In this issue of the Journal, two articles (1, 2) provide a rare developmental prospective on the brain processes involved in cognition and emotion in bipolar disorder. It is well established that bipolar disorder can have childhood onset. Indeed, retrospective studies have found that adult patients report childhood onset in the majority of cases (3). Furthermore, childhood onset is associated with greater severity of illness (3). The two studies in this issue are among the first to simultaneously study both youths and adults with bipolar disorder using the same diagnostic techniques and imaging methods. Both studies were supported by the National Institute of Mental Health intramural research program, in which there has been an emphasis on the importance of identifying the “narrow phenotype” of bipolar disorder (4). The pediatric patients in these two studies had a history of at least one hypomanic or manic episode with elevated mood or grandiosity as well as key DSM-IV-TR criterion B symptoms for mania. Thus, neither study included children with mood dysregulation, i.e., patients with chronic irritability/aggression who are often diagnosed with bipolar disorder not otherwise specified in clinical settings. The participants in these studies differed in comorbidity. Weathers et al. (1) excluded individuals with attention deficit hyperactivity disorder (ADHD), while two-thirds of the children and 12% of the adults in the Kim et al. study (2) did meet criteria for ADHD. It was appropriate to exclude individuals with ADHD from the Weathers et al. study because inhibitory control was being assessed, and thus high comorbidity with ADHD would have been a confound. The mean age at mania onset in the pediatric samples ranged from about 9 to 11 years, while the mean age at mania onset in the adult samples was about 21 years. Thus, the pediatric patients (mean age, 14 years) had been ill for about 3–5 years, while the adult patients (mean age, 40 years) had been ill for two decades.

To assess inhibitory control, Weathers et al. used the stop-signal task, in which the participant presses one of two buttons, depending on which letter appears on the screen. On certain trials, the background color changes to red, indicating to the participant not to press the button. The neurocircuitry of this task has been reasonably worked out (5). Stop cues activate the ventrolateral (inferior) prefrontal cortex; right-sided activation is associated with successful inhibition. During “response conflict” (the urge to press the button in response to a letter is contradicted by the red cue to stop), the anterior cingulate cortex is activated. Activation in the anterior cingulate cortex is further enhanced if inhibition fails. This anterior cingulate cortex activation during error is associated with a greater likelihood of improved performance on subsequent trials (6). Weathers et al. found an age group-by-diagnosis interaction in the anterior cingulate cortex during failed inhibition. Healthy child subjects had greater activation in this region than healthy adults, suggesting that the normal developmental pattern is a decrease in activity in the anterior cingulate cortex during conflict, perhaps reflecting greater efficiency of this process with age. In contrast, bipolar children exhibited deactivation in the anterior cingulate cortex during failed inhibition, while bipolar adults exhibited greater activation than healthy adults. Surprisingly, in this region, bipolar adults more resembled healthy children, while bipolar children showed the reverse pattern (decreased anterior cingulate cortex activation) relative to healthy children as well as to adults.

During successful inhibition, there were no effects of age group on activity in two other brain regions: the right nucleus accumbens and the left ventrolateral prefrontal cortex. Healthy subjects showed greater activation in these areas than bipolar patients. The nucleus accumbens has been associated with responsiveness to reward, but since there was no incentive associated with the task, it is unclear why there would be an effect on diagnosis. Excluding depressed patients did not change the result, ruling out that this subgroup is less responsive to reward as previously suggested in a study of adolescents with unipolar depression (7). Higher activation in the left ventrolateral prefrontal cortex has been associated with successful interference suppression in both ADHD and healthy comparison children (8). Despite the fact that none of the participants in the Weathers et al. study had ADHD, the alterations in the inhibitory circuit in these participants are strikingly similar to the circuitry changes observed in individuals with ADHD who do not have mood disorders (8, 9).

Kim et al. examined emotional processing using a simple task in which the participant is instructed to identify the gender of a person in a photograph. The participant is not told anything specific regarding the emotional expression of the person, and thus he or she implicitly processes the different emotions (fearful, angry, neutral). The amygdala is strongly activated by the presentation of fearful or angry faces. Indeed, this occurs even when the face is presented subliminally (10). In this study, regardless of the type of facial expression, bipolar children had greater right amygdala activation than both bipolar adults and healthy children, while bipolar adults and healthy comparison subjects did not differ from each other. Regardless of age group, bipolar patients had greater amygdala activity in response to fearful expressions than comparison subjects, while there were no differences between the groups in response to angry or neutral emotions. This suggests that in childhood, bipolar disorder is characterized by amygdala hyperactivity in response to all emotions, but in adulthood, there is a more specific overreactivity in response to fearful faces.

Given the high prevalence of ADHD in the pediatric bipolar sample in the Kim et al. study, it is important to bear in mind that recent work has shown that adolescents with ADHD without mood disorder also exhibit increased amygdala activity relative to healthy comparison subjects in response to presentation of angry faces as well as less connectivity between the amygdala and ventrolateral prefrontal cortex (11). Increased amygdala reactivity is also a feature of anxiety disorders (12). The key lesson of these two studies may be that we are not imaging diseases per se but the parts of the circuitry of inhibitory control and emotional responsiveness that are perturbed by the disorders. These two processes are most likely disturbed in a wide range of conditions, including ADHD and anxiety and mood disorders. The orange glow we see superimposed on the brain image is perhaps analogous to the white spots one sees on a chest X-ray of a patient with pulmonary disease. We know that region of the lung is disturbed, but we cannot say from the image alone whether the underlying process is bacterial, viral, cancerous, or toxic without further study. More complex, these brain circuits may not be disturbed in and of themselves but rather are reacting to some other disease process. For instance, the anterior cingulate cortex may become altered in both ADHD and bipolar disorder (and thus both disorders show poor inhibitory control), but different pathophysiological processes in each disorder ultimately alter functioning in this region. Thus, imaging may not provide a diagnostic test for a disorder but may provide a biomarker that relates to symptoms and can possibly be used to predict treatment outcome. As the authors of both articles point out, the next step involves longitudinal studies of children in the early stages of bipolar disorder (ideally before mood stabilizing treatment begins), with repeated neuroimaging through adulthood. The authors of these important studies are to be commended for setting us on this long road.