Australian and Newhttp://anp.sagepub.com/ Zealand Journal of Psychiatry
A meta-analysis of fMRI studies in healthy relatives of patients with schizophrenia Claire Scognamiglio and Josselin Houenou Aust N Z J Psychiatry published online 27 June 2014 DOI: 10.1177/0004867414540753 The online version of this article can be found at: http://anp.sagepub.com/content/early/2014/06/26/0004867414540753
Published by: http://www.sagepublications.com
On behalf of:
The Royal Australian and New Zealand College of Psychiatrists
Additional services and information for Australian and New Zealand Journal of Psychiatry can be found at: Email Alerts: http://anp.sagepub.com/cgi/alerts Subscriptions: http://anp.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav
>> OnlineFirst Version of Record - Jun 27, 2014 What is This?
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
540753 research-article2014
ANP0010.1177/0004867414540753Australian & New Zealand Journal of PsychiatryScognamiglio and Houenou
Research Australian & New Zealand Journal of Psychiatry 1–10 DOI: 10.1177/0004867414540753
A meta-analysis of fMRI studies in healthy relatives of patients with schizophrenia
© The Royal Australian and New Zealand College of Psychiatrists 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav anp.sagepub.com
Claire Scognamiglio1 and Josselin Houenou2,3,4,5
Abstract Objective: Genetically at-risk yet healthy relatives of patients with schizophrenia, sharing an important part of the genetic susceptibility to the disease, allow the study of neuroimaging endophenotypes. The aim of our study was to perform a meta-analysis of whole-brain functional magnetic resonance imaging (fMRI) studies that compared adult healthy relatives of patients with schizophrenia and controls. Methods: Twenty-one whole-brain fMRI studies were included (17 using cognitive tasks and four using emotional tasks), published between 2003 and 2013. These studies included 467 healthy relatives of patients with schizophrenia and 768 controls. To conduct the statistical analysis, we used the effect-size signed differential mapping software, a voxel-based meta-analytic approach. Results: In healthy relatives of patients with schizophrenia, we observed a general pattern of overactivation across the 21 fMRI studies in right-sided frontal, parietal and temporal regions compared to controls. This pattern was accompanied by an underactivation in the cingulate gyrus. Our analyses showed a very similar pattern during purely cognitive tasks; during emotional tasks, healthy relatives additionally overactivated the left parahippocampal gyrus. Conclusions: This fMRI pattern of prefrontal overactivation and hypoactivation of the cingulate gyrus may represent a candidate endophenotype for schizophrenia. Keywords Endophenotype, fMRI, healthy relatives, meta-analysis, schizophrenia
Introduction Schizophrenia is a chronic and multifactorial disorder, stemming from interplay of genetic vulnerability and environmental contributors. The lifetime risk for schizophrenia is approximately 0.5–2% in the general population, 6% in parents and 9% in siblings (Lichtenstein et al., 2009). The concordance rate for monozygotic twins is about 45% and 12% for dizygotic twins, suggesting a heritability around 64% (Lichtenstein et al., 2009). Molecular genetic studies reported several susceptibility genes possibly implicated in schizophrenia such as ZNF804 and DISC1 (Duff et al., 2013). Neuroimaging studies provide a useful insight on the etiopathophysiology of schizophrenia. The amount of neuroimaging research publications on schizophrenia has grown exponentially in the last years. Studies in patients with schizophrenia have consistently demonstrated anatomical abnormalities in cortical grey matter volume, involving frontal, temporal and parietal lobes, and left
Heschl gyrus (Vita et al., 2013). Concerning functional magnetic resonance imaging (fMRI), studies in patients have mainly focused on executive processing and facial affect processing. In a meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia, Minzenberg et al. (2009) reported a reduced activation of the prefrontal cortex (middle frontal, inferior frontal, 1Paris
Ile de France Ouest Medical School, Université Versailles SaintQuentin en Yvelines, Versailles, France 2UNIACT, NeuroSpin, I2BM, CEA Saclay, Gif-Sur-Yvette, France 3INSERM U955, Equipe 15 ‘Psychiatrie Génétique’, Créteil, France 4Fondation Fondamental, Créteil, France 5AP-HP, Hôpitaux Universitaires Mondor, Pôle de Psychiatrie, Créteil, France Corresponding author: Josselin Houenou, INSERM U955, Equipe 15 ‘Psychiatrie Génétique’, 40 rue de Mesly, Créteil 94000, France. Email: [email protected]
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 1
24/06/2014 12:47:04 PM
2
ANZJP Articles
superior frontal), anterior cingulate, basal ganglia and insula, and increased activation in several midline cortical areas in patients versus healthy controls (Minzenberg et al., 2009). With regard to emotional processing, a recent metaanalysis (Delvecchio et al., 2013) reported an underactivation in patients compared to controls throughout the entire facial affect processing network and an increased activation in visual processing regions (cuneus). Genetically at-risk yet healthy relatives of patients with schizophrenia are an excellent population in which to study neuroimaging endophenotypes. Endophenotypes were originally described as internal phenotypes that mediate the path between genes and the disease (Gottesman and Gould, 2003). Gottesman defined the endophenotypes as measurable components unseen by the unaided eye along the pathway between disease and distal genotype. They represent simpler clues to genetic underpinnings than the disease syndrome itself. They may be of neurophysiological, biochemical, endocrinological, neuroanatomical, cognitive, or neuropsychological nature. The most commonly accepted criteria for a genetically influenced endophenotype include trait identification in an objective and quantitative manner in patients before onset of the disorder and/or periods of remission; they should also run in families and be associated with an increased risk of clinical illness (Courtet et al., 2011). Endophenotypes thus provide critical evidence on the mechanistic pathways associated with genetic susceptibility. In addition, some authors have described resilience (Smieskova et al., 2012) as an ability to recover from change, and resilience factors as a protector for subjects with an at-risk mental state to develop psychosis. Unravelling such factors may prove fruitful in developing prevention strategies. Healthy relatives share 50% of the genetic susceptibility with patients. Studies in relatives are not biased by medication effects, number of episodes, and potential neurodegenerative illness progression – as in patients. In 2009, a qualitative review of 20 functional neuroimaging studies in healthy relatives, of executive functioning, language and memory paradigms, found that the right dorsal prefrontal region (Brodmann area (BA) 8/9/10/46) showed hyperactivity in 50% of contrasts and hypoactivity in 31% of contrasts (MacDonald et al., 2009). In 2010, a meta-analysis restricted to fMRI executive tasks reported that healthy relatives of patients had hypo- and hyperactivity in statistically overlapping right middle frontal regions (BA 9/10) (Goghari, 2011). This meta-analysis used anatomic likelihood estimation (ALE), a technique that does not take into account in the same analysis hyper- and hypoactivations, and also does not use the effect sizes reported. A recently developed method, effect-size signed differential mapping (ES-SDM) (Radua et al., 2012), allows these limitations to be overcome. The aim of this paper was thus to perform a meta-analysis of whole-brain fMRI studies of cognitive and emotional tasks in adult healthy relatives of patients with schizophrenia using ES-SDM in order to identify consistent functional
neuroimaging endophenotypes and resiliency markers of schizophrenia. We hypothesized that healthy relatives of patients with schizophrenia would exhibit abnormal activations in fronto-temporal regions, mainly during cognitive tasks, and in limbic regions during emotional tasks.
Materials and methods Studies selection A comprehensive literature search of whole-brain fMRI studies conducting comparisons between adult healthy relatives of schizophrenic patients (according to standardized diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV)) and healthy controls was conducted using two databases (PubMed, Google Scholar), using combinations of the following terms: relatives, twins, offspring, children, family studies, genetic risk, high risk, genetic liability, fMRI, schizophrenia, whole-brain. Articles found to be eligible on the basis of their title and, subsequently, abstracts were then selected to further determine suitability for inclusion in this study. Comprehensive reviews and reference lists were also checked to provide further studies. In case details such as coordinates were not included in the original publications, we contacted the corresponding authors for these details. We excluded studies that: (1) used exclusively a regionof-interest (ROI) approach; (2) did not report coordinates for the relevant contrasts and did/could not supply these when contacted; (3) were not reported in English; (4) were review articles, comments, and case reports; (5) used populations already included in another study; and (6) included relatives with significant symptoms. To perform the analyses, we extracted data on the characteristics of the studies and patients, measurements performed and results. For each report, we extracted the following items: author; year; sample size; description of study population (mean age, sex, family link between relatives and patients); fMRI characteristics (field strength in Tesla, Montreal Neurological Institute (MNI) or Talairach coordinates, the full width at half maximum (FWHM), software used, brain function investigated); and behavioural performance. In case several contrasts or tasks were present in the same study (e.g. different attentional tasks: sustained, selective and then dual attention task), we decided to choose only one task in order not to include the population twice. We opted to select, in an arbitrary way, the main result as described by the authors in their article (the one they were mainly discussing), or the task in the highest number of significant peak coordinates was found.
Statistical analysis Regional differences in activation during tasks between healthy relatives and controls were analyzed using ES-SDM
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 2
24/06/2014 12:47:05 PM
3
Scognamiglio and Houenou software (www.sdmproject.com), a voxel-based meta-analytic approach. The SDM methods have been described in detail elsewhere (Radua et al., 2011, 2012). In summary, SDM is a voxel-based meta-analytic approach that uses the reported peak coordinates in the different whole-brain studies. Once the coordinates are selected and converted into a common space, a map of the neurofunctional differences is separately recreated for each study using the peak t-values. This allows maps to be recreated of the effect size of the differences in blood oxygenation level dependent (BOLD) responses between healthy relatives and controls. ES-SDM uses hyperactivation and underactivation peaks on the same map. Studies are then combined with a random-effects model that allows sample size, intra-study variability and between-study heterogeneity to be taken into account (Radua et al., 2012). We conducted three analyses: the first one included all the different tasks that were used in the studies, regardless of their category (emotional or cognitive); the second one included only cognitive tasks; and the last one included emotional tasks. To test the robustness of the findings, these analyses were complemented with a jackknife sensitivity analysis. This analysis consisted of repeating the analysis and discarding (only) one study each time, and is used to assess the replicability of the results. We also looked for heterogeneity between studies by visually inspecting the Q statistics and funnel plots. Statistical significance was achieved with 50 permutations for all the analyses (standard permutation tests), as it was shown to have a high statistical stability. Results were then thresholded at p1 (peak height), and cluster extent greater or equal to 10 voxels; these thresholds have proven to yield consistent, valid and replicable results (Radua et al., 2012). The p threshold at 0.005 is uncorrected for multiple comparisons, based on the empirical validation detailed in this paper (Radua et al., 2012). Indeed, they proposed to use an uncorrected p=0.005 as the main threshold, as this was found to optimally balance sensitivity and specificity and to be an approximate equivalent to corrected p-value = 0.05 (indeed 0.025) in ES-SDM. Of course, this equivalence is only approximate, so to reduce the possibility of false positive results we suggest an additional z-based threshold (e.g. z>1). It would be associated to a clearly non-significant p-value under the standard normal distribution.
