Prolonged hemodynamic response during incidental facial emotion processing in inter-episode bipolar I disorder

Abstract

This fMRI study examined whether hemodynamic responses to affectively-salient stimuli were abnormally prolonged in remitted bipolar disorder, possibly representing a novel illness biomarker. A group of 18 DSM-IV bipolar I-diagnosed adults in remission and a demographically-matched control group performed an event-related fMRI gender-discrimination task in which face stimuli had task-irrelevant neutral, happy or angry expressions designed to elicit incidental emotional processing. Participants' brain activation was modeled using a "fully informed" SPM5 basis set. Mixed-model ANOVA tested for diagnostic group differences in BOLD response amplitude and shape within brain regions-of-interest selected from ALE meta-analysis of previous comparable fMRI studies. Bipolar-diagnosed patients had a generally longer duration and/or later-peaking hemodynamic response in amygdala and numerous prefrontal cortex brain regions. Data are consistent with existing models of bipolar limbic hyperactivity, but the prolonged frontolimbic response more precisely details abnormalities recognized in previous studies. Prolonged hemodynamic responses were unrelated to stimulus type, task performance, or degree of residual mood symptoms, suggesting an important novel trait vulnerability brain dysfunction in bipolar disorder. Bipolar patients also failed to engage pregenual cingulate and left orbitofrontal cortex-regions important to models of automatic emotion regulation-while engaging a delayed dorsolateral prefrontal cortex response not seen in controls. These results raise questions about whether there are meaningful relationships between bipolar dysfunction of specific ventromedial prefrontal cortex regions believed to automatically regulate emotional reactions and the prolonged responses in more lateral aspects of prefrontal cortex.

Figure 1

Brain rendering depiction of significant signal change to facial stimuli in the gender-identification fMRI task in the combined bipolar and control study groups. This map was constructed considering all main effect, temporal derivative, and dispersion derivative voxelwise data simultaneously for the best possible estimate of “activation” to task. Statistical results are thresholded at q<.05 False Discovery Rate corrections for searching the whole brain. Areas in red-yellow represent activation to all stimuli classes, whereas blue-turquoise represent negative BOLD signal change.

Figure 2

Peristimulus time histograms showing average study group hemodynamic response time-courses based around the voxel-centers of regions-of-interest showing significant (p<0.05 Holm-Bonferroni corrected) ANOVA group × basis set effects.

Figure 3

Peristimulus time histograms showing hemodynamic response to neutral, positive, and negative facial affective stimuli in bipolar and non-bipolar control groups in regions-of-interest showing a significant interaction of group × valence × basis set conditions in the mixed-effect ANOVA.