This paper describes how to use the emotional oddball task and fMRI to measure brain activation in children and adolescents at familial high risk for schizophrenia (FHR). FMRI was used to measure differences in fronto-striato-limbic regions during an emotional oddball task. Children with FHR exhibited abnormal functional activation during adolescence.
Adolescence is a critical developmental period where the early symptoms of schizophrenia frequently emerge. First-degree relatives of people with schizophrenia who are at familial high risk (FHR) may show similar cognitive and emotional changes. However, the neurological changes underlying the emergence of these symptoms remain unclear. This study sought to identify differences in frontal, striatal, and limbic regions in children and adolescents with FHR using functional magnetic resonance imaging. Groups of 21 children and adolescents at FHR and 21 healthy controls completed an emotional oddball task that relied on selective attention and the suppression of task-irrelevant emotional information. The standard oddball task was modified to include aversive and neutral distractors in order to examine potential group differences in both emotional and executive processing. This task was designed specifically to allow for children and adolescents to complete by keeping the difficulty and emotional image content age-appropriate. Furthermore, we demonstrate a technique for suitable fMRI registration for children and adolescent participants. This paradigm may also be applied in future studies to measure changes in neural activity in other populations with hypothesized developmental changes in executive and emotional processing.
Schizophrenia is a neurodevelopmental disorder with a known genetic component1,2 and with symptoms including deficits in both executive and emotional processing3,4. First-degree relatives are thought to be at an increased risk of developing schizophrenia, and have been shown to share some of these same neurocognitive deficits in both cognitive and social-emotional domains5. We therefore expect that brain activity in regions associated with executive and emotional processing may be altered in at-risk family members preceding the onset of clinical symptoms.
Previous studies have indicated that both adults with schizophrenia and adults at familial high risk show aberrant activity within executive and emotional processing networks; however it remains unclear how these changes come about during development. Demonstrating that these changes occur early in life will be a critical first step in understanding the pathophysiology of the disorder. Therefore, this study utilizes an emotional oddball paradigm during functional MRI (fMRI) scanning in order to measure brain activity during the completion of a task that requires both executive and emotional processing in adolescents who are at risk for developing schizophrenia. Oddball paradigms are frequently used to examine the function of fronto-striate circuitry in schizophrenia6 and in individuals with familial high risk7 by measuring selective attention processes allocated to task-relevant target stimuli. Here, a standard oddball task has been modified to include task-irrelevant aversive and neutral stimuli that have been shown to elicit changes in brain activity in patients with schizophrenia8.
This paper measures functional differences between healthy adolescents and adolescents at high familial risk for schizophrenia using an emotional oddball task. The task design is similar to that used by Fichtenholtz and colleagues9, but the selection of aversive emotional images has been modified to be appropriate for children between the ages of 9-18. The use of this task during functional MRI allowed for the identification of specific brain regions that showed patterns of hyperactivation and hypoactivation in children and adolescents with FHR for schizophrenia, in addition to age-related changes in neural activity during adolescent development.
The research techniques used during this study were approved by the institutional review boards (IRB) of Duke University and the University of North Carolina – Chapel Hill.
1. Imaging Task Design
Figure 1. Schematic of Task Design. This figure has been modified from Hart et al.20, with permission. Please click here to view a larger version of this figure.
2. Participant Setup and Scanning
3. Image Acquisition
4. Analyses
There were no differences between groups based on demographic characteristics20. Behavioral data indicated that the target detection task is at an appropriate level of difficulty for children and adolescents between the ages of 9-18. In the current study, controls correctly identified 82.36% of targets (S.D. = 0.14), and the familial risk group correctly identified 76.8% of targets (S.D. = 0.17). Both groups showed decreased accuracy when identifying emotional pictures compared to neutral pictures (F(1,40) = 5.63, p = 0.03).
The imaging data indicated that the experimental conditions led to significant activation in regions expected to be recruited during executive and emotional processing. Activation was seen in prefrontal, anterior caudate, insular, and posterior parietal areas during target trials and in the right amygdala, bilateral orbitofrontal cortex, fusiform cortex and visual cortical areas during aversive trials in both groups. Table 1 shows areas of significant activation in controls for each condition.
This paradigm also elicited significant differences in activation between controls and individuals with familial high risk for schizophrenia. The familial high risk group showed decreased activation in fronto-striate circuitry in response to target stimuli. Controls, in contrast, showed greater activation in the middle frontal gyrus and insula. Group differences between conditions are shown in Table 2 and Figure 2. The familial high risk group also showed different patterns of age-related activation compared with controls in response to target and aversive stimuli (Figure 3).
