripheral BDNF and neurocognitive phenotypes were minimal but we obtained significant effects for the reasoning and problem-solving domains; thus, higher levels of BDNF expression corresponded to better performance on reasoning/problem-solving tasks. The meta-analyses did not robustly establish an association between BDNF Val66Met polymorphism and neurocognition in schizophrenia. & 2015 Elsevier Ireland Ltd. All rights reserved.

Keywords: BDNF Val66Met Neurocognition Schizophrenia

Contents 1. 2.

3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.1. Search strategy and study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2.2. Categorization of neurocognitive tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2.3. Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3.1. Search results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3.2. Study 1: association between Val66Met polymorphism and neurocognitive functioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3.2.1. General intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 3.2.2. Processing speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3.2.3. Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 3.2.4. Verbal learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 3.2.5. Visual learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 3.2.6. Reasoning and problem solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 3.3. Study 2: Association between peripheral BDNF and neurocognitive functioning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 3.3.1. Attention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.3.2. Reasoning and problem solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.3.3. All neurocognitive phenotypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

n Corresponding author at: Department of Psychiatry, Weill Cornell Medical College, 21 Bloomingdale Road, White Plains, NY 10605, USA. Tel.: þ 1 914 997 5251; fax: þ 1 914 682 6906. E-mail address: [email protected] (A.O. Ahmed).

http://dx.doi.org/10.1016/j.psychres.2014.12.069 0165-1781/& 2015 Elsevier Ireland Ltd. All rights reserved.

Downloaded for Anonymous User (n/a) at Library of the Weill Medical College/Cornell University from ClinicalKey.com by Elsevier on November 19, 2017. For personal use only. No other uses without permission. Copyright ©2017. Elsevier Inc. All rights reserved.

2

A.O. Ahmed et al. / Psychiatry Research 226 (2015) 1–13

4.

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction A growing literature has established the relevance of neurocognitive deficits to the phenomenology of schizophrenia spectrum disorders both as an enduring feature of the illness, and as a predictor of clinical and functional outcomes (Green and Nuechterlein, 1999). Robust evidence suggests that neurocognitive deficits predate the onset of the first psychotic episode; are present among unaffected family members; and are linked to vulnerability genes for schizophrenia (Fitzgerald et al., 2004; Frommann et al., 2011; Wiener et al., 2013). What remains unclear however, is the degree to which neurocognition in schizophrenia is viable as a phenotype for studying specific genes and functional activity in protein pathways that may underlie phenomenological differences across the schizophrenia spectrum. The viability of neurocognitive phenotypes to such end is predicated on a demonstration of their association with sources of genetic variation that may alter neurochemical, neurophysiological, neurofunctional and/or other neuropathways relevant for neurocognition. The Val66Met single nucleotide polymorphism (SNP) on the Brain-Derived Neurotrophic Factor (BDNF) gene and its peripheral expression has gained strong interest in the last decade for its relevance to the incidence of schizophrenia and neurocognitive functioning (Chang et al., 2009; Kambeitz et al. (2012), Chen et al. (2006); Gratacòs et al. (2007); Hosang et al. (2010); Hwang et al. (2006); Iga et al. (2007); Kim et al. (2008); Perroud et al. (2008); Ribasés et al. (2003), Takahashi et al. (2008), Miyajima et al. (2008); Tsai et al. (2004), Huang and Reichart (2003), Gruber et al. (2012), Ho et al. (2007), Baig et al. (2010), Gray and Roth (2007), Witte and colleagues (2012); NevesPereira et al., 2005; Numata et al., 2006; Rosa et al., 2006). BDNF has also been linked to the incidence of other psychiatric syndromes – depression, bipolar disorder, anxiety disorders, suicidality, substance use, and eating disorders (Chen et al., 2006; Gratacòs et al., 2007; Hosang et al., 2010; Hwang et al., 2006; Iga et al., 2007; Kim et al., 2008; Perroud et al., 2008; Ribasés et al., 2003) and personality traits – neuroticism and introversion associated with greater risk for psychotic and affective disorders (Terracciano et al., 2010). In humans, the BDNF gene is located on the short arm of chromosome 11p13 between base pairs 27,676,439 and 27,743,604 (Pruunsild et al., 2007). The gene comprises 10 exons – nine 50 exons that include exons I, II, III, IV, V, Vh, VI, VII, and IX, that are alternatively spliced to one 30 encoding exon, exon IX. The exons code for 34 mRNA transcripts from nine BDNF mRNA forms (Pruunsild et al., 2007). The BDNF gene is characterized by several SNPs. The most extensively studied occurs at nucleotide 196 and is the rs6265 or Val66Met SNP (Pillai and Buckley, 2012). The Val66Met polymorphism, also described as a proregion G196A [guanine–adenine] polymorphism, results in a valine-for-methionine substitution at codon 66 on exon VIII. The valine-for-methionine substitution is associated with inefficiency in activity-dependent transportation of BDNF mRNA (Chiaruttini et al., 2004) and protein to dendrites (Chen et al., 2004; Egan et al., 2003), thus contributing to reductions in dendritic density, deficits in synaptic connectivity, and neurocognitive deficits (Chen et al., 2004; Egan et al., 2003).

10 10 11 11 11

Studies have shown that in both schizophrenia patients and non-clinical groups, carriers of the Met allele relative to Val homozygotes demonstrate reductions in their overall brain volume (Takahashi et al., 2008); hippocampal volume (Koolschijn et al., 2010; Molendijk et al., 2012; Smith et al., 2012); hippocampal activity (Egan et al., 2003); white matter integrity (Tost et al., 2013); and in frontal, temporal, and occipital gray matter volumes (Pezawas et al., 2004; Ho et al., 2006). The Met allele has also been linked to deficits in overall intellectual functioning and neurocognitive functions in schizophrenia patients (e.g., Egan et al., 2003; Vyas and Puri, 2012) and non-clinical samples (Miyajima et al., 2008; Tsai et al., 2004). Findings of the association between Val66Met and neurocognition have however been mixed in both clinical and non-clinical samples (Mandelman and Grigorenko, 2012). There are therefore questions about the robustness of the effects of BDNF Val66Met on neurocognitive functioning. Notwithstanding, the Val66Met polymorphism appears to have pleiotrophic properties for schizophrenia symptoms and its features. The BDNF protein is differentially expressed across several brain tissues and its expression levels in the central and peripheral nervous systems are influenced by nutrition, metabolism, behavior, and stress factors (Brietzke et al., 2012; Fuchikami et al., 2009; Gama et al., 2012; Griffin et al., 2011). BDNF is synthesized as its pro-isoform (pro-BDNF) after which it is proteolytically cleaved (i.e., N-terminal domain is removed) into its mature form within the neuron or after it is transported extracellularly (Barker, 2009). Whereas pro-isoforms such as pro-BDNF bind to p75 neurotrophin receptor (p75NTR), the mature neurotrophins bind to protein-kinase neurotrophin receptors – tropomyosine-related kinase (Trk) receptors (Huang and Reichart, 2003). BDNF binds to Trk B of the family of Trk receptors, a process that activates survival mechanisms in the central nervous system such as proliferation, growth, and neuroplasticity (Islam et al., 2009; Nguyen et al., 2009). In contrast, pro-isoforms such as pro-BDNF activate apoptotic pathways after they bind to p75NTR (Teng et al., 2005) and induce neuronal remodeling including axonal and dendritic pruning (Barker, 2009; Koshimizu et al., 2009). Studies have shown that BDNF and other neurotrophins play a role in neurogenesis, neuronal growth, survival, differentiation, maturation, neuronal migration, and positioning during brain development and influence dendritic growth, density, neural outgrowth, connectivity and neuroplasticity across the lifespan (Green et al., 2011). In people with schizophrenia, alterations in the intracellular trafficking and secretion of BDNF have been demonstrated in regions of the brain that are relevant to neurocognitive functioning including the hippocampus, frontal structures, cerebral cortex, striatum, cerebellum, the brain stem, hypothalamus, and the basal ganglia (Chen et al., 2004; Egan et al., 2003; Goto et al., 2011; Paz et al., 2006; Ray et al., 2014; Weickert et al., 2003; Wong et al., 2010). Although established that BDNF intracellular transport influences survival versus apoptotic processes in the CNS, and the Val66Met SNP predicts BDNF trafficking, clinical studies of BDNF are usually interested in peripheral measurements as indices of CNS activity. The degree to which peripheral measurements of BDNF in the serum and the plasma reflect BDNF activity in the CNS remains however an empirical question. Three studies have examined the association between Val66Met and peripheral BDNF. Zakharyan



and Boyajyan (2014) found evidence of decreased plasma BDNF in Met carriers in Armenians with schizophrenia. Aas et al. (2014) found similar reductions in Met carriers with schizophrenia in a Dutch sample. In contrast, Chen et al. (2014) found no differences between genotype groups with schizophrenia in plasma BDNF in a Hans Chinese sample. Several studies have compared peripheral BDNF expression levels in people with schizophrenia to healthy controls. These include comparisons of post-mortem brain samples, blood serum or plasma, and cerebrospinal fluid concentrations obtained from chronic schizophrenia and first-episode patients relative to controls (see Ahmed et al., 2013; Buckley et al., 2007a; 2007b; Green et al., 2011 for review). Overall, first episode and chronic patients demonstrate reduced BDNF concentrations in all peripheral media relative to controls but there is considerable variability in findings across studies (Ahmed et al., 2013; Buckley et al., 2007a; 2007b; Green et al., 2011). The present study used meta-analyses to investigate the role of the BDNF Val66Met SNP and the BDNF protein in neurocognition in people with schizophrenia. The study had two aims. The first was to examine the association between the Val66Met polymorphism and neurocognition in people with schizophrenia across several studies

3

and provide a quantitative index of differences between Met carriers and Val homozygotes. To further establish the relevance of BDNF for neurocognition, the second aim was to examine the association between peripheral expression levels of the BDNF protein and neurocognitive phenotypes. To this end, we were interested in correlational studies of peripheral BDNF and neurocognitive measures. Given that BDNF expression may be associated with the Val66Met SNP, a demonstration that peripheral BDNF predicts neurocognition in schizophrenia would provisionally establish BDNF activity as a mediator along the BDNF–neurocognition putative causal chain.