Results Studies selected Our literature search yielded 21 studies that included adult subjects, published between 2003 and September 2013 (Callicott et al., 2003; Choi et al., 2012; de Achaval et al., 2012; Delawalla et al., 2008; Filbey et al., 2008; Jamadar et al., 2013; Karch et al., 2009; Lee et al., 2010;
Li et al., 2007, 2012; Liddle et al., 2013; McAllindon et al., 2010; MacDonald et al., 2006; Sambataro et al., 2013; Sepede et al., 2010; Thermenos et al., 2013; van Buuren et al., 2011; Venkatasubramanian et al., 2010; Whalley et al., 2004; Woodward et al., 2007). We then categorized them into cognitive (n=17) and emotional (n=4) tasks. The characteristics of the studies included (author, year, population, field strength, etc.) are summarized in Table 1. Therefore, our meta-analysis includes the data of 467 relatives and 768 controls. Concerning the behavioural performances during the task, the authors noted no significant difference between relatives and controls in 10 studies. Nothing was specified in seven studies and there was a significant difference in four studies. The differences were that relatives executed tasks slower than controls (n=1), had a longer reaction time (n=1) or had more errors (n=2).
Meta-analysis including all 21 studies (Table 2) Our first analysis included all the 21 adult studies. We found in relatives of schizophrenic patients versus controls a significant hyperactivation in right-sided regions: middle temporal gyrus (BA37), inferior frontal gyrus (BA44), and superior parietal lobule (BA7). Hypoactivation clusters were restricted to the left cingulate gyrus (BA24) (Table 3). These results were highly replicable in the jackknife analysis. Results are displayed in Figures 1 (hyperactivation) and 2 (hypoactivation).
Cognitive processing task-related activations (Table 3) Seventeen whole-brain fMRI studies reported cognitive functioning task-related activity comparing healthy relatives of patients with schizophrenia and control groups. Those tasks included working memory tasks (n=4), go/ no-go (n=1), stroop (n=1), attentional processes (n=3), covert verbal initiation task (n=1), reaction time task (n=1), visual task (n=2), lexical decision task (n=2), executive task (n=1), and auditory detection task (n=1). Significant hyperactivations in relatives were found in seven clusters, encompassing right-sided clusters: inferior frontal gyrus (BA45), precuneus (BA7), middle and superior temporal gyrus (BA37 and 39), caudate, inferior parietal lobule (BA40), but also left precentral gyrus (BA6). By contrast, hypoactivations were only significant in the right cingulate gyrus (BA31) (Table 3).
Emotional processing task-related activations (Table 4) Four studies used tasks categorized as measuring emotional processing. The tasks were: eyes emotional theory of mind (n=1) and facial emotional valence discrimination (n=3). Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 3
24/06/2014 12:47:05 PM
ANP540753.indd 4
MacDonald et al., 2006 Callicott et al., 2003-1 Callicott et al. 2003-2 Sambataro et al., 2013 Thermenos et al., 2013 Jamadar et al., 2013
Filbey et al., 2008 Sepede et al., 2010
Delawalla et al., 2008 Woodward et al., 2007 Karch et al., 2009 McAllindon et al., 2010 Lee et al., 2010
Li X et al., 2007
Whalley et al., 2004
Cognitive tasks Choi et al., 2012 Liddle et al., 2013
Study name
23 25 65 43 27
Working memory
Working memory
Go/no-go task
Lexical decision task
Lexical decision task
6 11 21
33.6 (8.79) 34.7
11 11
40.78 (2.5)
25.2 (3.1)
36.6 (10.4)
36.6 (9)
34.4 (9)
33.2 (10.9)
53 34.4 (8.8)
36.0 (10.3)
36.9 (13.3)
12
21
21.3 (3.5)
21.7 (6.1)
26.8 (3.4)
20.71 (5.5) 20.4 (3.5)
Mean age, years (SD)
30
15
42
17 18
n
Visual backward masking performance Attention process Continuous performance task Cognitive task
Working memory Auditory target detection task Covert verbal initiation task Visual word decision task Continuous performance task Serial reaction time task Working memory Reaction time
Task
Relatives
Population characteristics
Table 1. Characteristics of the studies included in the meta-analysis.
4/23
13/30
24/41
11/14
6/17
7/14
2/4 5/6
11/10
4/7 11/0
5/7
14/16
7/8
17/25
9/8 7/11
Sex (M/F)
First-degree
Unspecified
–
Siblings
Siblings
First-degree
Unspecified Siblings
Siblings
Unspecified First-degree
Siblings
Siblings
Unspecified
Unspecified
Unspecified Siblings
Relationship
133
32
235
15
18
20
8 11
19
11 14
15
92
15
21
16 26
n
Controls
32.48 (1.05)
24.6 (2.8)
31.8 (9.6)
27.9 (8)
29.6 (7)
33.4 (8.4)
41 32 (5.1)
42.7 (9.0)
33.8 (9.20) 36.4
31.3 (11.2)
20.2 (3.4)
23.9 (5.2)
26.8 (2.7)
21.37 (2.8) 20.3 (4.9)
Mean age, years (SD)
5/3 5/6
14/5
4/7
68/65
13/19
113/122
6/9
11/7
10/10
–
10/5
39/53
7/8
13/8
9/7 15/11
Sex (M/F)
3
3
3
1.5
1.5
–
1.5 1.5
3
1.5 1.5
1.5
1.5
1.5
1.5
1.5 3
Field strength (Tesla)
(Continued)
MNI
MNI
MNI
Talairach
Talairach
Talairach
Talairach Talairach
MNI
Talairach Talairach
Talairach
Talairach
Talairach
Talairach
NA MNI
Referential
MRI characteristics
4 ANZJP Articles
Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
24/06/2014 12:47:05 PM
5
Talairach 14/2 24.4
1.5
MNI 7/ 18 27.5
3
MNI 6/6 29.25
3
Talairach – 8/6 28.4
Referential Field strength (Tesla)
16
We performed a SDM meta-analysis to identify convergence across standardized coordinates of 21 fMRI studies comparing healthy relatives of patients with schizophrenia and controls. The relatives demonstrated a pattern of overactivations in a right-sided fronto-parieto-temporal network. During cognitive tasks, these overactivations were mostly present in fronto-temporal regions, whereas in emotional tasks the left parahippocampal was overactivated and a wider range of bilateral regions was underactivated.
SD, standard deviation; M, male; F, female; MNI, Montreal Neurological Institute.
Siblings Venkatasubramanian et al., 2010
17
25.2
14/3
25 Siblings Van Buuren et al., 2011
24
29.4
8/16
12 Siblings 12
Facial emotional valence discrimination Facial emotional valence discrimination Facial emotional valence discrimination
31.25
4/8
14 Siblings 14
Emotional tasks de Achaval et al., 2012 Li et al., 2012
Eyes theory of mind
30.4
8/6
n Relationship Sex (M/F) n Study name
Task
Relatives
Population characteristics Table 1. (Continued)
Six significant clusters of overactivation were identified in the relatives compared to controls: the left sub-gyral (parietal) (BA40), right superior frontal gyrus (BA9), left lentiform nucleus (lateral globus pallidus), left parahippocampal gyrus (BA28), left precuneus (BA7) and the right middle temporal gyrus (BA39). We found hypoactivations in the right precentral gyrus (BA6), right inferior parietal lobule (BA40), left medial frontal gyrus (BA6), and the right inferior frontal gyrus (BA47) in the relatives versus controls (Table 4). The Q heterogeneity statistics of the different metaanalyses results were tested. This shows which brain regions present significant between-study heterogeneity as compared to the global set of voxels, even if they are only marginally heterogeneous. Our results were globally homogeneous. The jackknife analysis was performed to test only the coordinates which were significant in the primary meta-analysis, and all the regions replicate.
Discussion
Mean age, years (SD)
Controls
Mean age, years (SD)
Sex (M/F)
MRI characteristics
Scognamiglio and Houenou
Over- and underactivation of brain areas in healthy relatives of patients The general pattern of our results was that relatives of patients with schizophrenia had to recruit more neural resources during tasks (hyperactivation), especially in prefrontal areas. Taking into consideration the largely preserved behavioural performance of relatives of patients in the various tasks, our results suggest that the healthy relatives of patients may use some compensatory strategies. In other words, relatives would have to engage more neural effort in order to complete tasks by increasing the activation of brain areas engaged in a task or recruiting additional areas (Habel et al., 2010). Relatives may increase engagement of other processes to maintain task performance, such as attentional, mnemonic, and performance monitoring functions. This large pattern of overactivations is not present in a recent meta-analysis of fMRI studies exploring executive domains in patients with schizophrenia (Minzenberg et al., 2009). This study used ALE, therefore leading to reports of overactivations and hypoactivations in the same areas, such as the right medial frontal gyrus. This prevents us from directly comparing Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 5
24/06/2014 12:47:05 PM
6
ANZJP Articles
Table 2. Results of the SDM analysis for the 21 studies. Brain regions
z
p
Voxels
Brodmann area
Talairach coordinates x, y, z
Relatives > controls Right middle temporal gyrus 1.832 0.000069878 296 37 1.608 0.000304945 67 44 Right inferior frontal gyrus 1.360 0.001418369 34 7 Right superior parietal lobule Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≥ 1.138 and peak voxel z ≥ 1.000
46, –60, 2 52, 10, 20 18, –68, 56
Controls > relatives Left cingulate gyrus –2.019 0.000932306 146 24 Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≤ –1.411 and peak voxel z ≤ –1.000
–2, –2, 38
Table 3. Brain regions exhibiting significant activity across the full set of cognitive tasks. Brain regions
z
p
Voxels
Brodmann area
Relatives > controls Right inferior frontal gyrus 1.626 0.000162107 154 45 1.469 0.000525626 60 7 Right precuneus (parietal) 1.424 0.000714451 20 37 Right middle temporal gyrus 1.324 0.001358767 11 41 Right caudate (right transverse temporal gyrus) Right superior temporal 1.301 0.001560694 46 39 gyrus Left precentral gyrus 1.261 0.001966346 14 6 1.152 0.003710212 13 40 Right inferior parietal lobule Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≥ 1.097 and peak voxel z ≥ 1.000 Controls > relatives Right cingulate gyrus –2.319 0.000220681 216 31 Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≤ –1.443 and peak voxel z ≤ –1.000
our results. Nevertheless, several reports have shown that patients with schizophrenia may present a prefrontal inefficiency during cognitive tasks, leading to compensatory increased recruitment during tasks with high cognitive demands (Ettinger et al., 2011; Potkin et al., 2009). An argument that supports this hypothesis is that meta-analyses on studies of cognitive behavioural deficits in relatives show cognitive deficits in this population (Sitskoorn et al., 2004). Those results were observed while the subjects performed more difficult cognitive tasks, which can show that in those cases, compensatory mechanisms are no longer sufficient. In contrast, we found the cingulate gyrus to be consistently hypoactive across tasks in relatives of patients. These underactivations are also frequently reported in patients with schizophrenia (Adams and David, 2007). This pattern of overactivation of right prefrontal areas and underactivation of cingulate gyrus may thus represent an endophenotype of the disease, linked with the genetic susceptibility to schizophrenia. This interpretation is further supported by reports that prefrontal efficiency during a
Talairach coordinates x, y, z 54, 12, 20 14, –66, 52 46, –60, 2 32, –36, 4 56, –58, 18 –32, –18, 64 54, –32, 34 8, –8, 44
working memory task is correlated with a cumulative genetic risk score of schizophrenia (Walton et al., 2013) based on at-risk single nucleotide polymorphisms (SNPs) for the disease.