Figure 2. Activation Maps of Between-Group Differences. (A) Areas where the familial high risk group (n=21) showed greater activation than controls (n=21) during target processing. CAUD = Caudate; IFG = Inferior frontal gyrus; ITG = Inferior temporal gyrus. (B) Areas where controls showed greater activation than the familial high risk group during target processing. INS = Insula; MFG = Middle frontal gyrus; MTG = Middle temporal gyrus. (C) Areas where the familial high risk group activated more than controls during the Aversive > Neutral contrast. COC = Central opercular cortex. (D) Areas where the controls activated more than the familial high risk group during the Aversive > Neutral contrast. ACC = Anterior cingulate cortex; PC = Precuneus. This figure has been modified from Hart et al.20, with permission. Please click here to view a larger version of this figure.
Figure 3. Activation Maps of Age-Related Group Differences. (A) Areas with a greater positive correlation with age in the familial high risk group than in controls during target processing. ACC = Anterior cingulate cortex; INS = Insula; OFC = Orbitofrontal cortex; TH = Thalamus. (B) Areas with a greater positive correlation with age in controls than in the familial high risk group during the Aversive > Neutral contrast. IFG = Inferior frontal gyrus; PostCG = Postcentral gyrus; PreCG = Precentral gyrus. This figure has been modified from Hart et al.20, with permission. Please click here to view a larger version of this figure.
ST1. Within-Group Activation Foci in Controls (n = 21) | ||||||
MNI coordinates | ||||||
Region | Hemisphere | x | y | z | Max Z-value | Max p-value1 |
Target activation (p<0.05, false discovery rate corrected) | ||||||
Middle frontal gyrus / Frontal pole | B | -30 | -2 | 50 | 5.57 | <0.0000001 |
Inferior frontal gyrus | B | 46 | 12 | 32 | 5.41 | <0.0000001 |
Insula | B | -32 | 24 | 0 | 5.4 | <0.0000001 |
Precentral gyrus | B | -40 | -22 | 48 | 5.53 | <0.0000001 |
Thalamus | B | -12 | -16 | 12 | 5.03 | <0.0000001 |
Caudate | B | -12 | 12 | 4 | 4.07 | 0.000003 |
Putamen | B | 18 | 8 | 2 | 4.27 | 0.00009 |
Anterior cingulate / Paracingulate gyrus | B | 0 | 12 | 46 | 5.6 | <0.0000001 |
Posterior cingulate gyrus | B | 8 | -16 | 28 | 5.2 | <0.0000001 |
Superior / Middle temporal gyrus | B | 48 | -46 | 10 | 5.88 | <0.0000001 |
Fusiform / inferior temporal gyrus | B | -30 | -50 | -12 | 5.64 | <0.0000001 |
Superior parietal lobule / Supramarginal gyrus / Postcentral gyrus | B | 30 | -44 | 44 | 6 | <0.0000001 |
Lateral occipital cortex | B | 48 | -62 | 12 | 6.12 | <0.0000001 |
Aversive > Neutral activation (p<0.05, false discovery rate corrected) | ||||||
Inferior frontal gyrus | L | -44 | 14 | 14 | 3.16 | 0.0004 |
Frontal pole / Medial frontal cortex | B | -2 | 64 | 0 | 3.42 | 0.0005 |
Postcentral gyrus | L | -62 | -22 | 34 | 3.12 | 0.0004 |
Anterior cingulate cortex | B | -4 | 34 | 8 | 3.27 | 0.0002 |
Posterior cingulate gyrus | B | 0 | -44 | 28 | 3.26 | 0.0002 |
Inferior temporal / Fusiform gyrus | B | -44 | -44 | -14 | 3.03 | 0.0006 |
Angular gyrus | B | 46 | -64 | 8 | 3.42 | 0.0001 |
Supramarginal gyrus | L | -40 | -56 | 20 | 3.59 | 0.00005 |
Aversive activation (p<0.05, false discovery rate corrected) | ||||||
Amygdala | R | 22 | -4 | -18 | 2.86 | 0.001 |
Orbitofrontal cortex / Insula | B | 36 | 22 | -4 | 4.93 | <0.0000001 |
Middle frontal gyrus | B | 32 | 4 | 40 | 4.7 | <0.0000001 |
Frontal pole | B | -38 | 36 | 10 | 4.95 | <0.0000001 |
Anterior cingulate / paracingulate gyrus | B | 6 | 16 | 50 | 4.85 | <0.0000001 |
Posterior cingulate gyrus | B | 2 | -28 | 24 | 5.88 | <0.0000001 |
Thalamus | B | 18 | -26 | 2 | 5.44 | <0.0000001 |
Precentral gyrus | B | -44 | 8 | 34 | 4.54 | <0.0000001 |
Superior parietal lobule | B | -20 | -56 | 54 | 6.05 | <0.0000001 |
Lateral occipital cortex | B | -36 | -82 | 4 | 6.05 | <0.0000001 |
Occipital pole | B | -16 | -90 | 18 | 5.18 | <0.0000001 |
B, Bilateral | ||||||
1 Reported p-values are uncorrected, significant at FDR-corrected value of <0.05 |
Table 1. Within-Group Activation Foci in Controls (n=21). This table has been modified from Hart et al.20, with permission.