2. Method 2.1. Search strategy and study selection Studies for the meta-analysis were identified by searching PsycInfo (via EBSCO) and Medline (PubMed) for English language articles published in peer-reviewed journals between the years 2000 and 2013. The search strategy and reporting was informed by the PRISMA guidelines (Moher et al., 2009). Search terms were: “Schizophn”AND “BDNF” or “Brain Derived Neurotrophic Factor” AND “Neurocognitn” or “Cognitn.” Search results were limited to those studies that (1) were conducted with human subjects; (2) included individuals with a diagnosis of

Search Strategy: "Schizoph*"AND "BDNF" or "Brain Derived Neurotrophic Factors" AND "Neurocognit*" or "Cognit*."

2 databases searched: PsycInfo (via EBSCO) (n = 4,033) Medline (PubMed) (n = 3,134)

Abstracts reviewed for relevance to schizophrenia, BDNF, and neurocognition N = 155 Number of Articles Screened for Eligibility PsycInfo (via EBSCO) (n = 94) Medline (PubMed) (n = 61) Manual review of reference lists of identified studies = (0 non-redundant studies identified) N = 134 excluded n = 45 animal studies n = 14 reviews n = 75 no neurocognitive data or schizophrenia sample N = 21 Studies after initially excluding studies Contacted authors for neurocognitive data not available in their studies

N = 17 Studies included in the meta-analysis

Fig. 1. This figure demonstrates the search strategy that was used for the meta-analysis.


4


schizophrenia or related psychotic disorder – schizophreniform disorder, brief psychotic disorder, psychotic disorder not otherwise specified, delusional disorder, schizoaffective disorder; (3) conducted genotyping of the BDNF Val66Met SNP or obtained assays for peripheral measurement of BDNF; (4) included at least one assessment of neurocognitive functioning; and (5) provided data sufficient for the calculation of effect sizes (e.g., genotype group sample sizes, means, and standard deviations; or correlations of peripheral BDNF and neurocognitive data) for each neurocognitive test included in the study. We also chose to include studies that examined either serum or plasma BDNF although plasma BDNF levels are more variable than serum BDNF levels (Tsuchimine et al., 2014) because of the few overall studies. The reference section of selected articles was also searched for candidate studies for inclusion in the meta-analyses. No studies were identified that had not already been captured by the initial search strategy. To maintain consistency across studies that evaluated Val/Met carriers and Met/Met carriers separately, the Met heterozygotes and Met homozygotes were combined into one group. This step was completed to provide consistency between studies throughout the analysis. 2.2. Categorization of neurocognitive tests Neurocognitive tests were grouped according to the domains specified by the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) consensus cognitive battery (MCCB), where appropriate. The social cognition domain was excluded from the MCCB because none of the studies included in the final analyses included social cognition measures. Two additional neurocognitive phenotypes were included for the present analyses: language and general intelligence (g). The language domain was added because it emerged as a separate neurocognitive domain among the studies included in the present analyses, while the general intelligence domain was included to separately examine the association between general intelligence (g) and the Val66Met polymorphism. The final neurocognitive domains used in the present analyses were: (1) processing speed, (2) attention/vigilance, (3) working memory, (4) verbal learning, (5) visual learning, (6) reasoning and problem solving, (7) general intelligence, and (8) language. Individual neurocognitive measures from each study included in the meta-analysis were assigned to a neurocognitive domain by two raters with doctoral-level training and experience with neurocognitive assessments. There was high agreement between raters in the classification of neurocognitive measures into the separate neurocognitive domains (κ¼ 0.93). Disagreements on placement were resolved through discussion with a third party, and all authors agreed on the final consensus assignments. Measures for which a consensus could not be reached, that were judged to reflect more than one neurocognitive domain, or that did not meet the neurocognitive domain criteria were not included in the meta-analysis. Data were separated for the Val homozygote and Met carrier groups. In addition to cognitive data, the sample size, average age, percentage of male and female participants, and average level of education was collected for each group. When articles reported on more than one sample, each sample was treated as a separate study. 2.3. Data analysis The meta-analyses were completed using the MIX 2.0 Meta-analysis package (Bax, 2011; Bax et al., 2006). MIX was developed on a Microsoft Excel interface that it uses for calculation using Visual Basic as its programming language. Bax and colleagues conducted a validation study of the MIX program by comparing its results to those of STATA and Comprehensive Meta-analysis Version 2.0 when data from eight published meta-analyses were analyzed. They found that MIX produced results that were comparable to those of the other two programs with more extensive graphical outputs. The effect size was calculated for differences between carriers of the Met allele and Val/Val homozygotes using the mean, standard deviation, and sample sizes of the genotype groups. When the aforementioned statistics were unavailable, the effect sizes were estimated from the t, F, or exact p-values if these were available. Hedge's g was computed as the effect size measure given the relatively small sample size of studies included in the meta-analysis. Hedge's g is less positively biased when variance equality assumptions are violated (Grissom and Kim, 2005). To examine the association between peripheral BDNF and neurocognition, product-moment correlations between BDNF expression and neurocognitive variables were obtained within each study as measures of effect size. Across studies, the mean effect sizes, weighted for sample size and study variance, and a 95% confidence interval (CI) to estimate its degree of stability were also obtained. A Z statistic was computed to determine the significance of the effect size, and the Q statistic was calculated to estimate the degree of homogeneity/heterogeneity of the study effect sizes. Inconsistency across studies was indexed with the I2 metric. All analyses were performed using random effects models. Holm–Bonferroni sequential correction was applied to Z statistic p-values to correct for multiple comparisons of genotype groups and the multiple correlations of BDNF expression with neurocognitive phenotypes (Holm, 1979). A synthesis forest plot was used to represent the effect sizes of individual studies based on average effect size. A forest plot was also used to display which group performed better on each domain of neurocognition. Sensitivity analyses were used to examine the degree to which the inclusion of a given study inflates the overall effect size across studies. This process involved removing each individual study and

re-estimating the effect size to determine its individual effect on the overall mean effect size. To address publication bias or the “file drawer” problem – the propensity for journals to publish significant findings and reject studies that fail to reject the null hypothesis – the standard procedure for visual inspection and estimation of selectivity funnel plots were conducted. This process involved graphing the study effect sizes (mean difference or correlation magnitudes) on the x-axis against their standard error on the y-axis. In the absence of publication bias, a “funnel” distribution or standard errors may be observed. In contrast, a biased state produced a skewed or asymmetrical distribution of standard errors by effects. Sensitivity and selectivity plots and any plots not presented are available from the corresponding author. In the first meta-analysis there were too few studies in the Attention and Working Memory domains (one publication each) for them to be included in the comparisons of Met carriers and Val homozygotes. In the second meta-analysis that evaluated correlations between neurocognitive domains and BDNF peripheral expression, there were too few publications in processing speed, working memory, general intelligence, verbal learning, and visual learning (one publication each) for them to be examined as individual domains. There were enough samples to examine attention and reasoning/problem solving. We also examined the association of peripheral BDNF with all neurocognitive measurements.

3. Results 3.1. Search results Using PRISMA methodology summarized in Fig. 1, the search strategy resulted in 155 possible studies for inclusion in the metaanalysis. Of these studies, each was closely evaluated to meet full inclusion criteria. Of these possible studies, 45 were excluded for using animal samples. 89 studies were excluded for having insufficient neurocognitive data and failing to meet the criteria set in this meta-analysis. The authors of studies which met criteria 1–4 but did not provide data sufficient to calculate effect sizes (n¼21) were contacted via email to request those data. Data from one study (Gruber et al., 2012) was obtained in this fashion. The first analysis that compared neurocognitive testing domains between groups of Val homozygotes and Met carriers included 12 studies. The second analysis that examined the association between peripheral BDNF levels and neurocognitive phenotypes included five studies. Care was taken to ensure that studies published by the same group did not use the same or overlapping samples. Within individual studies however, the association between BDNF and each neurocognitive phenotype is frequently computed within the same sample (i.e., effect sizes are computed for general intelligence, verbal learning, working memory etc in the same sample). When the same article reports on the results of multiple independent samples, the result of each independent sample in the article is reported as an individual study. 3.2. Study 1: association between Val66Met polymorphism and neurocognitive functioning Table 1 identifies the studies included in the meta-analytic comparison of Met carriers and Val homozygotes with schizophrenia on neurocognitive scores. The table includes information about measures administered and sample characteristics (genotype frequencies, sex distribution, age, and education). In the meta-analysis, 67 comparisons of Met allele carriers and Val homozygotes on neurocognitive measures were drawn from 12 independent studies. These comparisons included 1890 people with schizophrenia of whom 1465 were men and 553 were women (average age¼31.85). The meta-analytic comparison included 972 Met allele carriers and 918 Val homozygotes. Effect size differences (Hedge's g) between Met carriers and Val homozygotes for each neurocognitive domain, significance tests, and homogeneity statistics are presented in Table 2. 3.2.1. General intelligence The general intelligence (g) domain yielded five studies (Egan et al., 2003; Ho et al., 2006; Chung et al., 2010; Lu et al., 2012; Smith et al., 2012) and eight samples for this meta-analysis. Within this



5

Table 1 Studies included in the meta-analysis of Val66Met and neurocognition in schizophrenia. Study

Val/Val n

Met carrier n

Gender

Age

Country

Education

Measure

Neurocognitive domain

Egan et al. (2003)

138

65

F¼ 42 M¼ 161

35 (9.33)

USA

13.9 (2.5)

WRAT reading comprehension WAIS-R Category fluency WCST perseverative errors

Language

Ho et al. (2006)

182

111

F¼ 80 M¼ 213

27.39 (9.21)

USA

12.87 (2.19)

WAIS-IV

General intelligence

Verbal memory domain Processing speed domain Problem solving domain

Verbal learning Processing speed Reasoning and problem solving Language Visual learning