Neural circuits As we assumed, relatives showed abnormal activations in prefronto-temporal networks during cognitive tasks but also in limbic regions during emotional paradigms. Fronto-temporal brain networks implicated in high-level cognitive processes are altered in these subjects. Indeed, during cognitive tasks, relatives overactivated regions such as the right inferior frontal gyrus (BA45), middle and superior temporal gyrus (BA37), and left precentral gyrus (BA6) but also parietal regions (precuneus and inferior parietal lobule). All these regions are implicated in cognitive processing and working memory. They belong to the ‘taskpositive network’. This network is activated during attention-demanding, cognitive tasks (Fox et al., 2005). A recent meta-analysis confirmed that working memory
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 6
24/06/2014 12:47:05 PM
7
Scognamiglio and Houenou
Figure 1. Result of the meta-analysis including the 21 studies: hyperactivations in relatives versus controls. (a) Axial view (sections z = 3, 13, 16, 19, 21, and 26 in MNI space); (b) lateral 3D right view.
Figure 2. Result of the meta-analysis including the 21 studies: hypoactivations in relatives versus controls. Sagittal view (sections x = −11, –3, 4, 7, 8, and 11 in MNI space).
relied largely on a bilateral fronto-parietal network including the inferior frontal gyrus (Rottschy et al., 2012). Our results suggest that during cognitive tasks, relatives of patients overactivated regions that are normally used during these tasks. No obvious additional region is recruited. During emotional tasks, the left parahippocampal gyrus was additionally activated. This region is implicated in the automatic emotional regulation (Phillips et al., 2008). Our finding of hyperactivity in the neural circuitry of emotion processing agrees with studies showing exaggerated reactivity to emotional stimuli in unaffected first-degree relatives of patients with schizophrenia (Myin-Germeys and van Os, 2007). For example, unaffected first-degree relatives, including siblings, show enhanced reactivity to daily life stressors as indicated by a more profound decrease in positive mood and increase in negative mood compared with control subjects in response to stressors (MyinGermeys et al., 2001), as well as an increased physiologic stress response to a metabolic stressor (infusion of the glucose analogue 2-deoxy-d-glucose) (Brunelin et al., 2008). Our result is in contrast with the findings in patients with schizophrenia who exhibit a decreased activation in the bilateral parahippocampal cortices during facial affect processing (Li et al., 2010). This suggests that the overactivation observed in healthy relatives may either be of compensatory nature or a resiliency marker. A striking result in our meta-analysis is that a vast majority of the overactivations observed in the healthy relatives of patients are right-sided for cognitive tasks. Though further exploration of this topic is needed, this suggests an altered balance between the hemispheres function. This
may be paralleled with the findings of a decreased cerebral lateralization and asymmetry in schizophrenia along with a higher prevalence of mixed- and left-handedness in schizophrenia (Sommer et al., 2001). This decreased lateralization is present for language but also motor functions (Altamura et al., 2012). A very recent meta-analysis of voxel base morphometry (VBM) studies in healthy relatives showed that increased grey matter volume was exclusively right-sided and observed in a fronto-parieto-temporal network including the precuneus, found to be overactive in our study (Xiao et al., 2013). This conjunction is in favour of a compensatory mechanism. Goghari et al. performed in 2011 a meta-analysis of fMRI studies in healthy relatives of patients with schizophrenia in comparison to controls, but restricted to executive tasks, using ALE, and including both whole-brain and ROI studies (Goghari, 2011). They did two analyses – one including all the studies and the second only the wholebrain analyses – and they compared both results. Most of the studies with ROI approaches focused on prefrontal regions and displayed reduced functional activations. This was in contrast with whole-brain studies in which both hyper- and hypoactivations were detected. The cause of this discrepancy is yet unclear. Thus, we chose not to include ROI approaches, to limit this potential bias. This study found hypo- and hyperactivity in the right middle frontal regions (BA 9/10) in relatives, and controls activated to a greater degree than relatives the right inferior frontal, right precentral, left posterior cingulate, left thalamus and right lingual regions compared to relatives. Relatives activated to a greater degree the right superior Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 7
24/06/2014 12:47:06 PM
8
ANZJP Articles
Table 4. Brain regions exhibiting significant activity across the full set of emotional tasks. Brain regions Relatives > controls Left sub-gyral (parietal) Right superior frontal gyrus Left lentiform nucleus
z 1.281 1.281 1.280
p
Voxels
0.001004239 0.001004239 0.001004239
26 106 12
Brodmann area
40 9 lateral globus pallidus Left parahippocampal gyrus 1.280 0.001093899 18 28 15 7 Left precuneus 1.274 0.001667052 29 39 Right middle temporal gyrus 1.270 0.002116378 Threshold parameters : Clusters of ≥ 10 voxels, with all voxels z ≥ 1.238 and peak voxel z ≥ 1.000 Controls > relatives Right precentral gyrus –1.041 0.000915607 102 6 66 40 Right inferior parietal lobule –1.023 0.001093642 23 6 Left medial frontal gyrus –1.011 0.001617213 20 47 Right inferior frontal gyrus –1.007 0.001801927 Threshold parameters : Clusters of ≥ 10 voxels, with all voxels z ≤ –0.988 and peak voxel z ≤ –1.000
frontal, right thalamus, right precentral, left inferior parietal and left precuneus regions compared to controls. Since the publication of this meta-analysis, 11 additional studies were published that we included in our study. In addition, because of the ALE algorithm, the authors had to conduct two separate meta-analyses for hyper- and hypoactivations. This leads to the contradictory finding of a hyperand hypoactivity in the right middle frontal regions. Furthermore, ALE does not take into account effect sizes; this is an important point, and it is supported by the SDM software in our analysis. A recent meta-analysis (Cooper et al., 2014) studied the fMRI studies in those at elevated genetic risk of developing schizophrenia and included only 13 studies. We included 21 studies; differences between these two meta-analyses are: first, the inclusion period was longer in our analysis (2003 to January 2012 versus 2003 to September 2013); and second, we contacted several authors (Choi et al., 2012; de Achaval et al., 2012; Sambataro et al., 2013) to get coordinates when they were not included in the publication. These differences may explain why we get very different results. Another difference is that we distinguished emotional and cognitive tasks in our analysis.
Limitations First, our meta-analysis does not include unpublished negative studies because of bias publication. Second, we recognize that there are a limited number of fMRI studies in relatives and especially only four fMRI studies on emotional tasks. We therefore have to interpret with care those results, and especially the generalizability of findings. Our global meta-analysis that included the 21 studies may have been driven by the cognitive tasks (n=17 studies). Third,
Talairach coordinates x, y, z –22, –48, 56 12, 46, 26 –24, –12, –6 –20, –14, –20 –6, –46, 48 50, –66, 10
54, –6, 32 40, –50, 56 –2, –20, 62 52, 28, –12
the acquisition protocols and MRI scanners of the included studies are heterogeneous. The meta-analysis is therefore complicated by the potentially confounding effects of differences between sites. Fourth, we did not include some studies because no coordinates were reported and the authors of these studies did not provide them when we contacted them. Fifth, it may be arbitrary to specify to which category a task belongs (emotional, executive, etc.). We therefore found it more relevant to include all tasks in the main analysis of this meta-analysis. In studies that used two tasks (different cognitive tasks, for example, with the same sample of population), we had to choose one task only, in order not to include a population twice in our analysis. We opted to take the main result or the results with more coordinates. This is questionable, but it would be worse to include a population twice. Sixth, the studies included may not represent the majority of the available literature on fMRI in healthy relatives of patients with schizophrenia as we did not include ROI studies. Seventh, it is important to discuss the fact that some of our results are also found in other disorders, such as prefrontal hyperactivation in bipolar disorder. Indeed, in a systematic review concerning studies about whole-brain fMRI of relatives of patients with bipolar disorder, Fusar Poli et al. (2012) observed that, independently of task, high-risk individuals showed increased neural response in the left superior frontal gyrus and in the medial frontal gyrus, among others. It is important to note that our results may not be specific to schizophrenia. Furthermore, common susceptibility factors may exist between bipolar disorder and schizophrenia (Fusar-Poli et al., 2012). Finally, there were not enough studies specifying exactly the relationship between relatives and patients with schizophrenia (i.e. siblings or children).