Table 2. Between-Group Differences in Activation | ||||||
MNI coordinates | ||||||
Hemisphere | x | y | z | Max Z-value | Max p-value1 | |
Targets | ||||||
Familial High risk > Controls (p<0.05, false discovery rate corrected) | ||||||
Frontal pole | B | 16 | 76 | 6 | 3.52 | 0.00007 |
Inferior frontal gyrus | L | -58 | 16 | 18 | 3.37 | 0.0001 |
Caudate | B | -14 | 20 | 10 | 3.2 | 0.0003 |
Inferior temporal gyrus | L | -52 | -44 | -20 | 2.94 | 0.0009 |
Controls > Familial High Risk (p<0.05, false discovery rate corrected) | ||||||
Middle frontal gyrus / Precentral gyrus | R | 48 | 8 | 34 | 3 | 0.0007 |
Frontal operculum cortex | L | -46 | 16 | -4 | 2.94 | 0.0009 |
Supplementary motor area | R | 18 | -16 | 40 | 3.02 | 0.0007 |
Insula | L | -34 | -18 | 4 | 2.94 | 0.0009 |
Precentral gyrus | B | 10 | -26 | 60 | 3.29 | 0.0002 |
Postcentral gyrus | B | 14 | -38 | 54 | 3.57 | 0.0001 |
Superior temporal gyrus | R | 54 | -6 | -4 | 3.18 | 0.0003 |
Middle temporal gyrus | R | 48 | -46 | 8 | 3.65 | 0.00004 |
Precuneus | R | 2 | -40 | 46 | 2.89 | 0.001 |
Lateral occipital cortex | B | -20 | -74 | 36 | 3.36 | 0.0002 |
Aversive – Neutral | ||||||
Familial High Risk > Controls (p<0.05, false discovery rate corrected) | ||||||
Central opercular cortex | R | 50 | -2 | 6 | 3.01 | 0.0007 |
Controls > Familial High Risk (p<0.05, false discovery rate corrected) | ||||||
Anterior cingulate cortex | L | -6 | 38 | 8 | 2.68 | 0.002 |
Precuneus | L | -10 | -54 | 36 | 2.7 | 0.002 |
B, Bilateral | ||||||
1 Reported p-values are uncorrected, significant at FDR-corrected value of <0.05 |
Table 2. Between-Group Differences in Activation Foci. This table has been modified from Hart et al.20, with permission.
The modified emotional oddball paradigm in the current study has been shown to elicit differences in neural recruitment during selective attention and emotional processing in children and adolescents at risk for schizophrenia. While existing paradigms using the emotional oddball task have been used to investigate neural changes in adult populations with psychiatric illness8, the current paradigm may be particularly useful for measurement of vulnerability markers in younger age groups.
There are several challenges inherent in performing fMRI studies with children and adolescents, including ensuring that the task is appropriate in terms of difficulty and content and in minimizing motion artifacts. Critical steps in the protocol include step 2.2, which requires that participants have a mock scan and practice session prior to the fMRI session. This step is particularly helpful for improving participants’ comfort and data quality. Additionally, the design of the behavioral task is critical to ensure that the task has an appropriate level of difficulty and that the selection of aversive images is appropriate for younger age groups. The current task design was successful in eliciting significant behavioral differences between the aversive and neutral conditions and had a moderate difficulty level for both high-risk and control groups.
A limitation of this protocol was that the task did not elicit a significant difference in amygdala activation between the aversive and neutral conditions. There was significant amygdala activation during the aversive condition, but the difference was not significant when contrasted with neutral stimuli. This finding is likely due to the selection of children-appropriate stimuli. Future modifications of the task could examine effects of emotional facial expressions, which may produce more robust differences in amygdala activation between emotional and neutral conditions21.
Other future applications of this technique include applying it to populations of children and adolescents who may be at risk for other psychiatric illnesses that may similarly affect executive and emotional processing. Several other psychiatric conditions, such as bipolar disorder and mood disorders, have been found to be associated with alterations in brain structure and function that reflect endophenotypes, or heritable markers associated with a disease22,23. This suggests the possibility that these intermediate changes in brain structure or function may potentially precede the onset of pathological symptoms in at-risk individuals. The use of the described paradigm in longitudinal studies and with children and adolescents may help to identify relevant endophenotypes for a particular condition to help to refine risk estimates.
The authors have nothing to disclose.
We thank Erin Douglas, Anna Evans, and Carolyn Bellion for their contributions to participant recruitment and clinical assessment. We also thank Michael Casp, Zoe Englander, Justin Woodlief, and James Carter for their contributions to data collection and analysis, and Robert M. Hamer for consultation on statistical analysis and editing of the manuscript. Finally, we thank the individuals and their families who participated in this study.
This study was supported by Conte center grant P50 MH064065 from the National Institute of Mental Health. Dr. Hart was supported by T32 HD040127 from the National Institute of Child Health and Human Development.