Language domain Visuospatial domain Rybakowski et al. (2006) 84

45

F¼ 63 M¼ 66

27.1 (9.6)

Poland

11.8 (2.1)

WCST perseverative errors WCST nonperseverative errors WCST cat completed WCST %CONC WCST 1st Cat N-back time N-back % correct

Ho et al. (2007)

Baig et al. (2010)

Chung et al. (2010) (Homicide sample)

74

39

14

45

19

34

F¼ 36 M¼ 83

F¼ 35 M¼ 23

M¼ 51

26.4 (6.77)

26.41 (3.17)

37.9 (6.6)

USA

Scotland

South Korea

NA

NA

12.6 (1.5)

Verbal learning

Speed/attention domain Problem solving domain Language domain Visuospatial domain

Processing speed Reasoning and problem solving Language Visual learning

Encoding – total # correct

Processing speed

Encoding – reaction time Retrieval recognition accuracy Retrieval reaction time

Processing speed Verbal learning Verbal learning

WAIS


RAVLT learning RAVLT delayed recall RAVLT delayed recognition RCFT copy RCFT immediate recall RCFT delayed recall WCST NCC

Verbal learning Verbal learning Verbal learning Visual learning Visual learning Visual learning Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving

WCST PE% 14

33

M¼ 50

38.5 (8.0)

South Korea

12.9 (1.9)

Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving Processing speed Reasoning and problem solving

Verbal memory domain

WCST PR%

Chung et al. (2010) (Non-homicide sample)

General intelligence Processing speed Reasoning and problem solving

WAIS


RAVLT learning RAVLT delayed recall RAVLT delayed recognition RCFT copy RCFT immediate recall RCFT delayed recall WCST NCC

Verbal learning Verbal learning Verbal learning Visual learning Visual learning Visual learning Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving

WCST PR% WCST PE%


6


Table 1 (continued ) Study

Val/Val n

Met carrier n

Gender

Age

Country

Education

Measure


Lu et al. (2012)

27

85

F¼ 53 M¼ 59

25.2 (4.3)

China

12.5 (3.6)

WAIS


Verbal IQ Performance IQ WCST cat completed

General intelligence General intelligence Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving

WCST perseverative errors WCST nonperseverative errors WCST %CONC Martinho et al. (2012)

Smith et al. (2012)

Zhang et al. (2012a, 2012b)

Gruber et al. (2012)

88

42

38

20

155

420

65

53

F¼ 68 M¼ 62

F¼ 20 M¼ 38

F¼ 79 M¼ 578

F¼ 77 M¼ 81

31.12 (10.88)

20.6 (4.7)

48.4 (13.7)

38.22 (13.23)

Brazil

Canada

China

Germany

o 8 years¼ 50% 48 years ¼ 50%

NA

9.0 (5.6)

13.77 (3.13)

Digit span forward

Working memory

Digit span backward Verbal fluency (COWAT)

Working memory Processing speed

IQ (NAART)


Logical memory immediate Logical memory delayed

Verbal learning Verbal learning

Immediate memory

Verbal learning

Attention Language Visuospatial/constructional Delayed memory

Attention/vigilance Language Visual learning Verbal learning

RAVLT learning

Verbal learning

RAVLT delayed recall RAVLT delayed recognition

Verbal learning Verbal learning

Note: COWAT ¼Controlled Oral Word Association Test; NAART ¼ North American Adult Reading Test; WCST¼ Wisconsin Card Sort Test; RAVLT ¼ Rey Auditory Verbal Learning Test; RCFT ¼Rey Complex Figure Test; WAIS¼ Wechsler Adult Intelligence Scale; WRAT ¼ Wide Range Achievement Test. In the WCST, PR¼ Perseverative Responses, PE ¼Perseverative Errors, CONC ¼Conceptual Level responses, and NCC¼ Number of Categories Completed

Table 2 Domains and effect sizes in BDNF studies. Domain

Samples (n)

Val (n)

Met (n)

Hedges's g (hg)

95% CI

Z

p

pc

Q-test

p-value of Q

General intelligence Processing speed Language Verbal learning Visual learning Reasoning & problem solving

8 8 4 17 9 19

518 388 641 1434 777 1032

467 732 549 999 495 1090

0.05 0.06 0.02 0.09 0.16 0.03

0.20 to 0.29 0.10 to 0.21 0.11 to 0.14 0.19 to 0.01 0.29 to 0.04 0.16 to 0.11

0.37 0.73 0.28 1.70 2.57 0.39

0.71 0.46 0.78 0.04 0.01 0.69

1.00 1.00 1.00 0.20 0.06 1.00

18.69 10.26 2.09 19.27 8.22 35.82

0.01 0.17 0.55 0.26 0.41 0.01

Note: Pc ¼ Holm–Bonferroni corrected p-values for multiple comparisons.

domain, no significant differences were observed between Val homozygotes and Met carriers (hg¼ 0.05, 95% CI¼ 0.20 to 0.29, Z¼0.37, p¼0.71). The effect sizes demonstrated a large (I2 ¼62.5%) significant heterogeneity (Q¼18.69, p¼0.01). The exclusion sensitivity test showed that removal of the study 1 sample in the Chung et al. (2010) study increased the aggregate effect size (hg¼0.13, 95% CI¼ 0.04 to 0.30, Z¼1.45, p¼0.14) but the effect size did not achieve significance. The exclusion selectivity test was unremarkable.

3.2.2. Processing speed The processing speed domain yielded 6 studies (Egan et al., 2003; Ho et al., 2006, 2007; Rybakowski et al., 2006; Baig et al., 2010; Martinho et al., 2012) and 8 samples for this meta-analysis. Within

this domain, no significant differences were observed between Val carriers and Met carriers (hg¼0.06, 95% CI¼ 0.10 to 0.21, Z¼0.73, p¼ 0.46). The effect sizes demonstrated moderate (I2 ¼31.79%) but non-significant heterogeneity (Q¼ 10.26, p¼ 0.17). The exclusion sensitivity test showed no propensity of the average effect size to change substantially due to the exclusion of any individual studies.

3.2.3. Language The language domain yielded 4 studies (Egan et al., 2003; Ho et al., 2006, 2007; Zhang et al., 2012a) and 4 samples for this metaanalysis. Within this domain, no significant differences were observed between Val carriers and Met carriers (hg¼ 0.02, 95% CI ¼ 0.11 to 0.14, Z¼ 0.28, p ¼0.78). No heterogeneity was



Author (year) Ho (2006) Ho (2007) Baig (2010) Baig (2010) Chung (2010) Chung (2010) Chung (2010) Chung (2010) Chung (2010) Chung (2010) Gruber (2012) Gruber (2012) Gruber (2012) Smith (2012) Smith (2012) Zhang (2012) Zhang (2012) Synthesis

Synthesis forest plot

Measure (CI) -0.07 (-0.31; 0.17) -0.43 (-0.8; -0.05) -0.23 (-0.78; 0.32) 0.02 (-0.52; 0.57) -0.01 (-0.63; 0.61) -0.05 (-0.67; 0.58) -0.25 (-0.87; 0.38) 0.05 (-0.58; 0.67) -0.04 (-0.66; 0.59) -0.26 (-0.88; 0.37) -0.34 (-0.71; 0.02) -0.09 (-0.45; 0.27) -0.23 (-0.6; 0.13) 0.65 (0.1; 1.21) 0.57 (0.02; 1.13) -0.09 (-0.28; 0.09) -0.07 (-0.26; 0.11) -0.09 (-0.19; 0.01) 1

7

0.5

0 hg

0.5

Sample size 293 119 58 58 48 48 48 47 47 47 118 118 118 58 58 575 575 2433 1

Weight % 12.54% 6.23% 3.17% 3.19% 2.52% 2.52% 2.51% 2.5% 2.5% 2.49% 6.49% 6.57% 6.54% 3.12% 3.15% 16.97% 16.98% 100%

1.5

Fig. 2. This figure depicts the forest plot for differences between Met carriers and Val homozygotes on the verbal learning domain.

Author (year)


Measure (CI)

Sample size

Weight %

Ho (2006)

0.28 ( 0.52; 0.04)

293

25.52%

Ho (2007)

0.38 ( 0.75; 0.01)

119

10.73%

Chung (2010)

0.22 ( 0.4; 0.85)

48

3.92%

Chung (2010)

0.24 ( 0.87; 0.38)

48

3.92%

Chung (2010)

0.25 ( 0.87; 0.38)

48

3.92%

Chung (2010)

0.32 ( 0.95; 0.31)

47

3.86%

Chung (2010)

0.35 ( 0.28; 0.98)

47

3.85%

Chung (2010)

0.21 ( 0.42; 0.84)

47

3.89%

Zhang (2012)

0.12 ( 0.31; 0.06)

575

40.39%

Synthesis

0.16 ( 0.29; 0.04)

1272

100%

1.5

1

0.5

0 hg

0.5

1

1.5

Fig. 3. This figure depicts the forest plot for differences between Met carriers and Val homozygotes on the visual learning domain.

detected (I2 ¼0.00%, Q ¼2.09, p ¼0.55) across effect sizes and the exclusion sensitivity test was unremarkable.

3.2.4. Verbal learning The verbal learning domain yielded 6 studies (Ho et al., 2006; Baig et al., 2010; Chung et al., 2010; Gruber et al., 2012; Smith et al., 2012; Zhang et al., 2012a) and 17 samples for this meta-analysis. Within this domain, although small differences were observed between Val homozygotes and Met carriers with Met carriers performing worse (hg¼ 0.09, 95% CI¼ 0.19 to 0.01, Z¼ 1.70, p¼0.04, see Fig. 2) the difference was not statistically significant following the application of Holm's correction procedure (Holm's p¼ 0.20). A Q-test for this domain indicated no significant differences in the effect sizes across studies (I2 ¼16.96%, Q¼19.27, p¼0.26). The exclusion sensitivity test however showed that the removal of the Smith and colleagues (2012)

samples from the meta-analysis produced a significant association with Met carriers performing worse (hg¼ 0.16, 95% CI¼ 0.28 to 0.04, Z¼ 2.68, p¼ 0.03).

