During performance of a WM task in which performance was comparable across groups, RELS showed (i) weak modulation (activation or deactivation from 0-back to 2-back condition) in the CV, insula, and amygdala/parahippocampal region, and (ii) exaggerated modulation in the frontopolar cortex and brainstem (from 0-back to 2-back) compared to that seen in controls. In the CV, group differences were due to significant differences in both the 0-back and 2-back conditions, while group differences in other regions were driven by exaggerated activity (brainstem, insula, and amygdala) or reduced activity (frontopolar cortex) in the 0-back condition in RELS (as indicated by our examination of parameter estimates for 0-back and 2-back conditions, and by 2-group t
-tests comparing 0-back and fixation conditions). In addition, RELS showed a nonsignificant trend toward reduced deactivation in the left OFC, slightly weaker than the significant finding in our previous study in adult RELS (13
). Results were observed despite the absence of statistical differences between groups on all demographic, neuropsychological, in-scanner performance, and most clinical variables; and many results were robust to ANCOVAs using vigor and IQ as covariates. However, all results should be viewed as preliminary due to the small sample size.
Our results suggest altered neural reactivity to a low-load cognitive-challenge in RELS which affects modulation in response to a higher-load cognitive challenge, consistent with reports of altered low-level task response in mood disorder populations (14
). The frontopolar findings largely replicate those of two BD studies of adult RELS (12
) and two fMRI studies of WM in BD (24
). Contrary to our hypotheses based on previous work (12
), there was minimal evidence of a link between negative-emotion related brain activity and task performance in RELS. For example, while tension/anxiety was related to faster 2-back reaction time in RELS (FDR-corrected p = 0.05), activity in the frontopolar cortex was only marginally related to WM performance (FDR-corrected p = 0.10). However, activity in many of the regions that differentiated the groups (frontopolar, insular, CV, and brainstem) was significantly related to negative mood in RELS, despite the fact that both groups exhibited mood scores in the normal range and did not differ on average (with the exception of vigor level, p = 0.06). Further, insular, frontopolar, and orbitofrontal activity during the 0-back condition was significantly related to mood scores in the whole sample.
Alterations in the CV are consistent with a study showing CV volume is associated with resilience to BD in RELS (63
), decreased CV signal during WM in BD (24
), and abnormal spectroscopic (64
) and structural findings in the CV in BD (65
). Altered CV activity may reflect homeostatic or compensatory adaptation in RELS which confers resilience to cognitive and emotional deficits seen in BD, consistent with the relationship of cerebellar volume and resilience to mood disorder in RELS (63
). Indeed, the CV participates in regulation of the midbrain/VTA–amygdala/parahippocampal components of the Papez circuit (67
), and plays a strong role in regulation of emotion (74
), long-term fear conditioning (80
), and Pavlovian conditioned cardiac responses (82
), even independent of the amygdala (81
). Further, CV transcranial magnetic stimulation (a putative treatment for mood disorders) directly affects amygdala/hippocampal activity (73
). Here, CV activity was related to vigor in the whole sample and to anger/hostility in RELS. While speculative, altered CV modulation in RELS may reflect preserved capacity for vigor and externalizing emotions (anger), which confer protection against depression and mood disorder.
In addition, CV activity in RELS may reflect a compensatory adaptation that preserves task performance in the context of altered task-related frontopolar modulation. Notably, lesions of the CV are associated with both executive function deficits (attention, WM) as well as cognitive, emotional, and locomotor changes seen in BD (76
). Thus, it is possible that inability to compensate for other abnormalities with CV activity may be one correlate of BD illness. Interestingly, the CV is sensitive to stress, glucocorticoids, and drugs of abuse (67
), which can be triggers to the onset of BD.
Many of the regions differentiating the groups (CV, insula, amygdala, raphe and other brainstem nuclei, and the VTA) participate in an extended visceromotor network modulating reward, fear, stress and autonomic/neuroendocrine responses, and motivation (10
). While pathophysiological mechanisms are unclear, abnormal modulation of this network could explain some of the abnormal cognitive, motivational, neuroendocrine, autonomic, and emotional symptoms in mood disorders (89
). The reticular formation, for example, plays pivotal roles in the regulation of several domains affected in mood disorders (arousal, attention, autonomic reflexes, motor functions, the wake/sleep cycle, and pain) (94
). The VTA (origin of the mesocortical and mesolimbic dopamine systems innervating several structures showing group differences here, such as the prefrontal, orbitofrontal, insular cortices, and amygdala) plays roles in the regulation of locomotor inhibition, arousal, affect, and drive (95
). The raphe nuclei are implicated in antidepressant action (96
) and pain inhibition, and are potentially relevant given genetic evidence of serotonergic abnormalities in BD (97
Finally, the findings of this study are also consistent with other recent findings of autonomic dysregulation in BD and other mood disorders, including (i) cardiovascular and vagal abnormalities (which pre-exist psychopharmacologic treatment) (37
); (ii) a relationship of maladaptive emotion regulation responses to lower vagal recovery in relatives of depressed persons (100
); and (iii) vagus nerve stimulation as a putative treatment for severe mood disorders (101
The findings of this study should be viewed as preliminary, due to the small sample size and exploratory nature of some of the results. Replication of the results in larger samples is needed. The result in the OFC, for example, did not meet our criteria for statistical significance. Though the resolution of fMRI is not sufficient to separate activations in small structures (i.e., in the pons and midbrain), the regions activated span multiple small structures and preliminary findings suggest a coherent network associated with regulation of autonomic arousal. In follow-up studies with larger samples, we will use ROIs derived from individual structural anatomy to look more precisely at the location of activations.