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 8
24/06/2014 12:47:06 PM
9
Scognamiglio and Houenou
Conclusion and perspectives In conclusion, our meta-analysis of fMRI studies comparing brain activations between healthy relatives of schizophrenic patients and controls identified abnormalities in brain networks implicated in cognition and emotion processing. We revealed a general pattern of overactivation in prefrontal areas and underactivation of the cingulate gyrus. This pattern may thus be a potential endophenotype of schizophrenia. To get further insight into this question, we suggest that future fMRI in healthy relatives explores emotional and social domains. These endophenotypes may prove useful as intermediate phenotypes to discover the genes associated with schizophrenia. Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Declaration of interests The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
References Adams R and David AS (2007) Patterns of anterior cingulate activation in schizophrenia: A selective review. Neuropsychiatric Disease and Treatment 3: 87–101. Altamura M, Fazio L, De Salvia M, et al. (2012) Abnormal functional motor lateralization in healthy siblings of patients with schizophrenia. Psychiatry Research 203: 54–60. Brunelin J, d’Amato T, van Os J, et al. (2008) Effects of acute metabolic stress on the dopaminergic and pituitary-adrenal axis activity in patients with schizophrenia, their unaffected siblings and controls. Schizophrenia Research 100: 206–211. Callicott JH, Egan MF, Mattay VS, et al. (2003) Abnormal fMRI response of the dorsolateral prefrontal cortex in cognitively intact siblings of patients with schizophrenia. American Journal of Psychiatry 160: 709–719. Choi JS, Park JY, Jung MH, et al. (2012) Phase-specific brain change of spatial working memory processing in genetic and ultra-high risk groups of schizophrenia. Schizophrenia Bulletin 38: 1189–1199. Cooper D, Barker V, Radua J, et al. (2014) Multimodal voxel-based metaanalysis of structural and functional magnetic resonance imaging studies in those at elevated genetic risk of developing schizophrenia. Psychiatry Research 221: 69–77. Courtet P, Gottesman II, Jollant F, et al. (2011) The neuroscience of suicidal behaviors: what can we expect from endophenotype strategies? Translational Psychiatry 1: e7. De Achaval D, Villarreal MF, Costanzo EY, et al. (2012) Decreased activity in right-hemisphere structures involved in social cognition in siblings discordant for schizophrenia. Schizophrenia Research 134: 171–179. Delawalla Z, Csernansky JG and Barch DM (2008) Prefrontal cortex function in nonpsychotic siblings of individuals with schizophrenia. Biological Psychiatry 63: 490–497. Delvecchio G, Sugranyes G and Frangou S (2013) Evidence of diagnostic specificity in the neural correlates of facial affect processing in bipolar disorder and schizophrenia: A meta-analysis of functional imaging studies. Psychological Medicine 43: 553–569. Duff BJ, Macritchie KA, Moorhead TW, et al. (2013) Human brain imaging studies of DISC1 in schizophrenia, bipolar disorder and depression: A systematic review. Schizophrenia Research 147: 1–13.
Ettinger U, Williams SC, Fannon D, et al. (2011) Functional magnetic resonance imaging of a parametric working memory task in schizophrenia: Relationship with performance and effects of antipsychotic treatment. Psychopharmacology (Berl) 216: 17–27. Filbey FM, Russell T, Morris RG, et al. (2008) Functional magnetic resonance imaging (fMRI) of attention processes in presumed obligate carriers of schizophrenia: Preliminary findings. Annals of General Psychiatry 7: 18. Fox MD, Snyder AZ, Vincent JL, et al. (2005) The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the United States of America 102: 9673–9678. Fusar-Poli P, Howes O, Bechdolf A, et al. (2012) Mapping vulnerability to bipolar disorder: A systematic review and meta-analysis of neuroimaging studies. Journal of Psychiatry & Neuroscience: JPN 37: 170–184. Goghari VM (2011) Executive functioning-related brain abnormalities associated with the genetic liability for schizophrenia: An activation likelihood estimation meta-analysis. Psychological Medicine 41: 1239–1252. Gottesman II and Gould TD (2003) The endophenotype concept in psychiatry: Etymology and strategic intentions. American Journal of Psychiatry 160: 636–645. Habel U, Chechko N, Pauly K, et al. (2010) Neural correlates of emotion recognition in schizophrenia. Schizophrenia Research 122: 113–123. Jamadar SD, Pearlson GD, O’Neil KM, et al. (2013) Semantic association fMRI impairments represent a potential schizophrenia biomarker. Schizophrenia Research 145: 20–26. Karch S, Leicht G, Giegling I, et al. (2009) Inefficient neural activity in patients with schizophrenia and nonpsychotic relatives of schizophrenic patients: Evidence from a working memory task. Journal of Psychiatric Research 43: 1185–1194. Lee J, Cohen MS, Engel SA, et al. (2010) Regional brain activity during early visual perception in unaffected siblings of schizophrenia patients. Biological Psychiatry 68: 78–85. Li H, Chan RC, McAlonan GM, et al. (2010) Facial emotion processing in schizophrenia: A meta-analysis of functional neuroimaging data. Schizophrenia Bulletin 36: 1029–1039. Li HJ, Chan RC, Gong QY, et al. (2012) Facial emotion processing in patients with schizophrenia and their non-psychotic siblings: A functional magnetic resonance imaging study. Schizophrenia Research 134: 143–150. Li X, Branch CA, Bertisch HC, et al. (2007) An fMRI study of language processing in people at high genetic risk for schizophrenia. Schizophrenia Research 91: 62–72. Lichtenstein P, Yip BH, Bjork C, et al. (2009) Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: A population-based study. Lancet 373: 234–239. Liddle EB, Bates AT, Das D, et al. (2013) Inefficient cerebral recruitment as a vulnerability marker for schizophrenia. Psychological Medicine 43: 169–182. McAllindon DP, Wilman AH, Purdon SE, et al. (2010) Functional magnetic resonance imaging of choice reaction time in chronic schizophrenia and first-degree relatives. Schizophrenia Research 120: 232–233. MacDonald AW 3rd, Becker TM and Carter CS (2006) Functional magnetic resonance imaging study of cognitive control in the healthy relatives of schizophrenia patients. Biological Psychiatry 60: 1241–1249. MacDonald AW 3rd, Thermenos HW, Barch DM, et al. (2009) Imaging genetic liability to schizophrenia: Systematic review of FMRI studies of patients’ nonpsychotic relatives. Schizophrenia Bulletin 35: 1142–1162. Minzenberg MJ, Laird AR, Thelen S, et al. (2009) Meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia. Archives of General Psychiatry 66: 811–822. Myin-Germeys I and van Os J (2007) Stress-reactivity in psychosis: Evidence for an affective pathway to psychosis. Clinical Psychology Review 27: 409–424.
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 9
24/06/2014 12:47:06 PM
10
ANZJP Articles
Myin-Germeys I, van Os J, Schwartz JE, et al. (2001) Emotional reactivity to daily life stress in psychosis. Archives of General Psychiatry 58: 1137–1144. Phillips ML, Ladouceur CD and Drevets WC (2008) A neural model of voluntary and automatic emotion regulation: Implications for understanding the pathophysiology and neurodevelopment of bipolar disorder. Molecular Psychiatry 13: 829, 833–857. Potkin SG, Turner JA, Brown GG, et al. (2009) Working memory and DLPFC inefficiency in schizophrenia: The FBIRN study. Schizophrenia Bulletin 35: 19–31. Radua J, Mataix-Cols D, Phillips ML, et al. (2012) A new meta-analytic method for neuroimaging studies that combines reported peak coordinates and statistical parametric maps. European Psychiatry 27: 605–611. Radua J, Via E, Catani M, et al. (2011) Voxel-based meta-analysis of regional white-matter volume differences in autism spectrum disorder versus healthy controls. Psychological Medicine 41: 1539–1550. Rottschy C, Langner R, Dogan I, et al. (2012) Modelling neural correlates of working memory: A coordinate-based meta-analysis. NeuroImage 60: 830–846. Sambataro F, Mattay VS, Thurin K, et al. (2013) Altered cerebral response during cognitive control: A potential indicator of genetic liability for schizophrenia. Neuropsychopharmacology 38: 846–853. Sepede G, Ferretti A, Perrucci MG, et al. (2010) Altered brain response without behavioral attention deficits in healthy siblings of schizophrenic patients: An event-related fMRI study. NeuroImage 49: 1080–1090. Sitskoorn MM, Aleman A, Ebisch SJ, et al. (2004) Cognitive deficits in relatives of patients with schizophrenia: A meta-analysis. Schizophrenia Research 71: 285–295. Smieskova R, Fusar-Poli P, Riecher-Rossler A, et al. (2012) Neuroimaging and resilience factors–Staging of the at-risk mental state? Current Pharmaceutical Design 18: 416–421.
Sommer I, Ramsey N, Kahn R, et al. (2001) Handedness, language lateralisation and anatomical asymmetry in schizophrenia: Meta-analysis. British Journal of Psychiatry 178: 344–351. Thermenos HW, Goldstein JM, Milanovic SM, et al. (2010) An fMRI study of working memory in persons with bipolar disorder or at genetic risk for bipolar disorder. American Journal of Medical Genetics. Part B, Neuropsychiatric Genetics 153B: 120–131. Thermenos HW, Whitfield-Gabrieli S, Seidman LJ, et al. (2013) Altered language network activity in young people at familial high-risk for schizophrenia. Schizophrenia Research 151: 229–237. Van Buuren M, Vink M, Rapcencu AE, et al. (2011) Exaggerated brain activation during emotion processing in unaffected siblings of patients with schizophrenia. Biological Psychiatry 70: 81–87. Venkatasubramanian G, Puthumana DT, Jayakumar PN, et al. (2010) A functional magnetic resonance imaging study of neurohemodynamic abnormalities during emotion processing in subjects at high risk for schizophrenia. Indian Journal of Psychiatry 52: 308–315. Vita A, De Peri L, Deste G, et al. (2013) Progressive loss of cortical gray matter in schizophrenia: A meta-analysis and meta-regression of longitudinal MRI studies. Translational Psychiatry 3: e275. Walton E, Turner J, Gollub RL, et al. (2013) Cumulative genetic risk and prefrontal activity in patients with schizophrenia. Schizophrenia Bulletin 39: 703–711. Whalley HC, Simonotto E, Flett S, et al. (2004) fMRI correlates of state and trait effects in subjects at genetically enhanced risk of schizophrenia. Brain 127: 478–490. Woodward ND, Tibbo P and Purdon SE (2007) An fMRI investigation of procedural learning in unaffected siblings of individuals with schizophrenia. Schizophrenia Research 94: 306–316. Xiao Y, Zhang W, Lui S, et al. (2013) Similar and different gray matter deficits in schizophrenia patients and their unaffected biological relatives. Frontiers in Psychiatry 4: 150.