3.2.5. Visual learning The visual learning domain yielded 4 studies (Ho et al., 2006, 2007; Chung et al., 2010; Zhang et al., 2012a) and 9 samples for this meta-analysis. Within this domain, small significant differences were observed between Val homozygotes and Met carriers (hg¼ 0.16, 95% CI¼ 0.29 to 0.04, Z¼ 2.57, p¼0.01, see Fig. 3) with Met carriers performing worse. The apparent statistical significance did not however survive the Holm's adjustment (Holm's p¼0.06). A Qtest for this domain indicated no significant differences in the effect sizes across studies (Q¼ 8.22, p¼ 0.41) and no effect of heterogeneity on the aggregate estimate (I2 ¼2.65%). The results of the exclusion


8


Table 3 Studies included in the meta-analysis of peripheral BDNF association with neurocognition. Author

Sample Size

Gender

Age M (S.D.)

Country Education M (S.D.)

Measure


Carlino et al. (2011)

40

F¼ 20 M¼ 20

48 (range: 42.25–58)

Italy

TMT

Reasoning and problem solving Attention/vigilance Attention/vigilance

13 (range: 10.25–15)

Digit symbol coding Digit span forward

Correlation with BDNF 0.55 0.36 0.36

Goto et al. (2011)

18

F¼ 9 M¼ 9

13–52

Japan

NA

WCST

Reasoning and problem solving

0.26

Niitsu et al. (2011)

60

F¼ 37 M¼ 26

35.9 (8.2)

Japan

13.8 (2.3)

Letter fluency

Processing speed

0.02

Category fluency WCST accomplished categories

0.04 0.00

Stroop Task-Part D Stroop Task-Part C DSDT – without distractor DSDT – with distractor

Processing speed Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving Processing speed Processing speed Attention/vigilance Attention/vigilance

0.09 0.15 0.05 0.06

Immediate memory

Visual learning

0.23

RBANS total Attention index Delayed memory Language Visuo-spatial

General intelligence Attention/vigilance Working memory Verbal memory Visual learning

0.12 0.09 0.07 0.02 0.01

Similarities

Reasoning and problem solving

0.19

Digit symbol coding Logical memory learning curve

Attention/vigilance Reasoning and problem solving Reasoning and problem solving Reasoning and problem solving

0.13 0.29

WCST perseveration errors TMT-Part A TMT-Part B

Zhang et al. (2012a, 2012b)

De Azua et al., 2013

251

45

F¼ 64 M¼ 187

F¼ 40 M¼ 50

52.1 (8.3)

24.3 (8.5)

China

Spain

9.9 (6.2)

Primary (11.1%) Obligatory (40%) Secondary (35.6%) University (13.3%)

Verbal paired asssociates learning curve I Verbal paired asssociates learning curve II

0.20 0.18 0.05

0.14 0.12

Note: DSDT ¼Digit Span Distractibility Test; WCST¼ Wisconsin Card Sort Test; TMT ¼ Trail Making Task. M ¼Male, and F ¼Female. Table 4 Domains and correlations in peripheral BDNF studies. Domain

Samples (n)

Cor (r)

95% CI

Z

p

pc

Q-test

p-value of Q

Attention Reasoning & problem solving All neurocognitive domains

6 10 25

0.04 0.18 0.09

0.21 to 0.12 0.09–0.29 0.04–0.15

0.52 3.61 3.19

0.60 0.0003 0.0014

0.60 0.001 0.003

13.51 11.01 40.44

0.02 0.28 0.02

Note: Pc ¼ Holm–Bonferroni corrected p-values for multiple comparisons.

sensitivity test however showed that the removal of the Ho et al. (2006) study would decrease the aggregate effect (hg¼ 0.12, 95% CI¼ 0.26 to 0.01, Z¼ 1.73, p¼ 0.08).

3.2.6. Reasoning and problem solving The reasoning and problem solving domain yielded 6 studies (Ho et al., 2006; Baig et al., 2010; Chung et al., 2010; Gruber et al., 2012; Smith et al., 2012; Zhang et al., 2012a) and 19 samples for this meta-analysis. Within this domain, no significant differences were observed between Val carriers and Met carriers (hg¼ 0.03, 95% CI¼ 0.16 to 0.11, Z¼ 0.39, p¼ 0.69). The effect sizes

demonstrated large, significant heterogeneity across studies (I2 ¼49.75%, Q¼35.82%, p¼ 0.007). The exclusion sensitivity test was however unremarkable. 3.3. Study 2: Association between peripheral BDNF and neurocognitive functioning The second analysis evaluated the correlation between neurocognitive domains and peripheral expressions of BDNF. For the second aim, we identified five studies and 25 correlations of peripheral BDNF and neurocognitive scores. The studies included in this analysis, the measures administered, and the sample



9


Author (year)

Measure (CI)

Gotoet al. (2009)

0.26( 0.23;0.65)

18

Carlinoet al. (2011)

0.55(0.29;0.74)

40

8.64%

Niitsuet al. (2011)

0 ( 0.25;0.26)

60

12.27%

Niitsuet al. (2011)

0.2( 0.06;0.43)

60

12.27%

Niitsuet al. (2011)

0.18( 0.08;0.42)

60

12.27%

Niitsuet al. (2011)

0.05( 0.21;0.3)

60

12.27%

de Azuaetal. (2013)

0.19( 0.11;0.46)

45

9.6%

de Azuaetal. (2013)

0.29(0; 0.54)

45

9.6%

de Azuaetal. (2013)

0.14( 0.16;0.42)

45

9.6%

de Azuaetal. (2013)

0.12( 0.18;0.4)

45

9.6%

0.19(0.09;0.29)

478

100%

Synthesis

0.4

0.2

0

0.2

0.4 cc

0.6

Sample size

0.8

1

Weight % 3.86%

1.2

Fig. 4. This figure depicts the forest plot for correlations between peripheral BDNF and reasoning/problem-solving measures.

Author (year) Goto et al. (2009) Carlino et al. (2011) Carlino et al. (2011) Carlino et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsu et al. (2011) Niitsuet al. (2011) Zhang et al. (2012) Zhang et al. (2012) Zhang et al. (2012) Zhang et al. (2012) Zhang et al. (2012) Zhanget al. (2012) de Azuaet al. (2013) de Azuaet al. (2013) de Azuaet al. (2013) de Azuaet al. (2013) de Azuaet al. (2013) Synthesis


Measure (CI) 0.26( 0.23;0.65) 0.55(0.29;0.74) 0.36( 0.6; 0.05) 0.36( 0.6; 0.05) 0.02( 0.24;0.27) 0.04( 0.21;0.29) 0 ( 0.25;0.26) 0.2( 0.06;0.43) 0.18( 0.08;0.42) 0.05( 0.21;0.3) 0.09( 0.17;0.33) 0.15( 0.11;0.39) 0.05( 0.21;0.3) 0.06( 0.2; 0.31) 0.23(0.11;0.34) 0.12(0; 0.24) 0.09( 0.03;0.21) 0.07( 0.05;0.19) 0.02( 0.1; 0.14) 0.01( 0.11;0.13) 0.19( 0.11;0.46) 0.13( 0.17;0.41) 0.29(0; 0.54) 0.14( 0.16;0.42) 0.12( 0.18;0.4) 0.09(0.04;0.15)

1

0.5

0

cc

0.5

Sample size 18 40 40 40 60 60 60 60 60 60 60 60 60 60 251 251 251 251 251 251 45 45 45 45 45 2469

1

Weight % 1.14% 2.45% 2.45% 2.45% 3.38% 3.38% 3.38% 3.38% 3.38% 3.38% 3.38% 3.38% 3.38% 3.38% 7.37% 7.37% 7.37% 7.37% 7.37% 7.37% 2.7% 2.7% 2.7% 2.7% 2.7% 100%

1.5

Fig. 5. This figure depicts the forest plot for the correlation between peripheral BDNF and all neurocognitive measures.

characteristics are summarized in Table 3. The meta-analysis for the second aim comprised 414 people with schizophrenia, including 292 men and 170 women (Average age¼39.54). Neurocognitive domains, correlations, 95% CI, Z, p, Holm's adjusted p, Q-test, and pvalue of each Q test are listed in Table 4. 3.3.1. Attention The Attention domain yielded four studies (Carlino et al., 2011; Niitsu et al., 2011; Zhang et al., 2012b; De Azua et al., 2013) and six samples. There was no association between BDNF expression and attention scores and those in the sample (r¼ 0.04, 95% CI¼ 0.21 to 0.12, Z¼ 0.52, p¼0.60). The correlation effect sizes were significantly heterogeneous across studies (I2 ¼62.98%, Q¼13.51, p¼ 0.02) but the exclusion sensitivity test was unremarkable. Removal of the sample from the de Azua et al. study did not significantly alter the average correlation effect size. 3.3.2. Reasoning and problem solving The reasoning and problem solving domain yielded 4 studies (Carlino et al., 2011; Niitsu et al., 2011; Zhang et al., 2012b; De Azua

et al., 2013) and 10 samples. There was a small, statistically significant correlation between reasoning and problem solving scores and BDNF expression levels (r¼0.19, 95% CI¼0.09 to 0.29, Z¼3.95, p¼0.000306, see Fig. 4). The correlation effect size remained statistically significant following the Holm's adjustment (Holm's p¼0.001). A Q-test for this domain indicated no significant differences in the effect sizes across studies (I2 ¼18.25%, Q¼11.01, p¼0.28). The exclusion sensitivity test showed no change in statistical significance when any of the studies was excluded from the analysis.

3.3.3. All neurocognitive phenotypes Combining all neurocognitive domains yielded 7 studies and 24 samples. There was a small but statistically significant correlation between all neurocognitive phenotypes and BDNF expression levels (r¼ 0.09, 95% CI ¼0.04 to 0.15, Z¼ 3.19, p ¼ 0.0014, see Fig. 5). The correlation effect size remained significant after the Holm's adjustment (Holm's p ¼0.003). A Q-test for this domain indicated significant differences in the effect sizes across studies (I2 ¼42.66%, Q¼40.44, p¼ 0.02). The exclusion sensitivity test was


10


however unremarkable with no study demonstrating a likelihood to change the statistical significance of the aggregate correlation.

