Neural activation during encoding of emotional faces in pediatric bipolar disorder

Abstract

Objective: Neurobiological understanding of bipolar disorder (BD) is limited by a paucity of functional magnetic resonance imaging (fMRI) research examining correlates of psychological processes. To begin to address these limitations, the current study tests the hypothesis that pediatric BD (PBD) subjects exhibit altered neural activation during encoding of emotional faces compared to typically developing controls.

Methods: Pediatric BD subjects (n=23; mean age=14.2+/-3.1 years) and controls (n=22; mean age=14.7+/-2.3 years) were matched on age, gender, and IQ. In this event-related fMRI study, subjects were scanned while viewing emotional faces and given a surprise recognition memory test 30 min postscan. Our main outcome measure was between-group differences in neural activation during successful versus unsuccessful face encoding.

Results: Pediatric BD youth exhibited reduced memory for emotional faces, relative to healthy comparisons, particularly on fearful faces. Event-related fMRI analyses controlling for this behavioral difference showed that PBD subjects, compared to controls, had increased neural activation in the striatum and anterior cingulate cortex when successfully encoding happy faces and in the orbitofrontal cortex when successfully encoding angry faces. There were no between-group differences in neural activation during fearful face encoding.

Conclusions: Our results extend what is known about memory and face emotion processing impairments in PBD subjects by showing increased fronto-striatal activation during encoding of emotional faces. Further work is required to determine the impact of mood state, medication, and comorbid illnesses on these findings.

Fig. 1

Probabilistic response reversal task. During initial acquisition trials (left), preferred stimulus (red bear) wins points and nonpreferred stimulus (blue cat) loses points. Participants receive feedback (win/lose 100 points) if their response is correct/incorrect respectively. During subsequent reversal trials (right), stimulus/reward relationship is reversed, so that previously preferred stimulus (red bear) is now nonpreferred and loses points, and the previously nonpreferred stimulus (blue cat) is now preferred and wins points. This figure illustrates the 100:0 condition when the preferred stimulus is rewarded 100% (and punished 0%). During the 80:20 condition, the preferred stimulus is initially rewarded 80% (and punished 20%), while the nonpreferred stimulus is rewarded 20% (and punished 80%). Then subsequent 80:20 reversal trials reverse this stimulus/reward relationship.

Fig. 2

Probabilistic response reversal significant diagnosis × phase interactions. All comparisons thresholded at p < 0.05 whole-brain corrected for multiple comparisons using AlphaSim implemented in Analysis of Functional NeuroImages software (AFNI). Left: neural clusters (orange) meeting this threshold. Right: mean blood oxygen level-dependent signal for selected contrasts with black = pediatric bipolar disorder (BD) and white = normal controls (NC); error bars = SD. ACQ = acquisition phase (sum of acquisition correct win and acquisition incorrect lose); REV = reversal phase (sum of reversal correct win and reversal incorrect lose); BA = Brodmann area.