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 10
24/06/2014 12:47:06 PM
A meta-analysis of fMRI studies in healthy relatives of patients with schizophrenia Claire Scognamiglio and Josselin Houenou Aust N Z J Psychiatry published online 27 June 2014 DOI: 10.1177/0004867414540753 The online version of this article can be found at: http://anp.sagepub.com/content/early/2014/06/26/0004867414540753
Published by: http://www.sagepublications.com
On behalf of:
The Royal Australian and New Zealand College of Psychiatrists
Additional services and information for Australian and New Zealand Journal of Psychiatry can be found at: Email Alerts: http://anp.sagepub.com/cgi/alerts Subscriptions: http://anp.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav
>> OnlineFirst Version of Record - Jun 27, 2014 What is This?
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
540753 research-article2014
ANP0010.1177/0004867414540753Australian & New Zealand Journal of PsychiatryScognamiglio and Houenou
Research Australian & New Zealand Journal of Psychiatry 1–10 DOI: 10.1177/0004867414540753
A meta-analysis of fMRI studies in healthy relatives of patients with schizophrenia
© The Royal Australian and New Zealand College of Psychiatrists 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav anp.sagepub.com
Claire Scognamiglio1 and Josselin Houenou2,3,4,5
Abstract Objective: Genetically at-risk yet healthy relatives of patients with schizophrenia, sharing an important part of the genetic susceptibility to the disease, allow the study of neuroimaging endophenotypes. The aim of our study was to perform a meta-analysis of whole-brain functional magnetic resonance imaging (fMRI) studies that compared adult healthy relatives of patients with schizophrenia and controls. Methods: Twenty-one whole-brain fMRI studies were included (17 using cognitive tasks and four using emotional tasks), published between 2003 and 2013. These studies included 467 healthy relatives of patients with schizophrenia and 768 controls. To conduct the statistical analysis, we used the effect-size signed differential mapping software, a voxel-based meta-analytic approach. Results: In healthy relatives of patients with schizophrenia, we observed a general pattern of overactivation across the 21 fMRI studies in right-sided frontal, parietal and temporal regions compared to controls. This pattern was accompanied by an underactivation in the cingulate gyrus. Our analyses showed a very similar pattern during purely cognitive tasks; during emotional tasks, healthy relatives additionally overactivated the left parahippocampal gyrus. Conclusions: This fMRI pattern of prefrontal overactivation and hypoactivation of the cingulate gyrus may represent a candidate endophenotype for schizophrenia. Keywords Endophenotype, fMRI, healthy relatives, meta-analysis, schizophrenia
Introduction Schizophrenia is a chronic and multifactorial disorder, stemming from interplay of genetic vulnerability and environmental contributors. The lifetime risk for schizophrenia is approximately 0.5–2% in the general population, 6% in parents and 9% in siblings (Lichtenstein et al., 2009). The concordance rate for monozygotic twins is about 45% and 12% for dizygotic twins, suggesting a heritability around 64% (Lichtenstein et al., 2009). Molecular genetic studies reported several susceptibility genes possibly implicated in schizophrenia such as ZNF804 and DISC1 (Duff et al., 2013). Neuroimaging studies provide a useful insight on the etiopathophysiology of schizophrenia. The amount of neuroimaging research publications on schizophrenia has grown exponentially in the last years. Studies in patients with schizophrenia have consistently demonstrated anatomical abnormalities in cortical grey matter volume, involving frontal, temporal and parietal lobes, and left
Heschl gyrus (Vita et al., 2013). Concerning functional magnetic resonance imaging (fMRI), studies in patients have mainly focused on executive processing and facial affect processing. In a meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia, Minzenberg et al. (2009) reported a reduced activation of the prefrontal cortex (middle frontal, inferior frontal, 1Paris
Ile de France Ouest Medical School, Université Versailles SaintQuentin en Yvelines, Versailles, France 2UNIACT, NeuroSpin, I2BM, CEA Saclay, Gif-Sur-Yvette, France 3INSERM U955, Equipe 15 ‘Psychiatrie Génétique’, Créteil, France 4Fondation Fondamental, Créteil, France 5AP-HP, Hôpitaux Universitaires Mondor, Pôle de Psychiatrie, Créteil, France Corresponding author: Josselin Houenou, INSERM U955, Equipe 15 ‘Psychiatrie Génétique’, 40 rue de Mesly, Créteil 94000, France. Email: [email protected]
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 1
24/06/2014 12:47:04 PM
2
ANZJP Articles
superior frontal), anterior cingulate, basal ganglia and insula, and increased activation in several midline cortical areas in patients versus healthy controls (Minzenberg et al., 2009). With regard to emotional processing, a recent metaanalysis (Delvecchio et al., 2013) reported an underactivation in patients compared to controls throughout the entire facial affect processing network and an increased activation in visual processing regions (cuneus). Genetically at-risk yet healthy relatives of patients with schizophrenia are an excellent population in which to study neuroimaging endophenotypes. Endophenotypes were originally described as internal phenotypes that mediate the path between genes and the disease (Gottesman and Gould, 2003). Gottesman defined the endophenotypes as measurable components unseen by the unaided eye along the pathway between disease and distal genotype. They represent simpler clues to genetic underpinnings than the disease syndrome itself. They may be of neurophysiological, biochemical, endocrinological, neuroanatomical, cognitive, or neuropsychological nature. The most commonly accepted criteria for a genetically influenced endophenotype include trait identification in an objective and quantitative manner in patients before onset of the disorder and/or periods of remission; they should also run in families and be associated with an increased risk of clinical illness (Courtet et al., 2011). Endophenotypes thus provide critical evidence on the mechanistic pathways associated with genetic susceptibility. In addition, some authors have described resilience (Smieskova et al., 2012) as an ability to recover from change, and resilience factors as a protector for subjects with an at-risk mental state to develop psychosis. Unravelling such factors may prove fruitful in developing prevention strategies. Healthy relatives share 50% of the genetic susceptibility with patients. Studies in relatives are not biased by medication effects, number of episodes, and potential neurodegenerative illness progression – as in patients. In 2009, a qualitative review of 20 functional neuroimaging studies in healthy relatives, of executive functioning, language and memory paradigms, found that the right dorsal prefrontal region (Brodmann area (BA) 8/9/10/46) showed hyperactivity in 50% of contrasts and hypoactivity in 31% of contrasts (MacDonald et al., 2009). In 2010, a meta-analysis restricted to fMRI executive tasks reported that healthy relatives of patients had hypo- and hyperactivity in statistically overlapping right middle frontal regions (BA 9/10) (Goghari, 2011). This meta-analysis used anatomic likelihood estimation (ALE), a technique that does not take into account in the same analysis hyper- and hypoactivations, and also does not use the effect sizes reported. A recently developed method, effect-size signed differential mapping (ES-SDM) (Radua et al., 2012), allows these limitations to be overcome. The aim of this paper was thus to perform a meta-analysis of whole-brain fMRI studies of cognitive and emotional tasks in adult healthy relatives of patients with schizophrenia using ES-SDM in order to identify consistent functional
neuroimaging endophenotypes and resiliency markers of schizophrenia. We hypothesized that healthy relatives of patients with schizophrenia would exhibit abnormal activations in fronto-temporal regions, mainly during cognitive tasks, and in limbic regions during emotional tasks.
Materials and methods Studies selection A comprehensive literature search of whole-brain fMRI studies conducting comparisons between adult healthy relatives of schizophrenic patients (according to standardized diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV)) and healthy controls was conducted using two databases (PubMed, Google Scholar), using combinations of the following terms: relatives, twins, offspring, children, family studies, genetic risk, high risk, genetic liability, fMRI, schizophrenia, whole-brain. Articles found to be eligible on the basis of their title and, subsequently, abstracts were then selected to further determine suitability for inclusion in this study. Comprehensive reviews and reference lists were also checked to provide further studies. In case details such as coordinates were not included in the original publications, we contacted the corresponding authors for these details. We excluded studies that: (1) used exclusively a regionof-interest (ROI) approach; (2) did not report coordinates for the relevant contrasts and did/could not supply these when contacted; (3) were not reported in English; (4) were review articles, comments, and case reports; (5) used populations already included in another study; and (6) included relatives with significant symptoms. To perform the analyses, we extracted data on the characteristics of the studies and patients, measurements performed and results. For each report, we extracted the following items: author; year; sample size; description of study population (mean age, sex, family link between relatives and patients); fMRI characteristics (field strength in Tesla, Montreal Neurological Institute (MNI) or Talairach coordinates, the full width at half maximum (FWHM), software used, brain function investigated); and behavioural performance. In case several contrasts or tasks were present in the same study (e.g. different attentional tasks: sustained, selective and then dual attention task), we decided to choose only one task in order not to include the population twice. We opted to select, in an arbitrary way, the main result as described by the authors in their article (the one they were mainly discussing), or the task in the highest number of significant peak coordinates was found.
Statistical analysis Regional differences in activation during tasks between healthy relatives and controls were analyzed using ES-SDM
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 2
24/06/2014 12:47:05 PM
3
Scognamiglio and Houenou software (www.sdmproject.com), a voxel-based meta-analytic approach. The SDM methods have been described in detail elsewhere (Radua et al., 2011, 2012). In summary, SDM is a voxel-based meta-analytic approach that uses the reported peak coordinates in the different whole-brain studies. Once the coordinates are selected and converted into a common space, a map of the neurofunctional differences is separately recreated for each study using the peak t-values. This allows maps to be recreated of the effect size of the differences in blood oxygenation level dependent (BOLD) responses between healthy relatives and controls. ES-SDM uses hyperactivation and underactivation peaks on the same map. Studies are then combined with a random-effects model that allows sample size, intra-study variability and between-study heterogeneity to be taken into account (Radua et al., 2012). We conducted three analyses: the first one included all the different tasks that were used in the studies, regardless of their category (emotional or cognitive); the second one included only cognitive tasks; and the last one included emotional tasks. To test the robustness of the findings, these analyses were complemented with a jackknife sensitivity analysis. This analysis consisted of repeating the analysis and discarding (only) one study each time, and is used to assess the replicability of the results. We also looked for heterogeneity between studies by visually inspecting the Q statistics and funnel plots. Statistical significance was achieved with 50 permutations for all the analyses (standard permutation tests), as it was shown to have a high statistical stability. Results were then thresholded at p1 (peak height), and cluster extent greater or equal to 10 voxels; these thresholds have proven to yield consistent, valid and replicable results (Radua et al., 2012). The p threshold at 0.005 is uncorrected for multiple comparisons, based on the empirical validation detailed in this paper (Radua et al., 2012). Indeed, they proposed to use an uncorrected p=0.005 as the main threshold, as this was found to optimally balance sensitivity and specificity and to be an approximate equivalent to corrected p-value = 0.05 (indeed 0.025) in ES-SDM. Of course, this equivalence is only approximate, so to reduce the possibility of false positive results we suggest an additional z-based threshold (e.g. z>1). It would be associated to a clearly non-significant p-value under the standard normal distribution.