4. Discussion The goals of the meta-analyses were to (1) examine the association between the BDNF Val66Met SNP and neurocognitive phenotypes; and (2) further examine the relevance of BDNF to neurocognition through an examination of the association between peripheral expressions of the BDNF protein and neurocognitive phenotypes. The metaanalyses included data from 17 studies that included 92 examinations of the association between a neurocognitive measure and BDNF. The meta-analysis that evaluated the association of the Val66Met SNP with neurocognition included 12 studies and 67 comparisons of Met carriers and Val homozygotes. The second meta-analysis included five studies and 25 examinations of the correlation between peripheral BDNF and neurocognitive scores. The first meta-analysis produced evidence of minimal or insubstantial differences between carriers of the Met allele and Val/Val homozygotes for most neurocognitive phenotypes. There were small differences on the verbal learning and visual learning domains with Met carriers performing worse on average but none of these differences survived the multiple testing correction procedure. Moreover, the Hedge's g effect size estimates were small. It is possible that in schizophrenia, the Val66Met variation on the BDNF gene has greater penetrance for tests of visual and verbal learning than other neurocognitive phenotypes including general intelligence. For the second meta-analysis, many neurocognitive phenotypes – working memory, verbal learning, visual learning, and general intelligence – were not represented enough in the literature within individual investigations of their association with peripheral BDNF expression to be evaluated. Most correlation effect sizes were in the minimal to small range with reasoning/problem-solving producing the largest and only statistically significant effect size. Higher levels of BDNF expression corresponded to better reasoning and problem solving abilities. Although overall neurocognition also had a statistically significant association with BDNF expression, this association appears to be driven mostly by reasoning/problem-solving measures. Overall, the sensitivity and publication selectivity analyses showed no effect of individual studies or publication bias on the study results. The results of the current meta-analysis are similar to those of Mandelman and Grigorenko (2012) with regard to the absence of robust, consistent associations between the Val66Met SNP and neurocognition. The differences between both studies are noteworthy – the current study focused exclusively on schizophrenia spectrum disorders whereas the studies included in the Mandelman and Grigorenko meta-analysis included non-clinical samples and a range of psychopathology. The current study also examined peripheral expressions of BDNF in people with schizophrenia as a further test of the association between BDNF and neurocognition. The current meta-analysis found some small associations between the BDNF Val66Met SNP and visual and verbal learning whereas this was absent in the Mandelman and Grigorenko meta-analysis. It may be that the impact of the Met allele on circumscribed aspects of neurocognition is only apparent in people with schizophrenia but not healthy samples or other psychopathology, which made up the majority of the Mandelman and Grigorenko meta-analysis. It may be that Val66Met effects on neurocognition are synergistic with the effects of other risk genes found exclusively in people with schizophrenia-related disorders. There remains the question of the status of neurocognitive domains as putative endophenotypes for Val66Met. The endophenotypic status has to be considered provisional until a similar meta-analysis can established an association between Val66Met and neurocognitive phenotypes in unaffected family members of people with schizophrenia.

The absence of robust associations between the BDNF Val66Met SNP and all neurocognitive phenotypes is puzzling. Given the role of the SNP in activity-dependent packaging and transport of BDNF protein across neurons (Egan et al., 2003), one would expect an influence on neurogenesis, synaptogenesis, neuroplasticity, and neurocognitive functioning downstream. The absence of robust association may be explainable by several factors. The impact of the Val66Met SNP on processes that influence neurocognition may be fairly minimal (or overstated) particularly on the backdrop of several other contributors to neurocognitive functioning. The influence of BDNF Val66Met on neurocognition may be better understood in relation to other polymorphisms on the BDNF gene and other sources of genetic variation that not only predict the incidence of schizophrenia but similarly contribute to neurocognitive functioning in an additive and/or synergistic way (Gray and Roth, 2007; Savitz et al., 2006). For example, the Val158Met SNP on the Catechol-O-Methyl Transferase (COMT) gene influences the activation patters of D1 and D2 receptors and dopamine activity in the prefrontal cortex. Individuals homozygous for the Val allele demonstrate higher COMT enzyme activity, and increased breakdown of dopamine, and by extension lower prefrontal dopamine activity and less efficient prefrontal function. Witte and colleagues (2012) found that a BDNF-by-COMT interaction may be critical not only for baseline neurocognitive functioning but the degree of cortical neuroplasticity as indicated by follow-up neurocognitive scores after inducing plasticity. Following stimulation, BDNF Met carriers performed worse on tasks of verbal learning than Val/Val homozygotes when all individuals are homozygous for the Val allele on the COMT. It is therefore possible that the effect of the Val66Met SNP on neurocognition is coupled with the effects of other genes that similarly impact neurocognition through independent or overlapping pathways. We did obtain small associations between the Val66Met SNP and verbal and visual learning phenotypes as well as a small association between BDNF expression and reasoning and problem solving. This pattern of association may be reasonable when the role of BDNF in neural circuits implicated in learning and executive functions is considered. With decreased expression of BDNF, people with schizophrenia experience less of its benefits on neuroprotection, neurogenesis, neural transmission, plasticity, and other survival processes (Egan et al., 2003; Goto et al., 2011; Paz et al., 2006; Van Winkel et al., 2008). The impact of decreased BDNF on synaptogenesis is particularly robust in the hippocampus, the site of learning and memory-associated long-term potentiation. There is evidence that both the hippocampus and frontal gyri are implicated in aspects of spatial information processing and both verbal and visual learning, whereas the frontal circuits are implicated in executive functions (Collins and Koechlin, 2012; Luo and Niki, 2003; Paz et al., 2006; Zeithamova et al. 2012). BDNF is highly expressed in the hippocampus and the frontal circuits; therefore the impact of the decreased BDNF may be more pronounced in these areas and more evident on neurocognitive tests that assess their functions. It is also possible that the association of peripheral BDNF with reasoning/problem solving is non-causal, rather reflecting separate, independent footprints of the schizophrenia syndrome or psychiatric illness. 4.1. Limitations The results of this study should be considered in light of some potential limitations. One limitation is that demographic variables, such as age and education level, were not available for each study. Another patient characteristic not available was a male–female separation of gender within the evaluated populations for the studies that were used. There is some evidence that the Val66Met genotype frequencies may differ by ethnicity with a higher frequency of the Met allele among people of Asians. It is unclear how the differences in genotype frequencies influence our results. We were unable to



investigate how ethnicity influenced specific neurocognitive phenotypes (i.e., working memory, language, etc.) due to the few available independent samples for such endeavor. The studies included in the meta-analysis often did not provide information about variables that may influence BDNF expression. Information about the form of treatment individuals were receiving (e.g., antipsychotic type, polypharmacy status), illness duration, substance use and other comorbidity patterns, course pattern (acute versus chronic) and other clinical variables were often missing. Previous research has indicated that these factors influence neurocognitive performance and may have provided stronger explanations in this study (Frommann et al., 2011; Wiener et al., 2013). These variables would have allowed a greater depth of comparison and additional comparisons in effect size across these variables could have been informative. Save the Zhang and colleagues study, the sample sizes represented in the meta-analyses were rather small. The problem of small sample sizes may have also contributed to a rather inconsistent genotype frequency distribution across studies. In the absence of a clear probability distribution for the Met allele, there remains the possibility that the sample sizes represented in the included studies may have created biased effect size estimates. There were even fewer studies available to examine the association of BDNF expression with neurocognition. We heard from only one author after contacting several to obtain information about the correlation between BDNF levels and neurocognitive variables. One can speculate that authors who obtained very low associations may be less likely to respond to our inquiries into their results. Another limitation is that we did not distinguish between studies that measured BDNF in serum samples versus plasma samples even though there is evidence that measurements may be less reliable in the latter (Tsuchimine et al., 2014). The possibility that the lack of distinction between serum versus plasma studies may have influenced the results cannot be ruled out. The rejoiner however is that the exclusion sensitivity analysis did not single out any individual study as exerting significant influence on the average correlation effect size. Finally, there were differences across studies included in the meta-analysis in BDNF ELISA kits used to measure BDNF concentrations. These kits may differ in their ability to detect and differentiate BDNF isoforms including proBDNF, matureBDNF, or truncated BDNF (Carlino et al., 2013). These limitations should be carefully considered for future studies of a similar nature, and efforts to reduce the impact of these limitations should be made. 4.2. Future directions It may be necessary to consider the association of BDNF and neurocognition outside the confines of diagnostic boundaries in the schizophrenia spectrum, particularly with evidence that several schizophrenia phenotypes may be dimensionally distributed in the general population and in clinical samples (Ahmed et al., 2012, 2012, 2013, in press). Examinations of the associations of BDNF with neurocognition within arbitrary diagnostic boundaries may underestimate the actual magnitude of the association of BDNF with phenotypes of an extended psychosis construct (Grove, 1991). In addition, future studies of BDNF and neurocognition would benefit from improved sample sizes, enough to ensure adequate statistical power. Larger samples would also ensure that the heterogeneity of neurocognitive performance in people with schizophrenia is adequately captured, thereby increasing the capacity to detect any true effects of BDNF on neurocognition (Joyce and Roiser, 2007). Future studies should also use neurocognitive measures that have enjoyed consensus as sensitive to neurocognitive impairments in schizophrenia (e.g., the MATRICS Consensus Cognitive Battery). The correlation of serum BDNF with reasoning/problem solving measures suggests that the upregulation of BDNF activity may be a putative target for cognitive enhancement in people with schizophrenia. Vinogradov

11

et al. (2009) demonstrated that cognitive training increased serum BDNF, and increased serum BDNF correlated with improvements in quality of life. Thus far, attempts at the psychparmacological enhancement of neurocognition in people with schizophrenia have been disappointing when enhancement strategies focus on action at single receptor sites (Ahmed and Bhat, 2014). The upregulation of BDNF to promote neurogenesis in conjunction with action at singlereceptor sites may offer a novel strategy for treating neurocognitive deficits in people with schizophrenia. 4.3. Conclusions In summary, the results of the present meta-analysis indicate that there are insubstantial differences between carriers of the Met allele and Val/Val homozygotes for most neurocognitive phenotypes. The meta-analysis produced evidence of small differences on the verbal and visual learning domains with Met carriers obtaining worse neurocognitive scores on average. While the statistical power coefficients were small, these results provide valuable information about neurocognitive decline and the likelihood of identifying those individuals that are at risk for further decline. Furthermore, these results suggest an association between peripheral BDNF and the reasoning and problem solving domain. While additional research that minimizes limitations is needed, these results provide useful insights into identifying potential risk for neurocognitive impairment for individuals with schizophrenia. Referencesn Aas, M., Haukvik, U.K., Djurovic, S., Tesli, M., Athanasiu, L., Bjella, T., et al., 2014. Interplay between childhood trauma and BDNF val66met variants on blood BDNF mRNA levels and on hippocampus subfields volumes in schizophrenia spectrum and bipolar disorders. Journal of Psychiatric Research 59, 14–21. Ahmed, A.O., Bhat, I.A., 2014. Psychopharmacological treatment of neurocognitive deficits in people with schizophrenia: a review of old and new targets. CNS Drugs 28 (4), 301–318. Ahmed, A.O., Buckley, P.F., Mabe, P.A., 2012. Latent structure of psychotic experiences in the general population. Acta Psychiatrica Scandinavica 125 (1), 54–65. Ahmed, A.O., Fridberg, D., Hanna, M., Buckley, P.F., 2013. Brain-Derived Neurotrophic Factor and neurocognitive profiles in the psychosis spectrum: findings in bipolar disorder and schizophrenia. In: Moore, N.B. (Ed.), Bipolar Disorders: Symptoms, Management, and Risk Factors. Nova Science Publishers, New York. Ahmed, A.O., Green, B.A., Goodrum, N.M., Doane, N.J., Birgenheir, D., Buckley, P.F., 2013. Does a latent class underlie schizotypal personality disorder? Implications for schizophrenia. Journal of Abnormal Psychology 122 (2), 475–491. Ahmed, A.O., Green, B.A., Buckley, P.F., McFarland, M.E., 2012. Taxometric analyses of paranoid and schizoid personality disorders. Psychiatry Research 196 (1), 123–132. Ahmed, A.O., Strauss, G.P., Buchanan, R.W., Kirkpatrick, B., Carpenter, W.T., 2015 in press. Are negative symptoms dimensional or categorical? Detection and validation of deficit schizophrenia with taxometric and latent variable mixture models. Schizophrenia Bulletin. http://dx.doi.org/10.1093/schbul/sbu163. Baig, B.J., Whalley, H.C., Hall, J., McIntosh, A.M., Job, D.E., Cunningham-Owens, D.G., Johnstone, E.C., Lawrie, S.M., 2010. Functional magnetic resonance imaging of BDNF val66met polymorphism in unmedicated subjects at high genetic risk of schizophrenia performing a verbal memory task. Psychiatry Research 183, 195– 201. http://dx.doi.org/10.1016/j.pscychresns.2010.06.009. Barker, P.A., 2009. Whither proBDNF? Nature Neuroscience 12 (2), 105–106. http: //dx.doi.org/10.1038/nn0209-105. Bax L., 2011. Professional Software for Meta-analysis in Excel Version 2.0.1.4. BiostatXL. Bax, L., Yu, L.M., Ikeda, N., Tsuruta, H., Moons, K.G.M., 2006. Development and validation of MIX: comprehensive free software for meta-analysis of causal research data. BMC Medical Research Methodology 6, 50. Brietzke, E., Mansur, R.B., Soczynska, J., Powell, A.M., McIntyre, R.S., 2012. A theoretical framework informing research about the role of stress in the pathophysiology of bipolar disorder. Progress in Neuro-Psychopharmacology and Biological Psychiatry 39 (1), 1–8. http://dx.doi.org/10.1016/j.pnpbp.2012.05.004. Buckley, P.F., Mahadik, S., Pillai, A., Terry Jr., A., 2007a. Neurotrophins and schizophrenia. Schizophrenia Research 94 (1-3), 1–11. http://dx.doi.org/ 10.1016/j.schres.2007.01.025. Buckley, P.F., Pillai, A., Evans, D., Stirewalt, E., Mahadik, S., 2007b. Brain derived neurotropic factor in first-episode psychosis. Schizophrenia Research 91 (1–3), 1–5. http://dx.doi.org/10.1016/j.schres.2006.12.026.

n

Indicates a study that was included in the meta-analysis.


12


Carlino, D., De Vanna, M., Tongiorgi, E., 2013. Is altered BDNF biosynthesis a general feature in patients with cognitive dysfunctions? Neuroscientist 19 (4), 345–353. n Carlino, D., Leone, E., Di Cola, F., Baj, G., Marin, R., Dinelli, G., et al., 2011. Low serum truncated-BDNF isoform correlates with higher cognitive impairment in schizophrenia. Journal of Psychiatric Research 45 (2), 273–279. http://dx.doi. org/10.1016/j.jpsychires.2010.06.012. Chang, H.-A., Lu, R.-B., Shy, M.-J., Chang, C.-C., Lee, M.-S., Huang, S.-Y., 2009. Brainderived neurotrophic factor Val66Met polymorphism: association with psychopathological symptoms of schizophrenia? Journal of Neuropsychiatry and Clinical Neurosciences 21 (1), 30–37. http://dx.doi.org/10.1176/appi. neuropsych.21.1.30. Chen, S.-L., Lee, S.-Y., Chang, Y.-H., Chen, S.-H., Chu, C.-H., Wang, T.-Y., et al., 2014. The BDNF Val66Met polymorphism and plasma brain-derived neurotrophic factor levels in Han Chinese patients with bipolar disorder and schizophrenia. Progress in Neuro-Psychopharmacology and Biological Psychiatry 51, 99–104. Chen, Z.-Y., Patel, P.D., Sant, G., Meng, C.-X., Teng, K.K., Hempstead, B.L., Lee, F.S., 2004. Variant brain-derived neurotrophic factor (bdnf) (met66) alters the intracellular trafficking and activity-dependent secretion of wild-type bdnf in neurosecretory cells and cortical neurons. Journal of Neuroscience 24 (18), 4401–4411. http://dx.doi.org/10.1523/JNEUROSCI.0348-04.2004. Chen, Z.-Y., Jing, D., Bath, K.G., Ieraci, A., Khan, T., Siao, C.-J., Herrera, D.G., Toth, M., Yang, C., McEwen, B.S., Hempstead, B.L., Lee, F.S., 2006. Genetic variant BDNF (Val66Met) polymorphism alters anxiety-related behavior. Science 314, 140– 143, http://dx.doi.org/10.1126/science.1129663. Chiaruttini, C., Vicario, A., Li, Z., Baj, G., Braiuca, P., Wu, Y., Lee, F.S., Gardossi, L., Baraban, J.M., Tongiorgi, E., 2004. Dendritic trafficking of BDNF mRNA is mediated by translin and blocked by the G196A (Val66Met) mutation. Proceedings of the National Academy of Sciences 106 (38), 16481–16486. n Chung, S., Chung, H.Y., Jung, J., Chang, J.K., Hong, J.P., 2010. Association among aggressiveness, neurocognitive function, and the Val66Met polymorphism of brain-derived neurotrophic factor gene in male schizophrenic patients. Comprehensive Psychiatry 51 (4), 367–372. http://dx.doi.org/10.1016/j.comppsych. 2009.10.003. Collins, A., Koechlin, E., 2012. Reasoning, learning, and creativity: frontal lobe function and human decision-making. PLoS Biology 10 (3), e1001293. http://dx. doi.org/10.1371/journal.pbio.1001293. n De Azua, S.R., Matute, C., Stertz, L., Mosquera, F., Palomino, A., De la Rosa, I., et al., 2013. Plasma brain-derived neurotrophic factor levels, learning capacity and cognition in patients with first episode psychosis. BMC Psychiatry 13 (1), 1–9. n Egan, M.F., Kojima, M., Callicott, J.H., Goldberg, T.E., Kolachana, B.S., Bertolino, A., et al., 2003. The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell 112 (2), 257–269. Fitzgerald, D., Lucas, S., Redoblado, M.A., Winter, V., Brennan, J., Anderson, J., Harris, A., 2004. Cognitive functioning in young people with first episode psychosis: relationship to diagnosis and clinical characteristics. Australian and New Zealand Journal of Psychiatry 38 (7), 501–510. http://dx.doi.org/10.1111/ j.1440-1614.2004.01403.x. Frommann, I., Pukrop, R., Brinkmeyer, J., Bechdolf, A., Ruhrmann, S., Berning, J., et al., 2011. Neuropsychological profiles in different at-risk states of psychosis: executive control impairment in the early – and additional memory dysfunction in the late – prodromal state. Schizophrenia Bulletin 37 (4), 861–873. http: //dx.doi.org/10.1093/schbul/sbp155. Fuchikami, M., Morinobu, S., Kurata, A., Yamamoto, S., Yamawaki, S., 2009. Single immobilization stress differentially alters the expression profile of transcripts of the brain-derived neurotrophic factor (BDNF) gene and histone acetylation at its promoters in the rat hippocampus. International Journal of Neuropsychopharmacology 12 (1), 73–82. http://dx.doi.org/10.1017/S1461145708008997. Gama, C.S., Canever, L., Panizzutti, B., Gubert, C., Stertz, L., Massuda, R., et al., 2012. Effects of omega-3 dietary supplement in prevention of positive, negative and cognitive symptoms: a study in adolescent rats with ketamine-induced model of schizophrenia. Schizophrenia Research 141 (2-3), 162–167. http://dx.doi.org/ 10.1016/j.schres.2012.08.002. Goto, N., Yoshimura, R., Kakeda, S., Moriya, J., Hayashi, K., Ikenouchi-Sugita, A., et al., 2011. Comparison of brain N-acetylaspartate levels and serum brain-derived neurotrophic factor (BDNF) levels between patients with first-episode schizophrenia psychosis and healthy controls. European Psychiatry 26 (1), 57–63. http://dx.doi.org/10.1016/j.eurpsy.2009.10.001. Gratacòs, M., González, J.R., Mercader, J.M., de Cid, R., Urretavizcaya, M., Estivill, X., 2007. Brain-derived neurotrophic factor Val66Met and psychiatric disorders: meta-analysis of case-control studies confirm association to substance-related disorders, eating disorders, and schizophrenia. Biological Psychiatry 61, 911– 922. http://dx.doi.org/10.1016/j.biopsych.2006.08.025. Green, M.F., Nuechterlein, K.H., 1999. Should schizophrenia be treated as a neurocognitive disorder? Schizophrenia Bulletin 25 (2), 309–319. Gray, J.A., Roth, B.L., 2007. Molecular targets for treating cognitive dysfunction in schizophrenia. Schizophrenia Bull 33, 1100–1119. http://dx.doi.org/10.1093/ schbul/sbm074. Green, M.J., Matheson, S.L., Shepherd, A., Weickert, C.S., Carr, V.J., 2011. Brainderived neurotrophic factor levels in schizophrenia: a systematic review with meta-analysis. Molecular Psychiatry 16 (9), 960–972. http://dx.doi.org/10.1038/ mp.2010.88. Griffin, É.W., Mullally, S., Foley, C., Warmington, S.A., O’Mara, S.M., Kelly, A.M., 2011. Aerobic exercise improves hippocampal function and increases BDNF in the serum of young adult males. Physiology and Behavior 104 (5), 934–941. http: //dx.doi.org/10.1016/j.physbeh.2011.06.005.