Results Studies selected Our literature search yielded 21 studies that included adult subjects, published between 2003 and September 2013 (Callicott et al., 2003; Choi et al., 2012; de Achaval et al., 2012; Delawalla et al., 2008; Filbey et al., 2008; Jamadar et al., 2013; Karch et al., 2009; Lee et al., 2010;
Li et al., 2007, 2012; Liddle et al., 2013; McAllindon et al., 2010; MacDonald et al., 2006; Sambataro et al., 2013; Sepede et al., 2010; Thermenos et al., 2013; van Buuren et al., 2011; Venkatasubramanian et al., 2010; Whalley et al., 2004; Woodward et al., 2007). We then categorized them into cognitive (n=17) and emotional (n=4) tasks. The characteristics of the studies included (author, year, population, field strength, etc.) are summarized in Table 1. Therefore, our meta-analysis includes the data of 467 relatives and 768 controls. Concerning the behavioural performances during the task, the authors noted no significant difference between relatives and controls in 10 studies. Nothing was specified in seven studies and there was a significant difference in four studies. The differences were that relatives executed tasks slower than controls (n=1), had a longer reaction time (n=1) or had more errors (n=2).
Meta-analysis including all 21 studies (Table 2) Our first analysis included all the 21 adult studies. We found in relatives of schizophrenic patients versus controls a significant hyperactivation in right-sided regions: middle temporal gyrus (BA37), inferior frontal gyrus (BA44), and superior parietal lobule (BA7). Hypoactivation clusters were restricted to the left cingulate gyrus (BA24) (Table 3). These results were highly replicable in the jackknife analysis. Results are displayed in Figures 1 (hyperactivation) and 2 (hypoactivation).
Cognitive processing task-related activations (Table 3) Seventeen whole-brain fMRI studies reported cognitive functioning task-related activity comparing healthy relatives of patients with schizophrenia and control groups. Those tasks included working memory tasks (n=4), go/ no-go (n=1), stroop (n=1), attentional processes (n=3), covert verbal initiation task (n=1), reaction time task (n=1), visual task (n=2), lexical decision task (n=2), executive task (n=1), and auditory detection task (n=1). Significant hyperactivations in relatives were found in seven clusters, encompassing right-sided clusters: inferior frontal gyrus (BA45), precuneus (BA7), middle and superior temporal gyrus (BA37 and 39), caudate, inferior parietal lobule (BA40), but also left precentral gyrus (BA6). By contrast, hypoactivations were only significant in the right cingulate gyrus (BA31) (Table 3).
Emotional processing task-related activations (Table 4) Four studies used tasks categorized as measuring emotional processing. The tasks were: eyes emotional theory of mind (n=1) and facial emotional valence discrimination (n=3). Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 3
24/06/2014 12:47:05 PM
ANP540753.indd 4
MacDonald et al., 2006 Callicott et al., 2003-1 Callicott et al. 2003-2 Sambataro et al., 2013 Thermenos et al., 2013 Jamadar et al., 2013
Filbey et al., 2008 Sepede et al., 2010
Delawalla et al., 2008 Woodward et al., 2007 Karch et al., 2009 McAllindon et al., 2010 Lee et al., 2010
Li X et al., 2007
Whalley et al., 2004
Cognitive tasks Choi et al., 2012 Liddle et al., 2013
Study name
23 25 65 43 27
Working memory
Working memory
Go/no-go task
Lexical decision task
Lexical decision task
6 11 21
33.6 (8.79) 34.7
11 11
40.78 (2.5)
25.2 (3.1)
36.6 (10.4)
36.6 (9)
34.4 (9)
33.2 (10.9)
53 34.4 (8.8)
36.0 (10.3)
36.9 (13.3)
12
21
21.3 (3.5)
21.7 (6.1)
26.8 (3.4)
20.71 (5.5) 20.4 (3.5)
Mean age, years (SD)
30
15
42
17 18
n
Visual backward masking performance Attention process Continuous performance task Cognitive task
Working memory Auditory target detection task Covert verbal initiation task Visual word decision task Continuous performance task Serial reaction time task Working memory Reaction time
Task
Relatives
Population characteristics
Table 1. Characteristics of the studies included in the meta-analysis.
4/23
13/30
24/41
11/14
6/17
7/14
2/4 5/6
11/10
4/7 11/0
5/7
14/16
7/8
17/25
9/8 7/11
Sex (M/F)
First-degree
Unspecified
–
Siblings
Siblings
First-degree
Unspecified Siblings
Siblings
Unspecified First-degree
Siblings
Siblings
Unspecified
Unspecified
Unspecified Siblings
Relationship
133
32
235
15
18
20
8 11
19
11 14
15
92
15
21
16 26
n
Controls
32.48 (1.05)
24.6 (2.8)
31.8 (9.6)
27.9 (8)
29.6 (7)
33.4 (8.4)
41 32 (5.1)
42.7 (9.0)
33.8 (9.20) 36.4
31.3 (11.2)
20.2 (3.4)
23.9 (5.2)
26.8 (2.7)
21.37 (2.8) 20.3 (4.9)
Mean age, years (SD)
5/3 5/6
14/5
4/7
68/65
13/19
113/122
6/9
11/7
10/10
–
10/5
39/53
7/8
13/8
9/7 15/11
Sex (M/F)
3
3
3
1.5
1.5
–
1.5 1.5
3
1.5 1.5
1.5
1.5
1.5
1.5
1.5 3
Field strength (Tesla)
(Continued)
MNI
MNI
MNI
Talairach
Talairach
Talairach
Talairach Talairach
MNI
Talairach Talairach
Talairach
Talairach
Talairach
Talairach
NA MNI
Referential
MRI characteristics
4 ANZJP Articles
Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
24/06/2014 12:47:05 PM
5
Talairach 14/2 24.4
1.5
MNI 7/ 18 27.5
3
MNI 6/6 29.25
3
Talairach – 8/6 28.4
Referential Field strength (Tesla)
16
We performed a SDM meta-analysis to identify convergence across standardized coordinates of 21 fMRI studies comparing healthy relatives of patients with schizophrenia and controls. The relatives demonstrated a pattern of overactivations in a right-sided fronto-parieto-temporal network. During cognitive tasks, these overactivations were mostly present in fronto-temporal regions, whereas in emotional tasks the left parahippocampal was overactivated and a wider range of bilateral regions was underactivated.
SD, standard deviation; M, male; F, female; MNI, Montreal Neurological Institute.
Siblings Venkatasubramanian et al., 2010
17
25.2
14/3
25 Siblings Van Buuren et al., 2011
24
29.4
8/16
12 Siblings 12
Facial emotional valence discrimination Facial emotional valence discrimination Facial emotional valence discrimination
31.25
4/8
14 Siblings 14
Emotional tasks de Achaval et al., 2012 Li et al., 2012
Eyes theory of mind
30.4
8/6
n Relationship Sex (M/F) n Study name
Task
Relatives
Population characteristics Table 1. (Continued)
Six significant clusters of overactivation were identified in the relatives compared to controls: the left sub-gyral (parietal) (BA40), right superior frontal gyrus (BA9), left lentiform nucleus (lateral globus pallidus), left parahippocampal gyrus (BA28), left precuneus (BA7) and the right middle temporal gyrus (BA39). We found hypoactivations in the right precentral gyrus (BA6), right inferior parietal lobule (BA40), left medial frontal gyrus (BA6), and the right inferior frontal gyrus (BA47) in the relatives versus controls (Table 4). The Q heterogeneity statistics of the different metaanalyses results were tested. This shows which brain regions present significant between-study heterogeneity as compared to the global set of voxels, even if they are only marginally heterogeneous. Our results were globally homogeneous. The jackknife analysis was performed to test only the coordinates which were significant in the primary meta-analysis, and all the regions replicate.
Discussion
Mean age, years (SD)
Controls
Mean age, years (SD)
Sex (M/F)
MRI characteristics
Scognamiglio and Houenou
Over- and underactivation of brain areas in healthy relatives of patients The general pattern of our results was that relatives of patients with schizophrenia had to recruit more neural resources during tasks (hyperactivation), especially in prefrontal areas. Taking into consideration the largely preserved behavioural performance of relatives of patients in the various tasks, our results suggest that the healthy relatives of patients may use some compensatory strategies. In other words, relatives would have to engage more neural effort in order to complete tasks by increasing the activation of brain areas engaged in a task or recruiting additional areas (Habel et al., 2010). Relatives may increase engagement of other processes to maintain task performance, such as attentional, mnemonic, and performance monitoring functions. This large pattern of overactivations is not present in a recent meta-analysis of fMRI studies exploring executive domains in patients with schizophrenia (Minzenberg et al., 2009). This study used ALE, therefore leading to reports of overactivations and hypoactivations in the same areas, such as the right medial frontal gyrus. This prevents us from directly comparing Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 5
24/06/2014 12:47:05 PM
6
ANZJP Articles
Table 2. Results of the SDM analysis for the 21 studies. Brain regions
z
p
Voxels
Brodmann area
Talairach coordinates x, y, z
Relatives > controls Right middle temporal gyrus 1.832 0.000069878 296 37 1.608 0.000304945 67 44 Right inferior frontal gyrus 1.360 0.001418369 34 7 Right superior parietal lobule Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≥ 1.138 and peak voxel z ≥ 1.000
46, –60, 2 52, 10, 20 18, –68, 56
Controls > relatives Left cingulate gyrus –2.019 0.000932306 146 24 Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≤ –1.411 and peak voxel z ≤ –1.000
–2, –2, 38
Table 3. Brain regions exhibiting significant activity across the full set of cognitive tasks. Brain regions
z
p
Voxels
Brodmann area
Relatives > controls Right inferior frontal gyrus 1.626 0.000162107 154 45 1.469 0.000525626 60 7 Right precuneus (parietal) 1.424 0.000714451 20 37 Right middle temporal gyrus 1.324 0.001358767 11 41 Right caudate (right transverse temporal gyrus) Right superior temporal 1.301 0.001560694 46 39 gyrus Left precentral gyrus 1.261 0.001966346 14 6 1.152 0.003710212 13 40 Right inferior parietal lobule Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≥ 1.097 and peak voxel z ≥ 1.000 Controls > relatives Right cingulate gyrus –2.319 0.000220681 216 31 Threshold parameters: Clusters of ≥ 10 voxels, with all voxels z ≤ –1.443 and peak voxel z ≤ –1.000
our results. Nevertheless, several reports have shown that patients with schizophrenia may present a prefrontal inefficiency during cognitive tasks, leading to compensatory increased recruitment during tasks with high cognitive demands (Ettinger et al., 2011; Potkin et al., 2009). An argument that supports this hypothesis is that meta-analyses on studies of cognitive behavioural deficits in relatives show cognitive deficits in this population (Sitskoorn et al., 2004). Those results were observed while the subjects performed more difficult cognitive tasks, which can show that in those cases, compensatory mechanisms are no longer sufficient. In contrast, we found the cingulate gyrus to be consistently hypoactive across tasks in relatives of patients. These underactivations are also frequently reported in patients with schizophrenia (Adams and David, 2007). This pattern of overactivation of right prefrontal areas and underactivation of cingulate gyrus may thus represent an endophenotype of the disease, linked with the genetic susceptibility to schizophrenia. This interpretation is further supported by reports that prefrontal efficiency during a
Talairach coordinates x, y, z 54, 12, 20 14, –66, 52 46, –60, 2 32, –36, 4 56, –58, 18 –32, –18, 64 54, –32, 34 8, –8, 44
working memory task is correlated with a cumulative genetic risk score of schizophrenia (Walton et al., 2013) based on at-risk single nucleotide polymorphisms (SNPs) for the disease.