Grissom, R.J., Kim, J.J., 2005. Effect Sizes for Research: A Broad Practical Approach. Erlbaum, Mahwah, NJ. Grove, W.M., 1991. When is a diagnosis worth making? A statistical comparison of two prediction strategies. Psychological Reports 69 (1), 3–17. Gruber, O., Hasan, A., Scherk, H., Wobrock, T., Schneider-Axmann, T., Ekawardhani, S., Schmitt, A., Backens, M., Reith, W., Meyer, J., Falkai, P., 2012. Association of the brain-derived neurotrophic factor val66met polymorphism with magnetic resonance spectroscopic markers in the human hippocampus: in vivo evidence for effects on the glutamate system. European Archives Psychiatry Clinical Neurosciences 262, 23–31. http://dx.doi.org/10.1007/s00406-011-0214-6. n Ho, B.-C., Milev, P., O’Leary, D.S., Librant, A., Andreasen, N.C., Wassink, T.H., 2006. Cognitive and magnetic resonance imaging brain morphometric correlates of brain-derived neurotrophic factor Val66Met gene polymorphism in patients with schizophrenia and healthy volunteers. Archives of General Psychiatry 63 (7), 731–740. http://dx.doi.org/10.1001/archpsyc.63.7.731. n Ho, B.-C., Andreasen, N.C., Dawson, J.D., Wassink, T.H., 2007. Association between brain-derived neurotrophic factor Val66Met gene polymorphism and progressive brain volume changes in schizophrenia. Am Journal of Psychiatry 164, 1890–1899. http://dx.doi.org/10.1176/appi.ajp.2007.05111903. Holm, S., 1979. A simple sequential rejective method procedure. Scandinavian Journal of Statistics 6, 65–70. Hosang, G.M., Uher, R., Keers, R., Cohen-Woods, S., Craig, I., Korszun, A., Perry, J., Tozzi, F., Muglia, P., McGuffin, P., Farmer, A.E., 2010. Stressful life events and the brain-derived neurotrophic factor gene in bipolar disorder. Journal of Affective Disorders 125, 345–349. http://dx.doi.org/10.1016/j.jad.2010.01.071. Huang, E.J., Reichardt, L.F., 2003. Trk receptors: roles in neuronal signal transduction. Annual Review of Biochemistry 72, 609–642. 10.1146/annurev.biochem.72. 121801.161629. Hwang, J.-P., Tsai, S.-J., Hong, C.-J., Yang, C.-H., Lirng, J.-F., Yang, Y.-M., 2006. The Val66Met polymorphism of the brain-derived neurotrophic-factor gene is associated with geriatric depression. Neurobiology of Aging 27, 1834–1837. http://dx.doi.org/10.1016/j.neurobiolaging.2005.10.013. Iga, J., Ueno, S., Yamauchi, K., Numata, S., Tayoshi-Shibuya, S., Kinouchi, S., Nakataki, M., Song, H., Hokoishi, K., Tanabe, H., Sano, A., Ohmori, T., 2007. The Val66Met polymorphism of the brain-derived neurotrophic factor gene is associated with psychotic feature and suicidal behavior in Japanese major depressive patients. American Journal of Medical Genetics B Neuropsychiatric Genetics 144B, 1003– 1006. http://dx.doi.org/10.1002/ajmg.b.30520. Islam, O., Loo, T.X., Heese, K., 2009. Brain-derived neurotrophic factor (BDNF) has proliferative effects on neural stem cells through the truncated TRK-B receptor, MAP kinase, AKT, and STAT-3 signaling pathways. Current Neurovascular Research 6 (1), 42–53. Joyce, E.M., Roiser, J.P., 2007. Cognitive heterogeneity in schizophrenia. Current Opinions in Psychiatry 20 (3), 268–272. Kambeitz, J.P., Bhattacharyya, S., Kambeitz-Ilankovic, L.M., Valli, I., Collier, D.A., McGuire, P., 2012. Effect of BDNF val(66)met polymorphism on declarative memory and its neural substrate: a meta-analysis. Neuroscience and Biobehavioral Reviews 36, 2165–2177. http://dx.doi.org/10.1016/j.neubiorev.2012.07. 002. Kim, B., Kim, C.Y., Hong, J.P., Kim, S.Y., Lee, C., Joo, Y.H., 2008. Brain-derived neurotrophic factor Val/Met polymorphism and bipolar disorder. Association of the Met allele with suicidal behavior of bipolar patients. Neuropsychobiology 58, 97–103. http://dx.doi.org/10.1159/000162356. Koolschijn, PCMP, van Haren, N.E.M., Bakker, S.C., Hoogendoorn, M.L.C., Hulshoff Pol, H.E., Kahn, R.S., 2010. Effects of brain-derived neurotrophic factor Val66Met polymorphism on hippocampal volume change in schizophrenia. Hippocampus 20 (9), 1010–1017. http://dx.doi.org/10.1002/hipo.20699. Koshimizu, H., Kiyosue, K., Hara, T., Hazama, S., Suzuki, S., Uegaki, K., Nagappan, G., Zaitsev, E., Hirokawa, T., Tatsu, Y., Ogura, A., Lu, B., Kojima, M., 2009. Multiple functions of precursor BDNF to CNS neurons: negative regulation of neurite growth, spine formation and cell survival. Molecular Brain 2, 27. n Lu, W., Zhang, C., Yi, Z., Li, Z., Wu, Z., Fang, Y., 2012. Association between BDNF Val66Met polymorphism and cognitive performance in antipsychotic-naïve patients with schizophrenia. Journal of Molescular Neuroscience 47 (3), 505–510. http://dx.doi.org/10.1007/s12031-012-9750-4. Luo, J., Niki, K., 2003. Function of hippocampus in “insight” of problem solving. Hippocampus 13 (3), 316–323. http://dx.doi.org/10.1002/hipo.10069. Mandelman, S.D., Grigorenko, E.L., 2012. BDNF Val66Met and cognition: all, none, or some? A meta-analysis of the genetic association. Genes, Brain, and Behavior 11 (2), 127–136. n Martinho, E., Michelon, L., Ayres, A.M., Scazufca, M., Menezes, P.R., Schaufelberger, M.S., et al., 2012. BDNF gene polymorphism, cognition and symptom severity in a Brazilian population-based sample of first-episode psychosis subjects. Revista Brasileira de Psiquiatria 34 (Suppl. 2), S219–S225. Miyajima, F., Ollier, W., Mayes, A., Jackson, A., Thacker, N., Rabbitt, P., Pendleton, N., Horan, M., Payton, A., 2008. Brain-derived neurotrophic factor polymorphism Val66Met influences cognitive abilities in the elderly. Genes Brain Behavior 7, 411–417. http://dx.doi.org/10.1111/j.1601-183X.2007.00363.x. Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., The PRISMA Group, 2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Medicine 6 (7), e1000097. Molendijk, M.L., van Tol, M.-J., BWJH, Penninx, van der Wee, N.J.A., Aleman, A., Veltman, D.J., et al., 2012. BDNF val66met affects hippocampal volume and emotion-related hippocampal memory activity. Translational Psychiatry 2 (1), e74. http://dx.doi.org/10.1038/tp.2011.72.



Neves-Pereira, M., Cheung, J.K., Pasdar, A., Zhang, F., Breen, G., Yates, P., et al., 2005. BDNF gene is a risk factor for schizophrenia in a Scottish population. Molecular Psychiatry 10 (2), 208–212. http://dx.doi.org/10.1038/sj.mp.4001575. Nguyen, N., Lee, S.B., Lee, Y.S., Lee, Y.S., Lee, K.-H., Ahn, J.-Y., 2009. Neuroprotection by NGF and BDNF against neurotoxin-exerted apoptotic death in neural stem cells are mediated through Trk receptors, activating PI3-kinase and MAPK pathways. Neurochemical Research 34 (5), 942–951. http://dx.doi.org/10.1007/ s11064-008-9848-9. n Niitsu, T., Shirayama, Y., Matsuzawa, D., Hasegawa, T., Kanahara, N., Hashimoto, T., et al., 2011. Associations of serum brain-derived neurotrophic factor with cognitive impairments and negative symptoms in schizophrenia. Progress in Neuropsychopharmacology and Biological Psychiatry 35 (8), 1836–1840. http://dx.doi. org/10.1016/j.pnpbp.2011.09.004. Numata, S., Ueno, S.-I., Iga, J.-I., Yamauchi, K., Hongwei, S., Ohta, K., et al., 2006. Brain-derived neurotrophic factor (BDNF) Val66Met polymorphism in schizophrenia is associated with age at onset and symptoms. Neuroscience Letters 401 (1–2), 1–5. http://dx.doi.org/10.1016/j.neulet.2006.02.054. Paz, R.D., Andreasen, N.C., Daoud, S.Z., Conley, R., Roberts, R., Bustillo, J., PerroneBizzozero, N.I., 2006. Increased expression of activity-dependent genes in cerebellar glutamatergic neurons of patients with schizophrenia. American Journal of Psychiatry 163 (10), 1829–1831. http://dx.doi.org/10.1176/appi. ajp.163.10.1829. Perroud, N., Courtet, P., Vincze, I., Jaussent, I., Jollant, F., Bellivier, F., Leboyer, M., Baud, P., Buresi, C., Malafosse, A., 2008. Interaction between BDNF Val66Met and childhood trauma on adult’s violent suicide attempt. Genes Brain Behavior 7, 314–322. http://dx.doi.org/10.1111/j.1601-183X.2007.00354.x. Pezawas, L., Verchinski, B.A., Mattay, V.S., Callicott, J.H., Kolachana, B.S., Straub, R.E., et al., 2004. The brain-derived neurotrophic factor val66met polymorphism and variation in human cortical morphology. Journal of Neuroscience 24 (45), 10099–10102. http://dx.doi.org/10.1523/jneurosci.2680-04.2004. Pillai, A., Buckley, P.F., 2012. Reliable biomarkers and predictors of schizophrenia and its treatment. Psychiatric Clinics of North America 35 (3), 645–659. http: //dx.doi.org/10.1016/j.psc.2012.06.006. Pruunsild, P., Kazantseva1, A., Aid, T., Palm, K., Timmusk, T., 2007. Dissecting the human BDNF locus: bidirectional transcription, complex splicing, and multiple promoters. Genomics 90 (3), 397–406. http://dx.doi.org/10.1016/j.ygeno.2007. 05.004. Ray, M.T., Shannon Weickert, C., Webster, M.J., 2014. Decreased BDNF and TrkB mRNA expression in multiple cortical areas of patients with schizophrenia and mood disorders. Translational Psychiatry 4 (5), e389. Ribasés, M., Gratacòs, M., Armengol, L., de Cid, R., Badía, A., Jiménez, L., Solano, R., Vallejo, J., Fernández, F., Estivill, X., 2003. Met66 in the brain-derived neurotrophic factor (BDNF) precursor is associated with anorexia nervosa restrictive type. Molecular Psychiatry 8, 745–751, http://dx.doi.org/10.1038/sj.mp. 4001281. Rosa, A., Cuesta, M.J., Fatjó-Vilas, M., Peralta, V., Zarzuela, A., Fañanás, L., 2006. The Val66Met polymorphism of the brain-derived neurotrophic factor gene is associated with risk for psychosis: evidence from a family-based association study. American Journal of Medical Genetics. Part B, Neuropsychiatric Genetics 141B (2), 135–138. http://dx.doi.org/10.1002/ajmg.b.30266. n Rybakowski, J.K., Borkowska, A., Skibinska, M., Szczepankiewicz, A., Kapelski, P., Leszczynska-Rodziewicz, A., et al., 2006. Prefrontal cognition in schizophrenia and bipolar illness in relation to Val66Met polymorphism of the brain-derived neurotrophic factor gene. Psychiatry and Clinical Neurosciences 60 (1), 70–76. http://dx.doi.org/10.1111/j.1440-1819.2006.01462.x. Savitz, J., Solmsm, M., Ramesar, R., 2006. The molecular genetics of cognition: dopamine, COMT and BDNF. Genes, Brain and Behavior 5, 311–328. Smith, G.N., Thornton, A.E., Lang, D.J., Macewan, G.W., Ehmann, T.S., Kopala, L.C., et al., 2012. Hippocampal volume and the brain-derived neurotrophic factor Val66Met polymorphism in first episode psychosis. Schizophrenia Research 134 (2–3), 253–259. http://dx.doi.org/10.1016/j.schres.2011.11.022. Takahashi, T., Suzuki, M., Tsunoda, M., Kawamura, Y., Takahashi, N., Tsuneki, H., Kawasaki, Y., Zhou, S.-Y., Kobayashi, S., Sasaoka, T., Seto, H., Kurachi, M., Ozaki, N., 2008. Association between the brain-derived neurotrophic factor Val66Met

13

polymorphism and brain morphology in a Japanese sample of schizophrenia and healthy comparisons. Neuroscience Letters 435, 34–39. http://dx.doi.org/ 10.1016/j.neulet.2008.02.004. Teng, H.K., Teng, K.K., Lee, R., Wright, S., Tevar, S., Almeida, R.D., Kermani, P., Torkin, R., Chen, Z.Y., Lee, F.S., Kraemer, R.T., Nykjaer, A., Hempstead, B.L., 2005. ProBDNF induces neuronal apoptosis via activation of a receptor complex of p75NTR and sortilin. Journal of Neuroscience 25 (22), 5455–5463. Terracciano, A., Tanaka, T., Sutin, A.R., Deiana, B., Balaci, L., Sanna, S., et al., 2010. BDNF Val66Met is associated with introversion and interacts with 5-HTTLPR to influence neuroticism. Neuropsychopharmacology 35 (5), 1083–1089. http: //dx.doi.org/10.1038/npp.2009.213. Tost, H., Alam, T., Geramita, M., Rebsch, C., Kolachana, B., Dickinson, D., et al., 2013. Effects of the BDNF val(66)met polymorphism on white matter microstructure in healthy adults. Neuropsychopharmacology 38 (3), 525–532. http://dx.doi. org/10.1038/npp.2012.214. Tsai, S.-J., Hong, C.-J., Yu, Y.W.-Y., Chen, T.-J., 2004. Association study of a brainderived neurotrophic factor (BDNF) Val66Met polymorphism and personality trait and intelligence in healthy young females. Neuropsychobiology 49, 13–16. http://dx.doi.org/10.1159/000075333. Tsuchimine, S., Sugawara, N., Ishioka, M., Yasui-Furukori, N., 2014. Preanalysis storage conditions influence the measurement of brain-derived neurotrophic factor levels in peripheral blood. Neuropsychobiology 69 (2), 83–88. http://dx. doi.org/10.1159/000358061. Van Winkel, R., Stefanis, N.C., Myin-Germeys, I., 2008. Psychosocial stress and psychosis. A review of the neurobiological mechanisms and the evidence for gene-stress interaction. Schizophrenia Bulletin 34 (6), 1095–1105. http://dx.doi. org/10.1093/schbul/sbn101. Vinogradov, S., Fisher, M., Holland, C., Shelly, W., Wolkowitz, O., Mellon, S.H., 2009. Is serum brain-derived neurotrophic factor a biomarker for cognitive enhancement in schizophrenia? Biological Psychiatry 66 (6), 549–553. http://dx.doi. org/10.1016/j.biopsych.2009.02.017. n Vyas, N.S., Puri, B.K., 2012. Evidence for an association between brain-derived neurotrophic factor Val66Met gene polymorphism and general intellectual ability in early-onset schizophrenia. Israel Journal of Psychiatry and Related Sciences 49 (2), 137–142. Weickert, C.S., Hyde, T.M., Lipska, B.K., Herman, M.M., Weinberger, D.R., Kleinman, J.E., 2003. Reduced brain-derived neurotrophic factor in prefrontal cortex of patients with schizophrenia. Molecular Psychiatry 8 (6), 592–610. Wiener, H., Klei, L., Calkins, M., Wood, J., Nimgaonkar, V., Gur, R., et al., 2013. Principal components of heritability from neurocognitive domains differ between families with schizophrenia and control subjects. Schizophrenia Bulletin 39 (2), 464–471. http://dx.doi.org/10.1093/schbul/sbr161. Witte, A.V., Kürten, J., Jansen, S., Schirmacher, A., Brand, E., Sommer, J., Flöel, A., 2012. Interaction of BDNF and COMT polymorphisms on paired-associative stimulation-induced cortical plasticity. Journal of Neuroscience 32, 4553–4561. http://dx.doi.org/10.1523/JNEUROSCI.6010-11.2012. Wong, J., Hyde, T.M., Cassano, H.L., Deep-Soboslay, A., Kleinman, J.E., Weickert, C.S., 2010. Promoter specific alterations of BDNF mRNA in schizophrenia. Neuroscience 169 (3), 1071–1084. Zakharyan, R., Boyajyan, A., 2014. Brain-derived neurotrophic factor blood levels are decreased in schizophrenia patients and associate with rs6265 genotypes. Clinical Biochemistry 47 (12), 1052–1055. Zeithamova, D., Schlichting, M.L., Preston, A.R., 2012. The hippocampus and inferential reasoning: building memories to navigate future decisions. Frontiers in Human Neuroscience 6, 70. http://dx.doi.org/10.3389/fnhum.2012.00070. n Zhang, X.Y., Chen, D.C., Xiu, M.H., Haile, C.N., Luo, X., Xu, K., et al., 2012a. Cognitive and serum BDNF correlates of BDNF Val66Met gene polymorphism in patients with schizophrenia and normal controls. Human Genetics 131 (7), 1187–1195. http://dx.doi.org/10.1007/s00439-012-1150-x. n Zhang, X.Y., Liang, J., Chen, D.C., Xiu, M.H., Yang, F.D., Kosten, T.A., Kosten, T.R., 2012b. Low BDNF is associated with cognitive impairment in chronic patients with schizophrenia. Psychopharmacology 222 (2), 277–284. http://dx.doi.org/ 10.1007/s00213-012-2643-y.