Neural circuits As we assumed, relatives showed abnormal activations in prefronto-temporal networks during cognitive tasks but also in limbic regions during emotional paradigms. Fronto-temporal brain networks implicated in high-level cognitive processes are altered in these subjects. Indeed, during cognitive tasks, relatives overactivated regions such as the right inferior frontal gyrus (BA45), middle and superior temporal gyrus (BA37), and left precentral gyrus (BA6) but also parietal regions (precuneus and inferior parietal lobule). All these regions are implicated in cognitive processing and working memory. They belong to the ‘taskpositive network’. This network is activated during attention-demanding, cognitive tasks (Fox et al., 2005). A recent meta-analysis confirmed that working memory
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 6
24/06/2014 12:47:05 PM
7
Scognamiglio and Houenou
Figure 1. Result of the meta-analysis including the 21 studies: hyperactivations in relatives versus controls. (a) Axial view (sections z = 3, 13, 16, 19, 21, and 26 in MNI space); (b) lateral 3D right view.
Figure 2. Result of the meta-analysis including the 21 studies: hypoactivations in relatives versus controls. Sagittal view (sections x = −11, –3, 4, 7, 8, and 11 in MNI space).
relied largely on a bilateral fronto-parietal network including the inferior frontal gyrus (Rottschy et al., 2012). Our results suggest that during cognitive tasks, relatives of patients overactivated regions that are normally used during these tasks. No obvious additional region is recruited. During emotional tasks, the left parahippocampal gyrus was additionally activated. This region is implicated in the automatic emotional regulation (Phillips et al., 2008). Our finding of hyperactivity in the neural circuitry of emotion processing agrees with studies showing exaggerated reactivity to emotional stimuli in unaffected first-degree relatives of patients with schizophrenia (Myin-Germeys and van Os, 2007). For example, unaffected first-degree relatives, including siblings, show enhanced reactivity to daily life stressors as indicated by a more profound decrease in positive mood and increase in negative mood compared with control subjects in response to stressors (MyinGermeys et al., 2001), as well as an increased physiologic stress response to a metabolic stressor (infusion of the glucose analogue 2-deoxy-d-glucose) (Brunelin et al., 2008). Our result is in contrast with the findings in patients with schizophrenia who exhibit a decreased activation in the bilateral parahippocampal cortices during facial affect processing (Li et al., 2010). This suggests that the overactivation observed in healthy relatives may either be of compensatory nature or a resiliency marker. A striking result in our meta-analysis is that a vast majority of the overactivations observed in the healthy relatives of patients are right-sided for cognitive tasks. Though further exploration of this topic is needed, this suggests an altered balance between the hemispheres function. This
may be paralleled with the findings of a decreased cerebral lateralization and asymmetry in schizophrenia along with a higher prevalence of mixed- and left-handedness in schizophrenia (Sommer et al., 2001). This decreased lateralization is present for language but also motor functions (Altamura et al., 2012). A very recent meta-analysis of voxel base morphometry (VBM) studies in healthy relatives showed that increased grey matter volume was exclusively right-sided and observed in a fronto-parieto-temporal network including the precuneus, found to be overactive in our study (Xiao et al., 2013). This conjunction is in favour of a compensatory mechanism. Goghari et al. performed in 2011 a meta-analysis of fMRI studies in healthy relatives of patients with schizophrenia in comparison to controls, but restricted to executive tasks, using ALE, and including both whole-brain and ROI studies (Goghari, 2011). They did two analyses – one including all the studies and the second only the wholebrain analyses – and they compared both results. Most of the studies with ROI approaches focused on prefrontal regions and displayed reduced functional activations. This was in contrast with whole-brain studies in which both hyper- and hypoactivations were detected. The cause of this discrepancy is yet unclear. Thus, we chose not to include ROI approaches, to limit this potential bias. This study found hypo- and hyperactivity in the right middle frontal regions (BA 9/10) in relatives, and controls activated to a greater degree than relatives the right inferior frontal, right precentral, left posterior cingulate, left thalamus and right lingual regions compared to relatives. Relatives activated to a greater degree the right superior Australian & New Zealand Journal of Psychiatry
Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 7
24/06/2014 12:47:06 PM
8
ANZJP Articles
Table 4. Brain regions exhibiting significant activity across the full set of emotional tasks. Brain regions Relatives > controls Left sub-gyral (parietal) Right superior frontal gyrus Left lentiform nucleus
z 1.281 1.281 1.280
p
Voxels
0.001004239 0.001004239 0.001004239
26 106 12
Brodmann area
40 9 lateral globus pallidus Left parahippocampal gyrus 1.280 0.001093899 18 28 15 7 Left precuneus 1.274 0.001667052 29 39 Right middle temporal gyrus 1.270 0.002116378 Threshold parameters : Clusters of ≥ 10 voxels, with all voxels z ≥ 1.238 and peak voxel z ≥ 1.000 Controls > relatives Right precentral gyrus –1.041 0.000915607 102 6 66 40 Right inferior parietal lobule –1.023 0.001093642 23 6 Left medial frontal gyrus –1.011 0.001617213 20 47 Right inferior frontal gyrus –1.007 0.001801927 Threshold parameters : Clusters of ≥ 10 voxels, with all voxels z ≤ –0.988 and peak voxel z ≤ –1.000
frontal, right thalamus, right precentral, left inferior parietal and left precuneus regions compared to controls. Since the publication of this meta-analysis, 11 additional studies were published that we included in our study. In addition, because of the ALE algorithm, the authors had to conduct two separate meta-analyses for hyper- and hypoactivations. This leads to the contradictory finding of a hyperand hypoactivity in the right middle frontal regions. Furthermore, ALE does not take into account effect sizes; this is an important point, and it is supported by the SDM software in our analysis. A recent meta-analysis (Cooper et al., 2014) studied the fMRI studies in those at elevated genetic risk of developing schizophrenia and included only 13 studies. We included 21 studies; differences between these two meta-analyses are: first, the inclusion period was longer in our analysis (2003 to January 2012 versus 2003 to September 2013); and second, we contacted several authors (Choi et al., 2012; de Achaval et al., 2012; Sambataro et al., 2013) to get coordinates when they were not included in the publication. These differences may explain why we get very different results. Another difference is that we distinguished emotional and cognitive tasks in our analysis.
Limitations First, our meta-analysis does not include unpublished negative studies because of bias publication. Second, we recognize that there are a limited number of fMRI studies in relatives and especially only four fMRI studies on emotional tasks. We therefore have to interpret with care those results, and especially the generalizability of findings. Our global meta-analysis that included the 21 studies may have been driven by the cognitive tasks (n=17 studies). Third,
Talairach coordinates x, y, z –22, –48, 56 12, 46, 26 –24, –12, –6 –20, –14, –20 –6, –46, 48 50, –66, 10
54, –6, 32 40, –50, 56 –2, –20, 62 52, 28, –12
the acquisition protocols and MRI scanners of the included studies are heterogeneous. The meta-analysis is therefore complicated by the potentially confounding effects of differences between sites. Fourth, we did not include some studies because no coordinates were reported and the authors of these studies did not provide them when we contacted them. Fifth, it may be arbitrary to specify to which category a task belongs (emotional, executive, etc.). We therefore found it more relevant to include all tasks in the main analysis of this meta-analysis. In studies that used two tasks (different cognitive tasks, for example, with the same sample of population), we had to choose one task only, in order not to include a population twice in our analysis. We opted to take the main result or the results with more coordinates. This is questionable, but it would be worse to include a population twice. Sixth, the studies included may not represent the majority of the available literature on fMRI in healthy relatives of patients with schizophrenia as we did not include ROI studies. Seventh, it is important to discuss the fact that some of our results are also found in other disorders, such as prefrontal hyperactivation in bipolar disorder. Indeed, in a systematic review concerning studies about whole-brain fMRI of relatives of patients with bipolar disorder, Fusar Poli et al. (2012) observed that, independently of task, high-risk individuals showed increased neural response in the left superior frontal gyrus and in the medial frontal gyrus, among others. It is important to note that our results may not be specific to schizophrenia. Furthermore, common susceptibility factors may exist between bipolar disorder and schizophrenia (Fusar-Poli et al., 2012). Finally, there were not enough studies specifying exactly the relationship between relatives and patients with schizophrenia (i.e. siblings or children).
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 8
24/06/2014 12:47:06 PM
9
Scognamiglio and Houenou
Conclusion and perspectives In conclusion, our meta-analysis of fMRI studies comparing brain activations between healthy relatives of schizophrenic patients and controls identified abnormalities in brain networks implicated in cognition and emotion processing. We revealed a general pattern of overactivation in prefrontal areas and underactivation of the cingulate gyrus. This pattern may thus be a potential endophenotype of schizophrenia. To get further insight into this question, we suggest that future fMRI in healthy relatives explores emotional and social domains. These endophenotypes may prove useful as intermediate phenotypes to discover the genes associated with schizophrenia. Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Declaration of interests The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
References Adams R and David AS (2007) Patterns of anterior cingulate activation in schizophrenia: A selective review. Neuropsychiatric Disease and Treatment 3: 87–101. Altamura M, Fazio L, De Salvia M, et al. (2012) Abnormal functional motor lateralization in healthy siblings of patients with schizophrenia. Psychiatry Research 203: 54–60. Brunelin J, d’Amato T, van Os J, et al. (2008) Effects of acute metabolic stress on the dopaminergic and pituitary-adrenal axis activity in patients with schizophrenia, their unaffected siblings and controls. Schizophrenia Research 100: 206–211. Callicott JH, Egan MF, Mattay VS, et al. (2003) Abnormal fMRI response of the dorsolateral prefrontal cortex in cognitively intact siblings of patients with schizophrenia. American Journal of Psychiatry 160: 709–719. Choi JS, Park JY, Jung MH, et al. (2012) Phase-specific brain change of spatial working memory processing in genetic and ultra-high risk groups of schizophrenia. Schizophrenia Bulletin 38: 1189–1199. Cooper D, Barker V, Radua J, et al. (2014) Multimodal voxel-based metaanalysis of structural and functional magnetic resonance imaging studies in those at elevated genetic risk of developing schizophrenia. Psychiatry Research 221: 69–77. Courtet P, Gottesman II, Jollant F, et al. (2011) The neuroscience of suicidal behaviors: what can we expect from endophenotype strategies? Translational Psychiatry 1: e7. De Achaval D, Villarreal MF, Costanzo EY, et al. (2012) Decreased activity in right-hemisphere structures involved in social cognition in siblings discordant for schizophrenia. Schizophrenia Research 134: 171–179. Delawalla Z, Csernansky JG and Barch DM (2008) Prefrontal cortex function in nonpsychotic siblings of individuals with schizophrenia. Biological Psychiatry 63: 490–497. Delvecchio G, Sugranyes G and Frangou S (2013) Evidence of diagnostic specificity in the neural correlates of facial affect processing in bipolar disorder and schizophrenia: A meta-analysis of functional imaging studies. Psychological Medicine 43: 553–569. Duff BJ, Macritchie KA, Moorhead TW, et al. (2013) Human brain imaging studies of DISC1 in schizophrenia, bipolar disorder and depression: A systematic review. Schizophrenia Research 147: 1–13.
Ettinger U, Williams SC, Fannon D, et al. (2011) Functional magnetic resonance imaging of a parametric working memory task in schizophrenia: Relationship with performance and effects of antipsychotic treatment. Psychopharmacology (Berl) 216: 17–27. Filbey FM, Russell T, Morris RG, et al. (2008) Functional magnetic resonance imaging (fMRI) of attention processes in presumed obligate carriers of schizophrenia: Preliminary findings. Annals of General Psychiatry 7: 18. Fox MD, Snyder AZ, Vincent JL, et al. (2005) The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the United States of America 102: 9673–9678. Fusar-Poli P, Howes O, Bechdolf A, et al. (2012) Mapping vulnerability to bipolar disorder: A systematic review and meta-analysis of neuroimaging studies. Journal of Psychiatry & Neuroscience: JPN 37: 170–184. Goghari VM (2011) Executive functioning-related brain abnormalities associated with the genetic liability for schizophrenia: An activation likelihood estimation meta-analysis. Psychological Medicine 41: 1239–1252. Gottesman II and Gould TD (2003) The endophenotype concept in psychiatry: Etymology and strategic intentions. American Journal of Psychiatry 160: 636–645. Habel U, Chechko N, Pauly K, et al. (2010) Neural correlates of emotion recognition in schizophrenia. Schizophrenia Research 122: 113–123. Jamadar SD, Pearlson GD, O’Neil KM, et al. (2013) Semantic association fMRI impairments represent a potential schizophrenia biomarker. Schizophrenia Research 145: 20–26. Karch S, Leicht G, Giegling I, et al. (2009) Inefficient neural activity in patients with schizophrenia and nonpsychotic relatives of schizophrenic patients: Evidence from a working memory task. Journal of Psychiatric Research 43: 1185–1194. Lee J, Cohen MS, Engel SA, et al. (2010) Regional brain activity during early visual perception in unaffected siblings of schizophrenia patients. Biological Psychiatry 68: 78–85. Li H, Chan RC, McAlonan GM, et al. (2010) Facial emotion processing in schizophrenia: A meta-analysis of functional neuroimaging data. Schizophrenia Bulletin 36: 1029–1039. Li HJ, Chan RC, Gong QY, et al. (2012) Facial emotion processing in patients with schizophrenia and their non-psychotic siblings: A functional magnetic resonance imaging study. Schizophrenia Research 134: 143–150. Li X, Branch CA, Bertisch HC, et al. (2007) An fMRI study of language processing in people at high genetic risk for schizophrenia. Schizophrenia Research 91: 62–72. Lichtenstein P, Yip BH, Bjork C, et al. (2009) Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: A population-based study. Lancet 373: 234–239. Liddle EB, Bates AT, Das D, et al. (2013) Inefficient cerebral recruitment as a vulnerability marker for schizophrenia. Psychological Medicine 43: 169–182. McAllindon DP, Wilman AH, Purdon SE, et al. (2010) Functional magnetic resonance imaging of choice reaction time in chronic schizophrenia and first-degree relatives. Schizophrenia Research 120: 232–233. MacDonald AW 3rd, Becker TM and Carter CS (2006) Functional magnetic resonance imaging study of cognitive control in the healthy relatives of schizophrenia patients. Biological Psychiatry 60: 1241–1249. MacDonald AW 3rd, Thermenos HW, Barch DM, et al. (2009) Imaging genetic liability to schizophrenia: Systematic review of FMRI studies of patients’ nonpsychotic relatives. Schizophrenia Bulletin 35: 1142–1162. Minzenberg MJ, Laird AR, Thelen S, et al. (2009) Meta-analysis of 41 functional neuroimaging studies of executive function in schizophrenia. Archives of General Psychiatry 66: 811–822. Myin-Germeys I and van Os J (2007) Stress-reactivity in psychosis: Evidence for an affective pathway to psychosis. Clinical Psychology Review 27: 409–424.
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 9
24/06/2014 12:47:06 PM
10
ANZJP Articles
Myin-Germeys I, van Os J, Schwartz JE, et al. (2001) Emotional reactivity to daily life stress in psychosis. Archives of General Psychiatry 58: 1137–1144. Phillips ML, Ladouceur CD and Drevets WC (2008) A neural model of voluntary and automatic emotion regulation: Implications for understanding the pathophysiology and neurodevelopment of bipolar disorder. Molecular Psychiatry 13: 829, 833–857. Potkin SG, Turner JA, Brown GG, et al. (2009) Working memory and DLPFC inefficiency in schizophrenia: The FBIRN study. Schizophrenia Bulletin 35: 19–31. Radua J, Mataix-Cols D, Phillips ML, et al. (2012) A new meta-analytic method for neuroimaging studies that combines reported peak coordinates and statistical parametric maps. European Psychiatry 27: 605–611. Radua J, Via E, Catani M, et al. (2011) Voxel-based meta-analysis of regional white-matter volume differences in autism spectrum disorder versus healthy controls. Psychological Medicine 41: 1539–1550. Rottschy C, Langner R, Dogan I, et al. (2012) Modelling neural correlates of working memory: A coordinate-based meta-analysis. NeuroImage 60: 830–846. Sambataro F, Mattay VS, Thurin K, et al. (2013) Altered cerebral response during cognitive control: A potential indicator of genetic liability for schizophrenia. Neuropsychopharmacology 38: 846–853. Sepede G, Ferretti A, Perrucci MG, et al. (2010) Altered brain response without behavioral attention deficits in healthy siblings of schizophrenic patients: An event-related fMRI study. NeuroImage 49: 1080–1090. Sitskoorn MM, Aleman A, Ebisch SJ, et al. (2004) Cognitive deficits in relatives of patients with schizophrenia: A meta-analysis. Schizophrenia Research 71: 285–295. Smieskova R, Fusar-Poli P, Riecher-Rossler A, et al. (2012) Neuroimaging and resilience factors–Staging of the at-risk mental state? Current Pharmaceutical Design 18: 416–421.
Sommer I, Ramsey N, Kahn R, et al. (2001) Handedness, language lateralisation and anatomical asymmetry in schizophrenia: Meta-analysis. British Journal of Psychiatry 178: 344–351. Thermenos HW, Goldstein JM, Milanovic SM, et al. (2010) An fMRI study of working memory in persons with bipolar disorder or at genetic risk for bipolar disorder. American Journal of Medical Genetics. Part B, Neuropsychiatric Genetics 153B: 120–131. Thermenos HW, Whitfield-Gabrieli S, Seidman LJ, et al. (2013) Altered language network activity in young people at familial high-risk for schizophrenia. Schizophrenia Research 151: 229–237. Van Buuren M, Vink M, Rapcencu AE, et al. (2011) Exaggerated brain activation during emotion processing in unaffected siblings of patients with schizophrenia. Biological Psychiatry 70: 81–87. Venkatasubramanian G, Puthumana DT, Jayakumar PN, et al. (2010) A functional magnetic resonance imaging study of neurohemodynamic abnormalities during emotion processing in subjects at high risk for schizophrenia. Indian Journal of Psychiatry 52: 308–315. Vita A, De Peri L, Deste G, et al. (2013) Progressive loss of cortical gray matter in schizophrenia: A meta-analysis and meta-regression of longitudinal MRI studies. Translational Psychiatry 3: e275. Walton E, Turner J, Gollub RL, et al. (2013) Cumulative genetic risk and prefrontal activity in patients with schizophrenia. Schizophrenia Bulletin 39: 703–711. Whalley HC, Simonotto E, Flett S, et al. (2004) fMRI correlates of state and trait effects in subjects at genetically enhanced risk of schizophrenia. Brain 127: 478–490. Woodward ND, Tibbo P and Purdon SE (2007) An fMRI investigation of procedural learning in unaffected siblings of individuals with schizophrenia. Schizophrenia Research 94: 306–316. Xiao Y, Zhang W, Lui S, et al. (2013) Similar and different gray matter deficits in schizophrenia patients and their unaffected biological relatives. Frontiers in Psychiatry 4: 150.
Australian & New Zealand Journal of Psychiatry Downloaded from anp.sagepub.com at UNIV TORONTO on July 5, 2014
ANP540753.indd 10
24/06/2014 12:47:06 PM
